Description
Platform startups on the Internet are an under-researched type of business, although some have demonstrated exceptionally high growth and global penetration (e.g., Google, Facebook, and Tinder). As growth-oriented new ventures are generally important for economic development, and given that their failure rate is generally high, the study focuses on a meaningful topic.
Joni Salminen
Sarja/Series A-12:2014
STARTUP DILEMMAS - STRATEGIC PROBLEMS
OF EARLY-STAGE PLATFORMS ON THE INTERNET
Turun kauppakorkeakoulu
Turku School of Economics
Custos: Professor Rami Olkkonen
Turku School of Economics
Supervisors: Professor Rami Olkkonen
Turku School of Economics
Pre-examiners: Professor Daniel Beimborn
Frankfurt School of Finance & Management
Professor Riitta Katila
Stanford University
Opponent: Professor Daniel Beimborn
Frankfurt School of Finance & Management
Copyright Joni Salminen & Turku School of Economics
The originality of this thesis has been checked in accordance with the University of Turku
quality assurance system using the Turnitin OriginalityCheck service.
ISBN 978-952-249-370-5 (print) 978-952-249-371-2 (PDF)
ISSN 0357-4652 (print) 1459-4870 (PDF)
Publications of Turku School of Economics, Series A
Suomen yliopistopaino Oy – Juvenes Print, Turku 2014
SUMMARY
Startup dilemmas - Strategic problems of early-stage platforms on the
Internet
Platform startups on the Internet are an under-researched type of business, alt-
hough some have demonstrated exceptionally high growth and global penetra-
tion (e.g., Google, Facebook, and Tinder). As growth-oriented new ventures
are generally important for economic development, and given that their failure
rate is generally high, the study focuses on a meaningful topic.
Platforms are defined as “places of interaction”, in which technology is em-
ployed to connect different user groups. Based on interaction, four platform
types are defined: 1) content platforms, 2) social platforms, 3) exchange plat-
forms, and 4) infrastructure. Their complements and usage motives vary; how-
ever, they inherit similar problems from the platform model.
The theoretical basis of this work derives from the literature on two-sided
markets. In addition, works analyzing platforms in the field of marketing, in-
formation system sciences, and network economics are included. Key con-
structs, many of which originate from earlier research on networks and stand-
ards, include network effects, critical mass, winner-takes-all, tipping, two-sid-
edness, and the chicken-and-egg problem. To complete the theoretical per-
spective, concepts on online specificities are also included; namely, user-
generated content, indirect monetization, and the freemium business model.
The methodological approach is grounded theory (GT) which is an induc-
tive method emphasizing a close connection of the researcher to the analyzed
material. GT includes coding, that is, raising the abstraction level by concep-
tualizing central themes (i.e., categories); constant comparison of novel find-
ings to those previously found; and theoretical sampling, that is, the systematic
attempt to challenge and extend ideas by collecting more material. A heavy
emphasis in this method is placed on the researcher’s ability to conceptualize
and theorize (i.e., theoretical sensitivity).
Most often, the method is applied to qualitative data, as also is the case in
this study. In total, 29 post-mortem stories written by founders of failed startup
ventures are analyzed. These data, publicly available on the Internet, comprise
the study’s principal material. Additional data include secondary startup inter-
views, interviews of six founders conducted by the author, and discussions at
numerous startup events in Finland, Sweden, and the USA over the four year
research process. Participation comprised discussions with founders in the at-
tempt to verify ideas on the emerging theoretical model, and gain valuable
industry insight.
In the analysis, four dilemmas of platform startups, emerged: 1) cold start
dilemma, 2) lonely user dilemma, 3) monetization dilemma, and 4) remora's
curse. The first two are variants of the chicken-and-egg problem; in a cold
start, there is a lack of content due to the lack of content, so users are unwill-
ing to join the platform. In the lonely user dilemma, there might be registered
users, but none are present at a given time or place; thus, there is no reason to
use the platform. In the monetization dilemma, users are given free access and
usage, but at the expense of revenue. When fees are introduced, the users flee.
In the remora's curse, a startup is able to solve the cold start dilemma by inte-
grating into a larger platform, but becomes vulnerable to its strategic behavior.
Essentially, the problems are interlinked. By solving the chicken-and-egg
problem through subsidization, a commonly applied strategy, a startup moves
toward the monetization dilemma and eventually fails for this reason. It might
also solve the problem by platform integration, or becoming a complementor
for a larger platform. It has been previously argued in the platform literature
that, if successful, the remora can perform envelopment, whereby it perma-
nently captures the host’s users. However, in this study it is argued that the
power dynamics do not favor the startup, which remains vulnerable to the
dominant platform’s opportunism. In this study, “selective integration” (i.e.,
content and value envelopment) is suggested as an alternative. In addition, the
merits and downsides of multihoming and the freemium model are discussed.
The study makes several contributions. First, the scope of the chicken-and-
egg problem, and also its solutions, is expanded to fit the realities of online
startups. This enables more useful approaches than most other studies focusing
on larger firms, exchange platforms, and pricing as a solution. Second, it is
shown that the strategic problems of early-stage platforms are connected,
which should be considered in studies. For practitioners, this implies the
recognition of the “dilemma roadmap” as a tool for strategic planning. Third, a
total of 19 different solutions are analyzed, and the requirements for a “perfect
solution” are characterized. Fourth, novel constructs are introduced for use and
further development by scholars. Finally, several avenues for further research
are put forward; for example, integration of founders’ biases into a theory,
expansion of platform theory, and the pursuit of more solutions.
Although “one size fits all” solutions are unlikely, theoretical analysis of
the solutions, even if complicated by reality, is a useful process to understand
the fundamental forces by which they are affected. Eventually, models can be
expanded to cover more aspects, thus enabling better solutions to emerge from
the cooperation of theory and practice.
Keywords: platforms, two-sided markets, startups, chicken-and-egg problem
TIIVISTELMÄ
Startup-dilemmat - Alkuvaiheen alustojen strategiset ongelmat
Internetissä
Tutkin alkuvaiheen Internet-alustojen strategisia ongelmia. Niitä on tutkittu
suhteellisen vähän, vaikka osa alustoista on saavuttanut poikkeuksellisen no-
pean kasvun (esim. Google, Facebook ja Tinder). Koska kasvuhakuiset yritys-
hankkeet ovat yleisesti ottaen tärkeitä taloudellisen kehityksen ja innovaatioi-
den kannalta ja koska niiden kuolleisuus on yleensä korkea, tutkimuksen aihe
on mielekäs. Aihe ei kuitenkaan kosketa pelkästään yhteiskuntaa, vaan myös
yksittäisiä yrityksiä ja yrittäjiä, jotka kamppailevat näiden strategisten ongel-
mien ja globaalin kilpailun parissa.
Alusta määritellään vuorovaikutusympäristöksi, jossa teknologia yhdistää
eri käyttäjäryhmiä. Käyttömotiivin perusteella työ jakaa Internet-pohjaiset
alustat neljään tyyppiin: 1) sisältöalustat, 2) sosiaaliset alustat, 3) vaihdanta-
alustat ja 4) infrastruktuuri. Vaikka niiden välillä on eroja, em. alustat jakavat
samat alustaliiketoimintamallin ongelmat.
Työn teoreettinen pohja on peräisin kaksipuolisten markkinoiden kirjalli-
suudesta. Lisäksi hyödynnetään markkinoinnin, tietojärjestelmätieteiden ja
verkostotaloustieteen kirjallisuutta. Keskeisiä käsitteitä ovat verkostovaikutuk-
set, kriittinen massa, "voittaja vie kaiken" -ilmiö, kaksi-suuntaisuus, ja muna-
kanaongelma. Em. kirjallisuushaarat ymmärretään tässä työssä alustakirjalli-
suutena. Internet-liiketoiminnan erityispiirteet, kuten käyttäjien tuottama si-
sältö ja epäsuora ansaintalogiikka, ovat myös mukana.
Metodina on grounded theory (GT), jonka soveltamiseen kuuluu koodaus,
eli käsitteellisen tason nostaminen aineiston keskeisiä teemoja nimeämällä ja
arvioimalla, jatkuva vertailu uusien ja edellisten löydösten välillä ja teoreetti-
nen otanta, eli luotujen teoreettisten konstruktioiden (teorian) systemaattinen
täydentäminen ja haastaminen lisämateriaalin avulla. Metodi painottaa vah-
vasti tutkijan kykyä käsitteellistää ja teoretisoida laadullista ja määrällistä ai-
neistoa, ts. teoreettista herkkyyttä. Useimmiten GT-menetelmää sovelletaan
laadulliseen aineistoon, ja niin tässäkin tutkimuksessa. Pääasiallisen aineiston
muodostaa 29 epäonnistumiskertomusta; lisäksi aineistoon kuuluu haastatte-
luja ja keskusteluja useissa startup-tapahtumissa Suomessa, Ruotsissa ja Yh-
dysvalloissa nelivuotisen tutkimusprosessin aikana.
Analyysin yhteydessä päätettiin keskittyä alustastartupeille ominaisiin on-
gelmiin, eli startup-dilemmoihin, erityisesti: 1) kylmän alun ongelma, 2) yksi-
näisen käyttäjän dilemma, 3) monetisointidilemma ja 4) remoran kirous. En-
simmäiset kaksi ovat muunnelmia muna-kanaongelmasta, jota on laajalti kä-
sitelty kirjallisuudessa. Kylmän alun ongelmassa sisällönpuute estää sisältöä
luovien käyttäjien alustaan liittymisen, ja näin yhdelläkään käyttäjällä ei ole
motiivia rekisteröityä. Yksinäinen käyttäjä saattaa sen sijaan olla jo rekiste-
röitynyt, mutta koska vastapuoli ei ole läsnä tietyssä ajassa tai paikassa, alus-
tan käytölle ei ole mahdollisuutta.
Monetisointidilemman mukaan käyttäjät houkutellaan alustaan tarjoamalla
ilmainen pääsy ja käyttö, mutta taloudellisen kannattavuuden kustannuksella.
Heti kun maksut otetaan käyttöön, käyttäjät pakenevat alustalta. Remoran ki-
rouksessa muna-kanaongelma on ratkaistu integroimalla suurempaan alustaan,
mutta vastineeksi joudutaan alttiiksi tämän alustajan omistajan strategiselle
käyttäytymiselle. Näiden dilemmojen analysointi, niiden sovittaminen aiem-
paan kirjallisuuteen sekä yritys löytää mahdollisia ratkaisuja ovat tämän työn
keskeistä antia.
Tutkimus tekee useita kontribuutioita. Ensinnäkin muna-kanaongelma laa-
jennetaan Internet-liiketoiminnan yhteyteen. Useimmat muut tutkimukset kes-
kittyvät suuriin yrityksiin, vaihdanta-alustoihin, ja hinnoitteluun muna-kana-
ongelman ratkaisuna. Toiseksi työ osoittaa, että alkuvaiheen alustojen strategi-
set ongelmat ovat sidoksissa toisiinsa. Yrityksille tämä merkitsee "dilemma-
tiekartan" hyödyntämistä strategisen toiminnan työkaluna, kun taas tutkijoille
se korostaa tarvetta lähestyä alustaongelmia kokonaisvaltaisesti. Myös mah-
dollisia ratkaisuja pohditaan laajalti: yhteensä tarkastellaan 19 eri strategian
soveltuvuutta dilemmojen ratkaisuun. Tutkimus esittää "täydellisen ratkaisun"
ominaispiirteet ja useita lupaavia mahdollisuuksia lisätutkimuksiin, mm. yh-
distämällä perustajien "harhat" osaksi teoriaa, laajentamalla alustateoriaa ja
etsimällä uusia ratkaisuja metodologisen pluralismin avulla.
Tutkimuksen mukaan muna-kanaongelmalla on kytkeytynyt luonne Inter-
net-alustojen yhteydessä; toisin sanoen yhden ongelman ratkaisu johtaa pian
toiseen. Tilannetta pahentaa ns. "kiitorata-efekti", jonka mukaan aloittelevalla
yrityksellä on rajallinen aika saavuttaa onnistumisia ennen sen lopettamista.
Olennainen lähtökohta on, että alusta ei kykene aina yhdistämään sopivia
käyttäjiä tai sisältöjä toistensa kanssa ja siten ratkaisemaan muna-kanaongel-
maa. Lisäksi ongelmaa ei voida ratkaista kerralla ja sitten jatkaa eteenpäin,
vaan yksinäisen käyttäjän dilemma seuraa alustaa niin kauan kuin se on ole-
massa. Ellei yritys tunnista ongelmia ajoissa, havaitse niiden välisiä yhteyksiä
ja sovella oikeita ratkaisuja, epäonnistumisen todennäköisyys kasvaa. Siksi
dilemmojen luonteen ymmärtäminen on ensiarvoisen tärkeää alustaliiketoi-
minnassa menestymisen kannalta.
Asiasanat: alustat, kaksisuuntaiset markkinat, startupit, muna-kanaongelma
ACKNOWLEDGEMENTS
I would like to thank the following persons: Professor Rami Olkkonen for
guiding me through the research process. His formidable attitude to teaching,
life, and academia has greatly inspired me. The pre-examiners, Professor
Daniel Beimborn and Professor Riitta Katila. Professor Aino Halinen-Kaila,
for guiding me through the doctoral studies. My colleagues, Mekhail Mustak
and Lauri Pitkänen, for their continuous support and fun times in the depart-
ment. (You guys will follow shortly!)
The founders who have tried and failed, but keep on trying. The fact they
shared their stories with the world enabled this research in the first place.
Special thanks are owed to Boost Turku Entrepreneurship Society and the
inspiring people I have met there. In addition, the founders I have met along
the way deserve respect for their hard work. Here are just a few of them, in no
particular order: Ajay Garg, Timo Herttua, Rasmus Kevin, Renato Kern, Tiina
J aatinen, Igor Burattini, Tatu Laine, J ose Teixeira, Marcos Tong, Ville Tapani,
Toni Perämäki, Linas Ceikus, J uho Vaiste, Ville Sirkiä, Timo Hänninen, Kari
Vuorinen, J anne Loiske, Ismo Karjalainen, Antero J ärvi, Ville Kaituri, J asu
Koponen, Camilla af Hällström, Antti Lundstedt, Hazzan Ajao, Denis
Duvauchelle, J eremy Boom, Guillaume de Dorlodot, Victor Vulovic, and Ben
Adamson. Thank you all. I hope to stay in touch!
The following have helped with administrative issues: Auli Rahkala-Toivo-
nen, Sanna Kuusjärvi, Riikka Harikkala, and J enni Heervä. These issues are
trickier than research problems, so I appreciate your help. I would also like to
thank Dr. Elina J aakkola, Professor Leila Hurmerinta, Dr. Hannu Makkonen,
Dr. Harri Terho, and Professor Nicole Coviello for their comments in research
seminars at the Turku School of Economics.
I would to express my gratitude to the following foundations for their sup-
port: Liikesivistysrahasto, J enny ja Antti Wihurin rahasto, Leonard Gestrinin
muistorahasto, OP Tukisäätiö, TOP-säätiö, Turun kauppaopetus-säätiö, Mar-
cus Wallenbergin säätiö, and Turun Kauppakorkeakouluseura. Without them,
there would be less knowledge in the world.
Finally, thanks to my family: to my mother Helena Salminen, for always
caring for me. She is still the best mother in the world. To my younger brother
Niko Salminen, for driving me to the airport when leaving for conference
trips, and also for other help. And to my father, Timo Salminen, for teaching
us that education matters. I wish all children could have the same lesson.
Turku, 14
th
September 2014
J oni Salminen
TABLE OF CONTENTS
SUMMARY
TIIVISTELMÄ
ACKNOWLEDGMENTS
1 INTRODUCTION ................................................................................................ 17
1.1 Research background ........................................................................ 17
1.2 Key concepts .................................................................................... 19
1.3 Research gap ..................................................................................... 20
1.4 Purpose and research questions ......................................................... 22
1.5 Positioning ........................................................................................ 26
1.6 Structure ........................................................................................... 31
2 METHODOLOGY................................................................................................ 33
2.1 Research strategy .............................................................................. 33
2.1.1 Introduction to research strategy ............................................ 33
2.1.2 What is GT? .......................................................................... 33
2.1.3 Why was GT selected as research method? ............................ 36
2.2 Research process ............................................................................... 38
2.3 Research data .................................................................................... 40
2.3.1 Data collection ....................................................................... 40
2.3.2 Selection criteria .................................................................... 42
2.3.3 Description of the startups ..................................................... 43
2.4 Analytical approach .......................................................................... 46
2.4.1 Coding process in GT ............................................................ 46
2.4.2 Application of GT in this study .............................................. 48
2.4.3 Coding guide ......................................................................... 51
2.5 Literature approach ........................................................................... 53
3 THEORETICAL BACKGROUND ....................................................................... 57
3.1 Concept of platform .......................................................................... 57
3.1.1 Platform theory and platform literature .................................. 57
3.1.2 Defining platforms ................................................................. 57
3.1.3 Markets vs. platforms ............................................................ 60
3.1.4 Mediation vs. coordination .................................................... 62
3.1.5 Direct and indirect effects of interaction ................................ 64
3.1.6 Networks vs. platforms .......................................................... 67
3.1.7 Websites vs. platforms........................................................... 67
3.2 Platform definition of this study ....................................................... 68
3.3 Typology for online platforms .......................................................... 70
3.4 Online platforms and user generation ............................................... 76
3.4.1 Why is UG included in the study? ......................................... 76
3.4.2 User-generated content .......................................................... 76
3.4.3 UG in online platforms .......................................................... 77
3.4.4 Ideal user-generation model .................................................. 78
3.4.5 Functional view to UG .......................................................... 80
3.4.6 Implications to startups .......................................................... 83
3.4.7 Limitations of UG ................................................................. 84
4 STARTUP DILEMMAS........................................................................................ 87
4.1 Introduction to dilemmas .................................................................. 87
4.1.1 What is meant by dilemmas? ................................................. 87
4.1.2 The use of dilemmas in this study .......................................... 87
4.2 Dilemmas in the platform literature .................................................. 89
4.3 Dilemmas emerging from analysis.................................................... 93
4.3.1 Results from the black box analysis ....................................... 93
4.3.2 Narrowing the focus of the study ........................................... 95
4.3.3 Chosen dilemmas and their treatment .................................... 97
4.4 Cold start dilemma ........................................................................... 99
4.4.1 Definition and exhibits .......................................................... 99
4.4.2 The literature ....................................................................... 105
4.4.3 Solution: Subsidies .............................................................. 111
4.4.4 Discussion ........................................................................... 115
4.5 Lonely user dilemma ...................................................................... 119
4.5.1 Definition and exhibits ........................................................ 119
4.5.2 The literature ....................................................................... 124
4.5.3 Solution: Remora ................................................................ 134
4.5.4 Discussion ........................................................................... 138
4.6 Monetization dilemma .................................................................... 142
4.6.1 Definition and exhibits ........................................................ 142
4.6.2 The literature ....................................................................... 148
4.6.3 Solution: Freemium ............................................................. 158
4.6.4 Discussion ........................................................................... 164
4.7 Remora’s curse ............................................................................... 168
4.7.1 Definition and exhibits ........................................................ 168
4.7.2 The literature ....................................................................... 180
4.7.3 Solution: Diversification ...................................................... 186
4.7.4 Discussion ........................................................................... 190
4.8 Summary and discussion on dilemmas ............................................ 193
5 SOLVING THE DILEMMAS ............................................................................. 201
5.1 Introduction .................................................................................... 201
5.2 Solutions ......................................................................................... 201
5.2.1 Exhibits ............................................................................... 201
5.2.2 Advertising .......................................................................... 203
5.2.3 Aggregation ......................................................................... 205
5.2.4 Community .......................................................................... 205
5.2.5 Exclusivity ........................................................................... 206
5.2.6 Facilitation ........................................................................... 207
5.2.7 Funding ............................................................................... 208
5.2.8 Get big fast .......................................................................... 209
5.2.9 Influencers ........................................................................... 209
5.2.10 Legitimacy .......................................................................... 210
5.2.11 Market-making .................................................................... 210
5.2.12 Marketing skills ................................................................... 213
5.2.13 Open source......................................................................... 214
5.2.14 Partnering ............................................................................ 216
5.2.15 Scarcity ............................................................................... 219
5.2.16 Search-engine marketing ..................................................... 220
5.2.17 Sequential approaches ......................................................... 221
5.2.18 Standalone value ................................................................. 227
5.2.19 Performance-based compensation ........................................ 228
5.2.20 Personal selling ................................................................... 228
5.3 Summary and discussion on solutions ............................................. 230
6 CONCLUSIONS................................................................................................. 237
6.1 Theoretical contribution .................................................................. 237
6.1.1 Addressing research gaps and questions ............................... 237
6.1.2 Expansion of the chicken-and-egg problem ......................... 239
6.1.3 Interrelatedness of platform dilemmas ................................. 240
6.1.4 Strengths and weaknesses of common solutions .................. 241
6.1.5 Conceptual expansion .......................................................... 243
6.1.6 Substantive theory: dilemmas of platform startups ............... 245
6.2 Managerial implications .................................................................. 247
6.2.1 Think, plan, and utilize the roadmap .................................... 247
6.2.2 Avoid the free trap............................................................... 248
6.2.3 Beware of theoretical UG and network effects ..................... 248
6.2.4 Beware of the internalization problem ................................. 250
6.2.5 Concluding advice ............................................................... 251
6.3 Marketing implications ................................................................... 252
6.4 Suggestions for further research ..................................................... 253
6.4.1 Comparison to success ........................................................ 253
6.4.2 More dilemmas.................................................................... 254
6.4.3 Strategic decision-making biases ......................................... 254
6.4.4 More solutions from practice ............................................... 256
6.4.5 Power dynamics .................................................................. 257
6.4.6 Synthesis of marketing and platform theories ...................... 258
6.4.7 Literature integration ........................................................... 259
6.4.8 Introducing other contexts ................................................... 260
6.5 Credibility ...................................................................................... 260
6.5.1 Evaluative criteria ............................................................... 260
6.5.2 Evaluation of credibility ...................................................... 261
6.5.3 Success with theory ............................................................. 264
6.5.4 Saturation ............................................................................ 265
6.5.5 Risks relating to data ........................................................... 267
6.5.6 Risks relating to method ...................................................... 277
6.5.7 Risks relating to researcher .................................................. 278
6.5.8 Generalizability ................................................................... 281
6.5.9 Overall assessment of credibility ......................................... 286
REFERENCES ......................................................................................................... 289
APPENDIX 1 CODING GUIDE ............................................................................ 323
APPENDIX 2 IS THE COLD START DILEMMA REALLY A DILEMMA? ....... 332
APPENDIX 3 SUPPORT FOR EARLY AND LATE LAUNCHES ....................... 333
APPENDIX 4 STRAUSSIAN EVALUATION OF CREDIBILITY ....................... 335
LIST OF FIGURES
Figure 1 Black box of startup failure ............................................................ 24
Figure 2 The platform literature ................................................................... 26
Figure 3 Research process............................................................................ 38
Figure 4 Historical positioning of the analyzed startups ............................... 46
Figure 5 Application of grounded theory ..................................................... 49
Figure 6 Difference between a reseller and a platform ................................. 62
Figure 7 Market coordination and platforms ................................................ 63
Figure 8 Interactions in an advertising-based online platform ...................... 65
Figure 9 Ideal user generation model ........................................................... 79
Figure 10 Exploratory outcomes – opening the black box of failure ............. 93
Figure 11 Strategic actions and their consequences ...................................... 98
Figure 12 Remora and envelopment ........................................................... 137
Figure 13 Weak remora.............................................................................. 175
Figure 14 Strong remora ............................................................................ 176
Figure 15 Dilemmas and associated problems ............................................ 195
Figure 16 Zigzag to a critical mass (Evans 2009a) ..................................... 225
Figure 17 Logic of “keep on trying”........................................................... 251
Figure 18 Spheres of applicability .............................................................. 282
Figure 19 A tentative formal theory ........................................................... 283
LIST OF TABLES
Table 1 Industry examples ........................................................................... 17
Table 2 Descriptions of analyzed startups .................................................... 44
Table 3 Examples from coding guide .......................................................... 52
Table 4 The literature keywords .................................................................. 54
Table 5 Definitions of a platform (i.e., two- or multisided market) .............. 58
Table 6 Types of network effects................................................................. 59
Table 7 Online platform types ..................................................................... 73
Table 8 Online platforms, interaction, and goals .......................................... 75
Table 9 Functional comparison of users and the firm................................... 81
Table 10 Analysis of dilemmas .................................................................... 96
Table 11 Exhibits of cold start dilemma .................................................... 101
Table 12 Too many consumers (of content) ............................................... 102
Table 13 Consumers and generators .......................................................... 103
Table 14 Basic solution of subsidization .................................................... 113
Table 15 Startup platform .......................................................................... 119
Table 16 Incumbent platform (with a critical mass) ................................... 120
Table 17 Exhibits of the lonely user dilemma ............................................ 121
Table 18 Exhibits of the monetization dilemma ......................................... 142
Table 19 Monetization dilemma simplified................................................ 143
Table 20 Price sensitive (both) .................................................................. 144
Table 21 Quality sensitive (A) ................................................................... 145
Table 22 Truth and assumptions ................................................................ 147
Table 23 Strategic pricing (adapted from Chakravorti & Roson 2006) ....... 156
Table 24 Remora’s choice ......................................................................... 170
Table 25 Exhibits of remora’s curse .......................................................... 171
Table 26 Risks of delegation ..................................................................... 174
Table 27 Applicability of dilemmas across platform types ......................... 193
Table 28 Exhibits from post-hoc analysis .................................................. 202
Table 29 Evaluating applicability of solutions ........................................... 231
Table 30 Addressing research gaps ............................................................ 237
Table 31 Answers to research questions .................................................... 238
Table 32 ‘Build it and they will come’....................................................... 255
Table 33 Evaluation of credibility.............................................................. 262
Table 34 Reasons for writing post-mortems............................................... 268
Table 35 Examples of self-attribution ........................................................ 270
Table 36 Examples of different interpretations .......................................... 274
GLOSSARY
Chicken-and-egg problem: the tendency of users not to join a platform if
others are not joining.
Complement (complementary good): an offering such as a service, program,
game, or other type of application provided by a first- or third-party in a plat-
form.
Complementor: a provider of a complement to a given platform.
Content platform: a platform for distributing, consuming, and sharing con-
tent.
Critical mass: the number and quality of members/complements required to
convince others to join a platform.
Cross-side network externality: seeindirect network effect.
Demand side: end users of a platform.
Direct network effect: a network effect applying to users of the same kind
that can be grouped as one based on their motives or interests (e.g., fans of the
same topic).
Envelopment: a strategy for one platform to capture users of another plat-
form.
Exchange platform: a platform connecting buyers and sellers.
Exclusivity: a rule requiring complementors to single-home (seesingle-hom-
ing).
Feedback loop: a positive or negative reinforcing effect according to which
members of a platform follow other members’ strategic actions.
First-party complement: a complement provided by the platform owner to its
platform.
Freemium: an Internet business model combining free and premium offer-
ings.
Indirect network effect: a network effect between different kinds of user that
should not be grouped as one based on their motives or interests (e.g., buyers
and sellers).
Installed user base: the number of users who have adopted a given platform.
Inter-operability: compatibility between platforms.
Inter-platform competition: competition between platforms.
Intra-platform competition: competition between same side members of a
platform.
Liquidity: amount of interaction occurring between members of a platform.
Monetization: converting free services into revenue.
Multihoming: diversifying strategy of platform members (i.e., both demand
and supply side) to commit to more than one platform by utilizing or distrib-
uting offerings.
Network effect: the more members/complements in a platform, the more use-
ful it is for its current and future users.
Platform: a place of interaction connecting two or more complementing
groups.
Single-homing: committing to only one platform (i.e., opposite of multi-
homing)
Social platform: a platform for interacting with other members due to social
motivation.
Standalone value: utility provided by the platform without its complements.
Supply side: complementors of a platform (seecomplementor).
Third-party complement: complements independent of the platform owner.
Two-sidedness: interaction of two groups complementing each other (e.g.,
buyers and sellers in exchange).
User-generated content (UGC): digital content provided by end users of a
platform.
Web 2.0: a term referring to interaction-enhancing features of the Internet.
Winner-takes-all: a competitive case in which the dominant player receives a
disproportionally high share of the total gains provided by a platform.
17
1 INTRODUCTION
1.1 Research background
The development of information technology in the late 1990s resulted in the
hype of e-commerce platforms (i.e., two-sided marketplaces connecting buy-
ers and sellers), only to be quickly followed by their demise (Evans 2009a).
However, the dotcom period spread seeds of a new beginning, and new plat-
form companies such as Google, Facebook, and Twitter quickly dominated
technology users’ everyday lives, and created new job types
1
and business op-
portunities. Many of them also command high returns: Eisenmann, Parker, and
Van Alstyne (2011, 1272) found that “60 of the world’s 100 largest corpora-
tions earn at least half of their revenue from platform markets.” The recent
surge of platforms has been observed in many industries, and businesses have
swiftly adopted the platform strategy and terminology (Hagiu 2009). Concepts
such asnetwork effects andcritical mass capture critical ideas relating to plat-
form-driven change in the business landscape.
The impact of platforms can be seen in the speed and scope of their growth.
Several platform startups have achieved fast growth and resulted in successful
exits for their founding teams (see Table 1).
Table 1 Industry examples
Platform Founded Type Exit*
eBay 1995 exchange platform IPO in 1998
Google 1998 content platform IPO in 2004
MySpace 2003 social platform Sold to News Corp. in 2005
LinkedIn 2003 social platform IPO in 2011
Facebook 2004 social platform IPO in 2012
YouTube 2005 content platform Sold to Google in 2006
Twitter 2006 social platform IPO in 2013
*IPO or trade sale
The success and vigor of platforms make them an interesting topic to study.
As shown by examples in Table 1, it is well established that platform startups
1
Consider a “search-engine marketer”, a profession that did not exist a decade ago.
18
can be immensely successful. However, it is also commonly known that many,
if not the majority, fail. In fact, observations on the hard reality of platform
startups are the inspiration for this study. These observations were made in the
local startup community in Turku, Finland, and in startup-focused online me-
dia, including influential technology and startup-related blogs, such as
TechCrunch
2
by Michael Arrington, AVC
3
by Fred Wilson, Startup Lessons
Learned
4
by Eric Ries, and also startup-oriented discussion forums such as
Quora
5
and Hacker News
6
; all of which include deep and interesting discus-
sions on various startup problems.
Post-mortem stories (i.e., failure narratives) of unsuccessful ventures are of
particular interest as they contain descriptions, written by their founders, of
problems faced by startups. Reading more than a dozen stories of startup
failure raised the author’s interest in the topic of failure, which evolved into
this study. Focusing on failure was deemed important for two reasons: 1) the
early stage of a startup firm is commonly termed the “valley of death” during
which many startups fail, and 2) success stories often include a “survivorship
bias”, as documented in the literature (e.g., Brown, Goetzmann, Ibbotson, &
Ross 1992). This reasoning led the author to believe that a better picture of
business challenges on the Web could be achieved by studying failures instead
of successes. If, indeed, the majority of startups fail, does it not make more
sense to study them rather than relying on data from the few successful ones?
According to this rationale, failures will help us to understand the Web as a
(sometimes hostile) business environment, reasons for why Internet companies
fail, and, ultimately, also some potential explanations for the rare successes.
The first focus of the study was on online business models, with the prem-
ise that the lack of a proper business model leads to a high failure rate. After
an initial inquiry, this assumption was rejected as there were, in fact, a re-
markable number of well-functioning business models, created especially after
the dotcom bust (see Rappa 2013 for a list). Therefore, it was concluded that
the lack of business models probably was not a major explanatory factor. After
reading a few post-mortem stories on the Web, the final research plan began to
take shape. It was clear that failure is a somewhat complex phenomenon and
involves variables at many levels, including commonly cited managerial
shortcomings, lack of marketing, and not solving real customer problems (see
e.g., Sharma & Mahajan 1980). Indeed, these are quite well understood rea-
sons for failure.
2
www.techcrunch.com
3
www.avc.com
4
www.startuplessonslearned.com
5
www.quora.com
6
news.ycombinator.com
19
In contrast, startup-specific business problems had 1) been examined less
extensively in previous research, and 2) not been well solved by practitioners,
as proven by the failure stories disseminated in the startup community. The
study, which initially was to identify critical success factors in different busi-
ness models that could explain why some firms succeed in online business
while others fail, therefore transformed into its current form. In fact, the final
topic of platform dilemmas emerged from the collected material (cf. Glaser &
Strauss 1967) as it became obvious that the majority of startups in the stories
were failed attempts to create a platform-based business. This discovery led to
synthesizing the practical problems into something more: startup dilemmas,
which require deep analysis and creativity to be solved.
Finally, the research conducted here includes a forward approach. Not only
were the problems defined, but also their solutions were actively sought by the
author while conducting research. McCarthy, Plantholt, and Riordan (1981)
wrote in their thesis, “Success versus survival?: the dilemma of high technol-
ogy firms” more than 30 years ago:
“The result of [earlier] studies has been to tell the reader:
This is what happened and,
This is how the companies responded.
We wanted a study that would tell us:
This is what happened and why.
This is how the companies responded, and
This is how the companies should have responded.”
A similar ambition has driven this study, in that it aims to provide useful in-
sight for founders and researchers by enhancing their ability to identify key
strategic problems and their interconnections in platform business.
1.2 Key concepts
The key concepts of this study are defined here. See the glossary for further
terminology relating to online business and platforms.
? Startup: an early-stage business organization. ‘Early-stage’ is defined as
no more than five years of age; ‘organization’ implies that the startup is not
necessarily incorporated; ‘business’ separates non-profit and open-source
projects from commercial ventures; and ‘marketing’ and ‘distribution’ refer to
the acquisition and serving of customers. In particular, a platform startup is a
startup attempting to create a platform-based business.
20
? Dilemma: a contradictory decision-making situation in which all alterna-
tives seemingly lead to an undesirable outcome (see Oxford Dictionary 2013).
In this dissertation, a dilemma is understood as a conceptualized strategic
problem, meaning that it has been given a name and definition. Whereas stra-
tegic problems emerge in idiosyncratic situations, dilemmas are generally
formulated and, thus, more abstract.
? Strategic problem: a decision-making problem of a strategic agent; in this
study, the platform startup. Strategy is defined in this dissertation as a chosen
course of action by a strategic agent that is therefore preceded by decision-
making. In general, any behavior that considers costs and benefits and makes a
choice, given the available information and assumptions, can be considered
strategic (Lyles & Howard 1988). Thus, the pros and cons of strategic prob-
lems are required to be weighed (Schwenk 1984). Lyles and Howard (1985,
131) describe these as “not the everyday, routine problems but the problems
and issues that are unique, important, and frequently ambiguous”. In other
words, strategic problems can be deep and complex.
? Platform: a place of interaction (cf. Evans 2003) that connects two or
more types of actor. Platforms are often associated with network effects (Katz
& Shapiro 1985), critical mass (Rohlfs 1974), and other associated constructs
discussed in Chapter 3. Many scholars interchangeably employ the terms
platform andtwo-sided markets (Rochet & Tirole 2003).
1.3 Research gap
Approximately a decade ago, platforms began to receive the attention of
scholars. Entering the vocabulary of economics, two-sided markets (Rochet &
Tirole 2003) have become the focus of increased research interest in the field
of industrial economics. Related terms such as app marketplaces and ecosys-
tem have gained popularity in other fields (see J ansen & Bloemendal 2013).
Overall, the implications of two-sidedness have spread from economics to
other disciplines such as information systems (Casey & Töyli 2012), strategic
management (Economides & Katsamakas 2006), and marketing (Sawhney,
Verona, & Prandelli 2005).
However, much remains to be discovered, in particular regarding platform
development (Piezunka 2011). These issues are more closely associated with
the platform business model than generic business problems that can be per-
ceived to apply across industries and firm sizes. Problems such as a lack of
marketing, running out of funds, changes in the business environment or
macro-economy, or management errors have been considered in the extant
literature (e.g., Miller 1977; Gaskill, Van Auken, & Manning 1993; Lussier
21
1996; Dimitras, Zanakis, & Zopounidis 1996). Similarly, there are multiple
studies dedicated to challenges faced by new ventures or startups (e.g.,
McCarthy et al. 1981; Zacharakis 1999; Honjo 2000; Azoulay & Shane 2001),
and also in the online context (Han & Noh 1999; Cochran, Darrat, & Elkhal
2006). |The strategic problems concerning platforms are less well-known. In
particular, five gaps exist.
First, not much is known concerning platform-specific business problems
beyond the chicken-and-egg dilemma. Despite some extensions to other stra-
tegic issues (see Chapter 4), the chicken-and-egg problem is perceived as the
fundamental issue in platform business (Evans 2002; Rochet & Tirole 2003).
However, as this study shows, there are other important problems faced by
platform startups.
Second, the perspective taken in the platform literature often neglects the
startup condition, mainly regarding the lack of resources or pricing power.
This tradition can be seen to stem from Farrell and Saloner (1985) and Katz
and Shapiro (1985), often cited by platform scholars, who focus on industry
standards and “monoliths”, not startup firms. For example, Farrell and Saloner
(1986) discuss the “penguin effect” relating to the adoption of standards, im-
plying that none of the players are willing to take the first step. Exceptions in
the more recent literature are Caillaud and J ullien (2003), Evans (2009a),
Evans and Schmalensee (2010), and Mas and Radcliffe (2011) who consider
the chicken-and-egg problem particularly from a startup/entrant perspective.
However, when strategies for solving the chicken-and-egg problem focus on
pricing (e.g., Caillaud & J ullien 2003) and advertising (see Chapter 4), they
might not be effective for startups lacking the means to execute either.
Eisenmann et al. (2011) propose envelopment as a strategy for capturing com-
petitors’ users (see Chapter 4.7). However, more strategies are needed.
Third, most studies relating to the chicken-and-egg problem are theoretical.
Mas and Radcliffe (2011) and Raivio and Luukkainen (2011) are exceptions
as they approach the problem through an empirical case study. Curchod and
Neysen's (2009) working paper is methodologically closest to this study as it
applies grounded theory. Although theoretical and analytical works have a lot
of intuitive appeal, empirical studies can help ground their concepts more
firmly in the reality of platforms.
Fourth, without closer examination on associated problems relating to its
antecedents or arising from its potential solutions, the chicken-and-egg prob-
lem is typically treated as being isolated. Such a narrow focus concerns most
other strategic problems in the platform literature. Strategy scholars such as
Lyles and Howard (1988) discuss interrelatedness of strategic problems. Thus,
a more holistic approach that recognizes the relations of strategic problems
and their solutions can be perceived as necessary.
22
Fifth, strategic solutions considered by the economist-dominated platform
literature are narrow and focus mostly on pricing (Shy 2011). The importance
of pricing in the online environment is negligible as de facto pricing of many
online platforms approaches zero in terms of both access and usage fees
(Teece 2010). Rochet and Tirole (2005), for example, make a case proving
that a platform can offer negative pricing to one market side and remain prof-
itable overall as a consequence of what Evans (2003) terms “internalizing the
externalities” of platform coordination. However, if entry pricing is set at zero
and the platform is still unable to attract users (i.e., solve the chicken-and-egg
problem), what can be done? It seems that answering this question requires an
answer not centered on pricing strategies.
There have also been calls by practitioners for the type of research at which
this study aims (The Entrepreneurial Enlightenment 2012):
"Most founders don’t know what they should be focusing on and
consequently dilute their focus or run in the wrong direction.
They are regularly bombarded with advice that seems contra-
dictory, which is often paralyzing."
Consequently, platforms’ strategic problems can be regarded as having both
empirical and theoretical relevance. In particular, startup founders, managers,
and investors are interested in learning more concerning specific challenges of
platform companies, as such insight can offer competitive advantage in their
respective markets.
1.4 Purpose and research questions
This purpose of this study is to address some of the previously mentioned re-
search gaps with appropriate research questions, and, in so doing, improve the
chances of platform startups to identify and solve central strategic problems
pertaining to the platform business model, thereby also increasing their
chances of survival.
Although platform startups are currently flourishing in the online market
space, every platform must bypass its early stage to become a viable and prof-
itable business. Therein lies the problem, as most startups ultimately fail
(Haltiwanger, J armin, & Miranda 2009; Watson & Everett 1999). To the au-
thor’s knowledge, whether platform businesses have a higher or lower failure
rate than other types of business has not been studied; however, startup ven-
tures tend to generally suffer from high failure rates. It can be assumed that
platform startups are no exception in this sense, and thus studying their prob-
lems forms the research purpose of this study.
23
The research problem can be formulated as the following research ques-
tions
7
:
RQ 1: What strategic problems are encountered by early-stage online plat-
forms?
RQ 2: How can the problems be conceptualized as dilemmas?
RQ 3: Are the dilemmas interrelated? If so, how?
RQ 4: How can the platform literature and founders’ experiences help solve
the dilemmas?
The research questions therefore relate to strategic problems of platform
startups, which are conceptualized as dilemmas. The interrelations of these
dilemmas are examined, and the study analyzes their potential solutions. Due
to the aforementioned high failure rate, it is meaningful to conduct such a
study that aims at improving the survival rate of platform startups by provid-
ing knowledge on potential challenges they are likely to face, and also offering
a basis for solution building.
By asking the first research question, the study extends beyond the chicken-
and-egg problem, and shows that there is more depth and complexity in plat-
form dilemmas than is generally considered in the literature. The study chal-
lenges the simplification of “getting both sides on board” as a solution (Evans
2002), and argues that even if this is accomplished, a platform does not neces-
sarily fulfill its economic goals in terms of becoming profitable. The study
investigates solutions beyond pricing and subsidies that are more suitable for
startups, given their constraints of time and resources. It is particularly useful
to see how theoretical approaches match the reality of startup founders, and
therefore the examined solutions stem from both the literature and empirical
material.
The second research question addresses conceptualization, which is a form
of abstraction; that is, moving from the particular to the general. Conceptual-
ization facilitates 1) communication of strategic problems, 2) their further
treatment, and 3) theory generation. Communication of novel concepts takes
place both among practitioners and scholars. When a theoretical concept has
reached a state of general knowledge within a field, communication relating to
the phenomenon becomes more efficient and advanced (Hunt 2002). The
benefits of conceptualization relating to further treatment can be seen, for ex-
ample, in the famous prisoner's dilemma (Axelrod 2006); defining this
7
Note that, due to the inductive nature of the study, the precise questions were formulated after
the analysis. Whatever their initial form, theoretical sampling of grounded theory tends to reshape
research questions (Urquhart, Lehmann, & Myers 2010).
24
problem and labeling it as it is has led many scholars to attempt to create
variations, find solutions, and apply it to different contexts. According to
Glaser (2002), conceptualization precedes theory generation, and is therefore
sine qua non in academic work. Instead of constantly reformulating the same
problem, practitioners are able to identify the situation in other contexts, and
therefore also consider general and particular solutions, proposed by others, in
their own context. In sum, conceptualization of startup dilemmas can assist in
bringing these benefits jointly to scholars and practitioners.
Furthermore, the identification, conceptualization, and analysis of platform-
specific strategic problems is a worthwhile research purpose because the for-
mulation of strategic problems influences the solution process
8
(Ackoff 1969;
Lyles 1981). As pointed out by strategic management scholars, strategic issues
are rarely isolated cases, and merit a wider perspective (Lyles & Howard
1988). By thoroughly understanding the problem and its associations, founders
are able to elicit appropriate solutions (Lyles 1981). By careful conceptualiza-
tion (i.e., naming and defining) of the strategic problems, this study raises the
abstraction level and provides a deep insight on them.
Although this study does not show exact relationships between specific
problems and outcomes, based on the material, it assumes that strategic prob-
lems impact the failure outcome; that is, discontinuance of business (Watson
& Everett 1993). The following figure displays the idea of a black box be-
tween the startup beginning and the failure outcome, of which the strategic
problems are a part.
Figure 1 Black box of startup failure
Solving strategic problems, therefore, is part of the process leading to the
outcomes of success or failure. Inference can also be drawn from Pawson and
8
Ackoff's (1969) elevator problem is a good example: if waiting for an elevator is defined as a
technical problem, the company needs to engineer faster elevators. If, however, it is defined as a
behavioral problem, people can be given an activity while they are waiting.
Startup Failure ?
Strategic problem Strategic problem
Strategic problem
Strategic problem
25
Tilley's (2009) model by stating that, in thecontext of online startups, strategic
problems act as mechanisms to the outcome of failure. Adopting this logic
highlights the importance of strategic problems as a research problem; as they
are assumed to be associated with failure, the ability of a firm to identify and
solve them through correct strategic choices is likely to have a positive impact
on the firm’s survival. In other words, solving all issues leads to a viable plat-
form in terms of both interaction and revenue.
To accomplish its purpose, the study is based on empirical evidence from
early-stage startups, not on incumbents or established industry firms that al-
ready have a stable position in the market, and can afford to solve the chicken-
and-egg dilemma and other strategic problems by mass marketing or other re-
source-intensive approaches. Therefore, the focus of the study is on early-
stage online startups that employ the Internet as their marketing and distribu-
tion channel, and follow the platform business model by enabling interaction
between two or more groups of users. The units of analysis are failed early-
stage Internet startups, and the empirical material comprises 29 post-mortem
reports by founders of failed Internet startups, originally published on the Web
(see Chapter 2). The material is analyzed by employing GT in an attempt to
answer the research questions, and thus fulfill its purpose.
The importance of the research purpose can be shown in many ways. Gen-
erally, it is accepted that startup companies are important for the economy
(Audretsch & Acs 1994): they create a large share of new jobs (Kane 2010),
develop innovations to improve people’s lives (Almeida, Dokko, & Rosenkopf
2003), redeploy resources by creative destruction (Schumpeter 1961), fill gaps
in customer needs, and tackle problems efficiently and relentlessly from new
perspectives (Shepherd & Kuratko 2009). Therefore, a study aimed at improv-
ing the conditions upon which startups will thrive can be regarded as im-
portant for 1) the society in general, 2) entrepreneurs and managers of plat-
form startups, and 3) investors seeking the best investment opportunities
among competing ventures.
At the same time, the research focus excludes particular types of issue out-
side its scope. Because the study focuses on problems of early-stage platform
startups on the Internet, other types of platforms and problems are beyond the
scope of this study. The excluded platforms exist in various forms; for exam-
ple, shopping malls, credit cards, and newspapers (Rysman 2009). Their
problems might be different from those of online platforms. Other excluded
problems include, for instance, general managerial problems (e.g., lack of ex-
perience) and general startup-related problems such as liability of newness
(Stinchcombe 1965). Details on how the researcher narrowed the focus on par-
ticular dilemmas can be found in Subchapter 4.3.2.
26
1.5 Positioning
This study is positioned to the platform literature, and particularly to its strate-
gic management stream (i.e., strategic management of platforms). Figure 2
demonstrates how the platform literature is understood in this study.
Figure 2 The platform literature
As depicted in the figure, the platform literature comprises:
· Economics literature specializing in two-sided markets (i.e., two-
sided platforms; multisided platforms).
· Information systems (IS) literature relating to electronic market-
places and mobile application marketplaces.
· Marketing literature focusing on consumer interaction within plat-
forms.
· Network literature on network effects and platform/network struc-
tures.
It is argued here that all of the previous streams contribute to the strategic
management of platforms, which also includes their design (Bakos &
Katsamakas 2008). A discussion on their contents and key areas of interest
follows.
It transpires that platforms are studied across disciplines. Rochet and Tirole
(2003) form the basis of the economics literature on two-sided markets. This
research tradition can be seen to originate from prior studies on network ef-
fects, standards, and technology adoption (Church & Gandal 1992; Farrell &
Saloner 1985; Katz & Shapiro 1986). The economists’ agenda often relates to
pricing, regulatory issues, and antitrust policies (Rysman 2009). However, it
Platform
literature
Economics
(two-sided
markets)
Marketing
(consumer
participation
platforms)
Information systems
(e-marketplaces,
app marketplaces)
Network literature
(network effects)
27
also relates to network effects that stem from the earlier literature on network
effects (Katz & Shapiro 1985), critical mass (Rohlfs 1974), and ‘tipping’
(Shapiro & Varian 1998).
In particular, the concept of tipping has been applied to inter-platform com-
petition, underlying the idea of Internet markets as winner-takes-all markets
(Noe & Parker 2005). Whether two-sided markets result in quasi-monopoly
situations and whether this is good or bad (Luchetta 2012) are central ques-
tions in this debate. Subsidies can be regarded as “dumping” in a one-sided
market, whereas they can be regarded as a necessity for creating liquidity in a
two-sided market (Evans 2009b). Another notable aspect for economic analy-
sis is that, in two-sided markets, both price levels and their structure are sig-
nificant (Rochet & Tirole 2003). For example, a sub-optimal solution in one
side (e.g., price subvention) can result in an improved overall solution regard-
ing the two-sided structure (Evans 2003). Thus, the economists’ agenda links
with themes such as pricing, inter- and intra-platform competition, coopetition,
and monopoly (Roson 2005; Birke 2008; Shy 2011).
Second, network theorists tend to consider one-sided platforms in which all
users are the same type (Wright 2004). This makes sense as they are interested
in graphing the network (Viégas & Donath 2004) or examining the diffusion
process (Valente 1995), and not necessarily the transactional implications or
quality of interaction occurring within the platform. Structurally, platforms
can be perceived as networks of connected actors, or “nodes” (Banerji & Dutta
2009). Rysman (2009, 127) notes that although, technically, the economics
literature on two-sided markets can be regarded as “a subset of network ef-
fects”, it tends to focus on pricing, whereas studies on network effects “typi-
cally focus on adoption by users and optimal network size.” In any case, net-
work effects in their positive or negative form remain a central concept (Shy
2011).
The contribution of network theory to the platform literature is exemplified
by Westland (2010) who combines network laws with willingness to pay. Net-
work theory does not usually consider strategic dimensions of platform man-
agement or two-sided implications; rather, it is interested in describing various
network structures, and explaining their growth and diffusion (e.g., Westland
2010). The overlap in the interests of network and platform theories can be
seen in adoption or diffusion. Although adoption can be studied from a social
perspective in line with Rogers’ (1995) seminal book, network theory tends to
focus on descriptive models as opposed to theorizing on the reasons for
diffusion/adoption. A different subset of studies from the information systems
(IS) tradition relates to technology acceptance, such as the technology ac-
ceptance model (Venkatesh & Davis 2000). These approaches have not been
considered in the platform literature that perceives network effects as the main
28
driver of adoption (Katz & Shapiro 1986), and, although they might provide
valuable insight on the complexity of adoption, they are also not considered in
this study.
Third, regarding the IS literature, Hyrynsalmi et al. (2012), Salminen and
Teixeira (2013), and J ansen and Bloemendal (2013, 3) address the recent
stream of mobile application marketplaces which they define as “an online
curated marketplace that allows developers to sell and distribute their prod-
ucts to actors within one or more multi-sided software platform ecosystems".
The focus has shifted from e-marketplaces (i.e., late 1990s to early 2000s) to
app markets. This shift has followed the change in business markets as espe-
cially consumers have adopted various app stores, and their significance has
therefore increased (Hyrynsalmi et al. 2012). The focus of earlier IS research
was often on business-to-business (B2B) exchange platforms, and especially
on the concept of liquidity (Evans 2009a). Contrary to earlier e-marketplace
research, modern app markets such as mobile phone applications are typically
business-to-consumer (B2C)-oriented (J ansen & Bloemendal 2013). A survey
on the electronic marketplace
9
literature can be found in Standing, Standing,
and Love (2010).
Fourth, marketers are interested in platforms. In 1998, Sawhney had already
highlighted the importance of moving from portfolio thinking to platform
thinking in his commentary for the Journal of the Academy of Marketing
Science, and argued that “marketers who master platform thinking may find
the 21st century to be a somewhat more inviting prospect.” Generally, the
marketing literature tends to focus on the consumer perspective of platform
interaction and strategies relating to marketing problems such as finding and
influencing particular lead users to propagate messages (Hinz, Skiera, Barrot,
& Becker 2011) or otherwise participate in platform interaction: co-creation
platforms (i.e., firms leveraging consumers in their value-creation activities) or
peer-marketing platforms for customers voicing their opinions. Hennig-
Thurau, Gwinner, Walsh, and Gremler (2004) studied consumer opinion plat-
forms, combining virtual communities and the traditional word-of-mouth liter-
ature. In particular, seeding is regarded as a viral marketing strategy to attract
the most prominent users to join a platform (Hinz et al. 2011). Marketing
studies overlap with network studies that aim to find the most influential
“nodes” (Hill 2006). However, marketing adds the actual persuasion of these
nodes to join the platform as "real people". Another special interest of
marketers is the relationship between a platform owner, advertisers, and
9
As discussed later, marketplaces are aspecial type, but not the only type, of platforms relying on
economic exchange as the form of interaction. For other types of platform, the interaction might take a
different form.
29
consumers in an advertising-enabled platform (e.g., Salminen 2010; Reisinger
2012).
Sawhney et al. (2005) examine the Internet as a platform for value co-crea-
tion with customers. This collaborative innovation can take advantage of the
Internet’s distinct features and be exploited, for example, in new product de-
velopment. Cova and Dalli (2009) discuss how marketing theory is developing
towards working consumers as it focuses on value co-creation and customer
participation. Platforms seem to offer opportunities for marketers to leverage
and monetize customer input, and marketing scholars have been showing in-
terest incrowds (Howe 2006) as resources. While marketing studies tend to be
applied, the idea of consumers as an extension of the firm can be perceived, at
a higher abstraction level, as linking to the Coasean theory of the firm (Coase
1937) and transaction cost analysis (Williamson 1981). The overlap shows
how economists and marketers are often interested in the same phenomenon,
although at a different degree of abstraction. In platform terminology, cus-
tomer participation can be expressed with the concept of user-generated con-
tent (UGC), or actions of a platform’s users, such as participating, writing,
uploading, and sharing (Daugherty, Eastin, & Bright 2008). As most of the
studied startups apply user generation (UG), this dissertation considers impli-
cations of the UG model. However, matching the approach with marketing
research (e.g., customer participation) is left for future studies.
The strategic perspective, taken by this study, is to conceive and evaluate
strategic choices for platform stakeholders (cf. Cusumano 2010). As a
perspective, it is not limited to any discipline but to all related works that per-
ceive actors as strategic in a platform context. While a lot of attention has been
paid to the platform owner’s perspective (Birke 2008), including management
of an installed base of users, standards, and complements (McIntyre &
Subramaniam 2009), research also discusses other stakeholders' strategies,
such as those of software developers (Salminen & Teixeira 2013), that is,
complementors. In addition, platforms’ end users are mostly assumed to react
to pricing (e.g., Rochet & Tirole 2003), the number and quality of comple-
ments, and network effects; in other words, the presence of other users (Evans
2002). Of special interest is users' adoption choice, which is perceived to be
constrained by exclusivity versus diversification (Roson 2005) and resource
constraints (Iacovou, Benbasat, & Dexter 1995).
Multihoming, the practice of participants to diversify their investments
across platforms, is an example of the strategic perspective in the platform lit-
erature (Armstrong 2006). Such behavior can be seen to occur in both the
supply- and demand-side, and its feasibility relates to cost functions; namely,
whether it is wise to commit to several platforms in terms of time, effort, and
financial cost, or whether focusing on one platform is sufficient to satisfy the
30
participant’s goals. Multihoming has been studied, for example, by
Hyrynsalmi et al. (2012) in mobile application markets. Another example of
the strategic perspective is the analysis of open versus closed platforms; that
is, which setting is more suitable for the platform owner/complementor
(Boudreau 2010; Eisenmann, Parker, & Van Alstyne 2009; Gawer &
Henderson 2007; Parker & Van Alstyne 2008). The strategic perspective
therefore focuses on platform stakeholders’ choices in reaching particular
goals and outcomes. An overview of strategic problems in the platform litera-
ture can be found in Chapter 4.
Due to the focus on startup dilemmas, or contradictory decision-making sit-
uations, this study is particularly positioned to the strategic management
stream of the platform literature. However, the study utilizes the related liter-
ature such as economics papers with a strategic focus; for example, determin-
ing the correct pricing, design, entry strategy, and subsidization. Such aspects
can contribute to solving startup dilemmas relating to online platforms. The
strategic perspective does not in itself exclude any actors as they can all be-
have strategically; users consider their own benefits, as do platform owners
and complementors.
Perhaps the distinctive feature here is that the question of which actors to
examine varies across streams of the platform literature. As economists are
interested in markets, they tend to examine buyers and sellers, or demand and
supply sides (Evans 2009a). IS researchers are currently paying increasing in-
terest to app marketplaces, so their two sides comprise app developers and end
users (e.g., Mian, Teixeira, & Koskivaara 2011). The strategic choice of a
marketing manager is how to allocate marketing budget (Fischer, Albers,
Wagner, & Frie 2011), whereas it is the CEO/founder’s problematic role to
organize the whole business. This study takes the latter perspective and fo-
cuses on dilemmas that are inherent to the startup’s business strategy, and not
only to its marketing strategy, pricing, or capability to attract developers. The
perspective is that of the startup/founder: What can it/he do to solve the
startup dilemmas and avoid failure?
In sum, it can be seen that platform studies across disciplines are interre-
lated and contain overlapping interests. For example, marketers are interested
in graph theory to find potentially influential users, and economists and mar-
keters share the interest of users/customers as “resources” or assets of the firm.
The chicken-and-egg problem is common to all; as economists aim to solve it
through pricing while marketers propose sales and persuasion as the solution,
these studies underlie adoption as the fundamental phenomenon. In a similar
vein, a strategic focus can be found across disciplines; it is more dependent on
which strategic problems within the platform are of interest to scholars and
what kind of solutions they examine. The present study focuses on strategic
31
problems grounded on the material, and can be perceived as critical for the
survival of platform startups on the Internet.
1.6 Structure
The study proceeds as follows. Chapter 2 describes the methodology, includ-
ing research strategy, research process, data collection and analysis, and also
the literature approach. The theoretical framework, including conceptual un-
derpinnings and their connection to platform literature, is discussed in Chapter
3. This chapter defines platforms as a concept, explains particularities of
online platforms, and presents critical assumptions; namely, user-generation
effects, which will be referred to in the empirical part.
Startup dilemmas in Chapter 4 contain the empirical part of this study, and
are based on post-mortem stories written by founders of failed startups. The
chapter investigates specific problems of platform startups that are conceptu-
alized into dilemmas and then analyzed with the help of the platform
literature. The treatment of each dilemma is divided into four sections:
definition and exhibits, literature review, potential solution, and overall
discussion. Chapter 5 elaborates further solutions based on the second-round
analysis, arising from both the empirical material and the scholarly literature.
Finally, Chapter 6 presents the contribution to theory and practice, further
research ideas, and evaluates the credibility of this study.
33
2 METHODOLOGY
2.1 Research strategy
2.1.1 Introduction to research strategy
This study aims at the creation of substantive theory (Glaser & Strauss 1965)
relating to strategic problems of platform startups on the Internet. This is ac-
complished by conceptualizing and increasing the abstraction level of the an-
alyzed post-mortem stories. The grounded theory (GT) methodology, outlined
by Glaser and Strauss (1967) is applied as an instrument of data collection and
analysis. The following sections explain the method, why it was chosen, and
how it was applied throughout the research process.
2.1.2 What is GT?
Grounded theory is aset of methods to systematically analyze empirical mate-
rial (Finch 2002). This data can be both quantitative and qualitative (Glaser
2004), although GT is most often associated with qualitative data (Kempster &
Parry 2011). Partington (2000) notes that the foundations of GT include
theoretical sampling, or a process of data collection guided by the emerging
theory andconstant comparison, or simultaneous coding and analysis of data.
Suddaby (2006, 634) confirms this perspective, and adds that “oth concepts
violate longstanding positivist assumptions about how the research process
should work.” This contradiction relates to the method’s history of countering
deductive methods in favor of theory generation from data (see e.g., Locke
1996, for a more detailed discussion). The coding process if more thoroughly
discussed in Chapter 2.4.
By nature, GT is an inductive
10
method, intended to help the researcher
elicit answers to his or her research problem from the empirical material
(Eisenhardt 1989). Contrary to deductive reasoning, in which the
presumptions are stronger and the researcher is narrowing the scope of
10
Note that by ‘inductive method’, an inductivetendency or emphasis is implied. Pure induction
and pure deduction, for that matter, are generally considered impossible; new ideas arise from their
combination, or abduction (Suddaby 2006).
34
inquiry, in inductive logic, the scope of inquiry is broader and central issues
are gradually revealed by scrutiny, which in GT is represented by the coding
process (Strauss & Corbin 1994). Eisenhardt (1989, 541) points out that, in
inductive studies, “researchers constantly compare theory and data-iterating
toward a theory which closely fits the data.” This fit between data and emerg-
ing concepts is perceived as important because it reduces the risk of the latter
being detached from empirical relevance (Eisenhardt 1989).
In contrast, by employing hypothetico-deductive logic, the researcher first
develops hypotheses; that is, assumptions concerning what is likely to happen
or be found in the analyzed data (Laudan 1981). After this, the hypotheses are
tested with a specific method such as experiments or statistical analysis, and
the results are discussed. The hypotheses are created either by observing real-
world phenomena or by analyzing the literature for theoretical gaps (Davis
2009). Grounded theory differs from this logic in at least three aspects.
First, there are no initial hypotheses that prove, disprove, or generate a the-
ory, and the theory is generated with the fewest presumptions possible (Glaser
& Strauss 1967). Second, GT does not rely on identifying a theoretical gap
prior to analysis (Heath & Cowley 2004). The defense of the method comes
from the self-proclaimed novelty of the phenomenon and idiosyncrasy of the
utilized material, as a consequence of theoretical sampling. The self-pro-
claimed novelty implies that there is indeed some reason for inquiry as not all
is yet known (Pandit 1996). This derives either from uniqueness of the data or
from newness of the phenomenon
11
. Although the literature is not employed as
a starting point, the relationship to previous theory needs to be considered, and
this is done subsequent to data collection and analysis (Goulding 2005).
According to Glaser (1978, 51), “[r]eading the theoretical literature should
be avoided when possible until after the discovered framework is stabilized”.
Therefore, in GT, the literature review is conducted after the formulation of
categories, which might seem unconventional for researchers trained only on
hypothesis testing (Kempster & Parry 2011). Finally, whereas research ques-
tions in hypothetico-deductive studies tend to be fixed, GT allows for chang-
ing the initial questions if found irrelevant in the field (Charmaz 1990). Such
flexibility is advantageous for theory generation as it reduces the impact of
preconceived constructs and encourages the discovery of new concepts.
Two points emerge from the previous explanations. First, it was posited that
GT is inductive by nature; second, there are no initial hypotheses. The two
words, emphasized in italics, have important implications. The nature of GT is
not pure induction, but more of abduction (as so-called inductive studies tend
11
For example, one cannot have conducted research on Web 2.0 startups prior to 2005 because the
concept did not exist. Thus, a substantive theory is a contemporary outcome, placed in time.
35
to be). This means deductive reasoning is employed in the course of the re-
search; but there is a strong emphasis on “letting the data speak for itself”
(Glaser 1978), as opposed to forcing it into preconceived hypotheses. How-
ever, over the course of the research process, central themes become more ap-
parent, at which point the researcher is encouraged to compare new findings
with intermediary conclusions; this is a form of deductive reasoning.
As an example, consider the following observation: “A cat is in a tree, be-
cause a dog chased it there.” The researcher can employ this piece of data to
formulate a general hypothesis: “Dogs don’t like cats.” To confirm this hy-
pothesis, however, he needs to conduct theoretical sampling. Thus, he seeks
more empirical descriptions on the relationship between cats and dogs, and
discovers the following description: “Today it was so nice to come home and
see my cat Jim and my baby dog Bozo asleep in the same basket.” Aha! Cats
and dogs can get along, so the original hypothesis is incorrect. This method of
comparing new and previous data through tentative assumptions enables cor-
rection and modification of our hypotheses. For example, we can observe a
specific condition and say that “dogs don’t like cats, unless they are accus-
tomed to them from an early age”, and then seek to validate or refute this hy-
pothesis through theoretical sampling. This simple example shows how hy-
potheses emerge from the analysis of data, not precognition; in other words,
they are grounded (Glaser & Strauss 1967).
Due to the somewhat vexing issue of induction/deduction, the GT method
can be best labeled as data-oriented. The rationale of data orientation is to
connect theory with the real world (Glaser & Strauss 1967). In other words, if
the only way to generate new theory were to examine existing theory, one
would never originate ideas “outside the system”. In contrast, the source of
theory can bein praxis or, in effect, the data, as opposed to its fitting into a
priori theoretical framework. Here it is not claimed that one or the other is
better; in the author’s opinion, the research gap can be found both in the liter-
ature andempiricism. For the latter, the challenge is to ensure the phenomenon
has not been exhausted prior to engaging deeply in research activities
12
while,
for the former, it is to ensure the topicality of the research purpose in real life.
Overall, GT aims to avoid theoretical exercises detached from real problems
(Glaser & Strauss 1967).
Thus, ideas come from informants themselves and are labeled to match
their use of language in vivo (Charmaz 2006). The sense-making of the
informants is translated by the researcher to match the discourse in the
12
This would waste resources as enough is already known on the topic, and the researcher would
be unable to provide new insights. In practice, however, novelty is debatable as many phenomena
occasionally reoccur in the literature without being rejected.
36
academic literature. This matching process is a requisite for a) finding the the-
oretical discourse in which the study can be positioned, and b) formulating the
theoretical contribution by employing established research language and con-
cepts (McGhee, Marland, & Atkinson 2007). If this translation is not per-
formed adequately, the researcher risks remaining isolated from academic dis-
course. The successful employment of GT results in a theory with unique as-
pects, although parts of it might overlap with existing theory (Glaser & Strauss
1967)
13
.
2.1.3 Why was GT selected as research method?
First, the researcher was interested in problems of post-dotcom Internet
startups, a phenomenon not well studied (see Chapter 1). When there are no
exact presumptions and the research topic is quite new, a method aiming to
discover central topics is beneficial (Glaser 1978). According to Finch (2002,
220), grounded theory fits well with “the development of novel knowledge
claims of under-researched phenomena.” As identified in the previous
chapter, there are several gaps relating to platform-creation activities, and
managers actively seek to understand why particular strategies work while
others do not. To determine the answer, theoretical analysis is needed.
Grounded theory (GT) is particularly useful when data are in qualitative
form and the researcher still seeks a systematic methodology (Glaser 2004).
GT gives good grounds for conceptualization and raising central topics and
patterns from the data (Charmaz 1990). These features are compatible with the
objectives of this study; thus, GT provides a good methodological match for
solving the research problem.
Second, the richness of the type of data on which this study is based is sim-
ultaneously both an advantage and a disadvantage. Qualitative data requires a
significant amount of sense-making and structuration (Suddaby 2006); how-
ever, the reduction process offers good grounds for theorization (Miles &
Huberman 1994). These properties support the choice of a method that aims to
generate explanations from data. GT is compatible with this need because it is
presented as a systematic method of analysis (e.g., Glaser & Strauss 1967;
Eisenhardt 1989; Kempster & Parry 2011) and often applied to qualitative data
(Goulding 2005).
Third, GT seemed a good fit for the author’s tendency to conceptualize.
Heath and Cowley (2004) argue that the researcher’s cognitive style should
13
Of course, the researcher needs to identify these overlaps and position his/her contribution
towards the extant literature. However, this process can take placeafter the analysis.
37
play a role in the selection of method. As explained in the previous chapter,
the research began with qualitative material from failed startups. The analysis
started by finding general themes in the material. As knowledge on different
methods increased, the author understood that he was in fact coding (e.g.,
Strauss & Corbin 1994). Shortly thereafter, the author learned about grounded
theory. As the method corresponded to the approach taken up to that point, it
was not a large step to adopt GT’s principles.
Fourth, the research construct, strategic problem, is not a fact that can be
quantified, observed, or measured as a variable in an empirical model. Nor can
it be regarded as a latent variable that might be constructed by utilizing other
variables; at least, not without complex interpretations. Rather, a strategic
problem is a concept, or a conceptual construction of reality. This study as-
sumes that strategic problems exist in the real world and, once defined in the
correct strategic situation, can be perceived as a relatively stable form of real-
ity by all interpreters with adequate understanding on their nature. Thus, stra-
tegic problems are situational patterns that emerge when specific contextual
conditions are met. Such an ontological position implies critical realism, in
which real events remain dormant until triggered by particular conditions,
upon which they become actual and, if observed, empirical (Partington 2000).
To produce such understanding that is required to identify and explain strate-
gic problems – as it is not obvious that everyone, even experienced managers,
will identify thema priori – GT has to offer a set of highly useful principles.
Fifth, GT enables an easy expansion from one topic dimension to another,
not being required to remain within the scope of the initial data (i.e., “all is
data”). In this study, this feature of GT shows in expansion from problems to
solutions; that is, additional interviews focusing on solutions and analysis of
online material relating to them. The approach is much different compared to
research designs where the data are static (i.e., remains as what is collected)
and then is employed to answer a priori research questions. In GT, a priori
research questions can change. In this study, the focus changed from business
models to failure to strategic problems, and was finally annexed by the discov-
ery of solutions. Therefore, GT's approach to see “all as data”, in addition to
its flexibility in terms of theoretical sampling and constant comparison, en-
couraged the researcher to be led by the phenomenon instead of his initial pre-
conceptions.
Relating to strategic issues of platforms, Gawer (2009) highlights the im-
portance of a firm’s capabilities and also industry- and firm-specific circum-
stances; that is, thecontext. An emphasis on context supports the choice of GT
as a methodology, as it is often applied to generate understanding on a re-
search problem in a particular context (Chesler 1987); that is, a substantive
theory (Glaser & Strauss 1965). For example, Orton (1997) reported the use of
38
GT in studying strategic change processes in loosely coupled systems. Kan
and Parry (2004) examined resistance to change in a hospital setting, and
identified paradoxical thinking as an influencer. According to Wagner,
Lukassen, and Mahlendorf (2010, 9), “grounded studies are especially
appropriate for gaining an initial understanding of complex transitions”; ar-
guably, strategic problems associated with startup failure can be categorized as
such.
The extant platform literature has approached the chicken-and-egg problem
(i.e., getting both sides on board) mainly from the pricing perspective, and
focused on analytical modeling (Piezunka 2011). Few inductive studies have
been conducted to understand the roots of the problem, or how it might be
solved (e.g., Birke 2008). This study provides a step in that direction. As will
be shown, mere pricing (i.e., levels or structure) is insufficient as a solution to
the cold start problem; in fact, several studied startups offered their products
for free, and still failed to gain growth. The lack of participation is only par-
tially explained by overly high prices; fundamentally, it is a much more com-
plex phenomenon. This study is geared towards the interpretations of failed
startup founders. In these stories, founders explain why their ventures failed.
The inductive nature of the study will provide a needed empirical grounding
for the treatment of strategic problems.
2.2 Research process
This chapter describes how the study was conducted, and how it evolved over
the course of the analysis. Figure 3 depicts the research process.
Figure 3 Research process
First, post-mortems were collected through online searches and by follow-
ing links from aggregators and curators of post-mortem stories [1]. Note that
Data
collection
Data
analysis
Literature
collection
Theoretical
integration
Post-hoc
coding
Constant comparison & theoretical sampling
1
2
3
4
5
Additional
interviews 6
Finalizing
7
0
39
post-mortems represent first-hand data for the analysis, whereas other readings
and discussions with founders in various startup events correspond to theoreti-
cal sampling of GT [0]. In turn, adjusting the learning from the literature, dis-
cussions, and interviews to preliminary findings corresponds to constant com-
parison of GT [0].
In the first phase of data collection, everything relating to failure of Internet
startups was retrieved. In the second phase, criteria for inclusion and exclusion
were developed and narratives were filtered, which is explained in the fol-
lowing subchapter. In total, 29 failure narratives remained at this point. Then,
to analyze the material [2], several phases of coding were conducted according
to the GT method (see Subchapter 2.4).
During the coding phases, strategic problems emerged as the key theme of
the study. Thereafter, the conceptualization of the dilemmas began. At this
point, various streams of the literature were collected and read [3] to
determine the study’s theoretical framework positioning. Note that at this
stage, several alternatives for placing the findings in the literature existed. The
studied research streams comprised the literature focusing on business models,
business failure, and platforms (two-sided markets). Eventually, based on the
author’s judgment, it was decided that strategic problems of the studied
startups, conceptualized as “dilemmas”, had most in common with the plat-
form literature. The research focus was therefore narrowed down, and a sys-
tematic integration of the platform dilemmas into the extant literature began
[4].
Consequently, a more thorough retrieval and review of the platform litera-
ture began (see Chapter 2.5). The researcher made multiple searches and col-
lected the literature by snowball sampling the found papers. The literature was
read keeping the theoretical constructs (i.e., dilemmas) in mind, and the syn-
thesizing of startup dilemmas and the platform literature began. This process is
described astheoretical integration by Urquhart, Lehmann, and Myers (2010).
Commonalities with the findings and extant theory could be found quite eas-
ily, which reassured the researcher that the correct literature had been chosen.
In other words, the literature and empirical accounts seemed to discuss the
same phenomena, despite utilizing different words.
In general, conducting the literature review and positioning after the analy-
sis is in strict accordance with the principles of GT (Strauss & Corbin 1994;
Glaser 2004). Such a choice is intended to facilitate inductive theory for-
mation: that is, to avoid preconceptions arising from the literature to shape the
conceptualization, understanding, and interpretation of the initial findings to
the degree where their originality is lost.
The author found it purposeful also to consider some potential solutions in
addition to the extensive analysis of the problems. However, solution finding
40
is an extension to the work; its main focus is on the dilemmas. To find addi-
tional solutions to the dilemmas, two steps were taken. First, post-hoc coding
was conducted after the literature analysis [5]; this time focusing on solutions
to the dilemmas. This step included coding of the original material for “what
if” statements in which the founders expressed what they would have done
differently, had they been given the choice. Second, the author decided to uti-
lize the principle of theoretical sampling by conducting additional interviews
[6] with founders.
After these efforts, the report was finalized [7]. The report was written in a
conventional format, outlining research questions, then methodology, the liter-
ature, and results. Suddaby (2006) mentions that this is a common way to re-
port a GT study. Although the exact research problems were formulated ex
post, presenting them in the introduction helps readers understand the study’s
purpose. The study itself began with no preconceived theory, as is the case
with GT studies (Glaser & Holton 2004), and the research gap or research
problem did not initially exist in the way described in Chapter 1. Prior to the
analysis, the researcher was interested in a different purpose than that which
emerged from the material over the course of the analysis. However, Suddaby
(2006) notes that although it can at times seem confusing, reporting GT by the
conventional “deductive study structure” is normal.
2.3 Research data
2.3.1 Data collection
The analyzed material comprises 29 failure reports by founders of failed
startups. The narratives, or “post-mortems” as termed by startups, were written
by founders to reflect the startup’s failure, in particular to identify reasons for
that failure. Thus, post-mortem is defined here as a story analyzing a failed
startup venture. The stories were collected from the Internet by following
links from various blog articles listing and publishing post-mortem analyses,
and conducting searches via Google search engine and two startup-centered
online communities.
Keyword phrases for Web searches included:
· startup failure story
· startup postmortem/post-mortem
· startup failure analysis
· business postmortem/post-mortem
· business failure analysis.
41
The data collection process began by gathering all post-mortem stories the
researcher could find. The search was conducted by finding aggregated blog-
posts listing startup failure stories and then following links to original posts,
similar to “snowball” sampling (see Biernacki & Waldorf 1981), and by per-
forming Web search queries. In particular, ChubbyBrain (2011) contained
links to several post-mortem stories. Following links, post-mortem stories
were captured for further filtering and analysis. Additionally, Google
14
was
utilized to find post-mortems; this is because Google’s search algorithms tend
to be the most accurate of current search engines (Uyar 2009), and its index of
Web pages is commonly judged as current and extensive (e.g., Gulli &
Signorini 2005).
Moreover, searches were conducted on two startup-focused online commu-
nities: Quora and Hacker News. These communities contain a substantial
number of discussions relating to Web startups, and also included discussion
threads on startup failure. Reading these discussions helped the researcher to
become familiarized with the phenomenon and find links to still new post-
mortem stories.
In addition to reading the post-mortems, the author sought additional ways
to deepen his knowledge on the startup industry or, as it is commonly termed,
the “startup scene”. The steps for doing so comprised the following.
First, approximately a dozen interviews by a startup-focused journalist
Andrew Warner
15
were read. As these interviews were freely available on the
website in transcribed form, they were read to find confirmation, contradic-
tion, or complements to the post-mortems’ findings. The interviews were a
good source of secondary data because they included both successful and
failed startups, and therefore provided useful background information on the
industry and the startups’ founders’ decision-making and ways of thinking.
Second, to deepen the knowledge on the sampled startups, the comment
sections of the post-mortem stories were read; the stories were published in
blogs, and therefore could be commented upon. There were some cases in
which other founders participated by questioning parts of the analysis or by
sharing their own stories. In addition, in a couple of cases, customers disa-
greed with the story, and also the content suppliers of one startup were bitter
(i.e., in platform terms, the “other side”). Although fascinating, analyzing the
discourse between founders and other interpreters was not the goal of this
study, so the researcher did not go deeply into the question of “who is right”.
14
www.google.com
15
www.mixergy.com
42
However, familiarization was enriched when founders’ replies brought further
clarification to the cases.
Finally, six additional interviews were conducted with founders. Each of
the interviews lasted for approximately an hour and was theme-based, the
theme being the strategies and tactics the founders had employed, or knew
about, in solving the chicken-and-egg dilemma. The founders were asked
questions such as “How are you solving the problem for side A(/B)?”, “What
is your most successful (/unsuccessful) solution?”, and “How are you planning
to grow in the future?” All founders were knowledgeable of the topic, and
could express advanced ideas relating to it. During the interviews, the author
made notes and mentally compared the emerging points to previous findings.
Later, the notes were integrated into the solutions section of this study.
Although the process of theoretical sampling can be continued for a very
long time (in fact, infinitely), within the frame of this study (i.e., its focus on
dilemmas; time constraints) six interviews, in addition to post-hoc coding,
were regarded as adequate for the discovery of solutions. Due to GT's ac-
ceptance of additional data collection and its comparison to earlier findings,
and also its acceptance of different types of data, the research process can be
continued in the future.
2.3.2 Selection criteria
All post-mortems were filtered for further analysis. The selection criteria com-
prised:
· Internet-based commercial venture, but not necessarily incorporated.
· Post-mortem written by one of the founders.
· Can be defined as a platform, connecting two or more groups.
· Established between 2004 and 2010 (i.e., Web 2.0, after the dotcom
period)
· No more than 60 months old (i.e., early-stage startup).
The “Internet-based” criterion stems from the research purpose, which is to
study online business, not offline-with-online-extension, or hybrids (i.e.,
“click-and-mortars”). A general definition of a platform was applied to iden-
tify appropriate startups; moreover, the process resulted in the emergence of
four online platform types.
Additionally, the depth and length of stories were considered, so that the
accepted stories had at least approximately 1,000 words to ensure some
“thickness” (Neilsen & Rao 1987). On average, a post-mortem story
43
comprised 3,037 words. Post-mortems were preferred to be as candid and
unbiased as possible, although this is a subjective measure; potential biases
will be considered later. The stories were not anonymously written as they
included authors’ names. To maintain somewhat consistent interpretations,
only stories written by founders were included; for example, there were some
that recounted interviews with founders, but these were judged less authentic
than had the founders actually written the stories. Tracking the authors in
social media services ensured authenticity of the stories. Most founders were
found via LinkedIn
16
, and they provided more information on their cases.
According to the previously mentioned principles, non-Internet businesses,
seemingly short and superficial stories, those not personally written by found-
ers, and those written in an editorial style or by a journalist were filtered out.
Filtering was conducted to limit the scope of study to self-reflection that was
inherently honest, authentic, and of some depth. However, for selection, incor-
poration (i.e., being a registered company) was not required as it was per-
ceived that this would rule out very early-stage startups on which the study
focuses.
2.3.3 Description of the startups
Overall, after 12 stories were excluded based on the aforementioned criteria,
29 failure stories remained for analysis. Short descriptions were written to
summarize the startups’ purpose in an easily understandable way. Such de-
scriptions facilitate the examination by third parties unfamiliar with the
startups; crystallization is also helpful for analytical purposes. Descriptions
were retrieved from two startup databases, CrunchBase
17
and ChubbyBrain
18
,
or, when neither of the databases contained data on a startup, Google search
engine was employed to find a description, preferably from the founder’s web-
site or blog. The general descriptions can be found in the following table.
16
www.linkedin.com
17
www.crunchbase.com
18
www.chubbybrain.com
44
Table 2 Descriptions of analyzed startups
Description Type Side A Side B
Backfence was a hyper-local, community-based
news and information service.
Content Local users Local users
Boompa was a social encyclopedia focusing on
motor vehicles.
Content Users Advertisers
Bricabox was a platform for creating a personal
social content site.
Social Users Users
ChubbyBrain captured and structured infor-
mation on innovation economy and startups.
Content Users -
Contrastream was a social music platform. Content Indie musicians Users
Devver aimed to turn desktop development tools
into cloud-based services.
Infra Users Developers
Diffle was a social networking site centered on
simple flash games.
Social Users Users
eCrowds combined Web content management
and social networking for SMEs.
Infra Consumers SMEs
eHarmony for Hiring aimed to match job-seek-
ers with job-providers.
Exchange J ob-seekers J ob-providers
EventVue was a tool for building conference
communities.
Social Conf.
participants
Conf.
participants
Imercive provided an IM marketing solution to
help brands’ consumer engagement.
Infra Brands Users
Kiko offered anyone a calendar to keep and
share online.
Social Users Users
Lookery helped social networks distribute their
data outside their Web sites.
Content Users Social
networks
Meetro was a location-aware IM client and real-
time social networking application.
Social Users Users
Monitor110 helped institutional investors ac-
cess, analyze, and monetize Web information.
Content Investors -
NewsTilt was a service for journalists to build an
online brand by engaging their readers.
Social Readers Journalists
Nouncer enabled real-time distribution of micro-
content to Web applications.
Infra Users Developers
Overto aggregated information from different
auction platforms to deliver better results.
Content Buyers/sellers -
Pixish was a platform for user-generated graphic
design work.
Exchange Designers Design-seek-
ers
PlayCafe was an online network that streamed
user-generated game shows.
Content Users Users
[Q&A startup] aimed at creating a marketplace
for selling and buying answers.
Content Askers Answerers
RiotVine was a social event guide for discover-
ing and sharing events with friends.
Social Visitors Event
organizers
SMSnoodle was an SMS based entertainment
channel for the Singapore region.
Content Content
providers
Users
SubMate enabled discovering new people and
things to do before and after commuting.
Social Commuters Commuters
Transmutable was a platform for doing 3D sim-
ulations on the Web.
Content Users -
Untitled Partners enabled fractional ownership
of art through cooperative purchasing.
Exchange Art-lovers Art-lovers
Verifiable wasa platform for data visualization. Content Users Users
Wesabe was a finance service for tracking per-
sonal spending patterns.
Exchange Consumers N/A
Xmarks offers a social bookmarking and syn-
chronization service.
Content Users Users
45
As can be seen, all platform types are presented. Their definitions are dis-
cussed in Subchapter 3.3. The frequency of platform types is as follows
(N=29):
· 13 content platforms
· 8 social platforms
· 4 exchange platforms
· 4 infrastructure platforms.
Most platforms studied are two-sided, but there are also one-sided plat-
forms, in which the users are not divided into two mutually complementing
groups. The average lifetime of a startup was 26 months.
The oldest startup was 57 months at the time of failure, the youngest 8
months
19
. The sample comprised both B2C and B2B startups, with the major-
ity being consumer-oriented startups. The mode of team size was 2.5 mem-
bers, with the largest team having 30 members and the smallest one member.
Most teams were male-dominant, and only two reported women in their team.
Approximately half of the founders (57%) were first-time founders, the rest
had earlier startup experience. All teams had technology experience, but only
38% reported prior marketing experience. The vast majority was US-based
startups; there was one startup from Poland and one from Singapore. Almost
all startups (86%) also applied either user-generation (UG) or aggregation as
their content creation model
20
, which makes UG (Chapter 3.4) highly
characteristic of this sample. Other characteristics include offering free ac-
cess/use of the platform, indirect monetization, and the freemium business
model. These features become relevant in Chapter 4.
The contemporary focus (see Figure 4) excludes dotcoms, several of which
were found among all retrieved post-mortems. Coincidentally, the selected
startups are a part of the so-called Web 2.0 era (O’Reilly 2005). The Web 2.0
period can be seen to start from around 2005 when the concept was first intro-
duced (O’Reilly 2005). The dotcom period is generally regarded to have oc-
curred in the late 1990s to the early 2000s (Razi, Tarn, & Siddiqui 2004;
Evans 2009a), including a strong hype cycle of unrealistic expectations for
Web platforms and e-commerce (Lieberman 2005), and then a quick demise
after a large share of these businesses failed to perform (Cochran et al. 2006).
19
Calculated from date founded to the date post-mortem was published.
20
A content model explains how the startup provides content for its users.
46
Figure 4 Historical positioning of the analyzed startups
The platforms in the dotcom era were mostly e-marketplaces (Wang,
Zheng, Xu, Li, & Meng 2008); since then, there have been considerable
changes in online startups’ business models (Rappa 2013) . While the first
wave of platforms included “importing” retail and B2B exchanges to the In-
ternet, Web 2.0 platforms offer purely digital services on their own (Aggarwal
& Yu 2012). This fact does not particularly reflect the research purpose (see
Chapter 1), although it does add to the topicality and novelty of the material
analyzed in this study.
With the relative “freshness” of the sample, the goal was to ensure that
problems remain topical. If problems were already solved in the “latest batch”
of Web startups, there would be no research gap in the empirical sense, which
would be a critical problem for GT that aims at usefulness of the resulting the-
ory (Glaser & Strauss 1967). Such a risk would be higher had the study in-
cluded Web 1.0 startups, and ignored the implicit learning occurring after, and
due to, them. In turn, the research gap in the literature stems from incomplete
understanding on the chicken-and-egg problem, its solutions, and derivative
problems (see Chapter 1.1).
2.4 Analytical approach
2.4.1 Coding process in GT
According to Strauss and Corbin (1994), after data collection, the researcher
should start by open coding; for example, reading through narratives, making
notes, and identifying themes and interesting phenomena. This process leads
to the creation of categories, or groupings of concepts, that appear to relate to
the same phenomenon (Glaser & Strauss 1967).
Dot-coms Web 2.0
ca. 1999-2001
Samplestartups (2004-2010)
ca. 2005à
47
GT reaches theoretical refinement through iteration; once themes begin to
emerge, the researcher re-reads the material while modifying conceptual
codes. Essentially, this leads to an index of codes, organized under categories
based on the nature of the phenomena being described in the text (Strauss &
Corbin 1994). A category can contain subcategories if the researcher interprets
the phenomenon as a hierarchy (Glaser & Strauss 1967).
Next, axial coding procedures are employed to compare extant codes to the
subcategories, and a selected part of the material is modified to reflect the core
category (i.e., selective coding). In general, axial coding refers to looking for
relationships between conceptual constructs, and the conditions in which they
take place; for example, they might coexist or appear only under specific cir-
cumstances (Strauss & Corbin 1994). The idea is to develop connections be-
tween the themes found in earlier coding (Strauss & Corbin 1994).
Through constant comparison, a grounded theorist derives the core category
from the material (Glaser 2004). The method is then employed to build con-
sistency through a mental process of comparing new coding to existing cod-
ing, so that it becomes part of a single theoretical framework (Goulding 2005).
A suggested approach to this is memoing (Strauss & Corbin 1994; Glaser
2004), which means noting down ideas in a form of meta-data.
Constant comparison reveals whether the new data provide anemergent fit
or not, thereby guiding theory generation (Glaser 2008). GT is an iterative pro-
cess in which new themes and relationships emerge, and the researcher is re-
quired to re-code the data (Gasson 2003). Finch (2002) describes this as mov-
ing from description to analysis, and from analysis to explanation.
Grounded theory arises from the interaction between researcher and data;
therefore, becoming intimate with the circumstances is helpful. According to
Glaser (1978), understanding the context can increase theoretical sensitivity,
which is a mixture of theoretical (i.e., the literature) and practical expertise,
and can improve the researcher’s judgment. Charmaz (1990) makes it clear
that topical knowledge improves the researcher’s ability to perform GT analy-
sis. The researcher’s inner ability to conceptualize in a meaningful way is
highlighted by both Glaser (1978) and Strauss and Corbin (1994) through the
concept of theoretical sensitivity.
Constant comparison is coupled with theoretical sampling by maximizing
similarities and differences between coded phenomena to find the boundaries
of theoretical constructs, not only what is apparent in a limited amount of
data
21
(Creswell 2008). The end result should be an abstract theory derived
from non-abstract data; that is, there is an increase in the level of abstraction
(Strauss & Corbin 1994). Together, theoretical sampling (i.e., finding
21
The requirement, therefore, is an adequate amount of data for all major variations to appear.
48
additional evidence to back up intermediary conclusions) and constant
comparison direct the researcher to verify whether the emerging theoretical
model holds as new data are collected, and to modify the model if necessary
(Elliott & Lazenbatt 2005).
A critical part of GT, as in much research, is deciding what to include or
exclude (Wagner et al. 2010). This cannot be known a priori, as central
themes are unknown until open coding. Once they emerge, one resorts to se-
lective coding; that is, discovering the central phenomenon (Glaser & Strauss
1967). The purpose of selective coding is to choose the themes and codes cen-
tral to the theory under development (Corbin & Strauss 1990). The idea is that
they are unknowna priori, as opposed to hypothetico-deductive logic, so that
the researcher does not know before looking what is central in the data (Heath
& Cowley 2004). Having strong presumptions, therefore, is a risk to finding
the core phenomenon in the data
22
. Once the core phenomenon (i.e., category)
emerges, alternative ways of looking at the material become less relevant, and
the work pivots around the central phenomenon (Strauss & Corbin 1994).
2.4.2 Application of GT in this study
In this study, the author maintained an open mind when first becoming ac-
quainted with the stories. The material was imported to QSR NVivo 10 (i.e., a
software package for qualitative data analysis), with which it was coded.
Figure 5 illustrates how GT was applied in this study.
22
In this sense, GT formulation is an inductive process; however, empiricists and rationalists both
agree that there is no pure deduction or induction because deduction by the researcher’s human mind
always carries some preconditions (i.e., no tabula rasa), as also does deriving conclusions from
empirical data. For a treatment on this topic see, for example, Perry and Jensen (2001); for abduction,
see the method section in Aarikka-Stenroos (2011).
49
Figure 5 Application of grounded theory
First, open coding was employed for familiarization with the material. At
this point, the process is exploratory and descriptive, and does not rely on hy-
potheses of earlier research, although there might be theoretical sensitivity
arising from the literatures’ knowledge, and professional and personal experi-
ence (Strauss & Corbin 1994). The process of constant comparison in this
study involved 1) analyzing failure narratives to build grounded theory and 2)
discussion with founders of other Internet startups outside the sample to verify
the intermediary conclusions. Therefore, the results of inquiry are consistent
with what founders deem important, but transcend individual accounts in
comparison and level of abstraction (cf. Finch 2002).
After open coding (Strauss & Corbin 1994), there was the choice to focus
on 1) startup failure, 2) strategic startup dilemmas, or 3) startup fallacies. Ini-
tially, startup failure was chosen as the focus because the author thought it was
an interesting topic. However, reading the literature eventually made it clear
that new venture failure has been quite extensively studied (see Chapter 1),
with results similar to those suggested by the open coding. Therefore, the
study risked confirming earlier findings and not generating much new
knowledge. On realizing this, the researcher became familiar with the platform
literature and decided to change focus. Therefore, “Dilemmas” was chosen as
the core category and other phenomena were excluded from this work.
Generally, GT studies tend to choose one core category, due to reasons of
manageability (Holton 2010). Dilemmas emerged as the core category in this
study due to the prevailing role of strategic problems in founders’ sense-mak-
ing. In particular, as strategic problems emerged as the most explanatory
theme, there was a shift from examining failure, as the core category, to stra-
tegic problems that were conceptualized as startup dilemmas. The kind of
Strategic problems
Strategic
problem 1
Strategic
problem 2
Strategic
problem 4
Strategic
problem 3
Axial coding
Core category
Data
Selective coding
Open coding
50
flexibility this change represents can be regarded as an advantage of the GT
method that encourages transformation in the course of analysis (Goulding
2005). It is better to change focus than to pursue a less fruitful topic (Corbin &
Strauss 1990). Moreover, the earlier analysis is rarely wasted but re-emerges
as insight for the new focus (Bauman 2010). For example, the initial focus on
failure factors helped the author associate the strategic problems with failure;
that is, ensure that they are important to the outcome.
In the axial coding phase, the researcher found both confirmation and ex-
ceptions, resulting in amending, extending, or refuting earlier assumptions; as
well as developed connections between the categories. Throughout the re-
search process, there was an iterative process of theory formation and constant
comparison of new ideas to earlier ones. This process of familiarization, meant
to increase theoretical sensitivity, included discussions with startup founders at
various startup events in Finland (Turku; Helsinki), Sweden (Stockholm) and
United States (San Francisco) between 2010 and 2014. The researcher partici-
pated in events where startup founders, investors, and enthusiasts gathered to
present business ideas and demonstrations, and to network. The events in-
cluded, among others, Good Morning Stockholm (2010; 2011); Slush Helsinki
(2011; 2012); Summer of Startups keynotes (Turku, 2011); Startup Day
(Stockholm, 2012); Launch Festival (San Francisco, 2014); and many other
startup events in Turku and San Francisco
23
.
The discussions, especially in the first years of conducting research, repre-
sent theoretical sampling in this study; that is, verification of intermediary
conclusions. As the study evolved into strategic dilemmas, the researcher be-
gan asking questions such as “What is your startup’s biggest problem at the
moment?” Answers categorically corresponded to one of the identified dilem-
mas, most typically relating to user acquisition. This led to belief that adequate
theoretical saturation had been accomplished. Overall, dialogue with founders
was comfortable as it is conventional to pitch (i.e., present business ideas) at
the events in which the author participated; thus, founders were in a ready
state of mind to discuss their startups. In addition, many of the founders were
interested when the author mentioned he was studying the failure of platform
startups, and they could relate to the dilemmas as explained by the author.
Moreover, the author was involved as an active member and board member
of the Boost Turku Entrepreneurship Society based in Turku (Finland) during
the period of data collection and analysis, and could therefore observe several
early-stage online startups in their infancy.
Overall, discussions with startup founders and also reading relevant back-
ground material increased the researcher’s domain-specific knowledge on
23
The author spent three weeks in San Francisco participating in local startup events.
51
online startups, thus facilitating the conceptualization of dilemmas. The par-
ticipation in startup events provided useful access to founders with different
business ideas, and enabled comparison with the emerging categories based on
the startups in the sample.
2.4.3 Coding guide
Strauss and Corbin (1994) set out an exact procedure, aparadigm model, for
GT analysis. In this study, the paradigm model requiring the analysis of con-
ditions, context, action, and consequences (Strauss & Corbin 1994) is regarded
as overly complex and not relevant to the research topic. Partington (2000, 95)
notes that this is a common concern and discourages the literal use of the
Straussian approach:
"The difficulty of applying universal grounded theory prescrip-
tions is borne out by experience with doctoral students working
in the field of organization and management who have attempted
to follow the Strauss and Corbin approach but have abandoned
it because of its bewildering complexity."
Glaser (1992) strongly attacks the Straussian approach for what he terms
“forcing conceptual categories”. This disagreement has been discussed exten-
sively elsewhere (e.g., Heath & Cowley 2004; Locke 1996), and will not be
repeated here. Instead of employing Strauss and Corbin’s (1994) paradigm
model, a coding guide is generated based on subcategories of the core cate-
gory; namely, “Dilemmas” as the core category with each dilemma as a sub-
category. Based on this classification, the material was re-coded. Table 3
shows examples from the coding guide.
52
Table 3 Examples from coding guide
Code Meaning Example
Coldstart
dilemma
Inability to get
content with-
out users.
underestimated the “Cold Start” problem, I read this article by
Bokado Social Design which talks about a big issue you face with a
social site, especially when it relies on user-generated content. The
value you provide to your users centers around the content on the
site, so to build a user-base you need a lot of content created by the
first users to kick-off the community.
Lonely user
dilemma
Inability to get
users without
other users.
If someone wasn’t online when you were online, they were no good
to you. While the real-time chat aspect of the application made for
some really serendipitous meetings, it also made it harder for people
to gauge the activity of their communities, especially if they logged
in at odd hours, people were set as away, etc.
Monetiza-
tion di-
lemma
Inability to
charge money
and get users.
For four years we have offered the synchronization service for no
charge, predicated on the hypothesis that a business model would
emerge to support the free service. With that investment thesis
thwarted, there is no way to pay expenses, primarily salary and
hosting costs. Without the resources to keep the service going, we
must shut it down.
The comprehensive coding guide can be found in Appendix 1. In total, 162
codes were utilized to discover meanings from the data. At the open coding
phase, general reasons for failure were coded. In the axial coding phase, they
were grouped into larger categories such as “Marketing”, “Team”, and “Busi-
ness model”. These described the founders’ reasons for failure. However, two
theoretically interesting categories also emerged at this stage: “Dilemmas” and
“Fallacies”, respectively referring to strategic problems and erratic thinking.
This is in line with GT, whereby coding proceeds from description to abstrac-
tion from time, place, and people (Glaser 2008).
The author was unable to find the approach of coding guide in the GT
methodology literature; the closest is Schmidt's (2004) description of employ-
ing a coding guide in the analysis of semi-structured interviews. Nevertheless,
this approach was useful
24
. Additionally, the online platform typology and
ideal UG model (see Chapter 3) reflect the conditional parameters that axial
coding, and also its paradigm model and conditional matrix, aim to discover
from the data. In other words, the spirit of GT is followed. At the same time,
this feature of opting towards more flexibility positions this work more closely
to classic GT, according to the prevailing interpretations of these two schools
(e.g., Heath & Cowley 2004; Locke 1996). Therefore, this study can be per-
ceived as being closer to the Glaserian school, advocating creativity instead of
rigor of analysis, although it does not explicitly subscribe to either school. In
fact, the commonalities of the two approaches seem to exceed their differences
24
It can be argued that writing a coding guide taps into the same cognitive processes as memoing;
namely, articulating and explicating the nature of the emerging constructs.
53
and, as shown by later editions of Strauss and Corbin’s book first released in
1994, they do not necessarily require the categorical following of their proce-
dure
25
. Therefore, it is not seen that two approaches are mutually exclusive
and, consequently, there is no need for a strict adherence to either at the ex-
pense of the other.
Finally, in the course of the analysis, the author applied game-theoretic il-
lustrations (i.e., strategic game situations) as an analytical tool. Regarding di-
lemmas as “games” facilitated their systematic analysis. These illustrations
can be seen in Chapter 4.
2.5 Literature approach
This chapter describes the process of the literature review. Note that the liter-
ature searches were conducted only after the initial analysis (see Chapter 2.2).
In other words, the analysis guided the selection of this particular theoretical
framework, and therefore positioning towards the literature. The overall work
is positioned to the platform literature, which can be perceived as a multi-dis-
ciplinary field. This enabled the researcher to selectively “borrow” the litera-
ture from other areas, such as entrepreneurship and strategic management; that
is, beyond the contribution of these disciplines to the platform literature. How-
ever, this study is positioned in the platform literature, from which theoretical
constructs are drawn.
The base concepts were as follows:
· Online platforms and ‘internet platforms’
· Platforms
· Two-sided markets and ‘two-sided platforms’
· Double-sided markets
· Dual-sided markets
· Multi-sided markets and ‘multisided markets’
· Multi-sided platforms and ‘multisided platforms’.
The concepts were deduced from the platform literature. Table 4 contains
keywords that were combined with the base concepts to produce search que-
ries.
25
“The analytic process should be relaxed, flexible, and driven by insight gained through
interaction with data rather than being overly structured and based only on procedures” (Corbin &
Strauss, 2008, p. 12).
54
Table 4 The literature keywords
Related dilemma Keywords
Cold start dilemma chicken and egg, chicken-and-egg, chicken-egg
user generation, user-generated content
Lonely user dilemma network effects, critical mass
Monetization
dilemma
monetization, monetization
freemium
Remora’s curse power, embedded
Freemium, a term originated by venture capitalist Fred Wilson in 2006 was
selected as a keyword because it has gained increasing interest from practi-
tioners and academicians (see e.g., Teece 2010, 2011), and many online
startups have adopted it as their monetization model. For the same purpose,
‘user generation’ was included. There were not many studies that referred to
these concepts in the platform context; however, studies conducted on other
contexts were chosen, which proved useful in positioning the dilemmas (see
Chapter 4). The literature searches were conducted with the Nelli search en-
gine (i.e., Turku University’s library system) that connects with the major lit-
erature databases including, for example, Science Direct, EBSCO, and Web of
Science. A special approach was employed to generate search queries, which
involved determining base concepts and combining them with keywords re-
lating to selected dilemmas. For example, if a base concept was ‘online plat-
forms’ and the dilemma-specific keyword was ‘power’, then the search query
would be ‘online platforms +power’. Phrase match of search words (e.g.,
“keyword”) was utilized, which resulted in more relevant hits than broad
match (i.e., keyword). All fields of articles were searched, including author-
specified keywords, title, and abstract.
The results were checked for relevance by reading their abstracts to elimi-
nate irrelevant articles (for a similar approach, see e.g., Wiltbank, Dew, Read,
& Sarasvathy 2006), leaving 302 articles that were saved to folders and read in
the process of analysis. The literature was then expanded based on reading the
articles, a form of snowball sampling. In particular, Publish or Perish software
was employed to retrieve articles
26
. This freeware software for Windows ena-
bles the user to run queries onGoogle Scholar, and shows up to 1,000 results
in one view. Additionally, it enables rapid sorting based on rating (i.e., number
of citations). According to Kousha and Thelwall (2007), Google Scholar is a
26
http://www.harzing.com/pop.htm
55
useful complement in retrieving research material, as it can find scholarly
works not included in academic databases.
Priority in selecting articles for a thorough reading was given to recent re-
search as interest in platforms is relatively new. Moreover, classic articles
were read to discover the origins of concepts and theory; for example, the
standards literature from the 1980s (e.g., Katz & Shapiro 1985; Farrell &
Saloner 1985), and network effects from Rohlfs (1974). The classics, and also
more recent seminal papers such as Rochet & Tirole’s (2003), were deduced
from the recent literature. The aim was to utilize the state-of-the-art platform
literature when positioning the dilemmas. Moreover, dissertations were con-
sidered, as some eminent platform theorists wrote their dissertations on plat-
forms (e.g., Hagiu 2008). Working papers were also included, although they
were retrieved beyond the database search.
According to Roson (2005), working papers form a considerable part of the
early (modern
27
) platform literature. Additionally, it was found that the highest
fit for this study were articles explicitly mentioning ‘online’, ‘internet’, or
‘platforms’ in their titles. Departing further from these concepts meant ab-
stracting from the context of online platforms and moving into more general
studies on the phenomenon. Merely including the theme keywords would have
generated a vast amount of the literature relating to the respective
phenomenon (e.g., ‘power’); however, the scope was kept within the platform
literature. Glaser (2004, 12) explicitly mentions that GT treats the literature
“as another source of data to be integrated into the constant comparative
analysis process.” Arguably, this has led to the selection of a theoretical
framework compatible with the emerged phenomenon.
27
“Early modern” refers to the interest following Rochet and Tirole’s seminal working paper in
2001, later published in 2003.
57
3 THEORETICAL BACKGROUND
3.1 Concept of platform
3.1.1 Platform theory and platform literature
Instead of a unified platform theory, scholars rely on similar constructs and
assumptions to study the particularities of platform business. The literature
focusing on the issues of two-sidedness, the chicken-and-egg problem (see
Chapter 4.4), critical mass, network effects, multihoming, and single-homing
(see Chapter 4.7), and associated constructs comprise what can be termed the
platform literature (Rochet & Tirole 2005; Roson 2005; Birke 2008; Shy
2011). These constructs also form the theoretical foundation of this study.
Platforms have been studied in several contexts. The following list contains
examples of some platforms that have been studied: online infomediaries
(Hagiu & J ullien 2011), mobile application marketplaces (e.g., Salminen &
Teixeira 2013), operating systems (Church & Gandal 1992), videogames
(Maruyama & Ohkita 2011), Yellow Pages (Rysman 2004), credit cards
(Rysman 2007), magazines (Kaiser & Wright 2006), and computer industry
(Gawer & Henderson 2007). More examples can be found, for example, in
work by Parker and Van Alstyne (2005). The different contexts are joined by
similar dynamics, including two-sided economics and network effects, which
are crucial for understanding the platform model. These dynamics are dis-
cussed in the following subchapter.
3.1.2 Defining platforms
A platform, or a two-sided market, can be defined in many ways. Table 5
shows definitions judged as the most important based on the literature review.
58
Table 5 Definitions of a platform (i.e., two- or multisided market)
Author(s) Definition
Evans (2003) “[multi-sided] platforms coordinate the demand of distinct groups of customers
who need each other in some way.”
J ullien (2005) “[Two-sided markets are] situations where one or several competing ‘platforms’
provide services that are used by two types of trading partners to interact and
operate an exchange.”
Rochet and
Tirole (2005)
“markets in which one or several platforms enable interactions between end-
users, and try to get the two (or multiple) sides ‘on board’ by appropriately
charging each side […] while attempting to make, or at least not lose, money
overall.”
Armstrong
(2006)
“Many markets involve two groups of agents who interact via ‘platforms,’ where
one group’s bene?t from joining a platform depends on the size of the other
group that joins the platform.”
Evans (2009b) “[Platforms] serve distinct groups of customers who need each other in some
way, and the core business of the two-sided platform is to provide a common
(real or virtual) meeting place and to facilitate interactions between members of
the two distinct customer groups.”
Gawer (2009) “Industry platforms are building blocks […] that act as a foundation upon which
an array of firms (sometimes called a business ecosystem) can develop comple-
mentary products, technologies or services.”
Rysman (2009) “Broadly speaking, a two-sided market is one in which 1) two sets of agents in-
teract through an intermediary or platform, and 2) the decisions of each set of
agents affects the outcomes of the other set of agents, typically through an exter-
nality.”
Hagiu and
Wright (2011)
“an organization that creates value primarily by enabling direct interactions
between two (or more) distinct types of affiliated customers”
First, based on the definitions, some platforms can be described as infra-
structure
28
rather than a market. Markets are what economists consider places
of exchange; that is, where people and companies trade goods and services
(for more definitions, see Diaz Ruiz 2012). Exchange, in other words, is one
form of interaction taking place in a platform of a particular kind (i.e., a mar-
ketplace), but it is not the only form, as will be shown in this study.
Second, a platform can take the shape of a network; consider, for example,
a telephone network or a social network on the Internet. However, it can also
be depicted as a repository of content, from which users retrieve content that
others have contributed. In sum, there are a large number of platforms with
various traits, although they share the same core (i.e., place of interaction
29
).
Third, as can be seen from Table 5, a characteristic commonly associated
with platforms is the presence of so-called network effects (Katz & Shapiro
1985). In a simple form, due to network effects, the more users a platform has,
28
“[A] base of common components around which a company might build a series of related
products” (Cusumano 2010).
29
Moreover, a place indicates a physical or virtual location.
59
the more valuable it becomes. When the platform is subject to direct network
effects
30
, a user’s benefit from utilizing a product increases with the number of
other users of the same kind (Shapiro & Varian 1998). Some physical net-
works, such as railroads or telephone networks, are classic examples of direct
network effects (e.g., Katz and Shapiro 1985). For example, as the railroad
network grows, more destinations become available to passengers. In a similar
vein, the more there are installed telephone connections, the more people one
is able to call. The required network size; that is, an adequate number of users
for a platform to serve its purpose of providing matches, is termed critical
mass.
In addition to direct network effects (i.e., relating to users of the same kind),
a platform can be subject to indirect network effects (i.e., relating to users of
another kind), which are essential for two-sidedness in the platform defini-
tions; that is, there are two distinct groups which influence each other (Rochet
& Tirole 2003). Moreover, an indirect network effect can be positive or nega-
tive, which is strictly a question of perception. Table 6 is based on Shy (2011)
who distinguishes between positive and negative, and direct and indirect di-
mensions.
Table 6 Types of network effects
Direct Indirect
Positive
Positive direct network effects
(e.g., telephone)
Positive indirect network effects
(e.g., auction)
Negative Negative direct network effects
(e.g., spam)
Negative indirect network
effects (e.g., advertisements)
Note that, due to perception, some network effects can be interpreted as
both positive and negative by different people (Shy 2011). For example, some
users might enjoy advertisements, whereas others find them disturbing
31
. Fi-
nally, indirect network effects can be asymmetric, so that one side of an inter-
action appreciates the presence of the other side more than that side appreci-
ates it. For example, if there are only a few buyers in a marketplace, new
buyers are more important to sellers thanvice versa. However, sellers are im-
portant for buyers, as there would be no marketplace without them.
30
Another term, network externalities, is sometimes employed to refer to the same phenomenon. In
this study, they are regarded as interchangeable. This is in line with common terminology in the
literature, while bandwagon effects is also employed. However, strictly speaking, “any economic
effect is an externality only if not internalized” (Farrell & Klemperer 2007, 2021).
31
The perception of advertising is affected by several factors, such as targeting and quality. Thus,
it is a good example of the relative nature of network effects.
60
More precisely, two-sided platforms “coordinate the demands of distinct
groups of customers who are dependent on each other” (Hagiu 2006). J ullien
(2005, 234) defines two-sided markets as “situations where one or several
competing ‘platforms’ provide services that are used by two types of trading
partners to interact and operate an exchange.” Consistent with these defini-
tions, Evans (2003) associates three properties with two-sided platforms: 1)
the presence of two distinct groups; 2) demand coordination benefits, whereby
one group increases the benefits perceived by the second group; and 3) the
necessity for an intermediary to “internalize the externalities”. According to
Evans (2002), it is crucial in two-sided markets to differentiate between price
structures
32
and price levels
33
.
The demand coordination benefits in this definition can be regarded as
equivalent to the concept of network effects
34
. Following Rochet and Tirole
(2003), it can be regarded as typical for platforms to subsidize one group of
users while making profit from the other group. However, because the two
groups experience cross-group linkages, it is not possible to isolate the profits
from the second group without the presence of the first group.
3.1.3 Markets vs. platforms
Hagiu and Wright (2011) state that the platform literature “has constantly
struggled […] with a lack of agreement on a proper definition”, continuing to
state that some authors have implied that retailers, such as grocers, supermar-
kets and department stores would be platforms. Indeed, the basic tenets
35
of
‘two-sidedness’ and ‘interaction between them’ can be satisfied with any mar-
ket, and therefore the platform literature would be no different from the earlier
way of understanding markets.
Relating to this conflict, Rochet & Tirole (2005, 2) refine their original
2003 definition because, based on it, “pretty much any market would be two-
sided, since buyers and sellers need to be brought together for markets to exist
and gains from trade to be realized.” For example, consider a hardware store
that deals with both suppliers and end customers; end customers go there be-
cause the store provides hardware which, in turn, is provided because the store
32
“How to divide the total price [of a transaction] between buyers and sellers” (Evans 2002, 46).
33
“What total price to charge [from] buyers and sellers” (Evans 2002, 46).
34
The concept of network effects differs from economies of scale in that the latter is regarded as a
feature of a single firm, whereas network effects generate benefits for the whole network of firms,
which are compatible with one another (Birke 2008). However, network effects can, in a sense, be
understood as demand- or supply-side economies of scale.
35
Two-sided platforms “serve two types of agents, such that the participation of at least one group
raises the value of participating for the other group” (Li, Liu, & Bandyopadhyay 2010).
61
is frequented by end customers. This and any type of mediated market ex-
change effectively follows the logic of network effects, and will be an “ordi-
nary” market. This definition is also visible in Evans (2009b, 4), who argues
that the fundamental role of platforms is “to enable parties to realize gains
from trade or other interactions by reducing the transactions’ costs of finding
each other and interacting.” In general, such a role can be regarded as being
close to that of a marketplace mediating supply and demand.
How, then, are platforms different from any other market? The platform lit-
erature provides an answer. In their later paper, Rochet and Tirole (2005, 2)
redefine a two-sided market as “one in which the volume of transactions be-
tween end-users depends on the structure and not only on the overall level of
the fees charged by the platform”. Such a particularity does not exist in a one-
sided market. In other words, this definition implies that the price structure
replaces price level as the key focus of interest. For example, one cannot con-
sider how free users are charged in freemium-based
36
online platforms (i.e.,
one side), but has to include paid users (i.e., two sides) to understand the mar-
ket. As there are expected network effects between the two groups,
influencing how price is distributed between them will either increase or
decrease the number of interactions.
This definition also avoids some of the other shortcomings. First, it refers to
users as opposed to trading partners, which is more appropriate for some non-
exchange platforms. In other words, the scope and type of interaction in plat-
forms exceeds the notion of exchange; it can be exchange, but it can also be
something else while still having indirect economic implications. By strict
definition, when the type of interaction moves from economic exchange to
other forms, the platform is no longer a marketplace. For example, it is not fair
to argue that a social network would be a marketplace, because users most of-
ten interact out of non-economic motives.
Are, therefore, all markets platforms? Essentially, “Yes”. As they require
both buyers and sellers to be present, they are two-sided platforms or two-
sided marketplaces. However, not all platforms are marketplaces. A market-
place is defined by exchange while, for example, a content platform host ac-
tivities relating to the content without engaging in exchange with other users.
However, although all platforms do not require economic exchange, they re-
quire some form of interaction. It might be discussion, sharing, content pro-
duction, and consumption, or more; thus, not all platforms are marketplaces.
36
Freemium, a portmanteau of ‘free’ and ‘premium’ (Wilson 2006), refers to one group of users
paying for a Web service and another utilizing it for free.
62
3.1.4 Mediation vs. coordination
Rochet and Tirole’s (2003) price coordination, however, is not foolproof as
their definition does not solve the problem of mediation; that is, how is inter-
action, such as exchange, organized in a platform. For example, consider a
merchant who subsidies some suppliers to sell products cheaper to end users;
here, price structure is affected and also the volume of products is likely to
change as consumers buy more due to low prices.
A satisfactory solution to this issue is offered by Hagiu and Wright (2011)
who distinguish resellers, or classic intermediation, from platforms, so that the
former deals with each market side separately; for example, a hardware store
first negotiates the inventory with suppliers, and then sells it to customers via
its retail locations. Essentially, the platform owner is an enabler of interaction,
but its active participation is not required for the participants to self-organize
interaction, as participants are engaged in direct communication. The situation
of intermediation versus coordination is illustrated in the following figure.
Figure 6 Difference between a reseller and a platform
In both cases, the presence of the other side is beneficial (i.e., there are net-
work effects); however, the coordination structure is different. In regular in-
termediation, the intermediary first creates dyadic relationships with both par-
ties individually, and only then enables the transaction. In a platform, the in-
termediary provides the platform for “open” interaction between the parties.
This has strong implications; for example, relating to customer power (i.e.,
who holds the customer relationship?) and the quality of interaction. As such,
consider how the platform owner is able to filter out negative externalities
(this topic is revisited in Subchapter 4.5.2.).
Regardless of it being in direct control of the interaction, the platform still
has to attract both participating sides (see Chapter 4.4), and enable their
interaction through some medium; for example, a website, bar, or a shopping
mall. Often, it also has to monitor the quality of interaction to prevent negative
Reseller
Supply-side
Platform
Demand-side
Supply-side
Demand-side
63
externalities
37
. The argument presented in Figure 6 is compatible with
Piezunka (2011) who differentiates between intermediation and coordination
as two distinct platform activities. Rysman (2009) follows a similar logic,
noting that in the intermediary model, which he terms a one-sided market, the
reseller takes ownership of the goods, whereas in a two-sided market the plat-
form owner enables transactions between exchange partners.
This distinctive feature has also been noted by other scholars (e.g., Luchetta
2012). Essentially, mediation by the platform is not necessarily of a transac-
tional nature, and the parties interact directly with one another. For example,
users send messages to one another, and the platform merely enables this di-
rect interaction. According to this perspective, shopping malls are platforms
because they enable direct interaction between shopkeepers and shoppers; su-
permarkets are not, because they are reselling the suppliers’ products. The im-
plication is that a shopping mall enables a direct relationship with the cus-
tomer, whereas supermarkets take control of customer relationships; such im-
plications are discussed in Chapter 4.7.
Moreover, platform coordination has some distinctive features. The evolu-
tion from market coordination to intermediation and, finally, to platforms can
be understood with the help of the following figure.
Figure 7 Market coordination and platforms
37
The exception is when the groups are self-coordinating, which follows the ideal UG model.
Side A Side B
Isolated
market
Intermediation
Platform
64
1. In the beginning, there are many isolated marketplaces where interaction
between market actors takes place (i.e., isolated marketplaces phase).
2. Then, there is a mediator that cuts the transaction costs relating to search,
negotiation, and interaction between market actors (i.e., mediated market-
places phase).
3. Finally, there is aunified platform where, again, parties self-coordinate;
however, here, the platform technology provides them with tools to cut trans-
action costs. The unified platform is independent of time and place, unlike
isolated marketplaces.
In this evolutionary perspective, the match-making model moves from dis-
intermediation to mediation to platform. In the first, matches are random and
ad hoc, whereas in a platform they are controlled by the platform technology.
Between these two, the intermediary cuts the number of connections required
from a market actor to successfully transact with another actor; these are clas-
sic intermediation benefits widely known in the strategic management and
marketing literature
38
(see e.g., Bergern Dutta, & Walker 1992; Bakos 1998).
In brief, the platform model enables scaling of market coordination without
losing the intermediation benefits, as self-organization is highly efficient
through the platform technology.
3.1.5 Direct and indirect effects of interaction
Luchetta (2012) employs the aforementioned distinction to divide platforms
into two-sided transaction and two-sided non-transaction markets. Indeed, the
distinction makes intuitive sense because a transaction does not always include
only the two sides; for example, in media markets where buyers of the media
space (advertisers) transact with both the advertising network and the consum-
ers seeing the ads. However, Luchetta (2012, 11) then claims that
"In media markets, the two sides are not necessary, they repre-
sent a business strategy. Television channels are a good exam-
ple: there are channels whose business model is two-sided, that
is based on free content and advertising revenues, alongside of
pay-per-view channels which earn revenues from subscription
fees."
38
If there arex supply agents andy demand agents in an isolated market, the number of potential
connections in an isolated market isx*y. In the intermediation model, they only need to deal with one.
The intermediation benefit is then (x*y)-1 for the whole market, and x-1 or y-1 for the agents,
respectively. The number of potential connections is the same in a platform as in isolated markets, but
it is assumed there exists a factor k, according to which both the overall efficiency and efficiency of an
agent to find matches is better than in the isolated market model.
65
However, this study disagrees with the argument, simply because of net-
work effects; that is, benefits derived by advertisers from displaying ads to
users. The advertising-based platform would not exist without advertising and
therefore, even assuming no direct interaction between advertisers and users
39
,
users derive indirect benefit from the existence of advertisers in the platform.
Moreover, it is not essential to delimit the type of interaction to transacting. In
an online media, interaction between advertisers and users takes place, for ex-
ample, through impressions (i.e., views), clicks, and email subscriptions.
Eventually, users might purchase; however, before that, the advertiser is inter-
ested in conversion-supporting actions in the sales process, as is generally un-
derstood in online marketing theory (Soonsawad 2013). Two sides are there-
fore necessary from the advertiser’s perspective.
However, if we think thoroughly, they are also necessary from the user’s
perspective. Namely, users benefit indirectly from the advertisers’ presence,
even if they might have direct negative “mind harm” from advertising. By de-
limiting ‘interaction’ to only taking place between users and content providers,
one misses the further layer of interaction between advertisers and content
providers. The logic is depicted in the following figure.
Figure 8 Interactions in an advertising-based online platform
39
This is not categorically the case; consider a user clicking an ad on the website, which is a direct
form of interaction. Further, the goal of advertising is to derive deferred interaction benefits “down the
road”; for example, when a user is next changing a car and recalls the banner ad.
User
2nd degree interaction
(primary motiveof
advertiser)
3rd degree interaction
(business logic)
Advertiser
Platform
owner
PLATFORM
(content)
1st degree interaction
(primary motiveof user)
66
Direct interaction is when advertising is shown to the user
40
. A third degree
of interaction is that which takes place between the advertiser and platform
owner
41
, termed business logic. In sum, advertising-based platforms can, ac-
cording to our reasoning, be termed platforms. However, the remark by
Luchetta (2012) is useful in that it recognizes the difference between a market,
as an entity, and a platform as a goal-driven firm. Other notable key distinc-
tions are that the platform interaction is not always exchange but that its goal
can be non-economic gains and that, in a platform triad, the user simultane-
ously might indirectly enjoy the advertiser’s presence, for enabling free con-
tent, and directly dislike advertising. This clearly fortifies the rule of no free
lunch and tends to be generally accepted by users (cf. Pauwels & Weiss 2008).
Therefore, the separating factor can be seen in the direct connection be-
tween users from different sides. In the hardware example, the suppliers have
already delivered products prior to a visit; it would be a platform if users could
see the inventory in advance and order directly from the suppliers. Ordering
from the retailer, as opposed to ordering directly from the supplier via a web-
site, does not constitute a platform but a reseller
42
. Second, the platform differs
from a standard market with intermediaries, in that a standard market advo-
cates ‘one-to-many’ structures (see e.g., Dwyer, Schurr, & Oh 1987), whereby
the market-maker is transacting with suppliers and end users separately at dif-
ferent touch points.
In contrast, in a platform, the parties interact with each other at one touch
point; that is, not with the platform but still via the platform, which is a dis-
tinction from dyadic transactions. In platforms, the major task for the platform
owner is to build liquidity; referring to the number of interactions, sometimes
this might involve creating a critical mass of participants on both sides (Evans
2009a).
This solves the conceptual problem as in the hardware example, buyers and
sellers would not be in direct interaction with one another and, therefore, the
market would not be a platform. The argument is consistent with Hagiu and
Wright's (2011, 2) definition: "[Multisided platforms] enable direct interac-
tions between the multiple customer types which are affiliated to them." As
such, platforms are not intermediaries in the traditional sense; rather, they are
places of interaction, offering facilitating service or features to their users.
40
However, consider the limitations; for example, banner blindness (Benway & Lane 1999) makes
users effectively ignore advertising, in which case the desired interaction benefit does not materialize.
This works both ways; as noted by Evans (2002), the platform is unable to tax transactions beyond the
platform (i.e., the value capture problem), no matter how large the deals made by users and
advertisers.
41
Note that ‘platform owner’ is employed interchangeably with ‘platform sponsor’ in this study.
42
In this type of interaction, the intermediary is directly interfering in transactions between users in
a way other than offering the medium.
67
3.1.6 Networks vs. platforms
How do platforms differ from networks? Structurally, a platform can be un-
derstood as a set of interlinked nodes (e.g., Westland 2010). It is only when we
segment the network users into various groups when the two-sided or multi-
sided dynamics of platforms become relevant. For example, a one-sided plat-
form is the same as any network in terms of network effects: the more there is
any type of users, the more new users are willing to join. The same does not
apply in two- or multisided platforms in which the extant users need to be of a
complementing (i.e., different) type to encourage joining (i.e., positive indirect
network effects). For example, if there are many sellers in a platform but few
buyers, adding more sellers does not increase other seller’s willingness to join;
in fact, it might be reduced by such an increase (i.e., negative direct network
effect). Therefore, while platforms, like almost any social construct, can be
modeled and understood as a network, there are particular dynamics for which
perceiving platform startups as platform startups and not network startups is
meaningful. The analysis is likely to become more useful as a consequence of
such a decision.
3.1.7 Websites vs. platforms
Another defining question is: what differentiates a platform from a regular
website? Following the earlier argumentation, in a normal website, a visitor
does not interact with other visitors. If this is so, then the websiteis a platform.
In practice, the interaction has a purpose, such as engaging with content-re-
lated activities, social interaction, or exchange. As the reader might remember,
this also clarifies why a shopping mall is a platform but a department store is
not. It also marks why research is needed; the nature of interaction taking
place in platforms is believed to be different to that in an intermediary setting.
For example, consider two firms: ActivityGifts
43
and Gidsy
44
. Both ecom-
merce sites sell experiences
45
. In both cases, the end “product” is sold by the
startup and provided by a supplier; however, who coordinates the exchange is
crucially different. In Gidsy, the buyer contacts the service provider directly,
and the website is merely a platform where anyone (i.e., users) can join and
place experiences for sale. In contrast, ActivityGifts first contracts individual
suppliers and then resells their services on the website, taking care of the
43
www.activitygifts.com
44
www.gidsy.com
45
E.g., a tandem jump in Prague.
68
customer interface (i.e., an intermediary model). This has considerable impli-
cations for the two firms; for example, ActivityGifts’ model is much more dif-
ficult to scale up than Gidsy’s model because it requires adding additional
sales personnel to contract suppliers, and customer service personnel to man-
age the customer relationships and support, whereas Gidsy’s model might re-
quire more trust from buyers, as suppliers are not verified, and leave the plat-
form owner vulnerable to direct transacting between buyers and sellers outside
the platform (i.e., the value capture problem). In sum, ActivityGifts is a re-
seller (i.e., a place of buying) and Gidsy is a platform (i.e., a place for interac-
tion that is, in this, case buying)
46
. A website, whether providing products or
not, must enable direct interaction between actors to be definable as a plat-
form.
3.2 Platform definition of this study
A few notions arise from the previous definitions that influence how platforms
are defined in this work. First, “trading partner” (J ullien 2005) is a narrow
conceptualization of the activity taking place in some platforms, especially at
the consumer side. If the basic unit is interaction
47
as opposed to exchange, it
is possible to examine two-sided interactions between users of the same or
different kind, users and platform, and users and advertisers. Instead of trading
partners, therefore, this study mostly refers to ‘users’ as actors in online plat-
forms. However, user is not a synonym for customer, and therefore the defini-
tiontwo groups of customers is different from our purpose; in practice, either
of the groups might be treated differently based on their ability or willingness
to pay (WTP).
Second, there is a problem with perceiving the growth of indirect network
effects as the growth of the number of actors in the complementing part. This
feature is inherited from the early network effects literature focusing on in-
dustry standards (Katz & Shapiro 1985), whereby the number of participants is
restricted, and from network theories such as Metcalfe’s law
48
, which focus on
the growth of an infrastructure such as the Ethernet or telephone networks.
The fallibility of size equals critical mass will be discussed in Subchapter
4.5.2.
46
However, both have the same revenue model: they make sales and retain a commission prior to
forwarding payments to suppliers.
47
Any type of activity necessitating two or more people to take place. Note that ‘interaction’, in a
similar sense, is also employed by Rochet and Tirole (2005).
48
The value of a network increases proportionally ton
2
, whenn is the number of individual inter-
connected nodes.
69
Based on the previous discussion, platforms in this study are defined as
follows:
A platform is a place of interaction among one or many groups
of users whose interaction benefits, known as network effects, the
platform owner aims to increase and monetize.
The ‘place’ attribute is similar to Evans' (2009b) definition of platforms as
physical or virtual meeting places for two distinct groups. The place can be a
marketplace, in which case the interaction is in a form of exchange between
buyers and sellers, or it can be a content platform in which users create content
such as discussions or video. Furthermore, it can be a social network where
users engage in social interaction. What makes Internet platforms interesting,
from both practical and theoretical perspectives, is their immense potential for
scaling (as in: increase in size). For example, Facebook grew from zero to one
billion users in eight years (Shaughnessy 2012). Clearly, this potential opens
new types of business opportunity. Startups are among the first to experiment
with these opportunities in platform markets. Their role is either to create new
platforms or join existing ones as a complement; both strategies are considered
in this study.
Monetization in this definition is similar to “internalizing externalities”
(Evans 2002; Rochet & Tirole 2003), in that the platform extracts rents for the
coordination benefits it provides for its members. As noted, delimiting the
‘two groups’ required for platforms to buyers and sellers (e.g., Li et al. 2010)
is not appropriate in the online context because other divisions are equally rel-
evant, such as, for example, free users and paid users, users and advertisers,
and contributors and consumers of content. Buyers and sellers are associated
with exchange platforms in this study, and other groups under their proprietary
platforms. Therefore, the scope of platforms extends beyond markets, and the
definition by Hagiu and Wright (2011) is the most influential for the definition
of a platform in this study. Note, however, that this does not render the two-
sidedmarket literature obsolete; most of it applies even when the interaction in
the market is other than economic exchange.
Critical to this notion, in the context of online platforms, is that match-
making is performed through programmatic means; that is by algorithms
matching complementing parties (e.g., buyers and sellers; men and women)
based on the criteria and other information they have willingly given, or in-
formation that is retrieved from their behavior or other context to facilitate
match-making. These actions of the platform owner aim to increase the num-
ber of matches and subsequent interaction; outcomes that are associated with
the viability of the platform, both in terms of liquidity and monetary gains.
70
3.3 Typology for online platforms
This section presents the types of platform examined in this study, and relates
them to other platform types studied in the literature. Classification of plat-
form types can be employed to understand their strategic problems. In partic-
ular, different complements and motives of participation are associated with
different platforms. Because complements are associated with the strength of
network effects (Parker & Van Alstyne 2010), and motives with the use of a
platform (Boudreau & Lakhani 2012), recognizing that these vary by the type
of platform might lead to important discoveries for their strategic manage-
ment.
Platform classifications relevant to this study can be divided into a hierar-
chy of three domains:
General platforms ? Technology-enabled platforms ? Online platforms
These domains are now explored in more detail. First, Evans (2003) identi-
fies three types of platform: 1) market makers, 2) audience makers, and 3) de-
mand coordinators. Market makers create an environment for economic ex-
change in which parties are involved in transactions with each other. Market
makers reduce transaction costs relating to searching for and negotiating with
trading partners (Evans 2003). Second, audience makers create matches be-
tween advertisers and end users to enable advertisers to send messages to their
desired target audience that comprises users of the platform. The platform
owner needs to determine how much advertising is allowed in the platform,
especially when it interrupts the consumers’ usage (Rysman 2009). Third, de-
mand coordinators enable members to interact by providing services in the
background, such as operating systems and payment cards (Evans 2003).
Gawer (2009) divides platforms into 1) internal platforms, addressing one
firm’s offerings (e.g., Sony Walkman); 2) supply-chain platforms, addressing
chain-wide execution (e.g., the Renault-Nissan alliance); and 3) industry plat-
forms where the focus is on industry-level coordination (e.g., Microsoft Win-
dows). A similar typology is presented by Piezunka (2011) who distinguishes
that streams of the literature tend to focus on 1) product platforms; 2) industry
platforms; and 3) two-sided markets. The difference is marked by the role of
the platform sponsor: in the first, it offers products; in the second, it coordi-
nates complements and end-users who might directly interact outside the plat-
form; and in the third, it coordinates two sides interacting in a platform.
One of the most cited general platform classifications is that by
Schmalensee and Evans (2007). It includes 1) exchanges, 2) advertiser-sup-
ported platforms, 3) transaction platforms; and 4) software platforms.
71
Exchanges include coordinating dyadic interactions between buyers and
sellers; any stock exchange would qualify. On the Web, for example, auction
websites such as eBay are included. Examples of advertiser-supported
platforms include TV or radio stations that show content for free and monetize
by selling advertising space. Online equivalents are, for example, search
engines such as Google (see Salminen 2010). Transaction platforms, such as
credit card providers, mediate transactions between merchants and consumers;
PayPal is a good example on the Internet. Software platforms are tied to
specific hardware, and sometimes referred to with the concept of ecosystem;
for example, mobile application platforms such as Nokia’s Ovi Store. Huang,
Ceccagnoli, Forman, and Wu (2009) define ecosystems as communities of
innovation networks in which industry leaders coordinate collective efforts of
developers and other partners towards shared goals.
Second, technology-enabled platforms mentioned by Saha, Mantena, and
Tilson (2012) include computer and mobile operating systems, online advertis-
ing networks, job boards, real estate brokers, electronic marketplaces, and
payment systems, both mobile and online, such as Paypal. Technology plays a
role in facilitating connections between supply and demand sides, and also
computes optimal routes and allocations between parties, a property that is
useful when mapping potential connections. Internet platforms are built on top
of Web technologies that are typically based on open standards, thus excluding
competitive strategies based on patents and standards, and enabling social
connections and scaling effects associated with technology (for illustration,
see Horowitz & Kamvar 2010).
Third, online platforms are a subset of technological platforms; they can
also be termed Internet platforms (Sawhney et al. 2005) due to the fact that the
Internet is the medium through which participants interact. Consequently,
Web technologies play a major role in how match-making is executed by the
platform owner, and also in how the platform scales; for example, very rapidly
extending across national boundaries. Some authors also employ the term
electronic intermediary (Bakos 1998). However, intermediation refers to a
value chain deviating from the market-making function; for example, enabling
parties to independently find one another (Evans 2003). Gazé and Vaubourg
(2011) distinguish online auctions, traveling intermediaries, online media,
massive multiplayer online role-playing games, and e-payment platforms.
Saha et al. (2012) mention the following online platforms: 1) electronic la-
bor markets, 2) ecommerce sites, 3) online advertising platforms, 4) online
auctions, and 5) group-buying platforms. Caillaud and J ullien (2003) point out
that online platforms are able to monitor individual transactions, and therefore
charge transaction fees tied to the number of interactions, not only access fees.
Gazé and Vaubourg (2011) discuss another trait they perceive as typical for
72
online marketplaces: side-switching, which is changing roles from buyer to
seller as it is easy, reversible, and has no financial cost. However, it is unclear
if this feature only applies to online markets as, also in brick-and-mortar cases,
one can act as a seller and buyer within the same market space (consider e.g., a
flea market).
Conceptual classifications can be regarded as arbitrary
49
because typologies
can employ alternative criteria while classifying the same phenomenon; nei-
ther being wrong nor correct (Kotha & Vadlamani 1995). For example,
Schmalensee and Evans (2007) employ criteria such as the form of product
(i.e., software platform), business model (i.e., advertiser-supported platform),
and the operating level (i.e., transaction platform, which can be regarded as a
form of infrastructure). Thus, we can posit that it is difficult to define plat-
forms in a mutually exclusive way as they might be embedded, so that, for
example, a software platform provides advertiser-supported products.
There are some reasons why the extant classifications are insufficient in the
context of this study. First, although we can detach the embedded platform
from its parent and examine it in isolation, depending on whether we are inter-
ested in the infrastructure level, business model, or type of interactions taking
place within it, based on particularities in these dimensions, it can be argued
that online platforms merit their own classification. For example, software
platforms (Schmalensee & Evans 2007) would combine hardware and soft-
ware, which is not relevant in pure online business
50
. This liaison arises from
the hardware-software paradigm introduced by Katz & Shapiro (1985), and
relates to the complementarity of the two; hardware being more valuable ac-
companied by useful software, whereas, clearly, software cannot be run with-
out hardware. The result is a different kind of chicken-and-egg problem than
the one analyzed in this study, and is often implied when explaining computer
industry dynamics
51
(e.g., Boudreau & Hagiu 2009).
Second, many classifications employ a revenue model as the defining fac-
tor. However, the revenue model only explicates how the platform generates
revenue (e.g., by advertising); this information is not very relevant for solving
the chicken-and-egg problem. Much more central, in this respect, is examining
why users join and participate in platform interaction, and even pay for access
49
In the sense that, although consistent mutually exclusive items, classifications can be equally
valid but different.
50
However, as previously noted, similar dynamics to other two-sided markets can be assumed;
thus, the platform literature is highly relevant. Furthermore, some strategies applied in the mobile
application markets by key players are associated with the fact they are also hardware vendors.
Therefore, even if the application marketplaces can be modeled in isolation as their own two-sided
markets, in some cases, theories might want to consider the hardware liaison.
51
Clearly, the platform perspective would explain the difference between rivals through the
concept of network effects: one is able to leverage them while the other is not.
73
and usage. For this purpose, the classification presented next considers poten-
tial motives of users of online platforms.
In an attempt to provide a specific classification for online platforms, com-
bining both the digital environment and the two-sided structure, the following
classification is proposed. Overall, it is based on the grounded theory analysis
and thus on the nature of the startups examined in this study.
Table 7 Online platform types
Type Sides Focus of interaction Industry example
Exchange platform Buyers and sellers Exchange motive / transac-
tions
eBay
Content platform Creators and con-
sumers of content
Content creation and con-
sumption
Google
Social platform (one-sided) Social motive Facebook
Infrastructure Providers and devel-
opers
Enable other products and
services
The Internet
This classification matches well with the nature of startups sampled in this
study (see Table 2). Note that the infrastructure platform is a special case
which involves no interaction between parties. Infrastructure is compatible
with Gawer and Henderson’s (2007) definition
52
of a platform as a structure to
build on top of, but not with Hagiu and Wright’s (2011). In this study, interac-
tion between parties involved in the platform is regarded as more important
than hardware-level interaction, and thus the infrastructure model will not be
considered in later parts of this study.
The exchange platform connecting buyers and sellers is the most widely
documented case in the platform literature, and most authors refer to it when
discussing platforms. Based on the literature review, especially in the field of
economics, the focus is on marketplaces (i.e., exchange platforms). Some au-
thors see a platform in the sense of infrastructure (i.e., enabling to be built
upon), which is not the most fruitful approach when considering how the plat-
form owner can solve business problems as a strategic agent.
There can be some overlap between the types. For example, Kim and Tse
(2011) analyze knowledge-sharing platforms, which would classify either as
52
“We define a product as a ‘platform’ when it is one component or subsystem of an evolving
technological system, when it is strongly functionally interdependent with most of the other
components of this system, and when end user demand is for the overall system, so that there is no
demand for components when they are isolated from the overall system” (Gawer & Henderson 2007,
54).
74
content or exchange platforms depending on the type of interaction. If re-
spondents are provided payments, not “payments in-kind” as in Mungamuru
and Weis (2008), the interaction is exchange, and therefore the “laws of ex-
change” should apply. If, however, interaction is voluntary and driven by in-
trinsic motivation such as the status of knowing a lot, it is a content platform
and visitors are interested in receiving content benefits. Furthermore, if there
is a relatively stable community and users engage in lengthy discussions and
roles, the platform can be classified as a social platform (Mital & Sarkar
2011).
The overlap problem can be solved by splitting the user motives into pri-
mary and secondary motives. For example, a person searching the Web to find
information has a primary ‘search intent’ (cf. Schlosser, White, & Lloyd
2006), even though he/she might end up sharing the results of the search in a
social platform. Therefore, even when they relate to content, primary motiva-
tions to visit a social platform can be of a social kind. For example, sharing
content is arguably more about sharing (i.e., ‘social intent’) and less about
content. As the purpose of this platform is to provide content, most people
visit it because of that. However, some people might visit it for the primary
purpose of sharing, or some other motive for social interaction. Thus, in our
typology, exchanging content is a spillover effect, not the primary motive for
participating in the platform.
This discussion is not merely semantic. In some cases, the secondary mo-
tive can become an even more powerful predictor of conversion than the ac-
tual motive of visiting the platform. For example, Oestreicher-Singer and
Zalmanson (2009, 39) observe that “in the context of music content, commu-
nity activity is more strongly associated with the likelihood of subscription
than is the music consumption itself.” Furthermore, the interplay between
content and social features can be a critical part in finding solutions to thecold
start dilemma through spillover effects (see Chapter 4.4) and remora’s curse
(Chapter 4.7)
53
. If the motive for participation is known, the platform owner
can provide appropriate incentives, including implicit incentives (e.g., social
satisfaction) and explicit incentives (e.g., monetary compensation), or a com-
bination of both (Ren, Park, & van der Schaar 2011). A place of exchange can
be regarded as being subject to different rules and “economic laws” (i.e., type
of reciprocity associated with interaction
54
) than a content platform.
Therefore, motives for interaction differ in content platforms and social
platforms. Different motivations are also applied in Luchetta's (2012) typology
53
Throughenvelopment (Eisenmann et al., 2011) or embedded platforms, in which a platform of a
different type is built on top of the host to gain access to its user base.
54
For example, Porter (2004) argues that “[r]elationships in networked-based communities are
often of short duration and driven by utilitarian needs.”
75
of platforms. By adopting his style of presentation, online platforms can be
specified as follows.
Table 8 Online platforms, interaction, and goals
Platform Interaction User A Goal of A User B Goal of B
Content Consumption Consumer To consume
content
Contributor To contribute
content
Social Communication Individual A To connect
with B
Individual B To connect
with A
Exchange Transaction Buyer To buy Seller To sell
It follows that the purpose for interaction is symmetric for social platform
users, and the resulting problems are coordinating problems (e.g., who is
online) that the platform will efficiently solve (e.g., by storing messages). The
same applies to an exchange platform; the needs are symmetric, although in-
versely, so that they complement each other. However, participants in a plat-
form relying on user-generated content face some asymmetry; there are users
interested in consuming content and others who produce it. This suggests
some dynamics that will be considered in Chapter 4.4; namely, the goals of the
two sides cannot always be solved by simple match-making.
In sum, exchange platforms address buyers and sellers whose primary in-
teraction is a transaction (i.e., trade). Content platforms comprise users con-
tributing content and users consuming it
55
, and content-related activities (e.g.,
consumption, reading, writing, and watching) are the type of interactions for
which the platform exists. Social interaction (i.e., discussions, chats, messag-
ing, and communication) defines social platforms and their users. A social
platform is one-sided when the user is only interested in others sharing his/her
traits or interests, whereas, in a dual-sided social platform, the user seeks a
complementing party.
55
Whether they are seen as one heterogeneous group or two distinct groups (i.e. one- or two-sided
platform) is another arbitrary choice enabled by flexibility of the two-sided framework.
76
3.4 Online platforms and user generation
3.4.1 Why is UG included in the study?
In this chapter, the author outlines an ideal user generation model to capture
the potential benefits of user generation (UG) targeted by startup founders.
Wishful UG effects are characteristic to the startups in the sample, and influ-
ence the emergence of strategic problems and their solutions. User generation
in the theoretical treatment enhances the outcome more than omitting this
critical aspect, which is why UG is considered a part of the startups’ business
logic.
The lack of desired behavior from users is, based on analyzing the post-
mortem stories, associated with the failure outcome, and therefore needs to be
considered in this substantive theory of strategic problems. The online plat-
form typology and the ideal UG model are Internet-based specificities, and
considering them deepens the perspective on platform strategies in the online
business context.
3.4.2 User-generated content
In the literature, UG effects are often approached through the concept of user-
generated content (UGC), also commonly utilized by startup founders. A
widely accepted definition (e.g., Hermida & Thurman 2008; Ochoa & Duval
2008; Banks & Deuze 2009) can be found in Wunsch-Vincent and Vickery
(2007), defining UGC as a) content made available through the Internet, b)
involving creativity by end-users of a Web service, and c) non-commercial
motives of creation.
Earlier papers relating two-sided markets/platforms with UGC include
Albuquerque, Pavlidis, Chatow, Chen, and J amal (2012), who compare firm-
initiated promotion with user-initiated promotional activities, and Evans
(2009b), Ren et al. (2011), and Calvano and J ullien (2012) who consider UGC
as a means to connect consumers and advertisers. Yoo (2010) also mentions
UGC in his treatise on network providers and two-sided markets, but only to
differentiate it from peer-to-peer networks. Kim and Tse (2011, 42) give an
example of UG effects in a platform context:
"[A] knowledge-sharing platform may overcome an initially
small membership by starting with some prebuilt knowledge da-
tabase to attract new questioner members. New questions posted
by these questioners will attract some answerers, whose answers
will in turn attract more questioners. This cyclic process can
77
help a knowledge-sharing platform to overcome the chicken-
and-egg problem that is commonly found in a general two-sided
market."
However, none of these papers consider the theoretical implications of UG
in online platforms, or fully extend its meaning beyond content-related activi-
ties. First, considering content only as content, as opposed to a complement,
does not capture its implications for online startups. Following Chapter 3.1,
apart from installed user base, complements are associated with network ef-
fects; thus, the more content in a content platform, the more valuable it is to its
users, all else being equal (cf. Varian 2003). At a general level, the same ap-
plies to interaction: its volume positively correlates with the usefulness of a
platform (see Chapter 4.5).
Second, the functions of UG are not limited to user acquisition as is typi-
cally considered (e.g., Kim & Tse 2011). In contrast, the users are seen to
adopt more roles that influence the viability, growth, and success of a plat-
form. These functions are discussed in later sections. Third, even if Web 2.0
marked a “revolutionary” disruption of the earlier “version” of the Web (see
O’Reilly 2005), the notion of UG has not developed much since. For example,
Beuscart and Mellet (2009) note that Web 2.0 is associated with UGC sites,
blogs, social news sites, and social network sites. UG, therefore, is commonly
perceived as “video and photo sharing” or similar activities, the meaning of
which to the platform is not regarded as critical. However, the importance of
UG seems obvious when observing the scale and diffusion of online platforms
in the real world. For example, consider Facebook, originating from a small
community and, a few years later, having over one billion registered users
(Shaughnessy 2012)
56
. Such results were obtained without advertising, very
minimal user support and staff-per-user ratio (Kirkpatrick 2010), instead rely-
ing on UG effects.
3.4.3 UG in online platforms
When, therefore, considering UG not from a static perspective but as a com-
plement, a new definition is needed. As network effects are highly important
for two-sided markets, a definition for UG relating to platforms is proposed:
User generation is provision of content or other direct or indi-
rect benefit through actions of a user of an online platform to
56
Edwards (2013) notes, correctly, that a portion of this includes fake profiles and double
registrations; nevertheless, the growth is phenomenal.
78
others users characterized by the platform owner’s attempt to
monetize it.
This definition is compatible with the two-sided framework in the following
ways. First, it states that users provide benefit for other users, thus implying
network effects (e.g., Katz & Shapiro 1985). Second, the benefit can be either
direct or indirect (Shy 2011), based on the type of interaction (e.g., communi-
cation vs. trade) and the mechanism of benefit provision; for example, peer-
marketing can increase the installed user base, which indirectly benefits other
users. It can be seen that members of the same sub-group or dyad of a platform
benefit from direct interaction, whereas participants included in another group,
such as advertisers, the platform owner, and other potential stakeholders,
might derive indirect benefit from the existence of the user base (cf. Clements
2004). Third, it includes the aspect of the platform owner aiming to monetize
the UG outcomes, which is compatible with Rochet and Tirole’s (2005) and
Evans’ (2003) perceptions on a commercial platform and the “internalizing of
externalities”.
After establishing this definition, the totality of UG effects, which are no
longer limited to content, is explored. In this conceptual inquiry, suitable theo-
retical frameworks are applied, reaching beyond the immediate platform liter-
ature. This is necessary to base our theoretical argument on the appropriate
literature. As noted, the platform literature is lacking in this aspect, as it does
not consider UG effects with the same gravity that, as the analysis revealed, is
shown by platform startups. Based on the grounded theory (GT) analysis,
founders implicitly assume these UG benefits in their platform strategy.
3.4.4 Ideal user-generation model
User generation was an essential concept for the studied online platforms, and
its emergence can be employed to explain inadequate strategic responses by
founders. To understand the logic applied by founders, we generate a compre-
hensible model going beyond what is understood in prior platform research as
thepotential benefits of UG (i.e., UG effects).
‘Ideal’ UG refers to properties associated with UG in the online context.
Namely, it is the optimal model at which startups aim, but which most will not
reach, as proven by the sample. Central to this idea is that UG aims to replace
functions typically assumed by the firm, and delegates them to the user base.
While the full treatise of this notion goes beyond the scope of this study, the
79
user base is, in the ideal sense, utilized to extend the “boundaries of the firm”
57
(Coase 1937). Namely, startups applying this implicit and theoretical model
aim at 1) minimizing human intervention and labor cost through designing a
technological solution (i.e., the platform); 2) facilitating the creation of
socially desirable high-quality content, or activating a virtuous cycle of
network effects; 3) scaling their technology and business beyond the limits of
the startup’s internal resources with the help of platform users; and 4), above
all, delegating critical business functions to the user base.
The following figure depicts the ideal model of UG effects.
Figure 9 Ideal user generation model
The ideal UG model is associated with network effects, so that the content
actually forms a complement in the platform, making it more valuable to all
users (Arroyo-Barrigüete, Ernst, López-Sánchez, & Orero-Giménez 2010). In
other words, the first set of users’ actions is expected to initiate the virtuous
cycle. More precisely, in the ideal UG model, it is implied that the user inter-
nalizes some of the critical tasks of the startup that, in turn, by externalizing
them, will reach better operational efficiency. The user base is, in theory, able
to produce effects several magnitudes larger than the startup’s resources would
enable; thus, it can be regarded as the startup’s resource. This can be proven
by assuming that the cost to the platform owner to create content, marketing,
customer service, and other functional activities necessary to launch and
maintain the platform increases with, for example, labor costs and marketing
investments, and thus has a realistic limit (i.e., budget), whereas similar
57
It goes beyond the study’s scope as the study cannot prove that firms/startups systematically
delegate their critical functions to users, which is why the model is ideal or theoretical. However,
startups that apply state-of-the-art methods to survive seem to indicate this possibility. This is
definitely a topic for further research.
UGC
content
e-WOM
new users
P2P support
support
Social indicators
conversions/ revenue
USER
GENERATION
80
activities without transaction cost due to platform coordination, self-
organization, or labor cost are effectively cost-free for the platform owner.
Therefore, harnessing UG and network effects leads to a virtuous cycle in
which the willingness to adopt increases as a function of earlier accepted in-
vitations, which then increases the willingness, for example, to send more in-
vitations (Trusov, Bucklin, & Pauwels 2008). This is compatible with network
effects as defined by Farrell and Klemperer (2007, 2007): “there are network
effects if one agent’s adoption of a good (a) benefits other adopters of the
good (i.e., a ‘total effect’) and (b) increases others’ incentives to adopt it (i.e.,
a ‘marginal effect’).” Through UGC, the startup aims to increase the ‘total
effect’, while user-generated user acquisition is planned to increase the ‘mar-
ginal effect’. Given that the diffusion/propagation has an upper boundary on
the limits of a market (Salminen & Hytönen 2012), the outcome can become a
winner-takes-all situation (Noe & Parker 2005) in which the platform replaces
its rivals as a side-effect of expansion (cf. Facebook replacing MySpace). This
effect is noted by Arroyo-Barrigüete et al. (2010, 643), who describe it as a
“re-alimentation schema that makes strong products ever stronger (virtuous
circle) and weak products ever weaker (vicious circle).” This exponential in-
flationary trait can be explained by the "small world" characteristics of the
Internet that hosts online platforms (Schnettler 2009), and the desire to propa-
gate messages, such as invitations and content. Zhang and Zhu (2011) find that
social effects lead to an increase in contribution when the installed base of
Wikipedia increases, therefore supporting UG’s virtuous cycle.
The ideal UG model is grounded in the data; namely, in the assumptions of
the founders. They implicitly and explicitly devise their businesses to support
the idea of UG as this is perceived to be the interaction taking place in the
platform. Indeed, the concepts of interaction and UG are closely associated
and refer to the same phenomenon, which is the activity taking place in the
platform. Note that the ideal UG model differs greatly from what is
understood in the literature by “user-generated content”. Namely, the literature
examines UG almost proprietarily as a function of content creation, whereas
this study’s definition, derived from the two-sided platform literature, extends
far beyond mere content.
3.4.5 Functional view to UG
To provide a more granular perspective to the idea, general functions of a firm
are related to UG. The following list parallels an organizational structure (i.e.,
the firm) in which functions mostly have a clear purpose (Mathur 1979). The
analogy becomes even more distinct when presenting a functional comparison.
81
Table 9 Functional comparison of users and the firm
User function Firm function
Content creation Content production
Moderation Quality control
User acquisition Marketing
Support Customer service
Feedback Market research
The paralleled functions are general functions in which many firms, in-
cluding platforms, need to engage. A platform needs to provide content or li-
quidity (Evans 2009a); it needs to moderate the quality of UGC or other com-
plements for spam (Moh & Murmann 2010), or low-quality complementors
(Boudreau & Hagiu 2009); it needs to provide support/customer service for its
users/customers (Rochet & Tirole 2003); it should take user requests into con-
sideration when modifying platform design (Stanoevska-Slabeva 2002) and,
thus, spontaneous feedback by the community or platform users is a form of
market research; and, most importantly, it needs to acquire users/customers,
that is, conduct marketing (Eisenmann, Parker, & Van Alstyne 2006). These
are general functions that need to be organized in one way or another; accord-
ing to the UG model, users are given most if not all of these tasks.
As explicated, users obtain control over content generation that enables not
only cost-free production, from the startup’s perspective, but a quality that
matches the audience’s tastes
58
. To ensure quality, users monitor each other
and report negative behavior. By automating the system to respond to user re-
ports, the startup avoids any labor relating to quality control
59
. Moreover, in a
platform demonstrating community traits, users face social disproval from
other users as a consequence of misconduct, which they are therefore likely to
avoid (Sheridan 2011). In exchange, users help each other to learn the plat-
form, and also mediate commonly agreed rules of behavior. Support can arise
from earlier platform-specific investments by other users who behave in an
altruistic way (Boudreau & Lakhani 2012).
Arroyo-Barrigüete et al. (2010, 644) refer to the “learning network effect”,
which “derives from the fact that an increase in network size will increase the
number of users with specific knowledge of the related technology.” Essen-
tially, expert users provide a form of “after-sales service” to new users,
58
Kim and Tse (2011, 41) describe this effect: “[a]s a result of knowledge sharing between
members, knowledge-sharing Web sites have an accumulated knowledge database of answered
questions that attracts people who have questions.”
59
In its simplest form, there are report functions. It is more complex when the user becomes a
moderator; that is, an active agent who scouts the platform for low-quality interaction.
82
thereby increasing the platform diffusion. According to this argument, the
presence of peer support can facilitate bothex ante adoption andex post inter-
action.
User-generated invitations, a form of peer marketing (see e.g., Smith,
Menon, & Sivakumar 2005) are, in fact, also beneficial for other users, given
they match their preferences; if not, they will be interpreted as unsolicited
messages (i.e., spam). This is because they reduce the recipients search cost
for interesting content. Therefore, sharing links of content among peers is an
efficient dissemination mechanism and, in theory, resolves the need for any
other marketing
60
, which will be revisited in Chapter 4. It is widely acknowl-
edged that peer-to-peer propagation plays a critical role in the diffusion of
most online platforms that are currently dominant, and that this effect relates
to UG (Albuquerque et al. 2012, 406):
"[C]ontent creators, besides populating the platform with mate-
rials, serve as marketing agents by advertising their own content
or generating referrals and links to uploaded content in other
websites. Given the interconnectedness and viral community
structure of the Internet, the relation between marketing activi-
ties by the firm and the decisions of content creators is likely to
play an essential role in the development of most user-generated
content platforms."
Moreover, consider the customer acquisition function that, as per the ideal
UG model, relies at least partly on search engines providing a marketing chan-
nel by automatically indexing content, and thereby providing free organic traf-
fic (i.e., visitors) to the website, a process here termedsearch-engine external-
ity
61
. Host platforms can provide a stream of users; that is, act as marketing
channels based on actions of the startup and in respect to the platform type
(e.g., content platforms ? Google traffic; social platforms ? Facebook traf-
fic). This strategy is similar to envelopment (Eisenmann et al., 2011), and its
merits and risks are discussed in Chapter 4.7.
60
Interaction that provides direct network value is described by Oestreicher-Singer and Zalmanson
(2009, 14): “small acts of structured contribution that can be perceived as adding value to the user’s
own content consumption but that can also add value to the community […] for example, tagging
content with keywords to ease its discovery, or rating content in order to promote its popularity and
reputation.”
61
This is an externality as it is not a reason for interaction between content consumers and content
creators. For the startup, it represents an externality that can be internalized.
83
3.4.6 Implications to startups
What does the ideal model imply for a platform startup? First, the immediate
costs of providing the platform are radically reduced. Second, effective emer-
gence of UG implies self-organization, similar to projects such as Wikipedia
(see Stvilia, Twidale, Smith, & Gasser 2008) and Linux (Benkler 2002). Third,
transaction costs relating to coordination of the business are externalized,
meaning that the users will begin to coordinatefor the firm (Hagiu 2006).
In particular, UG can reduce the cost of scaling; namely, increasing the
connections and activity in a platform. Imagine the platform owner’s cost
structure comprising user acquisition, denoted a, platform maintenance, de-
notedm, and user support (i.e., customer service), denoteds. To grow the user
base, the platform owner incurs the cost of a multiplied by each acquired user.
If the user base grows by a factor of k, the support cost also grows by this
factor. However, if the user acquisition and support functions are performed
by the current user base at their own expense (i.e., time), only the fixed
62
cost
m remains for the platform owner. By applying UG, the platform owner is able
to maintain itsneutrality, thus avoiding costly marketing and support activities
while monetizing the increased usage of the platform driven by UG’s expo-
nential dynamics and network effects. For example, a content platform exists
to produce and disseminate content to other users. In a regular website, the
owner creates the content; in a platform, it originates from the users. The
startup as a platform owner is not a ‘side’ of the market but still benefits from
the content, as it helps to attract more users on which the platform owner can
capitalize (i.e., monetize).
As previously described, self-moderating effects reduce a startup’s work-
load when scaling to millions of content units. By reducing the transaction
costs of its members, a platform is able to attract users (Hagiu 2006). This fact
has been established in the platform literature. However, less attention have
been paid to other costs. In particular, introducing UG reveals its importance
for online platforms, but also makes it easier to understand the founders’ logic
in pursuing UG in their platform strategies. Hagiu (2006, 2) claims that “any
MSP [multisided platform] performs one or both among two fundamental
functions: reducing search costs and reducing shared transaction costs among
its multiple sides.” By combining parties of interaction, the platform is able to
lower the cost of finding matches, negotiating, and validating their quality. In
a UG platform, these activities originate from the user base. Search costs for
users are lowered because they are able to find new interesting content or
62
In this example, platform maintenance cost is assumed to be fixed; for example, startups
applying cloud-based hosting can convert their server costs from fixed to variable.
84
connections, and the benefit is sustainable because the platform keeps auto-up-
dating as a consequence of other users’ actions. Further, the platform does not
need to invest in advertising because users promote the service to other users.
Finally, startups can integrate mechanisms into their platform design to
nurture UG activities, such as 1) community building, 2) viral mechanisms, 3)
search-engine externalities, and 4) frictionless sharing (Darwell 2013). If these
efforts are either built-in to the product or come at minimal cost by inviting
friends and sending emails to gather the critical mass, then it is assumed that
marketing is free and the customer acquisition cost equals zero. These kinds of
economy makes it possible to replace paid match-making services. Moreover,
indirect monetization might be a requisite to 1) maintain social norms, not
economic norms (Fehr, Kirchler, Weichbold, & Gächter 1998), and 2) encour-
age reciprocity; as users get a free platform, they might feel the need to con-
tribute or “pay” through UG. Susarla, Oh, and Tan (2012) studied diffusion in
a UGC platform (i.e., YouTube) and concluded that diffusion is influenced by
social contagion rather than user heterogeneity. Clearly, user-to-user dynamics
play a role in the adoption/diffusion of a platform, and therefore should be
considered in relation to the chicken-and-egg problem.
3.4.7 Limitations of UG
In sum, within the online platform, users are expected to engage in multiple
roles such as content creation, moderating (i.e., quality control), customer ac-
quisition (i.e., inviting other users), and providing support (i.e., customer ser-
vice). As a result, the required input from the platform owner is, in theory,
greatly reduced through coordination features programmed into the platform.
Indeed, in an ideal situation, the platform becomes self-sustaining and self-
propagating, while the platform owner is still able to internalize some of the
benefits from user interaction, typically indirectly (e.g., by selling data or ad-
vertising space), whether the interaction is content creation and consumption,
exchange, or social interaction.
Clearly, this is where the problems begin, as the ideal UG model often re-
mains just that – ideal. The rest of this study, especially the section discussing
dilemmas, demonstrates some of the central challenges in applying UG as a
critical part of a platform startup’s business model. The ideal model rarely, or
almost never, materializes
63
, and in practice the platform owner’s intervention
63
The examples provided earlier suffer from survivorship bias, which means that they are not
adequate for generalizing empirical meaning. However, nothing prevents taking them as part of a
theoretical model, given that the limitations are properly understood.
85
is needed in most parts to guarantee progress and frictionless operation. How-
ever, the few examples, in addition to assumptions of transaction-costless
functioning, of the ideal platform are compelling thoughts. Therefore, alt-
hough it might seem counterintuitive with hindsight, ignorance concerning
marketing
64
as exhibited by several founders in the sample, and reported after-
wards
65
, has logic; had users accepted those critical functions, the startup
might have succeeded. The ideas that the users replace marketing and that
search costs are categorically low are debunked in Chapter 4.5 and Subchapter
4.6.2 respectively; “no need for marketing” is challenged in Chapter 5; and the
negative effects of removing quality control from the platform owner are
discussed in Subchapter 4.5.2. Therefore, despite its theoretical merits, the
ideal UG becomes dangerous when applied literally.
Overall, the ideal model leaves many aspects uncovered and relies on unre-
alistic assumptions of users’ willingness and ability to manage critical func-
tions of the firm. In its pure form, the winner-takes-all outcome can also be
unrealistic, as it fails to consider multihoming behavior and other aspects of
adoption apart from network effects, such as differentiation through features
or marketing. Nevertheless, it helps us to understand how platform startups ap-
proach strategic thinking. Understanding UG is important as it is associated
with startups’ strategic decision-making. For example, consider Facebook,
Twitter, and Google that have accumulated users not by paying but by peer
effects. These success stories formreference points that orientate founders
towards choosing similar elements in their own startup
66
. This effect is close
to an anchoring bias (Bunn 1975) according to which decisions are made
based on prominence and not averages or suitability of the case context.
Finally, some UGC-based companies (e.g., Twitter) are struggling with
monetization, which indicates that even a working UG model might not be
sufficient to guarantee economic sustainability. As such, popularity does not
indicate profitability (see Chapter 4.6), and solving the chicken-and-egg prob-
lem through UG might leave a firm vulnerable to other strategic dilemmas.
With this premise in mind, the study will now move to discuss the strategic
problems of platform startups in more detail.
64
“The next issue to tackle was marketing. How do we make them aware of it? We decided to use
blogs. What better way to expand than to piggyback on an existing network? […] It rarely works.
Everyone wants to do it, but it isn’t easy to get bloggers to write about something.” (May 2007).
65
“We fell into the ‘build it and they will come’ school of thought (although even when they came,
we still weren’t in good shape).” (The Chubby Team 2010).
66
“When it came to certain website design elements, we didn’t know what customers wanted […]
and so instead, we thought ‘let’s take elements from sites we like and tweak them’ and we’ll get the
same magical effects on our site that they’ve gotten. Wrong. Features don’t work in a vacuum. They
work because you take time to understand your customer and then build features to accommodate
them.” (The Chubby Team 2010).
87
4 STARTUP DILEMMAS
4.1 Introduction to dilemmas
4.1.1 What is meant by dilemmas?
A ‘dilemma’ is a situation of conflict, in which a decision maker usually faces
two mutually exclusive choices that both lead to a seemingly undesirable out-
come. In the Oxford Dictionary (2013), dilemma is defined as “a situation in
which a difficult choice has to be made between two or more alternatives, es-
pecially ones that are equally undesirable.” Although, in everyday life, indi-
viduals often face contradictory decision-making situations, researchers in ac-
ademia tend to model decisions through preferences and weights; thus, out-
comes that are perceived more costly are avoided while those with higher ex-
pected gains will be sought (e.g., Layard, Layard, & Glaister 1994). In
psychology, one speaks of cognitive dissonance, a state of contradictory emo-
tions relating to situations, persons, or outcomes (Festinger 1962).
In economics, scholars examine various tradeoff situations (e.g., Ball et al.
1988; Cohen & Klepper 1992). In particular, economists apply game theory to
examine actors’ strategic choices under a set of assumptions; for example, the
prisoner’s dilemma (see Axelrod 2006) is a famous game-theoretic problem
that is structurally close to the cold start dilemma, presented in Chapter 4.4, in
that participants are driven to a dissatisfactory solution. In the strategic man-
agement literature, strategic or wicked problems (Mason & Mitroff 1981) are
characterized by associations with other problems, recursive feedback, envi-
ronmental uncertainty, ambiguity in definition, conflicting tradeoff in their
solutions, and societal constraints upon theoretically effective solutions (Lyles
& Howard 1988). Proper definition of a problem and assessment of the strate-
gic situation are regarded as important for finding a solution (Klein 2012), alt-
hough individuals are perceived to be constrained by their cognitive capabili-
ties andbounded rationality (Simon 1956).
4.1.2 The use of dilemmas in this study
Here, the concept of dilemma will be applied to examine various challenges to
platform startups. More precisely, four startup dilemmas are conceptualized in
88
this study that derive from the material through the inductive grounded theory
approach, and formed through discussions with founders (i.e., contextualiza-
tion), and the support from the theoretical framework (see Chapter 2). There-
fore, the origin of these dilemmas is inductive while their conceptualization
and treatment follow a deductive process, based on the aforementioned con-
textualization and support from the literature.
A typical dilemma for online companies is the monetization of their offer-
ings. The assumption by founders can be termed expectation of free; more
precisely, they believe consumers prefer free Web services to paid ones be-
cause, it is argued, consumers are very reluctant to pay for digital goods. Note,
this is an assumption that drives founders’ thinking, regardless of whether it is
true or not. If this assumption is true, startups monetizing their products di-
rectly risk commercial flight as soon as fees are introduced. Due to hyper-
competition (D’aveni 1994) andlow switching cost
67
, startups have few means
with which to generate lock-ins. However, they need both users and revenue.
Therefore, they offer the product for free to solve the dilemma of adoption,
and everything is presumably fine.
Except that, consequently, they fall into afree-rider trap; a paradoxical sit-
uation in which the success of a product, measured by installed user base,
leads to growing economic losses
68
. In other words, the demand for a Web
service is high while financial returns are low due to free provision (Lee &
Brandyberry 2003
69
). Formally, we can argue for the logic of Shy (2011), in
that startups form a belief of willingness to pay; if all startups assume that
willingness to pay is zero (i.e., demand is zero with any price), this is a case of
equilibrium and the belief becomes self-fulfilling (Evans 2002).
In this simple illustration, three lessons can be found. First, that dilemmas
are associated with assumptions in real human situations. Second, that they are
often related, so that one problem and its solution lead to a new problem; thus,
there is a need for new thinking. Third, that solutions are most often tradeoffs
between “two evils” or the selection of only one desirable option, such as
when selecting between users or revenue. Finally, one can observe that
dilemmas offer a fruitful ground for various conceptualizations; for example,
thefree-rider trap. This dimension makes them appealing to various groups,
67
Switching cost is defined here according to Shy (2011, 120): “When firms capture market share
before they encounter competition, the network effects that are associated with their installed bases
generate switching costs, which are the costs of switching from one brand to another incompatible
brand”. Refer to an alternative definition in Subchapter 4.6.1.
68
This is easy to prove by assuming that each user costs to acquire and serve while producing no
revenue in return. In such a scenario, exponential growth leads to exponential losses.
69
”There are ample examples where commonly employed metrics (unique visitors, page views,
sales, etc.) suggest success whilecompanies struggleto obtain profitability” (Lee & Brandyberry
2003, 10).
89
both researchers and managers. In fact, similar assumptions are applied in
Chapter 4.6, in which the ‘monetization dilemma’ is discussed in depth.
4.2 Dilemmas in the platform literature
Several dilemmas have been identified in the platform literature. The most
important of them (Rochet & Tirole 2003), the chicken-and-egg problem, is
part of this dissertation, separated into cold start and lonely user dilemmas.
This issue is discussed thoroughly in Chapters 4.4 and 4.5. This section will
provide a literature overview on other strategic problems in the platform liter-
ature.
In general, Eisenmann et al. (2006) mention three challenges faced by a
platform owner: 1) the pricing problem, or setting prices so that overall profit
is optimized and takes two-sided dynamics into account; 2) the winner-takes-
all problem, which is topical for platforms not dominating due to a tendency of
markets to tip, which is the tendency of one system to dominate its rivals in
popularity after gaining an initial edge (Katz and Shapiro 1994); and 3) the
envelopment problem, involving rival platforms integrating the platform as
part of their offering and thus capturing users. These issues are discussed in
the following chapters with regard to dilemmas that emerged from the mate-
rial. Envelopment can be perceived as a solution for startups to fight dominant
platforms with regard to the cold start dilemma, whereas the price-setting
problem relates to the monetization dilemma (see Chapters 4.4 and 4.6).
Cennamo and Santalo (2013) discuss two particular problems: coring ver-
sus tipping and positioning dilemma. Coring implies exclusive contribution
(e.g., apps exclusivity) by complementors to a specific platform
70
. If the com-
plementor gives exclusive rights to the platform owner, he/she loses the op-
portunity to multihome; therefore, there is conflict of interest between the plat-
form owner who prefers exclusive complements and the complementors who
prefer multihoming to maximize profits. The more there are a) exclusive com-
plements and b) complements overall, adding to intra-platform competition,
the less feasible it is for new entrants to join; thus, the coring dynamics are
against tipping dynamics (Cennamo & Santalo 2013).
In a similar vein, Lee (2013) discusses exclusivity as a strategic problem;
when possible, forcing exclusivity is beneficial to the platform owner.
However, this is done at the expense of competitiveness. If gains from an
70
Gawer and Cusumano (2008) define the terms as follows: “‘Coring’ is using a set of techniques
to create a platform by making a technology ‘core’ to a particular technological system and market.
‘Tipping’ is the set of activities that helps a company ‘tip’ a market toward its platform rather than
some other potential one.” Cennamo and Santalo (2013) employ these terms in an applied sense.
90
exclusive platform fall short of combined gains from other smaller platforms,
rational complementors will switch. Therefore, any case with several equally
or near-equally strong rival platforms that requires exclusivity from third-party
complementors might be ineffective. Modern app marketplaces, for example,
tend not to require exclusivity (Hyrynsalmi et al. 2012). Multihoming is also
typical for video game platforms, in which game makers publish their titles on
many platforms simultaneously (Idu, van de Zande, & J ansen 2011). However,
there are exclusive first-party titles that do not prevent third-party publishers
from multihoming (Clements & Ohashi 2005). However, as the platform
engages in direct competition with its complementors by offering first-party
supply, there is a conflict of interests. If first-party titles comprise the majority
of sales within a platform, third party vendors have less incentive to join than
if there were no exclusive first party titles. This, as argued, forms a dilemma
of first-party exclusivity versus non-exclusivity (Lee 2013).
Another strategic problem relating to complementors is how collaborative
versus competitive it should be in terms of complementors (Economides &
Katsamakas 2006). By definition, a platform can be either neutral or competi-
tive (Hsiao 2003). If the platform owner competes with application providers,
future providers have less incentive to join as, when given a choice, they are
likely to avoid predatory platform owners. However, assimilation through ac-
quisition might be regarded as preferable from an economic perspective, as
proved by several purchases by dominant online platforms (e.g., Facebook
acquiring Instagram; Google acquiring J aiku). Huang et al. (2009, 3) refer to
this problem as “the fine line that platform sponsors must walk between max-
imizing profits and leaving sufficient residual profit opportunities to encour-
age complementary innovation.” They mention that absorbing complements
can increase a platform owner’s profit in the short term while discouraging
other complementors from making platform-specific investments. The strate-
gic problem of the complementor is avoiding to be absorbed or made obsolete
by integration into the core platform (Huang et al. 2009).
A related problem discussed by Hagiu and Wright (2011) is disintermedia-
tion; when the platform has performed its duty and matched two member
groups (e.g., buyers and sellers), the two can, in some cases, continue their
interaction without utilizing the platform, thereby eliminating the possibility of
lifetime revenue. Thus, the following strategic dilemma can be formulated: if
the platform owner enables transparency and uncontrolled communication
among its members, it receives more interaction because it is easier for mem-
bers to interact; this is beneficial for growth but leads to loss of lifetime gains.
However, if the platform enforces non-transparency and strict control on
communication between members, it can retain interaction within the platform
at the cost of interaction levels (Hagiu & Wright 2011).
91
Relating to quality, Wu and Lin (2012) discuss the problem of governing
diversity. There are multiple problems associated with diversity, which is de-
sired by the demand side but problematic for the supply side. One problem is
that the more competition focuses on a particular niche, the less overall benefit
the platform owner receives. Competition is likely to drive down prices and
platform owner’s profits, insofar as they depend on the pricing of its comple-
ments (e.g., through revenue sharing), while intensive focus on particular cat-
egories of complements foregoes long tail effects; that is, larger sales volume
based on diverse tastes and needs of end users. The second problem is the is-
sue of quality. The platform’s reputation is affected by its complements’ spill-
over effects, so that reputable and popular complements elevate a platform’s
image, whereas low-quality complements reduce its appeal to end users. These
problems are not dilemmas, as they lack contradiction. However, they require
proper strategic response in controlling quality without repelling complement-
ors, and encouraging variety in terms of niches and categories to fulfill differ-
ent end-user needs and thus reap long tail benefits. Wu and Lin (2012) propose
discriminatory support based on quality to enhance heterogeneity that, in their
model, leads to higher overall profits for the platform owner.
The positioning dilemma assumes two rival platforms, a generalist and a
specialist, focusing on mass markets and niches respectively (Cennamo &
Santalo 2013). To differentiate from the competition and create a distinct po-
sitioning in the minds of users, the platform must decide between the two
approaches. If it chooses the generalist, it will lose distinction, and a potential
niche market. However, if it chooses the specialist approach, it risks losing
users who are interested in both generalist and specialist content. Because of
winner-takes-all dynamics, users are inclined to adopt the generalist platform,
with widest selection of content (Cennamo & Santalo 2013).
Reisinger (2004) mentions a strategic problem relating to subsidization, a
common strategy in two-sided markets. The competitive dynamics can lead to
a prisoner’s dilemma situation in which competing platforms set negative
prices, thus eroding their profits (Reisinger 2004). However, it is unclear
whether this is simply a manifestation of competition in general, in that com-
petition tends to lower prices and profits, or a unique problem for platforms.
Nevertheless, its effect can be seen in the analyzed startups that typically ap-
plied an indirect monetization model without, however, a working plan to ex-
tract sufficient revenue from either side (see Chapter 4.6). Relating to indirect
monetization, and particularly to the audience maker model (Evans 2003), a
special strategic dilemma takes place when the demand side perceives adver-
tising negatively but is employed as a monetization model. Logically, the
more advertising the end users see, the more revenue the platform owner
earns, although at the cost of end users’ dissatisfaction (Anderson &
92
Gabszewicz 2006). Therefore, the dilemma involves setting the level of
advertising so that it fulfills both economic goals and, if not serving, at least
not repelling users from the platform.
Church and Gandal (2004) identify four types of demand-side issues relat-
ing to platforms: 1) coordination problems, 2) tipping/standardization, 3)
multiple equilibria, and 4) lock-in. They explain coordination problems from
the customer’s perspective, so that a customer choosing the wrong platform or
standard risks “being stranded” as the expected network effects do not actual-
ize (Church & Gandal 2004). There is a coordination problem because the
customers cannot communicate their willingness to joinex ante, and therefore
each is hesitant to join. As can be seen, this is indeed the chicken-and-egg
problem (e.g., Evans 2009a). Customers cannot redeploy their platform-spe-
cific investment towards adoption of another platform. It is unlikely, however,
that the platform owner would be able to exercise power because users simply
have no incentive to stay, despite any sunk cost. Thus, thehold-up problem is
unlikely to arise (cf. Klein 1998). The choice of the platform is a demand-side
strategic problem; it helps understand why users are cautious when adopting
platforms. In tipping, after a particular threshold, one platform becomes domi-
nant and all users convert to being its customers (Shapiro & Varian 1998).
Tipping becomes a problem if inferior technology is chosen, in which case
the opportunity cost is the loss of superior technology in achieving platform
users’ goals (Church & Gandal 2004). The Qwerty keyboard layout is an often
employed example that, according to some, is not the optimal layout in terms
of writing speed but is practically impossible to replace due to its wide adop-
tion; that is, network effects (Parker & Van Alstyne 2010). Once a standard
has been widely diffused, it is hard to abolish; however, before that, its dis-
semination is difficult due to the chicken-and-egg problem. In the literature,
this is referred to as the standardization problem (Besen & Farrell 1994;
Weitzel, Beimborn, & König 2006). Multiple equilibria is the opposite of tip-
ping, so that customers are unable to commit to any competing platforms due
to fear of choosing the wrong one (Church & Gandal 2004). In this case, all
competing platforms lose as adoption is delayed to the last possible moment.
Finally, lock-in can become an issue for users adopting the winning design
(Church & Gandal 2004). Multiple types of power play can arise; for example,
the aforementioned hold-up problem whereby the platform owner can raise
prices as long as the switching cost remains higher or there are no de facto re-
placements, or the quality of the platform’s operations or technology might
suffer due to lack of competition. These effects are similar to monopoly, and
are naturally associated with locked-in customers (Farrell & Klemperer 2007).
As can be seen, the platform literature has discussed a variety of strategic
problems relating to strategic choices of platform owners, complementors, or
93
demand-side users. The dilemmas presented here were identified through a
literature inquiry and represent the current state of research. However, any
number of new dilemmas can be created based on alternative situations. This
study focuses on four specific strategic dilemmas that are presented in the
following chapter. The four dilemmas emerged from the GT analysis, and are
chosen because they represent the issues identified by the studied platform
startups’ founders. Moreover, they seem to respond to the platform literature,
in which the chicken-and-egg problem typical for startups is central.
4.3 Dilemmas emerging from analysis
4.3.1 Results from the black box analysis
The following figure reveals a model of the “black box” of failure (Chapter
1.2) based on grounded theory (GT) analysis. After deciding to focus on
dilemmas (i.e., emergence of the core category), selective coding was con-
ducted to find support for dilemmas, with new dilemmas also being found.
Figure 10 Exploratory outcomes – opening the black box of failure
Startup
Biases Dilemmas
Failure
· Pioneer’s dilemma
· Cold start dilemma
· Lonely user dilemma
· Monetization dilemma
· Remora’s curse
· Pivot dilemma
· Peter Pan’s dilemma
· Juggernaut dilemma
· Illusion of scale
· Illusion of free
· Technology bias
· Build it and they will come
· Dog food blindness
· Sunk code fallacy
· Reference point bias
”Bl ack Box”
94
The purpose of Figure 10 is to show the larger framework in which the cho-
sen dilemmas are rooted. Briefly, the dilemmas are defined as follows:
· Pioneer’s dilemma: if the startup launches too early, it will pay the
pioneer’s cost and is likely to fail due to insufficient resources; if it
launches too late, it is unable to capture users from incumbents.
· Cold start dilemma: without content, users are unwilling to join and
generate content.
· Lonely user dilemma: without other users available at a given time,
users are unable to use the platform.
· Monetization dilemma: if access and usage of a platform is provided
for a fee, users are unwilling to join; if access and usage is free, the
platform is economically non-viable.
· Remora’s curse: if users or content is sourced from a host platform,
the cold start problem can be solved; however, at the loss of power
relating to customer relationships, monetization, and so on.
· Pivot dilemma: if the startup accommodates its user’s wishes in prod-
uct development, it loses focus; if it does not, it loses the user.
· Peter Pan’s dilemma: if the startup accepts external funding, it loses
decisive authority and becomes vulnerable to hasty decisions; if it
does not, it loses against competitors with funding.
· Juggernaut dilemma: due to lack of legitimacy, the startup is unable
to convert enterprise clients which would grant it legitimacy.
Following earlier research outlining failure as a combination of reasons
(Lussier 1996), it can be stated that the failure of the sampled startups com-
prises 1) general business problems (e.g., management issues; lack of market-
ing), 2) startup-related problems, arising from the fact of being a startup (e.g.,
“liability of newness”), and 3) platform-specific problems.
“Illusions”, which were mentioned by some founders, are perceived as fal-
lacies and observed also potentially to exist in other cases. It was also found
that founders typically associated biases (i.e., their own thinking errors) as rea-
sons for why they could not properly address the dilemmas, or even identify
them in time. They are perceived to relate to dilemmas because they affect the
assumptions of strategic decision-making. For example, assuming that all us-
ers prefer freeness over quality will more likely lead to a monetization di-
lemma than a contrary premise.
The identified fallacies are defined as follows:
95
· Illusion of scale: the tendency of startup founders to assume online
businesses require less effort to succeed than offline businesses.
· Illusion of free: the non-validated assumption that users are unwilling
to pay for online products.
· Technology bias: the tendency of startup founders to assume that all
startup problems can be solved by technological means.
· Build it and they will come: the tendency of startup founders to as-
sume that the product will market itself.
· Dog food blindness: the refusal of accepting fault in one’s product.
· Sunk code fallacy: the tendency of startup founders to refuse to make
drastic business changes (i.e., pivots) due to the time and effort spent
making the current version of the product.
· Reference point bias: the tendency of startup founders to assume that
successful implementation of a particular strategy or tactic in another
context would automatically work in their context (e.g., “because it
works for x, it will work for us”).
Due to limitations on the scope of this study, fallacies were left for further
research. It was considered that including them would 1) take away the focus
of dilemmas, and 2) expand the required theoretical basis to become too exten-
sive for one study. In other words, to maintain depth of the analysis, it was not
perceived possible to thoroughly discuss dilemmas and biases, and so the latter
are only briefly discussed as preliminary observations.
Further clarification in the next section will explain why a subset of prob-
lems was chosen for detailed treatment. Consistent with GT principles (Glaser
2004), all dilemmas and illusions were captured by the author. Particular
names, including “remora”, “cold start”, “dog food”, “build it and they will
come”, and “sunk code” were taken from founders’ post-mortems and industry
terminology.
4.3.2 Narrowing the focus of the study
As can be seen, the GT analysis identified many phenomena that remain out-
side this report. All research can be regarded as a tradeoff leading to the neces-
sity of restraining the research focus (Eisenhardt & Graebner 2007), and
focusing on dilemmas was simply the author’s choice. The author preferred a
deeper focus on dilemmas, albeit this decision omitting the treatment of biases
that, according to the analysis, are equally important when considering the
failure outcome.
96
The analysis showed that platformstartups struggle with many other prob-
lems relating to their startup nature (e.g., Wasserman 2013). However, given
its positioning, this study focuses on platform-specific problems. Most other
startup problems are well documented in the literature. For example, liability
of newness (Bruderl & Schussler 1990; Freeman, Carroll, & Hannan 1983;
Singh, Tucker, & House 1986; Stinchcombe 1965) is associated with the prob-
lem of legitimacy (i.e., “J uggernaut dilemma”). The entrepreneurship literature
has analyzed problems of adaptation and related turnaround strategies (e.g.,
Boyle & Desai 1991; Hofer 1980; Melin 1985).
In a similar vein, the strategic management literature has identified glitches
between venture capitalists and founders. Also, growth pains such as thecash
flow problem
71
(Mears 1966; Wilcox 1971) are associated with Peter Pan’s
dilemma. Katila, Rosenberger, and Eisenhardt (1998) studied the “shark’s di-
lemma”; that is, how a startup can collaborate with a larger organization while
retaining its competitive advantage. Pioneer’s advantages and disadvantages,
and also those of early movers, have been extensively covered in the literature
(e.g., Agarwal & Gort 2001; Golder & Tellis 1993; Kerin, Varadarajan, &
Peterson 1992; Lieberman & Montgomery 1988; Robinson, Fornell, &
Sullivan 1992).
The following table classifies the strategic dilemmas based on their applica-
bility.
Table 10 Analysis of dilemmas
Dilemma Specific to plat-
form startups
Specific to
startups
Specific to online
business
Applies to any
business
Cold start x
Lonely user x
Monetization x x
Remora’s x
Pivot x
Peter Pan’s x
Pioneer’s x
J uggernaut x
The cold start dilemma can be regarded as a specific problem for platforms,
regardless of whether they are online or offline (see Chapter 4.4). The lonely
user dilemma relates not only to activating users, but also to time
72
; thus, it is a
problem of real-time social services. Monetization is a general problem of
71
The cost of customer acquisition needs to be covered instantly, while customer lifetime revenue
is received in the future. Therefore, the faster the company grows, the more it accumulates loss.
72
Finding available matches at any given time complicates coordination, and thus aggravates the
chicken-and-egg problem.
97
online offerings, and also applies to platforms, although not necessarily to all
startups beyond Internet markets. Remora’s curse applies when the platform
startup employs the remora strategy to obtain users or content from another
platform. The pivot dilemma
73
applies to all businesses, but is not a specific
problem of platform startups. Peter Pan’s dilemma is a problem for startups
that need to decide whether to remain small and be consumed by competition,
or grow big and be consumed by expenses. The pioneer’s dilemma relates to
launching an unfinished product and failing to gain adoption, or waiting for it
to be perfected and losing competitive advantage. Similarly, the juggernaut
dilemma is proprietary to startups: they cannot get customers due to a lack of
legitimacy, and due to lack of customers, they cannot get legitimacy (cf.
Stinchcombe 1965).
Detailed treatises on all dilemmas, although enticing, would have severely
fragmented the study as they clearly connect with multiple streams of the liter-
ature. In other words, breadth was sacrificed for depth. This decision was rein-
forced by the fact that it proved difficult to find a common denominator that
would have enabled building a unified theoretical framework, as now has been
achieved by relying on the platform literature. Finally, due to the relative re-
cency of the platform/two-sided markets literature, it was concluded that there
is more room for contribution than other identified streams, especially on the
strategic management of platforms.
4.3.3 Chosen dilemmas and their treatment
Following the aforementioned rationale, the focus of this study is on strategic
problems proprietary to platform startups on the Internet, particularly on the
following dilemmas:
· Cold start dilemma
· Lonely user dilemma
· Monetization dilemma
· Remora’s curse.
These dilemmas will be discussed in detail in the following sections, while
other dilemmas are omitted. The presentation of dilemmas follows the struc-
ture of: 1) definition and exhibits, 2) the literature positioning of the dilemma,
and 3) solutions derived from theory. Solution is intuitively defined as a
73
If a startup heeds customer feedback when developing a product, it loses its raison d’être, or its
original vision; if it ignores customer feedback, it loses the customers.
98
solution to a problem that, in this case, satisfactorily solves one or both parts
of the dilemma.
It is argued that by solving the cold start dilemma through subsidization
(e.g., offering free access and usage), the startup will face the monetization
dilemma, whereby it is unable to capture economic value from the interaction
taking place in the platform. Whereas, when solving the lonely user dilemma
by applying the remora model (i.e., ‘envelopment’ in the platform literature),
the startup faces what is termed ‘remora’s curse’ (i.e., dependence of the host
platform). The latter condition bears similarity to the classic hold-up problem,
which is explained in Subchapter 4.7.2. Cold start and lonely user dilemmas
are understood as different realizations of the chicken-and-egg problem pre-
sented in the dissertation’s introductory chapter.
The following figure illustrates the idea.
Figure 11 Strategic actions and their consequences
Therefore, the following strategic decisions apply:
1. When facing the cold start dilemma, the startup solves it by subsidiza-
tion or remora.
2. When facing the lonely user dilemma, the startup solves it with therem-
ora model or subsidization.
3. When facing the monetization dilemma, the startup solves it with the
freemium model.
4. When facing remora’s curse, the startup solves it by diversifying.
The selection of solutions arises from the literature and analysis of the em-
pirical material. Subsidization is commonly considered a solution to the cold
Cold start
dilemma
Lonely user
dilemma
Monetization
dilemma
Remora’s curse
Subsidization Remora
Freemium Diversifying
Types of chicken-
and-egg problem
Derivative
problems
99
start problem in the platform literature (e.g., Rochet & Tirole 2005), while the
remora model is conceptually similar to the envelopment strategy presented by
Eisenmann et al. (2011). Freemium, however, is a special form of subsidiza-
tion that is commonly applied by Web startups (Wilson 2006; Niculescu &
Wu 2013). Notice that subsidization and the remora model can both be applied
in relation to the cold start and lonely user dilemmas. In the following sub-
chapters, this particular order has been chosen for the purpose of presentation,
that is, not to repeat their treatment.
Diversifying is synonymous to multihoming, which is a central concept in
platform theory (Armstrong 2006). Also, because there is generally a high de-
gree of interoperability between Web platforms, for example, through applica-
tion programming interfaces, or APIs (see Rochet & Tirole 2003), both envel-
opment and multihoming are common in online markets (Mital & Sarkar
2011). Therefore, the considered solutions depict both platform theory and
practice. However, they have not been integrated into one framework in the
extant literature.
4.4 Cold start dilemma
4.4.1 Definition and exhibits
The cold start dilemma is a specific problem for content platform startups re-
lying on UG. The dilemma can be defined as follows: when there is a lack of
existing content, no users are motivated to create new content, and so there
remains a lack of content. As a result, the ideal UG model fails, the platform
will fail, and the startup will fail. These assumptions will be examined next.
First, it is assumed that existing content has a relationship with new content;
that is, the reason why other content is created. Note that the content also in-
volves externalities that are not other content, such as sharing and ‘liking’.
These spillover effects will be discussed in Subchapter 4.4.2.
Second, the cold start dilemma might differ through the assumption of
multi-sidedness, depending on whether or not user homogeneity is assumed:
1) in a one-sided content platform, users provide content that is beneficial for
the same type of users, and 2) in a two-sided content platform, users provide
content that is beneficial for other types of user. The interaction between user
groups defines the type of the platform, which is logically derived from the
fact that users’ interests vary: for example, buyers seek sellers, not other buy-
ers in an auction platform; males (typically) seek females in a dating site;
however, people interested in mobile phones are looking for other people of a
similar type in a mobile phone discussion forum. This is relevant due to
100
motivational factors; namely, creating a community, which will be revisited
when discussing solutions. If the interests of the users are common, the plat-
form can be defined as a community, and therefore tactics to acquire users
from this particular niche should varyvis-à-vis a mass audience.
The question of motives and incentives is paramount for getting the desired
response (see Table 11 [2]). Based on the ideal UG model (see Chapter 3.4),
the goal is that the content and actions of first-arrived users lead to the re-
cruitment of second-generation users either directly (e.g., invitations) or indi-
rectly (e.g., content indexed by search engines), as opposed to the startup ac-
quiring new users, which requires marketing investments and, potentially,
skills (Table 13, [6]). Building such assets can be prevented by thebuild it and
they will come fallacy (Table 13, [2]), defined as a tendency of technology-
oriented founders to avoid marketing.
There are two types of participation behavior: contribution and consump-
tion
74
. Contribution is feasible if expected benefits are larger than the cost of
contribution. Consumption has a lower cost but also a low switching cost due
to a generally high number of alternative sources of content (i.e., substitutes)
on the Internet. Further, the benefit of consumption arises from the
informative or entertainment properties of the content; the supply side benefit
comes from search engine externalities, which is compatible with online
search behavior (Hsieh-Yee 2001), and also the social spillover effect, such as
sharing or commenting on the content. The difference in participating
behavior enables analysis of the setting as a two-sided platform, whereas
including single-user motivation would result in a one-sided platform.
The analyzed startups report the cold start problem as follows.
74
Note the similarity to types of monetization behavior: joining without paying (i.e., free users)
and paying for joining (i.e., customers).
101
Table 11 Exhibits of cold start dilemma
Example
[1] "(We) underestimated the “cold start” problem […] especially when it relies on user-generated
content. The value you provide to your users centers around the content on the site, so to build a
user-base you need a lot of content created by the first users to kick-off the community." (Dickens
2010).
[2] "We fell into the “build it and they will” come school of thought (although even when they came,
we still weren’t in good shape). Users didn’t review because there was no enlightened self-
interest for them to do so. Nobody wanted to edit our data for the same reason." (The Chubby
Team 2010).
[5] "[A] place where people can find and provide information about private companies/startups, and
also review them. Awesome idea, right?! WRONG! Where the heck are we going to get all this
data about startups?" (The Chubby Team 2010).
[6] "[H]aving to consistently find new content was probably the biggest hit to my motivation for the
site. As much as I loved indie music it was draining to constantly find new albums to post up."
(McGrady 2008).
[7] "[The startup] was a product designed to connect journalists with readers. As such, we had two
sets of customers, which means we need to do customer development twice. I spent a great deal
of time designing the ultimate solution for journalists, and almost no time on what readers
wanted. As such, I didn’t really know what to make, or what to say to the journalists about what
they should write." (Biggar 2010).
[8] "Our proposition was made even more complicated because we were trying to create a market-
place. When a magazine opens for submissions, you’re submitting to that magazine. But [the
startup] was one step removed – anyone could make an assignment. So even if you trusted [us],
you didn’t necessarily trust the person who posted the assignment." (Powazek 2008).
[9] A close friend reading my early business presentation told me he liked it a lot but was worried
about one line. He said it sounded like I’m building infrastructure and that is going to create
problems with most investors. At the time I was focused on building a website and the comment
didn’t register. Now, with more perspective, I can certainly appreciate the advice. (Hammer
2008).
[10] In community-generated media, trust is everything. When you ask for submissions, contributors
go through an instant internal calculation: “Do I trust these people with my work?” When your
site is brand new, you’ve got no record to rely on. And with more shady “user-generated
content” schemes popping up every day, people have their defenses up. (Powazek 2008).
As noted, and demonstrated by exhibits, a cold start is a specific problem
for Internet startups trying to leverage UGC [1]. Examples include discussion
forums, blogs, and various crowdsourcing services, in which the startup offers
a platform for discussion or other forms of social interaction; for example,
dissemination of information, pictures, videos, or ratings [5]
75
. Such startups
depend on relevant and updated content (e.g., articles, comments, pictures,
reviews, and ratings) to acquire visitors, convert them into repeat users, and
encourage them to produce more content
76
.
75
For example, YouTube is a content platform, so is Flickr. Whether they are one-sided or two-
sided platforms is arbitrary.
76
In contrast, in-house content-generation does not require users to actively produce content;
albeit, startups following this model might enable UGC, hoping for UG benefits.
102
Consequently, if there is little or no UGC in the platform, new participants
have no or only small incentive to join; if no participants join, no new content
is created, and so forth. As no visitors and new content are created, there is no
reason for the platform to exist and the startup will fail. In other words, the
cold start dilemma is a variation of the well-known chicken-and-egg problem,
and quite a typical reason for platform startups to fail.
The cold start dilemma can be demonstrated with a simple game. The fol-
lowing considers a one-sided platform in which users are all the same type,
and gain benefit from each other’s participation.
Table 12 Too many consumers (of content)
C
2
Contribute Not
C
1
Contribute 0, 0 -1, 1
Not 1, -1 0, 0
In the game, contributing can be unpleasant for consumers of content; thus,
they incur a cost (i.e., -1). This cost is a function of time, effort, and uncer-
tainty concerning the usefulness of contributing (see Table 13, [10]). Moreo-
ver, if a consumer contributes content, another consumer will not return the
favor but simply consumes the content, thus gaining a payoff of +1. If both
parties contribute, the effort and benefit cancel each other out. Both receive a
payoff of zero. However, this is not stable equilibrium as each party has an
incentive to improve his/her position by not contributing (i.e., moving from 0
to +1). Because not contributing yields the same benefit as contributing if the
other side makes the same choice, players might be indifferent to contribution.
The safest strategy, which minimizes potential cost, known as minimax (see
Camerer 2003), is not to contribute, as contributing risks a negative payoff.
Therefore, users will not contribute when they expect others not to return
the favor. However, more importantly, they might not contribute especially
when they expect others to contribute.
Now consider changing the game into a two-sided platform, where there are
two different groups of users, both of which derive additional benefit from
complementing interactions.
103
Table 13 Consumers and generators
G
Contribute Consume
C Contribute 2, -1 0, -1
Consume 3, 1 0, 0
Generators enjoy contributing (i.e., creating content), although it is more
valuable to them when there are consumers of content. In contrast, they are
indifferent to consuming content. Consumers of content do not prefer contrib-
uting, so it is costly to them: however, they derive benefit from the content
produced by generators. This is clearly demonstrated in Table 13, case [7],
where the startup focused on one side while neglecting the other; as a conse-
quence, it was unable to provide network effects for either.
Generators receive intrinsic benefit from contributing and are indifferent to
other players’ contributions, but not consumption. In fact, consuming content
is a type of invited free-riding: the creator of the content wishes it to be con-
sumed, and is indifferent to whether others create content or not; although,
they might appreciate “side payments” such as praise and criticism. Further, in
contrast to the previous example, the payoff for the consumer no longer relates
to his/her own choice of contributing and, given he/she has information on the
generators’ payoffs, the consumer will always prefer not to contribute. A sta-
ble equilibrium
77
is when generators contribute and consumers do not.
Furthermore, this explains why merely joining a platform is insufficient in
the absence of active usage. The users who join will quickly churn if the plat-
form is “cold”. This observation is critical in terms of determining which ac-
tion to follow: joining or participating
78
. Therefore, it becomes important for
the startup to find contributors and offer them a convenient platform. Contrib-
utors and consumers can have complementary needs, which is why both par-
ties are needed. Moreover, the startup needs both to consider the critical mass
of any users (see Chapter 4.5) and also the correct proportion of participants.
This is because, in a two-sided setting, the types are interdependent, and rather
than preferring the existence of a similar kind of participant, users prefer a dif-
ferent kind to join.
A two-sided platform, in particular, functions on reciprocal utility: the util-
ity of groupA for groupB is in proportion to the utility of groupB to groupA
77
Neither party would gain a better payoff by switching.
78
As noted by one of the founders (Roseman 2010): “good Google foo [website traffic] won’t save
you. You need that traffic to translate into a community. A visit is not an interesting statistic,
especially in a business that requires the community to produce content.”
104
(for details, see Subchapter 4.4.2). The chicken-and-egg dilemma is therefore
associated with the quality, amount, and type of activity of respective groups
to their counterparts. Advertising is a special variation as there is no content
without advertisers
79
; without content, no visitors; and without visitors, no
benefit for advertisers. Therefore, although often considered negative network
effects vis-à-vis end users, advertisers can, in fact, generate indirect utility by
funding content creation. However, this does not apply under UG; thus, ad-
vertisers are not considered in the dilemma.
Further, a two-sided platform requires strategies for managing both coun-
terparts, whereas a one-sided platform considers the needs of only one set of
customers or users (Table 13, [6]). The implication is that the startup needs to
consider both sides in its strategies. Finally, although the startup can mediate
interaction between two parties, there needs to be a degree of mutual trust for
the interaction to take place, as exemplified in Table 13, [8]. This trust might
not automatically transfer from the platform to all of its users.
Often, the monetization model for content platforms is indirect due to high
competition in most content markets, and also the considerations presented
below; thus, the content is provided for free and monetization is achieved by
showing advertisements (see monetization dilemma, Chapter 4.6). Note that
UG introduces some restrictions to monetization; firms cannot readily charge
for amateur content due to uncertain quality, expected unwillingness to pay
(refer to Subchapter 4.6.1), and resistance from the users creating the content
who might feel that their rights are violated if their content is monetized with-
out revenue sharing. The typical monetization model is therefore indirect:
content platforms delegate content creation to users instead of utilizing the
firm’s own resources, design a process through which the content is re-used to
attract new users such as search engines and social sharing functions, and then
monetize the content through indirect revenue models, typically advertising.
Due to the problem between UGC and direct monetization, that is, the
startup cannot directly monetize the free content provided by users without
consumers’ retaliatory effects (i.e., churn), and by providers of the content
(i.e., resistance to charge for their input), the strategies for monetization re-
main limited. In fact, even after solving the cold start dilemma, a startup needs
to solve the monetization dilemma to become what can be termed a viable
business. Moreover, platforms dealing with real-time interaction need to con-
sider the lonely user dilemma.
79
As the firm is unable to provide content without indirect monetization; note there is an inverse
proportion to advertising, so that users typically respond negatively to its increment.
105
Finally, Appendix 2 includes a brief meta-discussion on the definition of
cold start; namely, if it can be considered a dilemma based on the definition
given here.
4.4.2 The literature
The chicken-and-egg problem of acquiring content, participants, or liquidity
80
is widely recognized in the platform literature, and also the preceding litera-
ture on e-marketplaces (see e.g., Caillaud & J ullien 2003; Parker & Van
Alstyne 2005; Sun & Tse 2007; Kim & Tse 2011; Raivio & Luukkainen
2011).
In fact, the chicken-and-egg dilemma is an inherent consequence of the
two-sided nature of a platform due to the requirement of “getting both sides on
board” (Evans 2002), or an “essential feature of two-sided market analysis”
(Luchetta 2012, 11). Therefore, the phenomenon is not novel; however, rela-
tively little empirical research is focused on its solutions. Furthermore, it is
often neglected by economic models that assume simultaneous entry (Hagiu &
Spulber 2012), and consider pricing the key strategy for encouraging entry
(Piezunka 2011).
The cold start dilemma can be described as a coordination problem, in
which either all or no users adopt a platform (Farrell & Klemperer 2007). This
is similar to herd behavior (see Banerjee 1992) or circular logic, as a user’s
action depends on that of collective action, and can lead to tipping in an in-
dustry-wide setting (see Katz & Shapiro 1994). As a theoretical extreme, it
displays how network effects can dominate adoption if all other determinants
of adoption are ignored. However, it does not apply well to real contexts.
Farrell and Klemperer (2007) give an example of a photography market, com-
prising photographers and film developers who both favor each other, in
which they insert more realistic assumptions, so that groups are not making
all-or-nothing choices; rather, some users might adopt and others not in re-
sponse to individual preferences and heterogeneity among users, and negative
network effects termedintra-group congestion.
This network term, congestion, translates into negative network effects in
the platform context. Also, Shy (2011) notes that network effects are not al-
ways positive, and thus do not always increase willingness to adopt. For ex-
ample, consider a network in which users are evaluated as being, for instance,
“shady” or unreliable. Such a network tends to attract a similar kind of
80
The termliquidity is often employed in the context of e-marketplaces (e.g., Ordanini, Micelli, &
Di Maria 2004), and refers to transaction volumes.
106
participant while high-quality users will refrain from adoption
81
. With regard
only to the platform, the perceived quality of participants is likely to play a
role of adoption. The startup therefore needs to pay attention to not merely
attracting any users but such users who increase the propensity of desired
users (i.e., the target market) to join. This granularity is often neglected when
discussing “traffic driving” and “community building” as strategies for
attracting users. Negative network effects are discussed more thoroughly in
Subchapter 4.5.2.
Although the chicken-and-egg problem lacks a substantial amount of em-
pirical work, it has been focused on by some studies; for example, Mas and
Radcliffe (2011) studied it in the context of a mobile payment platform in a
developing country, Funk (2006) contrasted J apanese and Western efforts in
building a mobile Internet, and Raivio and Luukkainen (2011) documented
challenges relating to Open Telco, a project inviting mobile carriers to collab-
orate. Mas and Radcliffe (2011) identify three factors that prevent payment
platforms from scaling up: 1) lack of effective network effects, 2) the “sub-
scale trap” (i.e., cold start dilemma), and 3) lack of trust from both parties.
They argue (ibid., 305) that
"At first, all these elements work against a deployment. The ben-
efit to a customer of joining the system is minimal when few oth-
ers are connected (network effects) and the merchant network is
not sufficiently dense […] to meet their […] needs. Meanwhile,
merchants remain reluctant to tie-up scarce working capital […]
because they do not yet see enough demand from customers […]
And customers lack trust in the system, because they know few
people who can vouch for the service."
The literature refers to a critical mass (e.g., Rohlfs 1974; Evans &
Schmalensee 2010), which can be defined as the point where the perceived
cost of participation is lower than the benefits of participation, and the benefits
of generating content are superior to only consuming it (for a more detailed
approach to critical mass, see Subchapter 4.5.2). Two points are important
relating to the timeliness of a critical mass: 1) user expectations, and 2) com-
petitive dynamics. User expectations are important because if users are pre-
sented with a “cold platform” they might quickly notice it to be of no use and
81
A real example might beblack hat search engine optimization communities, which basically aim
to divert search engine algorithms by applying unethical practices such as link farms. The existence of
such communities is strictly a negative externality for honest search-engine optimizers, as the “rotten
apples” ruin the reputation of the industry and force search engines to tighten their rules concerning
optimization.
107
never return
82
. The cold start period is formulated by Evans (2009a, 102) as
the “ignition phase”, in which
"[C]ustomers are trying the platform and assessing its value;
these early adopters will stop coming back, and stop recom-
mending it to their friends, if the platform does not grow quickly
enough."
This approach would seem to be compatible with the notion of a ‘one-shot
game’, which users either join instantly or not at all, and which, in turn, can
become problematic when associated with the “perfect product” fallacy
83
. Es-
sentially, while the beginning of a platform startup is critical, it is questionable
whether, for fear of not generating a critical mass, its launch should be delayed
or not.
Second, in a competitive setting where rivals compete for the same users,
whichever reaches a critical mass first can become the dominant platform. In
the case of strong network effects, it is expected that reaching a critical mass
will lead to tipping (Shapiro & Varian 1994), in which users of competing
platforms switch to the focal platform, and thus the market is what is termed
winner-takes-all (Sun & Tse 2007). In contrast, the critical mass can be re-
garded as isolated from competition, when assuming multihoming and less-
than-strong network effects, so that internal consistency
84
of the user base is
sufficient, regardless of coexisting platforms.
Moreover, trust creates a relationship with the legitimacy of new ventures
(Stinchcombe 1965) and the vast body of the literature in which it is discussed
(for a review of online context, see Grabner-Kräuter & Kaluscha 2003). Trust
is, however, implied in the two-sided markets literature through expectations.
An example of negative expectations is given by Galbreth, March, Scudder,
and Shor (2003, 316):
"If buyers and sellers are skeptical of the prospects of an e-mar-
ketplace, these expectations might be lowered [which] could
cause participation levels to move toward the empty e-market-
place equilibrium as opposed to the internal one. […] The gen-
eral lack of confidence in Web-based initiatives in recent years
could explain why many e-marketplaces failed to grow to the
projected participation levels ? when potential participants ex-
pect the e-marketplace to fail, it will."
82
A strategy for overcoming this is to avoid a large-scale launch (i.e., marketing launch) prior to
reaching a critical mass (see Chapter 4.8).
83
A tendency to delay launch until the product is ready. See Appendix 3 for comparison between
late and early launch.
84
Internal consistency refers to the possibility of a user finding a match with a relative ease.
108
In fact, expectations are in the nature of the dilemma; if expected adoption
is low, then it remains low and the expectation becomes a self-fulfilling
prophecy (Shy 2011). In contrast, if the platform gains more powerful advo-
cates, expectations are higher and the prophecy takes a positive direction, even
hype. The case of hype was present in the earlier stage of online platforms,
namely the dotcom era. Evans (2009a) argues that the failure of dotcom plat-
forms was due to three reasons: 1) existing bilateral relationships and other
offline arrangements “got the job done”, so there was no need for online
platforms; 2) participants perceived that e-marketplaces aimed to depreciate
their value. Thus, they mainly focused on price competition between
participants, which dilutes brand and services; and 3) lack of liquidity. Due to
the fact that sellers were skeptical about joining, buyers lost interest, and there
was no reason for the platforms to exist
85
.
Rohlfs (1974) and Funk (2006) refer to the cold start dilemma as astartup
problem
86
and associate it with direct or indirect network effects. Funk (2006)
studied how J apanese firms were able to overcome this problem when intro-
ducing a mobile internet, whereas Western companies failed to do so, and dis-
covered that this was due to 1) generating entertainment content by smart
partnering, and 2) offering the content via a micro-payment system in which
fees were collected and redistributed through a standard system. Evans (2009a,
102) associates this with a critical mass by stating that “[t]he challenge that
catalyst entrepreneurs face is how to achieve the critical mass necessary for
ignition […] over some reasonable space of time.” This thought will be revis-
ited in Chapter 4.8.
Schilling (2009, 195) notes that “if the user must invest considerable effort
in learning to use a computer platform, he/she will probably choose to invest
this effort in learning the format they believe will be most widely used.” This
approach, again, highlights the importance of network effects in creating plat-
form pull. Furthermore, it links the cold start problem to adoption barriers.
Adoption of technology, or ‘technology acceptance’, has been widely studied
within information systems science (for a literature survey, see e.g.,
Mäntymäki 2011) and especially by economists who consider benefits versus
costs; ultimately, leading to preference concerning the platforms as not all of
them can be adopted. Their adoption costs can include psychological re-
sistance, time, and effort, which might be crucial for some users, even to the
extent that time is scarcer than money. Some startups highlighting free models
as the primary motive for adoption might apply a sub-optimal pricing strategy,
which will be discussed in Chapter 4.6.
85
This analogy to dotcoms, ten or so years later, will be revisited in Chapter 4.6.
86
Rohlfs (1974, 18): “how to attain such a user set, starting from a small or null initial user set.”
109
Both the number of participants per se and their proportion need to be con-
sidered. This is especially important for match-making platforms such as da-
ting services. An example relating to content platforms, in which the reasons
for participation relate to information goals, might be a knowledge-sharing
platform, here a sub-type of a content platform, in which there needs to be the
correct proportion of questions and answers (Kim & Tse 2011), and providers
of both to guarantee a sustaining process of UG. Similarly, an exchange plat-
form needs to make sure there are sufficient buyers to interest sellers andvice
versa (Teece 2010). Therefore, even regarding content, and not only social
connections, the startup faces coordination issues of a match-making type.
Rationality of adoption does not only reflect the current situation (i.e.,
weighing costs and benefits as they are), but also the future prospects of the
platform; namely risk and uncertainty. Katz and Shapiro (1986, 824) formulate
this position as follows:
"n the presence of network externalities, a consumer in the
market today also cares about the future success of competing
products.[…] total benefits derived from it will depend, in part,
on the number of consumers who adopt compatible products in
the future."
This is also asserted by Köster (1999) who asserts that adoption depends on
both historical interactions (i.e., existing base of content or users) and ex-
pected interaction (i.e., expected bases). Although both Katz and Shapiro’s
(1986) and Köster’s (1999) definitions referred to other complementary dura-
ble goods, digital content is compatible with the implications; essentially, life-
time projections and long-term survival of the chosen platform are likely to
influence the contribution decision. Equally, the demand-side will find this
important, as finding a reliable source of content reduces search costs (e.g.,
time spent employing search engines).
Another point to note from Katz and Shapiro’s (1986) quotation is technol-
ogy-specificity, which is rooted in the notion of asset specificity (Riordan &
Williamson 1985). Specificities might have weight in the adoption decision;
for example, as many startups eventually fail (Haltiwanger et al 2009), users
might experience greater than normal doubts concerning contributing to or
joining them, given they are aware of failure rates. In the case of demise, a
user loses the time and effort invested in learning a new system (i.e., the
learning curve is a platform-specific investment), incurs search costs for find-
ing a replacement, and might experience loss of private data in some cases
87
.
87
Although the author must note that all failed startups observed during the research period
provided a decent export function prior to closing down, this seems to be the industry standard for a
‘graceful exit’.
110
Finally, the user might not only take a passive stance in evaluating the risk of
platform demise, but might take action to influence it; thus showing support,
although based on the notion of improving his/her own gain. Due to awareness
of inter-platform competition, the propensity of user recommendations, also
termed peer marketing, viral marketing, or simply word-of-mouth, can actu-
ally increase, thus enabling the startup to reach the ideal user-generated cus-
tomer acquisition. As put by Katz and Shapiro (1986, 831), “given the network
externalities, each consumer wants all other consumers to purchase his
favored technology.” Therefore, in theory, network effects can encourage a
user to promote the platform to his/her peers; the more who join, the more
useful it also becomes for him/her.
As established, the market does not exist if the chicken-and-egg problem
remains unsolved, as there is no reason for either party to interact (Evans
2002). Moreover, the adoption by one group triggers a cascading growth as the
benefits of adding a new member come both from that new member and the
influence this member has on attracting more members. Evans (2002, 76)
notes that “f we assumed the base of sellers were important to attracting
buyers, (and vice versa), the indirect benefits would be even greater because a
buyer joining the system would induce additional sellers to join (and so on),
which would generate additional indirect benefits on both the seller and buyer
sides.” Hence, we are close to the definition of viral growth (e.g., Salminen &
Hytönen 2012), which is essentially exponential growth due to one member
inviting more than one new member, and so on
88
.
Note that if we do not assume two-sidedness (i.e., distinct user groups) and
network effects (i.e., interdependent interaction), the cold start problem would
not be compatible with the chicken-and-egg problem in the literature. This is
because, if the content platform is a one-sided market, the platform owner
(i.e., startup) might simply provide the content, as is done by media
broadcaster and news websites, among other content portals, and there would
be no cold start problem. This would render the study’s treatment a trivial
exercise. To counter this, we can distinguish 1) two distinct user groups (i.e.,
consumers and contributors of content) and also 2) interdependence (i.e.,
network effect) between them, which is mediated by UG, so that the
consumers derive benefit from the content produced by contributors.
Whether contributors derive benefit from the presence of consumers is ar-
guable; it is possible that the nature of this benefit is intrinsic motivation, de-
sire to create, or some form of altruism. The motives to contribute to platform
88
The simplest definition for viral growth isx*y >1, wherex describes the number of users invited
(by base user) andy the number of accepting users who join (Salminen & Hytönen 2012). When all
users successfully invite more than one new member, the growth is viral.
111
development have been studied, in particular, relating to open-source plat-
forms: Schilling (2009, 202) asserts that “in the software industry, individual
programmers may work on an open-source software program because it re-
sults in solutions to their own problems, provides an opportunity to interact
with peers and improves their reputation as experienced programmers.” Sim-
ilar, but varied, motives can be found in the UG context, although more re-
search is needed. There is an emerging body of the literature relating to the
motives of the crowd which comes close to this purpose; for example, Dow et
al. (2011), Kittur et al. (2013), Pitkänen and Salminen (2012), Zhao and Zhu
(2012), Zheng, Li, and Hou (2011). Drawing from this literature would open
opportunities in utilizing crowds to solve the cold start dilemma.
4.4.3 Solution: Subsidies
Subsidization is the most commonly considered solution in the platform liter-
ature. By definition, a side is subsidized “when the price it faces is lower than
the price it would face in an independent market” (Bakos & Katsamakas 2008,
173). In the lonely user dilemma, to increase possibilities for interaction, the
startup might subsidize one of the user sets for joining. With regard to the cold
start dilemma, the startup might subsidize contributors of content in exchange
for their efforts, or developers for creating extensions or content creation tools.
Because developers or other contributors might require payment in exchange
for their contribution, subsidies can also include negative prices (Parker &
Van Alstyne 2005).
For example, Evans (2002) argues that “providing low prices or transfers to
one side of the market helps the platform solve the chicken-and-egg problem
by encouraging the benefited group’s participation – which in turn, due to
network effects, encourages the non-benefited group’s participation.” In a
similar vein, Spulber (2010, 7) argues that “incentives induce strategic partici-
pation which resolves the cross-market coordination problem.” Free offerings
are not a new invention. Rohlfs (1974, 33) had already proposed that “[t]he
most direct approach is to give the service free to a selected group of people
for a limited time.” Rohlfs (ibid., 33) further argues that the initial user base
must be sufficiently large to achieve a critical mass, and notes that “half
measures are worse than useless”, because demand will be zero without a
critical mass of users. This feature, stemming from network theory, has been
adopted by the platform literature, so that “[a]n important characteristic of
multisided markets is that the demand on each side vanishes if there is no de-
mand on the others, regardless of what the price is” (Evans, Hagiu, &
Schmalensee 2006).
112
Although Piezunka (2011) specifies that the platform owner might not give
negative prices because of a potential moral hazard problem (Gawer &
Henderson 2007), negative prices have been observed, for example, by Parker
and Van Alstyne (2005), and cited by Mas and Radcliffe (2011) as a strategy
utilized by Paypal. Mas and Radcliffe (2011) argue that negative prices (i.e.,
paying for users to join) helped PayPal achieve a critical mass faster than
would have been enabled only by zero prices. Other startups have applied rec-
ommendation fees to incentivize their users to promote the service to their
peers (Libai, Biyalogorsky, & Gerstner 2003). However, this can be defined as
marketing rather than subsidization, which aims to get users to adopt the ser-
vice by not charging them for its use
89
(Lyons, Messinger, Niu, & Stroulia
2012). Free or negative pricing is possible, although with the precisely identi-
fied threat of spamming.
Moreover, the startup might begin by subsidizing one side and, when it is
secured, then moving the subsidization to the other side. The goal is to reach a
state in which no subsidy is needed as the platform has become self-
sustaining. Caillaud and J ullien (2003) refer to this strategy as divide-and-
conquer, by which the market is divided into two markets (i.e., two-sided
markets) with one being captured by subsidization, after which the other side
will follow, enticed by network effects. However, subsidization might become
a permanent strategic choice to maximize participation in a two-sided market
(Rochet & Tirole 2003). In the context of online startups, subsidization refers
to free models in which the users, or a segment of them, are not charged.
Complete free offerings, or non-paid access and usage of the platform, are
referred to as freefying, whereas offering a free version and a paid version
between which the users can move (i.e., convert or downgrade) is termed
freemium (see Chapter 1.5).
Consider basic subsidization. In a sequential form, the startup subsidizes
one party who is more unwilling to join; after securing participation, the other
side can be acquired, for example, through online marketing. Subsidies are, for
example, offering free trial (i.e., direct monetization) and free premium mem-
bership (i.e., mixed monetization). The problems of freefying are discussed
elsewhere (see Chapter 4.6). However, essentially, freefying does not elimi-
nate the cost of adoption although it eliminates the economic part of it; neither
is it an effective competitive strategy, as it is easy to copy and can advocate
unwillingness to pay. Freefying also creates the monetization dilemma when
the startup is unable to capture economic value from its user base.
89
A more advanced form is a dual-sided referral incentive; for example, Dropbox offers additional
storage space for both the referred and referring user.
113
Table 14 Basic solution of subsidization
side A
Not willing Willing
side B Not willing Subsidize DN*
Willing DN* DN*
*Do nothing
The startup only needs to subsidize when either party, or both, are unwilling
to join. For example, if A is willing to join and B is not, the strategy is to sub-
sidize B and do nothing with A. If both are willing, the startup does not need
to give subsidies, which would indicate very strong demand for the platform.
If both are unwilling, the startup needs to subsidize both, which would indicate
either low demand or high competition.
However, complications arise: what should be done when both parties re-
fuse to join, and the subsidized price is already set at zero? Such is the case
with free models
90
. In this case, as proven with our sample, a potential fallback
is failure, or otherwise applying alternative solutions to solve the dilemma. A
third option might be to set a negative price, and pay users to register, alt-
hough this strategy would be poor for many reasons. For example, the quality
of entrants might be low, the costs of adding users grow linearly, and the
problem of active use impacted by users not joining due to intrinsic motiva-
tion
91
.
Moreover, subsidization can quickly become too costly if the market is
large, as it typically is in online consumer markets
92
. Belleflamme and
Toulemonde (2004) show that even in B2B markets with a limited number of
participants, a platform subsidy can make the platform unprofitable to a severe
degree, which is due to its linear property: each member needs to be subsi-
dized to the same extent. Therefore, the solution would be to subsidize until a
critical mass is reached (e.g., by covering this as a form of marketing expense)
and then allow UG to take over user acquisition, as per the ideal UG model.
Another solution would be discriminant subsidies in which the amount of sub-
sidy varies based on the expected utility of the user, (i.e., to other users in a
single-market platform and to the other side of the market in a two-sided
90
Free models refer to free access and usage (i.e., freefying) and the freemium model.
91
Clearly, the cost of adding users is negative, even without including potential subsidies, when
having considered user acquisition costs, such as advertising and other marketing costs. Therefore,
negative price attached with indirect monetization (i.e., free users) and marketing costs can quite
easily be detrimental from a financial perspective.
92
Consider that the total cost of subsidies, c, would be dependent on a fixed cost x (per user) that
would grow exponentially by factor a alongside exponential growth of user basey (subject to a), so
that c = (xy)
a
. Hence, exponential growth would lead to an exponential cost of subsidization.
114
platform); in other words, paying some users more to join. However, this
would raise questions of fairness and might be counter-productive.
The question of subsidies can also be turned around by considering the pro-
spective members’ perspective; perhaps opinion leaders could be recruited by
offering them other types of incentive, for example, of a social nature. Eventu-
ally, the tactics might vary but the principle remains: members who produce
the most content are more valuable in content platform, and members who are
socially connected and willing to propagate other users to join are valuable to
the social platform
93
. Moreover, the user is not necessarily acting in accord-
ance with economic rationality principles when creating content or joining a
platform; even an exchange platform can involve social motives for participa-
tion. Alas, the type of utility they seek might be more fragmented and hetero-
geneous than generally understood by startups or platform theory. Tapping
into social motives might, in these cases, provide gains that exceed the effect
of financial incentives or cost savings.
Furthermore, moving from free to paid products can become problematic.
As noted by Brunn, J ensen, and Skovgaard (2002), penetration pricing is a
common tactic in one-sided markets, although it can have an effect on long-
term profitability. Introducing fees (i.e., ‘bait and switch’), if the form of sub-
sidization is freefying, can be difficult as parties rebel against going from “free
to fee” (Teece 2010). The early platform literature established that the plat-
form sponsor (i.e., platform owner) actively promotes the platform, not only
by subsidizing the cost of its adoption. For example, Katz and Shapiro (1986,
822) define a platform sponsor as one who “is willing to make investments to
promote it”. Further, they argue that when two rival technologies exist, if one
is promoted and the other is not, the promoted one can rank higher in adop-
tion, regardless of whether it is superior
94
in some objective comparison (e.g.,
features). Therefore, subsidization alone cannot be regarded as the optimal
solution for the cold start dilemma as factors other than price of usage influ-
ence adoption. Further, the cost of subsidization might be lost due to rivals
competitively reducing their rates.
Moreover, there is the question concerning which side to subsidize? First,
Belleflame and Toulemonde (2004, 6) argue that “in several categories of two-
sided markets, most agents of one side of the market arrive before most agents
of the other side.” Hagiu (2006, 721) points out that “in the software and vide-
ogame markets, most application developers join platforms (operating systems
and game consoles) before most users do.” In social platforms, the order of
93
The conclusion, therefore, matches that drawn by Li and Penard (2013), in that quality can
replace quantity in the early stage of a platform.
94
This was the case in the early 2000s when Sega first launched its Dreamcast console, but players
hesitated to adopt it due to Sony’s clever promotion of the soon-to-be-launched Playstation 2.
115
entry does not seem to be relevant. However, if this division is assumed, con-
tent creators in content platforms need to arrive first as content must logically
exceed its benefits. In general, the side that derives less benefit from partici-
pation should be subsidized (Curchod & Neysen 2009) as the risk of non-
adoption is assumed greater. However, Parker and Van Alstyne (2005, 1503)
assert that either both sides can be the target in the context of free distribution,
or the side that “contributes more to demand for its complement is the market
to provide with a free good” when network effects are high. This notion does
not consider the risk of adoption but rather the benefit for profit maximization.
Second, some users are more valuable than others, in terms of their network
utility. The two-sided literature terms these marquee customers (Rochet and
Tirole 2003), while in the social networking literature, prominent users are
termed prestige nodes (Evans 2009a); and in marketing such terms as
influencers (Gillin 2009), early adopters (Rogers 1995), or opinion leaders
(Flynn, Goldsmith, & Eastman 1996) are employed; that is, users with whom
many people want to connect. These users influence others to join the
platform.
Therefore, influencers generate significant direct or indirect network
effects; thus, it pays to subsidize their entry (Niculescu & Wu 2013).
Therefore, platforms need to identify and recruit influencers early on both
sides. Finally, Parker and Van Alstyne (2005) assert that the structure of subsi-
dies depends on the industry. They give some examples (ibid., 1496), for
example video streaming services and advertising, in which subsidizing
consumers is an industry norm; and operating systems and videogames, in
which the developers are subsidized while consumers are paying. Therefore,
there might be no universal solution to which party receives subsidies, as
industry conventions need to be taken into account.
4.4.4 Discussion
Subsidization can solve the cold startup problem under some circumstances,
but only as a local solution; there is thefree beer effect based on which giving
a free platform is not viable business unless it is successfully monetized. Price
is not the only matter influencing adoption although it will necessarily limit
the scope of price-related strategies; therefore, even negative pricing might not
be sufficient to solve the cold start dilemma. Subsidization might also lead to a
free-rider problem if the product is structured to enable both free and paid us-
age (e.g., freemium). This takes place by assuming satisficing behavior (Si-
mon 1956) and positive willingness to pay. The user is then able to receive
benefit from the platform without economic cost, while the startup incurs a
116
freefying loss, given that there are users who would have been willing to pay if
no free option was presented. Startups opting for freefying thus perceive the
risk of adoption as greater than the risk of deferred monetization, potentially
by an indirect monetization model
95
.
Consider that the startup is aware of the cold start problem, which, at least
intuitively, it is in most cases. To kick-start the content platform, it produces
some initial content in the hope of initiating the user discovery process that
will lead to users finding, reacting to, and sharing content forward, thereby
recruiting new users. However, if this strategy fails, and the crowds fail to
materialize, the startup will find that it needs more content or that a type of
pivot is required.
However, if its in-house resources do not scale to match content production
or, perhaps, it lacks specific expertise for content production, it might consider
UG, representing a “magic bullet”, the solution to the content problem. How-
ever, this study has already established that UG is not a solution to the cold
start problem (Table 11), but rather the consequence of a need or interest.
Therefore, given that the root cause lies elsewhere, the cold start problem is
not resolved, and the startup returns to the beginning.
It is therefore crucial that a startup recognizes the limits of in-house content
generation, unless its deliberate purpose is to develop a content-production
organization, which is a strategy sometimes termed ‘inbound marketing’
96
. In
a business model reliant on UG, however, in-house content provision cannot
easily extend beyond kick-off due to its costliness; in fact, in the long run, it
would dissolve the benefits of UG if the content community was not be self-
sustained. For leveraging the potential of Internet users in terms of content,
UG is simply superior to any alternatives
97
, which are aggregation and what
Mark Zuckerberg terms frictionless sharing (Darwell 2013), a concept refer-
ring to automated sharing of activities conducted online. At the time of writ-
ing, this approach is in its infancy and facing major resistance regarding pri-
vacy issues.
The route to a solution can, in fact, arise from the fact that not only existing
content, or users, make up for the decision to contribute, or join, but that
expectations of future interactions can attract users to perform first interactions
with the platform. This abstracts the requirement of network effects from
quantity or quality to a signaling problem, essentially an issue of marketing
communications. If the startup is able to communicate the vision of the
95
This is commonly known as “searching for a business model”.
96
The cold start dilemma does not concern a firm that employs in-house content generation,
because it will start deterministically producing content from the beginning. Its challenges relate more
strongly to traffic generation and marketing, both being closely associated in online marketing.
97
Consider replacing all user-generated content in Facebook with editorial content.
117
platform in a credible fashion to, for example, early adopters or “influencers”
and employ this communication to commit them to performing first
interactions with the platform, in this case creating content, the cold start
dilemma is in theory solved. However, user acquisition that relies on content,
as in the ideal UG model, is a major problem, which is why credible signaling
without first-interaction commitment would be fruitless. Thus, a positive
expectation is preceded by awareness, generated by some form of marketing:
Sequence: marketing actions ? awareness ? positive (or negative)
expectation ? adoption (or non-adoption)
In some sense, it would be sensible for the startup to find a mixed solution
in terms of one that is pure. For example, hiring an in-house community man-
ager to coordinate content generation with partners and ensure initial attempts
to achieve a critical mass are successful. Further, the process of content dis-
semination needs to be well considered, so that 1) the startup has ready access
to social media platforms where it can disseminate the content, referred to as
an ‘integrated marketing communications strategy’ (see Mangold & Faulds
2009), and 2) that the website is optimized for participation and sharing to be
as frictionless as possible. However, at the same time, founders need to
acknowledge that technically frictionless does not imply socially frictionless
sharing. Ries (2011) argues that users might assume a social risk related to
sharing their behavior, and therefore the expected propensity to share does not
necessarily materialize.
In theory, the cold start dilemma can be solved by integration into an exist-
ing platform that supplies the much needed users who will generate content.
However, there can be 1) strong intra-platform competition and 2) misalign-
ment of goals between the startup and platform owner, which result in rem-
ora’s curse (Chapter 4.7). Equally, in theory, freefying will remove the obsta-
cle of purchasing as it sets the price at zero. However, it ignores the fact that
the cost of adoption not only includes a financial cost but also the time and
effort of learning a new product (i.e., changing behavior). In addition, when
the solution is effective and new users join, the startup faces the problem of
monetizing the user base. Once free, it is hard to revert to offering paid prod-
ucts without a considerable churn in the user base. As argued throughout the
study, users are not synonymous with customers. This conundrum is discussed
in Chapter 4.6.
A horizontal strategy would be to increase the scope of topics (i.e., econo-
mies of scope) to drive adoption. A vertical strategy would be to find oppor-
tunity niches; that is, unserved content areas, unserved social niches, or buyers
and sellers not being adequately served. Finally, differentiation in terms of
118
features (in simple terms: better execution) can perform an opportunity to grab
customers despite network effects, which is often put forward as the reason for
both Google’s and Facebook’s success, whereby they both provided better
solutions in comparison to alternatives with a critical mass, including Yahoo
and MySpace, both of which are slower and more complex. Thus, in a struc-
tural sense, there are likely to be aspects that are impossible to account for in
the design of a platform such as execution and features, which can have a
stronger influence than network effects
98
. These can either work in favor of or
against the incumbent, depending on whether it has a better product or not.
More research is needed, however, on the critical success factors of competing
platforms’ execution strategies.
It has been discovered that successful adoption addresses overcoming
change resistance relating to routines (Oreg 2003), and that time and effort are
comparable to financial cost (i.e., money) as factors determining the outcome
(Webster 1969). Although the removal of financial cost through freefying can,
in an economic sense, reduce the overall cost of adoption, it cannot be re-
moved as time and effort lead to the necessary existence of a learning curve
(Yelle 1979) that, as stated, is a factor of adoption. This is the reason why the
cold start problem, interpreted as a problem of adoption, cannot be solved by
freefying alone. Even if the subsequent monetization problem were solved, the
adoption problem would potentially return to haunt the startup. Finally, the
potential discrepancy in perception is crucial as it explains why startups per-
ceive freefying as an answer to cold start problems, which is a logical conclu-
sion assuming that the adoption cost only includes financial cost, whereas us-
ers would, in reality, require assistance to decrease the learning curve. Clearly,
from an economic perspective, the learning curve can be increased if expected
benefits are high. Unfortunately, not all startups create products for such a
need that a user is willing to spend considerable time and effort on learning
their systems.
The introduction of negative network externalities that increase with harm-
ful activity is a potential risk. In a two-sided market with indirect
monetization, Internet startups often resort to advertising as their monetization
model. However, advertising typically represents a negative indirect network
effect for end users, which is a manifestation of the “cat and mouse” game
between advertisers and consumers because the latter desire to escape the
former. In contrast, in a one-sided setting, spam and fake profiles result in
direct negative externalities for both parties.
98
As expounded by one of the founders (Dickens 2010): “We were offering information on great
albums and community voting. But other sites like Last.fm and Hype Machine were offering the actual
music. That was a competitive advantage that’s hard to beat, and we lacked a significant user base to
convince enough people.”
119
4.5 Lonely user dilemma
4.5.1 Definition and exhibits
Generally, for users to join a social platform, they expect to find other individ-
uals using it. If none can be found, there is little or no incentive to join the
platform. The logic is equivalent to the cold start dilemma. In contrast, once
the first generation of users have signed up, new users are enticed to join
through the connections of the first group, and so on; startup founders refer to
theviral effect or simply exponential growth. The logic is based on the notion
that the total benefit generated by a social platform can be measured through
the number of connections between users (cf. Metcalfe’s law; see Briscoe,
Odlyzko, & Tilly 2006), and the frequency and quality of activity within these
connections (i.e., the network effects).
The principle of users’ mutual expectations can be demonstrated with a
simple game.
Table 15 Startup platform
S
2
Join Not
S
1
Join S
2
, S
2
-1, 0
Not 0, -1 0, 0
S
2
is the potential number of interactions between members of the platform,
and marks the network effect. Albeit being a bad proxy for network value (see
Subchapter 4.5.2), it is easy to quantify and represents an upper limit for
interactions (Aggarwal & Yu 2012). In other words, parties potentially draw
symmetric benefit from each other’s presence, and payoffs are equal.
J oining has a cost, if not financial then time and effort, which is why ex-
pected non-participation of another party leads to both not joining. Both par-
ties would be advantaged by joining but as it is risky for each of them to do so,
the outcome might be both not joining. This is referred to as the coordination
problem in game theory (Van Huyck, Battalio, & Beil 1990), and describes
well the lack of legitimacy to which the new platform is subject.
In contrast, consider an incumbent platform. This example demonstrates the
importance of a critical mass.
120
Table 16 Incumbent platform (with a critical mass)
S
2
Join Not
S
1
Join S
2
, S
2
S
2
-1, -1
Not -1, S
2
-1 -1, -1
In this case, the incumbent platform already provides a critical mass of us-
ers (or content) for interaction, which is why the dominant strategy for each
party is to join. Even when other users do not join, the entrant receives benefit
from the existing base of users (S
2
-1). Because both parties have the incentive
to join, it is also the Pareto-dominant equilibrium. Therefore, it is much more
difficult for a new platform to attract entrants than for an existing platform
with a critical mass. Consequently, even in the presence of multihoming (see
Chapter 4.8) and low switching cost, the startup can fail to rally users.
However, the basic chicken-and-egg problem becomes more complicated
when introducing dynamic factors, such as time and place. Consistent with the
definition of the cold start dilemma, the lonely user dilemma can be defined as
follows:
In a social platform, when there are no existing users, no new
user will have a motivation to join. Additionally, when there are
no active users at a given time or place, no other users will use it
at that time or place.
If a user has no contacts in a social service, the perceived benefit of the ser-
vice equals zero for that particular user at that particular time or place, re-
gardless of the number of registered users or “static” critical mass, such as
content, that is always available. In practice, these platforms can include social
platforms such as chat services requiring simultaneous presence of parties, and
location-based services whereby the interacting parties need to be available at
the same time and also in the same place.
In the cold start dilemma, the focus is on recruiting new users (e.g., to gen-
erate content) and keeping them active in UG activities. In the lonely user di-
lemma, the focus is on acquiring users for social interactions taking place be-
tween individuals and groups, and keeping this interaction active (i.e., the
problem of active use) while considering the effect of time. As such, at any
given time, not only on average, the platform must have a critical mass to pro-
vide matches and thus be useful
99
.
99
Consider the Facebook platform: if in some given time frame, all of a user’s friends were offline
and had not updated their statuses, eventually the user would permanently stop using the service,
regardless of how many registered users there are.
121
Therefore, the requirement of a critical mass is much more extensive than in
the case of static content
100
. In other words, the demand-side benefit in social
platforms is derived from social interaction (i.e., social exchange) instead or
more or less static content, with the source being topicality, information, en-
tertainment, or other properties of the content. For example, unlike communi-
cation between friends, reviews and videos are not social interaction in a fun-
damental sense
101
. In a content platform, users enter the website for the sake of
the content (e.g., news, reviews, articles, and videos), whereas the lonely user
dilemma is typically associated with social network sites in which availability
of others is conditioned by time and/or physical location. Table 17 exhibits the
dilemma.
Table 17 Exhibits of the lonely user dilemma
Exhibit
[1] "I think you need a critical mass in any community and we didn’t quite achieve that critical
mass. I mean, who wants to go into a forum when there’s really nobody to talk to?" (Warner
2009).
[2] "Lastly, the “real-time problem”. This one is similar to the location problem in that if someone
wasn’t online when you were online, they were no good to you. While the real-time chat aspect
of the application made for some really serendipitous meetings, it also made it harder for people
to gauge the activity of their communities, especially if they logged in at odd hours, people were
set as away." (Bragiel 2008).
[3] "We launched our product and got all of our friends in Chicago on it. We then had the largest
papers in the area do nice detailed write-ups on us. Things were going great. We had hundreds
of active users and you could feel the buzz around it. […] The problem, we would soon find out,
was that having hundreds of active users in Chicago didn’t mean that you would have even two
active users in Milwaukee, less than a hundred miles away, not to mention any in New York or
San Francisco. The software and concept simply didn’t scale beyond its physical borders."
(Bragiel 2008).
[4] "The weakness of the hub strategy was the market players never arrived at the same time. Sellers
would flock but there would be no buyers, or buyers would flock and there would be no sellers."
(Anonymous founder).
[5] "The real tests come at moments like we had about a week after our initial launch. Lots of
people dropped by, told us they loved the site, and didn’t come back. So, there we were, left with
one big question that lead to endless others: Why aren’t they coming back? Is something too
confusing? Is our idea a bad one? Do we just wait and see if they come back later? Do we need
to build another tool?" (Karjaluoto 2009).
100
Here it is assumed that social interaction expires much more rapidly. However, content also
expires. If no new content is added, after a while users will stop using it, although the remaining
content would continue to provide benefit; this is not the case for social interaction. However, in some
cases thetopicality of content approaches the temporality required by social exchange. Consider, for
example, a news portal in which all content has to be fresh.
101
Note that this does not exclude spillover effects between content and social interaction. In fact,
these are generally requisite for UG effects to occur.
122
We can deduce that the number of users required to initiate the self-replica-
tion UG process is often referred to as a critical mass, both in the literature
(see the following chapter) and by practitioners [1].
Thecoordination problem [4] is distinguished from the real-time problem
[2] based on the notion of time. Coordination fails as a result of an overall lack
of participants in the other side. The real-time problem might mean that there
is a potential critical mass in the other side, but that they are momentarily in-
active
102
. As noted previously, the real-time aspect is emphasized in the lonely
user dilemma due to immediacy of social interaction. More precisely, coordi-
nation can relate to the participants’ different needs, which requires under-
standing both sides well and managing their expectations. The timing of con-
verting users can be critical here; thus, if the technology is premature, per-
suading users to join can cause a major disappointment. Basing the platform
design on the premise of self-organization might not take place in reality.
Furthermore, in social environments, match is not simply a question of the
number of members in group A or B, but also their quality (i.e., compatibility).
Match might require a special type of user property relating to, for example,
demographics or offline relations. Not all counterparties willing to interact
will regularly provide a match
103
.
The real-time problem, if defined as ‘getting users on board’, suggests that
solving the cold start problem is insufficient to solve the lonely user dilemma;
that is, registering to a platform does not automatically lead to active use,
without which, the platform will gradually die regardless of adding new users.
This is termed churn in marketing and is parallel to pouring water into a
bucket with a hole. Thus, while loyalty is low, increasing customer base will
only increase cost, relating to lifetime value, as customers constantly abandon
the service. In other words, users of a real-time service need to be simultane-
ously present or coordination will fail. This is crucially different from static
content, whereby coordination is much less affected by timeliness.
The transferability problem [3] implies that a predominant user base in
context A (e.g., location) cannot automatically be generalized as a critical
mass in context B (i.e., another location), even when it fits the notion of criti-
cal mass in its primary context. In particular, the problem relates to hyper-
local platforms such as location-based services. Although our exhibit
102
How is the real-time problem different from the lonely user dilemma? The former is a
manifestation of the latter, in which time is the match-making criterion. However, the lonely user
dilemma can be manifested in relation to other match-making criteria, such as physical location and
preferences. In both cases, the user is “lonely” without an adequate match.
103
However, this type of differentiation also exists in exchange platforms. Consider the following
criteria for match; for example, item being transacted, condition, reputability, and location of the other
party. In general, however, users are more selective in engaging in social interaction with “strangers”
than transacting with them.
123
addresses location, the transferability problem itself can be generalized into
any context in which one group is so distinct from another that direct network
effects will not emerge across the two groups.
Therefore, the startup needs to consider its match-making role
104
and
emphasize user-acquisition based on the development of dynamics between
the groups. For example, if there is a shortage of either female or male mem-
bers in a dating service, more users of the required gender need to be recruited.
When there is disconnection between user bases, for example, niche division
or geographical distance (i.e., local social networks), there is a shortage of
synergy between user segments; that is, no positive network effects arise even
when the groups are connected. Hence, each segment needs to be built indi-
vidually due to proprietary network externalities to that community, although
the transferability problem [3] will not be overcome unless propagated by
members of the community. The benefits for a startup involved in building
multiple communities are therefore limited to learning gains, which can
facilitate replication of critical success factors, and also potential reputation
and brand spillover effects when users in another community become aware of
the platform’s existence, and perhaps start acquiring its community.
Furthermore, users’ homogeneity, defined as similarity of interests, demog-
raphy, or other feature that increases similarity, might influence the perceived
utility of the network by an individual user. These features can include, for
example, location, online status, and similar preferences. Diversification is
needed if the service is match-making between opposite groups with users
looking for counterparts (i.e., buyers for sellers; men for women); thus, homo-
geneity tends to be counterproductive in two-sided markets
105
. However, in
one-sided platforms, users derive benefit from similar users joining the ser-
vice, which implies direct network effects. They might also appreciate com-
plements by other firms, such as plugins, games, or additional content by third
parties within a platform ecosystem, which reflect indirect network effects.
Conversely, when there are users in one side (A) competing for members on
the other side (B), each additional user in A in fact reduces the incentive for
similar users to join. Strictly speaking, assuming that match-making exhibits
rivalry in that connections between members in A and B exclude other
104
Essentially, a marketplace platform is a mediator between two parties, often supply- and
demand-side, so that it offers auxiliary benefits in addition to matching (i.e., coordinating), such as
payment options and vouching.
105
In a sense, this is a trivial observation. The definition of two-sidedness entails the idea that the
groups are distinct. Therefore, similarity merely parallels this state with some actual criteria for
distinctness.
124
connections, this increases competition
106
. Therefore, a high number of
members in A represents negative (i.e., direct) network effects for a prospec-
tive member of A as they are competing for the same resources (i.e., members
of B). The final decision to join is affected by the difference between
perceived negative network effects (i.e., the level of competition) in
comparison to perceived positive network effects (i.e., the number/attractive-
ness of group B). Implications such as these will be further discussed in the
following literature subchapter.
4.5.2 The literature
The problem of active use is expressed in Albuquerque et al. (2012, 407):
“there are usually two (or more) stages in the decision to participate in a user-
generated platform. Users must first opt to visit the site, and once in the site,
they must decide to generate and/or consume the available content.” In other
words, acquiring a user is not alone sufficient to guarantee interaction benefits
to the other side. A similar conclusion is drawn by Xia, Huang, Duan, and
Whinston (2007) who distinguish between the user decision to adopt/join
AND to continue use . Therefore, it is acknowledged that the adoption choice
is not the only requisite for a permanent solution to the chicken-and-egg
problem.
The lonely user problem, therefore, is not only limited to encouraging reg-
istrations or other forms of subscription/joining, from which a startup might
infer that optimizing registration pages (i.e., landing page optimization) is a
top priority, but also to activity taking place after the user has enrolled. As
presented by Boudreau and Hagiu (2009, 171), “Facebook must then activate
the ‘social graph’. Beyond simply establishing linkages among members, it
must keep these linkages active, fresh and compelling”. This is a problem as,
after joining, the success in fact depends on the community’s or platform’s
activity, which is the source of new users. In fact, if the replication rate or viral
coefficient drops below one, the exponential growth stops (Salminen &
Hytönen 2012) and, considering churn, the user base can begin to decrease.
Not only this, but it is also possible that users estimate the degree of activ-
ity, as it is part of their utility function, prior to joining, and employ this esti-
mation, based on the activity they see when visiting the platform for the first
time, as an adoption factor. This would mean that a low anticipated frequency
106
This might or might not be the case, depending on the strategy of users. For example, in a dating
site, a user might stop creating connections to potential dates after finding “the one”. However, it is
also possible that he/she might continue to create further connections.
125
of activity decreases the expected utility. It has been previously argued in this
study that expected benefits are proportional to expected costs; as such, more
time and effort are spent on adopting platforms that are perceived to be genu-
inely useful. This might seem to be a trivial statement but, in fact, it supports
the idea of showing social activity to prospective users in an attempt to per-
suade them to join, which calls for a different strategy to the walled garden, or
bowling pin, strategy often applied to explain Facebook’s success (e.g.,
Spulber 2010)
107
.
Second, the real-time problem can be approached through the concept of
synchronicity. Porter (2004) defines it as the “degree to which a medium ena-
bles real-time interaction”. She distinguishes between 1) synchronous and 2)
asynchronous interaction; the former requires that parties are present simulta-
neously, and therefore corresponds to the definition of the real-time problem.
Asynchronous platforms, such as online forums, enable users to browse and
create messages at their convenience. Porter (ibid.) mentions that a particular
interaction design does not necessarily lead to interactivity among users, as
users might not behave as expected. Hence, the real-time problem can arise if
users are not utilizing real-time features to their advantage. The requisite of
timeliness makes it much harder for the startup to generate matches (see Sub-
chapter 4.5.1) than in a registration-based system.
According to Caillaud and J ullien (2003), users might utilize several service
providers in a case where one platform does not provide a match. Hagiu
(2006) employs the example of Match.com to elaborate the need for registra-
tion and data collection necessary to create matches through some form of
permanence. Overall, when introducing social interaction, as opposed to con-
tent, and time, as opposed to constantly available content, the chicken-and-egg
problem assumes a more complex form.
There are also limitations of network effects both as a concept and as an
automatic solution to chicken-and-egg problems in two-sided markets pre-
sumed by some startup founders. A definition, in the context of social net-
works, is given by Mital and Sarkar (2011, 380): “the probability of a new
user subscribing to an application is proportional to the number of the appli-
cation’s existing users. Thus social networking sites exhibit network effects”.
Network theory has created many constellations of network value. One of the
best known is Metcalfe’s law, stating that the value of a network is propor-
tional to n
2
, where is n is the number of nodes connected in the network
107
Essentially, the question relates to two alternative choices: 1) hide the interaction or show it and
2) let the content be indexed by search engines or not. What has worked for Facebook might, in fact,
be countered by the totally different open garden strategy. As such, this is a vivid example of context
influencing the implication.
126
(Gilder 1993). Later, the law was criticized for the fact that not all connections
are in active use (Briscoe et al. 2006).
For example, Odlyzko and Tilly (2005) refer to the concept of gravity,
which means that local connections are more valuable than those that are more
distant, basically negating the assumption of uniform value. As Samanta
(2009, 3) notes, simply multiplying the market sides is not a realistic measure
for network effects: “in a large network, such as the internet or a credit card,
with billions of potential transactions between buyers and sellers, most are not
used at all. Therefore, it would be wrong to assume that the volume of trans-
actions per buyer will grow linearly with an increase in number of sellers”.
The aim here is not to go too deeply into the discussion of the nature of “net-
work laws”, but to consider how they apply to online platforms.
In fact, the network value debate parallels that in the platform literature. As
such, calls for other criteria in addition to network size have been made by
several authors (e.g., Farrell & Klemperer 2007; Suarez 2005; Birke 2008). In
particular, economists have discovered the limitations of employing the size of
a user base as the predominant proxy of network effects. In his literature sur-
vey, Birke (2008, 24) argues that
"A […] departure from the assumption that total network size
matters can be found in some of the newer empirical papers on
network effects which argue that social networks are mainly
local and that local geographical network size is therefore the
relevant network measure."
Consider that network effects are dependent on contextual factors such as
an industry or market, even a company, as a user base of one company, can
interact more than that of another company and therefore be more valuable.
Suarez (2005, 719) examines network effects at the industry level and argues
that
"[A]n industry that features very strong ties could simply annul
classical network effects […] an industry with moderately strong
ties may allow for both strong ties and classical network effects
to be significant. Finally, an industry in which weak ties pre-
dominate would de facto revert to the classical case, a monolith
that cannot be broken into parts on the basis of tie strength."
Ties are implied by Suarez to increase change resistance (see Coch &
French 1948) or switching costs (Farrell & Klemperer 2007). It is simply
possible that existing industry relationships, or ‘inertia’, exceed the expected
benefits of network effects. In online consumer platforms, barriers for switch-
ing might also involve habitual or behavioral elements relating to learning
costs (Farrell & Klemperer 2007) and loyalty; the latter of which being possi-
bly based on seemingly irrational logic such as brand preference (see
127
Thompson & Sinha 2008). For example, Maicas, Polo, and Sese (2009) refer
to personal network effects which underscore relativity. According to Maicas
et al. (ibid.), personal networks influence switching behavior. Intuitively, it
can be concluded that personal networks are most likely to affect adoption and
usage of a social platform.
In particular, the limitation of network effects in social platforms is noted
by Boudreau and Hagiu (2009, 171): “members care only about their relevant
network rather than the aggregate network [so] growth is about expanding a
mosaic of social networks rather than scale per se.” This argument will be
discussed below. Second, consider Evans’ (2002) requirements for network
effects: 1) one agent’s adoption of a good, product, or service benefits other
adopters, and 2) “his adoption increases others’ incentive to adopt.” Evans
(ibid.) refers to these two effects as the total effect and marginal effect. How-
ever, both assumptions can be contested in online contexts. First, consider in-
clusion of negative network effects. An example is mentioned by Boudreau
and Hagiu (2009, 171): “Facebook has the challenge of minimizing negative
interactions on its platform, ranging from irrelevant interactions, those that
are inappropriate to the context, all the way to ‘fraudsters’ and illicit activ-
ity.”
Clearly, all platform types considered in this study are subject to negative
externalities emerging from low-quality participation: in content platforms, it
is a risk of spam and low-quality content, in social platforms the aforemen-
tioned negative interactions, and in exchange platforms, the risk of frauds and
scams. Consequently, a startup is obliged to monitor the quality of its user
base. In the ideal model, this is assumed to be the task of the self-organizing
user base. However, as argued, the ideal in real cases rarely occurs. In fact, the
quality of a platform’s user base can be defined as a determining factor of in-
teraction.
For these reasons, individuals might prefer to keep their personal core net-
work small (Mital & Sarkar 2011). This, again, is not taken into account when
modeling only positive network effects and network size. In fact, the reverse
can take place: privacy can explain why some social platforms are more suc-
cessful than others, and why open inclusion is not always the optimal choice
(cf. Boudreau 2010). If network effects are tied not only to quantity, but in-
volve qualitative aspects, the idea of benefits being proportional to thenumber
of users is not fully compatible with the notion of the positive nature of net-
work effects. For example, consider spam (i.e., unwanted email messages):
when it increases due to more participants joining email markets, the basic
premise of network effects would imply that user benefits increase, but clearly
this is not the case. Even by restricting the application of the premise to
128
positive network effects, the implications might not be as fruitful as originally
thought.
For example, when a large number of medium-quality content is produced,
the marginal effect on overall benefit is much smaller than if the same content
was of high quality, or, if medium-quality users join a network, other prospec-
tive users are less interested than if high-quality users join
108
. Therefore, alt-
hough strictly speaking “true”, and the network benefit is added proportion-
ally, the proportion is mediated by quality. Obviously, quality is hard to de-
fine
109
, especially as it can differ according to the preferences of individual
users. In sum, applying the notion that network effects increase in proportion
to quantity, of users or content, can be regarded as unnecessarily delimiting
110
.
Although the economic literature tends to focus on positive network effects
(see e.g., Birke 2008), users in online markets are easily affected by negative
effects such as the aforementioned spam (Hinde 2003) and harassment in so-
cial platforms, cluttered or obtrusive advertising on content platforms (Rumbo
2002), and the risk of fraudsters in an exchange platform. Therefore, the issue
of negative network effects is crucial in the online context, and should be un-
derstood as a potential cost for adoption, or even a barrier.
Moreover, the relationship between installed user base and propensity to
adopt is not straightforward. However, heterogeneity of a given user base can
impose strong effects on a user’s willingness to adopt. For example, consider a
20-year-old user who is looking for a date in a large network, A, but each and
every one he/she finds is unsuitable. Then he/she switches to platformB with a
user base of less than half of network A’s, and finds a date immediately.
Clearly, there are noreal network effects for her in network A, even if it has a
larger user base than network B. Of course, matching involves a possibility of
chance, but perhaps network A was targeting elderly users of age 60–70, and
this did not match with the searcher’s intent. Regardless of the reason, it is
insufficient to assume that network effects are uniform across all users, and
that they originate only from the size of the user base.
Similarly, we can consider a one-sided platform where convention argues
that adding a new user benefits other users of the same type. However, by ap-
plying the previous logic from a two-sided market, the marginal increase of
network utility by adding user x in sidek of the platform for user y in network
sidef (i.e., the opposite side) will only be positive if the additional user is a
108
Note that the medium condition is deliberately framed, because low would indicate a negative
network effect, whereas medium is a small positive network effect.
109
See Reeves and Bednar (1994) and Zeithaml (1988) for some attempts.
110
According to the author’s understanding, however, the misconception is not due to the investors
of these laws but to the fact they have been later redeployed to contexts missing fit. For example,
Metcalfe’s law originally described Ethernet connections, not social networks.
129
suitable match (e.g., of the correct kind, quality, type, or age). In other cases,
the increment is either neutral
111
or negative. For example, Farrell and
Klemperer (2007, 1974) consider that
"sers of a communications network or speakers of a lan-
guage gain directly when others adopt it, because they have
more opportunities for (beneficial) interactions with peers."
This condition, albeit intuitively making sense, only applies if individuals
are interested in communicating with speakers of other languages. If one has,
for example, learned a language only to communicate with a spouse, which
might be the case in international marriages, again, adding new members will
not increase the network utility in praxis
112
. Additionally, note the concept of
‘peers’ in the definition. It might be that the new speaker of the language is a
baby on the other side of the world; clearly, proximity in the network influ-
ences how realistic it is that network effects become useful.
Even in electronic networks, such as the Internet through which all users
have a theoretical ability to connect with everyone else, private social net-
works are often concentrated, for example, by region, age, or preferences
(Thelwall 2008), as if to mirror offline proximity conditions. We rarely see
random connections in social life, but they havepurpose that leads to unique,
and perhaps unpredictable, network topographies also in a relatively friction-
less network, such as the Internet. In other words, we cannot automatically
assume that the existence of network size suffices for adoptionper se, and that
if it suffices for user x, it will also suffice for user y. Therefore, the ‘chicken’
might not purely be in the size or structure of the network, but in the underly-
ing differentiating factors
113
of targeted users such as, for example, age, loca-
tion, or preferences.
Banerji and Dutta (2009, 605) come to the same conclusion by stating that
“positive externalities arise from the specific patterns of interaction between
groups of users” as opposed to those that are general. However, given indirect
network effects, the issue is slightly more complicated than this. While direct
utility derived from a distant user might be zero, or diminishingly low, the
111
In fact, from the three types of network effect (i.e. positive, neutral, and negative), neutral,
which is when the user is indifferent to the other side, can be subtracted. As an aggregate condition
this cannot apply, otherwise all users would be indifferent and there would be no interaction, and,
thus, no platform. However, neutrality intuitively applies at an individual level. For example, merely
adding some random member to Facebook is unlikely to increase a user’s interest in the platform.
112
Note that on average, someone from the language community is likely to be interested in
communicating with the new person; therefore, in aggregate, the benefit (slightly) increases.
However, this does not affect the skeptic’s or spouse’s benefit from the network, as it is not the
network that made him/her adopt the language.
113
In fact, even in Farrell and Klemperer's (2007) definition, this is considered beneficial. However,
this elaboration is included to highlight the nature of “beneficial”, to avoid the misinterpretation of
size of user base as being synonymous with success in achieving network effects.
130
indirect effect of having a lot of “zeros” adds up to a user base, which provides
useful network externalities even when a user is interested in interacting with a
selected few. This is elaborated by Arroyo-Barrigüete et al. (2010, 646) who
employ the example of Microsoft Messenger in stating that “at a global level,
[adding new irrelevant users] would have an influence due to the fact that, if
there are a large number of users, the system will be improved over succeed-
ing versions (indirect network effects).” Hence, the pool of other users, alt-
hough individually meaningless to the user, is beneficial as a group.
As in the previous example, locality (in a geographic sense) can be associ-
ated with social factors and therefore to the strength of network effects. Suarez
(2005), referring to Rogers’ (1995) innovation studies, asserts that a local
community might refuse to adopt a technology with a larger overall user base.
This seems to confirm the idea of the transferability problem from one context
to another: a local optimum does not generalize to a global optimum. In other
words, as observed by some startups in the sample, major dominance in one
market might not endure when expanding to other markets. Again, the size of
user base does not matter per se, and a smaller community with high
consistency
114
can endure external pressures to adopt technological innova-
tions.
Moreover, Suarez (2005, 712) claims that thestrength-of-ties perspective is
commonly employed in social network theory to imply superiority or inferior-
ity of connections, so that “relationships among the different actors in a net-
work can be broadly classified into some basic types: strong versus weak ties
and direct versus indirect ties.” If adopting this perspective, weak ties would
seem to indicate less significance to platform adoption
115
. This principle was
also observed by Rohlfs (1974): “If an individual’s demand is contingent on a
few principal contacts’ being users, there may exist many small self- sufficient
user sets.” Indeed, users behave differently than telephone networks, in which
the total utility provided can be parallel to the number of connections it is able
to create between randomly connecting users. However, even there the usage
of those connections is not uniform, or totally random; for example, some
connections are more actively utilized, whereas others are more rarely uti-
lized
116
. Therefore, there is purposefulness behind the usage of pre-existing
connections; similarly, there is determinism in the way users generate
114
Defined as strong network effects, and no multihoming.
115
However, Birke (2008) gives an example concerning why the availability of remote nodes is
important. Consider, for example, the emergency number; although rarely utilized, access to it is
important to users of the telecommunications network.
116
“Texas and Maine may have less to communicate”, as noted by Briscoe et al. (2006).
131
connections
117
, such as friendships, in social platforms. For example,
friendship requests from unknown individuals tend to be rejected on Facebook
(Pempek, Yermolayeva, & Calvert 2009). Alternatively, the reverse can occur,
and the user actively employs social platforms to find people previously
unfamiliar to him or her. It is this ‘hidden intent’ that complicates modeling
network effects.
Consequently, it is important to note that network effects areone criterion
for adoption, and therefore cannot fully explain it. The superiority of a product
can explain cases such as Google overcoming Yahoo, and Facebook over-
throwing MySpace, even in the presence of network effects. This principle is
mentioned by Farrell and Klemperer (2007, 2012) who elaborate the Qwerty
versus Dvorak case by asserting that “f the penalty is small, switching […]
could be privately inefficient for already-trained QWERTY typists even with-
out network effects. And evidently few users find it worth switching given all
the considerations including any network effects.” In a similar vein, Suarez,
(2005, 711) notes that an “obvious explanation” for deviations from expected
network effects are other factors, such as price or technological characteristics.
In other words, network effects are not the only source of benefits derived
from adopting a platform; the platform might have stand-alone value
(Kristiansen, 1998), or its adoption is the consequence of word-of-mouth.
Thus, adoption does not result from network effects but from social effects,
such as the bandwagon effect (Henshel & J ohnston 1987), or from the plat-
form’s stand-alone value. This suggests startups should not overly rely on
network effects as the ultimate goal but also focus on other areas, such as
technology, differentiation, and marketing. All of these are potential variables
explaining network effects, as opposed to network effects magically appearing
from user interaction. This separation principle is at times forgotten also in the
literature, which considers network effects to fully explain adoption while
wondering why, in the presence of strong network effects
118
, one competitor
overcomes another.
The confusion between differentiation and network effects might arise from
the specifications of analytical models; for example, differentiation is not al-
ways considered. In such models, network effects portray a situation in which
two equally differentiated and equally marketed platforms are competing,
which might not always be the case in the real world, by pricing and size of
117
Rochet and Tirole (2005, 5) stress this point through the concept of usage: “The cardholder and
the merchant derive convenience benefits when the former uses a card rather than cash; a caller and
a callee benefit from their communication, not per se from having a phone; and so forth.”
118
This is a special problem of applying the formal models to empirical contexts. In a formal sense,
strong network effects are defined so that “[n]etwork effects are strong if they outweigh each
adopter’s preferences for A versus B, so that each prefers to do whatever others do” (Farrell &
Klemperer, 2007, 2018).
132
user base. Logically, such models tend to give results that favor the im-
portance of network effects.
It is possible to take alternative approaches to defining network effects; for
example, consider a proposed definition:
"In general, the higher the strength of network effects in a two-
sided platform (defined as propensity to find a match), the
smaller the initial user base required to grow."
This definition overlaps with viral marketing theory, in which the growth
idea is approached, for example, through the concept of aviral coefficient (see
Salminen & Hytönen 2012). In such a model, the propensity to send and ac-
cept invitations defined whether the network experiences exponential growth.
Coincidentally, perceived network effects would affect both propensities, so
that users are more likely to send invitations, as added users increase their
benefit, and accept them because of the benefits provided by the existing net-
work. Again, however, we stumble upon the ‘minimum requirement’ concept
(i.e., critical mass), as the expected benefits of the network are considerably
lower if the user base is insufficient to convince invitees to join. Defining the
network effects as the propensity of a user to find a match relating to his/her
intent avoids the ‘quantity versus quality’ problem. In this case, both can con-
tribute positively or negatively to the emergence of network effects.
The idea of propensity (i.e., probability) is somewhat compatible with
Roson (2005) who argues that, from a demand perspective, two sources for
network externalities can be identified: 1) single interaction externality, in
which “matching quality improves when more alternatives become available”
and 2) multiple interaction externality, in which every user gets a benefit from
every interaction by other pairs. However, Roson (ibid.) also assumes that
quality is improved by the number, not type, of participants.
Finally, we consider the notion of critical mass, as it can be regarded as
requisite for the presence of network effects. A two-part definition is offered
by Suarez (2005, 718). First, “A critical mass occurs at the point at which
enough adopters have chosen a particular technology that the technology’s
further rate of adoption becomes self-sustaining.” This indicates self-propaga-
tion that was found to be central in the ideal UG model. Suarez (ibid.) adopts
the second part from Katz and Shapiro (1992): “the system with a lower in-
stalled base enjoys a significant advantage for instance, newer and superior
technological capabilities”. Therefore, incumbents are seen to possess excess
inertia (Farrell & Klemperer 2007) regardless of their quality. There are recent
cases in the online market, however, that neglect this ‘inertia’. Most currently
dominant platforms employed as examples in this study have been
133
early-movers but not first-movers, first-mover disadvantage can arise, for
example, as a result of technological inferiority
119
. Evans and Schmalensee
(2010, 21) show that “even without fixed costs or economies of scale, platform
businesses typically need to attain a critical mass when they are launched in
order even to survive”, which would indicate that a ‘go big or go home’
strategy is suitable for these markets, and that heavy investments in early
marketing to acquire a user base would be required. A critical mass can
therefore be defined as the condition between functioning (i.e., realized) and
non-functioning (i.e., theoretical) network effects, so that:
No critical mass ? no network effects
Critical mass ? positive/negative network effects
This is based on the assumption that the network must have some type of
minimum participation, not necessarily relating to quantity, before it can pro-
duce benefits or matches for users. This is also the position of Shapiro and
Varian (1998, 184) who argue that: “Network externalities make it virtually
impossible for a small network to thrive. The challenge is to overcome the
collective switching costs that such a network requires to grow.” In the net-
work economics literature, successful achievement of network effects, in a
competitive setting with no multihoming, is often termedtipping (e.g., Katz &
Shapiro 1994), meaning that once a rival technology reaches a particular
degree of adoption, all industry participants migrate to support that technol-
ogy. Such is the case concerning standards (e.g., Farrell & Saloner 1985).
When the winning design emerges and industry participants become aware
of its predominance, they will start supporting it and abandon other standards
(Katz & Shapiro 1994). Therefore, multihoming can exist, in this type of set-
ting, only until the dominant technology has been decided, after which all
players will single-home. However, if there is interoperability between tech-
nologies, which is not the case for mutually exclusive standards, the behavior
might be different and the market might eventually become oligopolistic; that
is, comprise many standards or platforms (e.g., Hagiu and Wright 2011). More
precisely, this can be seen to be the case for many online platforms that all
have internal consistency in terms of critical mass (i.e., users who are com-
patible with each other to the extent that matches can easily be created), but
none has an absolute dominance of the market
120
.
119
Technical problems are commonly acknowledged as one of the reasons for Friendster losing to
other social networks.
120
For example, there are several competing online dating platforms, which is possible due to users
multihoming or otherwise preferring one platform to another.
134
However, at the same time, this is contrary to Shapiro and Varian’s (1998)
argument, which supports the notion that industries with strong network ef-
fects gravitate to the leading platform. The difference is in the notion of ‘in-
dustry’ versus ‘market’, and standards and compatibility. If we change the unit
of analysis from an industry, in which it makes inarguable sense to employ
compatible technology through standardization, to different market verticals,
we can better understand the outcome in the case of many two-sided markets,
including those online.
Nevertheless, a critical mass cannot be seen to equal stable market domi-
nance, although it might imply niche dominance, assuming lower competition.
Niches, therefore, are local targets as opposed to mass markets. In the study’s
sample, a startup faced special problems of hyper-local communities (see the
previous chapter). In particular, the issue is that dominating one local niche
does not automatically lead to advantages when moving beyond the niche; that
is, network effects do not generalize. This becomes a problem when a partic-
ular niche is insufficient for viability or the goals of the startup.
Thus, thenon-transferability of network effects can become an issue in the
differentiation strategy, if segmentation is too narrow, or if markets are natu-
rally isolated (e.g., cities in some circumstances) or exhibit social dissimilari-
ties. Hence, the existence of the transferability problem, as described in the
definition, has been noted in the literature.
4.5.3 Solution: Remora
As noted in Subchapter 4.5.2, the practice of a platform drawing benefit from
the user base of another platform bears similarity to envelopment in the plat-
form literature. It is described as follows (Eisenmann et al. 2011, 1271):
"Envelopment entails entry by one platform provider into an-
other’s market by bundling its own platform’s functionality with
that of the target’s so as to leverage shared user relationships
and common components. Dominant firms that otherwise are
sheltered from entry by standalone rivals due to strong network
effects and high switching costs can be vulnerable to an adjacent
platform provider’s envelopment attack."
Envelopment, in this sense, is an aggressive strategy that aims to replace the
target platform by making its pre-emptive assets, namely high switching cost
and strong network effects, ineffective. However, as previously noted, the
remora strategy aims at becoming a complement instead of a substitute. Due to
power imbalance in a remora setting, the remora is not aiming to replace but
accommodate the host (see Figure 12).
135
However, Eisenmann et al. (2011) name the following industry examples of
envelopment: PayPal ? eBay; Google Docs/Chrome/Android ? Google
search. Therefore, envelopment can take place as a market diversification
strategy in which the platform draws from its own user base in a different ver-
tical. Clearly, Google is a strong example of such an expansion, as it currently
provides more than a dozen services that more or less relate to its core func-
tionality (i.e., search). In a similar vein, Ries (2011) notes that “many […] vi-
ral products didn’t really build their own working ecology: they colonized
someone else’s. That was true for PayPal cannibalizing eBay, YouTube and
MySpace, and could still be true of Slide, Zynga, or RockYou – we’ll see.” The
“colonization”, as argued, is beneficial for both parties in the case of comple-
ments.
Even in eBay’s case, PayPal provided a useful auxiliary service that was
missing from eBay’s own selection. eBay tried to introduce its own version,
termed BillPoint, which is an example of substitution by the host, but failed
due to better marketing strategies employed by the remora. According to Mas
and Radcliffe (2011, 311):
"PayPal faced a constant fight with their ecosystem host, eBay,
once eBay realised that some of the value from their customers
was going to PayPal. As the owner of the platform, eBay sought
to derive significant advantage from integrating its own payment
engine into its marketplace website. […] Ultimately, PayPal
knew that their continued success was going to be dependent on
eBay’s not shutting them out of their auction website entirely, so
they sought increasingly to diversify from eBay auctions […],
and eventually sold out to eBay."
This case demonstrates well the hazards of a remora. First, competition
from the host platform that, after discovering the remora is grabbing a dispro-
portionate amount of its user value, reacts by introducing a substitute, then, the
danger of being denied access and, finally, being absorbed by the host, in a
form of acquisition. It also demonstrated diversification as a means to counter
the threat of the host, which will be discussed in the following chapter. Nota-
ble industry cases include Twitter’s acquisition of Tweetdeck (Parr 2011b) and
Facebook’s acquisition of Instagram (Constine & Kutler 2012). In Facebook’s
case, its strategic goal was to make the acquired company a complement for its
own platform, thus enabling better photo sharing. In Twitter’s case, the ra-
tionale had to do with substitution. In other cases, the remora is not so lucky,
with the host eventually absorbing its product ideas as features in the subse-
quent release.
Eisenmann et al. (2009, 225) elaborate the problem of substitution-through-
absorption:
136
"Dependency also came with a danger […] Many software exec-
utives wondered if they could trust ambitious Microsoft employ-
ees with sensitive information. Executives […] had seen the in-
novative features of prior software show up as features in later
versions of Microsoft’s products [and] wondered if their em-
ployees’ conversations with Microsoft’s technical staff would
contribute to seeding a future competitor."
Whereas, in a standards and technology setting, a firm can protect its intel-
lectual property rights (Church & Gandal 2004), abstract ideas are not patenta-
ble in most countries. Features absorbed by Microsoft include, according to
Parker and Van Alstyne (2008), disk defragmentation, encryption, media
streaming, which is also employed as an example by Eisenmann et al. (2011),
and Internet browsing. Absorbing has also been documented in the Web envi-
ronment (Parker & Van Alstyne 2008, 2): “Whether through internal develop-
ment or acquisition, coercive or not, platforms such as Apple, Facebook,
Google, Intel, Microsoft, and SAP have routinely absorbed valuable features
developed by ecosystem partners.” Parker and Van Alstyne (2008) go on to
state that absorbing developers’ innovations can reduce their incentives to
continue developing for the platform, given they remain uncompensated
121
,
and can lead to developers exiting the platform. A case of developer flight in
the videogames market is described by Eisenmann et al. (2009, 151):
"If an incumbent has been too aggressive in extracting value,
demand- and supply-side users may rally around entrants […]
When it dominated the console market, Nintendo dealt with
third-party game developers in a hard-fisted manner. Conse-
quently, developers were pleased to support Sony when it
launched the PlayStation console in 1996."
Consequently, it is a useful tactic to attract remoras for the host not only be-
cause they increase network benefits for the demand-side, but because they
can provide ideas on how to improve the core platform. At the same time,
over-exploitation of their ideas provides a negative incentive to continue col-
laboration. A similar effect can take place in the demand-side where, for ex-
ample, privacy issues might become a concern
122
.
Exclusion, or denial of access, is similar to a counter-envelopment strategy
depicted by Raivio and Luukkainen (2011, 79): “The envelopment threat re-
fers to a case where we have several platform providers and common
121
In contrast, some startups welcome acquisition as an exit strategy, and therefore cases of
absorbing-through-acquisition would increase the incentives of such startups to join.
122
This was noticed by Facebook as a competing project termedDiaspora gained much success in a
crowdfunding platform; Diaspora aimed to attack Facebook’s privacy-related weaknesses by
promising a more secure platform, although it later failed.
137
customers. In this situation one provider may try to exclude other platforms
from the market.” It might not necessarily be that the platforms are competing
in the same market, but if the host perceives the remora taking advantage of it
without reciprocity, it can resort to denial of access. For example, such was the
case of Craigslist (i.e., general exchange platform) cutting access of AirBnB
(i.e., specialized exchange platform), a peer-accommodation service
123
. In
particular, the value of content can be seen in a platform strategy termed
walled garden (e.g., Berners-Lee 2010), in which the platform owner restricts
the visibility of information to search engines
124
that are known profiteers of
capturing economic value from third-party content.
According to Eisenmann et al. (2009), specific results from the hold-up
problem might include 1) limiting quality of cross-platform transactions, 2)
raising prices, and 3) charging for interoperability rights. In particular, the
platform host might deliberately “limit the quality of cross-platform transac-
tions to maintain differentiation” (Eisenmann et al. 2009, 138). Charging for
interoperability rights as a strategy for the host to improve its margins can in-
clude charging startups for API units
125
. In turn, a startup is initially a weak
platform that hinders its ability to utilize the same strategy of monetizing API
usage. The following figure illustrates how the standard remora strategy dif-
fers from envelopment.
Figure 12 Remora and envelopment
Consider that, as soon as the number of complements exceeds a particular
threshold (e.g., they fulfill categories needed for inter-platform competition),
the group itself is enough for the platform owner, and individual developers
become expendable. This logic can be employed to explain power asymmetry
123
In this case, cutting access to posts would negateaggregation, whereas preventing auto-posting
would remove automatic access to users.
124
For example, most Facebook content cannot be accessed by Google; therefore, Google started its
own social networking platform, Google+.
125
For example, such a practice is operated by Google (i.e., Google AdWords).
Envelopment
Remora
138
between the developer community, which is fragmented and competing, and
the platform owner that is concentrated and a monopoly within the platform.
When dissatisfied, the host can more or less replace individual complements
as new entrants are interested in taking their place. However, if the entire pop-
ulation turns against the host for some reason, the momentum will be reversed
and the host will quickly lose popularity. In the case of envelopment, the rela-
tionship is more hostile, and the counterparty is interested in replacing some or
all of the host’s functions.
This chapter has shown how remora’s curse relates to the literature. We
have employed theoretical constructs such as hold-up to understand the prob-
lem, and cited industry examples of both the remora strategy and the host’s
opportunistic behavior. However, we have also argued that the host’s oppor-
tunism is curbed by 1) multihoming behavior and 2) the risk of adverse selec-
tion to restrain its use of power.
4.5.4 Discussion
When the number of users is zero and the benefit derived from the product by
the user depends on the number of other users (i.e., direct network effects
apply), the benefit provided by the product to additional users also equals zero.
Therefore, no rational users join. This notion is applicable to both one- and
two-sided markets
126
. Within two-sided markets, under indirect network ef-
fects, when users of one (i.e., supply) side have no match from the other (i.e.,
demand) side, there is no incentive to join, regardless of the number of other
users in the same side, andvice versa.
As the number of relevant contacts increases, so does the perceived benefit
of joining the network, even up to the point where the user feels social pres-
sures to join. Both buyers and sellers need to be present in an exchange plat-
form, or marketplace, otherwise the service provides no benefit for the user. In
contrast, when users fail to actively utilize the platform, it becomes a cold
platform, which explains why MySpace lost to Facebook, regardless of the
critical, although static, mass acquired. It also implies that a critical mass
should not be measured by the number of connections (i.e., Metcalfe’s law),
but as the number of interactions across time. The number of interactions indi-
cates active use, which is as substantial a problem for a startup as the cold start
problem.
126
Their difference is that, in one-sided markets, users are of the same type (e.g., friends in a social
network), whereas a dual-sided market addresses two different types of user (e.g., sellers and buyers)
who are influenced by one another.
139
The literature survey provided useful approaches to understanding the na-
ture of the lonely user dilemma. Central to this is that the first models focused
on how the network appears to an outsider, not how the connections are em-
ployed. It was later discovered that the structure and usage of a social network
are closely tied to one another; the connections emerge between people who
share interests or are also associated in offline interactions, which forms the
potential of the network usage. Consequently, realized, not theoretical, inter-
action is required for the UG benefits at which a startup aims. Inarguably,
without active use, the potential of the platform does not sustain, which is de-
picted as liquidity in the early platform literature and as theproblem of active
use here; possible even when the cold start problem is solved.
The problem of active use is, in fact, quite important as it relates to sus-
taining the perception of utility. While strong network effects can quickly
grow a user base, with low loyalty the platform faces considerable churn,
which leads to a reverse effect of exponential growth, a sudden decay. For ex-
ample, if lead users or top contributors of content exit the platform, their fol-
lowers might easily follow, after which the followers of those followers exit,
and so on. Diffusion and churn can therefore be perceived as being subject to
herd behavior. This obviously implies that the startup is advantaged by keep-
ing the community active, after it has been established.
The real-time problem adds a temporal component to the lonely user di-
lemma. Put simply, at any moment in time, there needs to be a critical mass of
users active in the platform. As the potential moments of time extend well be-
yond storing the interactions (e.g., messages saved in an inbox), potential con-
nections are much scarcer than in a non-real-time social network. Essentially,
the startup has a major coordination problem, as people do not necessarily
connect to the platform at the same time as other users. At best, this pattern is
hard to control
127
. The real-time problem concerning simultaneous entry of
sides is consistent with the literature. If the requirement of simultaneous pres-
ence is strong, there is no solution other than efficient coordination of user
flows; for example, based on the hour of day, day-part advertising, and focus-
ing on the side lacking members. There is likely to be a threshold up to which
the lack of counterparts is tolerated, with the user retrying at a later time.
Additionally, even the usefulness of content can have a temporal nature, be-
cause new and different content is required by new and different users. Re-
gardless of this “timeliness of utility”, the lifetime of content can be regarded
as long, even infinite. While it remains in the Web server, it exists, as opposed
to users who log out, and can be indexed and shown by search engines,
127
An exception would be when the platform is tied to a specific event (e.g., a football game); then,
all participants would self-coordinate to be simultaneously present.
140
thereby providing visitors with consistent value
128
. Obviously, some pieces of
content (e.g., “classics”) hold their utility longer than others.
Moreover, it was found that the conceptualization of network effects,
namely their contrast to the size of the user base, might be problematic. Un-
derstanding the motive to join seems to be associated with deviations of eco-
nomic rationality, even if optimizing social utility replaced profit-seeking.
Therefore, altruistic behaviors and group dynamics can come into play. The
critical mass phenomenon, the author believes, cannot be explained only by
network effects and the growth of utility. Other phenomena such as the band-
wagon effect and herd behavior are most likely involved. Whether they are
perceived as rational or not, is a different matter.
If network size is not the correct metric to indicate network effects, and lo-
cality has been perceived as more important, what does this imply? In a sense,
that 1) expectation of close proximity (e.g., city; local community) is useful
for the startup, but 2) under UG, the network structure spawns from the user
base. These are complementary remarks in the sense that, given the infor-
mation based on which one can assume greater consistency among the
network unit, the startup should target that unit as opposed to employing mass
marketing, even when mass markets are the goal
129
Relating to other reasons for adoption, consider a user joining the dominant
platform instead of a startup platform. In some cases, the motive for non-
adoption might not lie within the stronger network effects of the dominant
platform but the in the fact that it offers a better solution. This is an important
notion, which arises from the marketing assumption that people join platforms
to satisfy a need, as opposed to the platform assumption that they join because
there are network benefits. Here, theories can confuse the direction of causal-
ity; namely, that network effects would be the sole motive for avoiding
switching while, in reality, loyalty is a more complex phenomenon
130
. As
noted in Subchapter 4.5.1, loyalty is a critical factor in sustaining the level of
active use, which is also a requisite for the realization of network effects.
The platform literature refers to marquee users, whereas this type would be
described in the marketing literature as opinion leaders, individuals character-
ized both by a large number of connections and their associatedsocial capital.
However, the essential conclusion is that marquee users are not only able to
increase the network effects of a given platform but regular “J ohn and J ane
128
An illustration of the lifetime value of content can be seen in J acob Nielsen’s (1998) article, in
which he describes how visitors to his popular blog articles have increased over time.
129
However, for example, mass advertising can bring legitimacy to some platforms, which is
especially important in the B2B context. An interesting case study on the topic can be found in Mas
and Radcliffe (2011).
130
Other criteria such as usability, trust, and features of the platform are likely to affect
adoption/loyalty.
141
Does” can increase the utility of the network as they provide a better match for
other “regular” users, which contrasts somewhat to the traditional marketing
convention of targeting ‘lead users’
131
. Network effects are a coordination
game, not a pure game of numbers.
In sum, through a literature survey, it has been shown that there are specific
cases in which quantity does not apply as a proxy to network effects. These
include, at least, the following:
· The presence of negative network effects (e.g., spam; socially
undesirable connections).
· Whenever the user base is non-homogeneous; that is, in most cases of
content and social connections in which both access and quality of
interaction are important.
· When there areother motives for adoption, so that network effects do
not explain platform adoption, switching, multihoming, or refusal to
adopt to the full extent.
Thus, the effectiveness of network effects in accumulating a user base de-
pends on circumstances such as industry, quality, interoperability, and multi-
homing behavior. Users act purposefully and are influenced by social motives;
modeling their connections without taking motives into account portrays an
inaccurate image. The author has also proposed an alternative approach that,
rather than merely the number, involves the probability of matching as a proxy
for network effects. This approach is conceptually more accurate as it gener-
alizes across a large variety of perceptional factors that influence network ef-
fects such as, for example, demographics, preferences, intents, and location.
In brief, it can be concluded that not only the number of connections is of
consequence, so is their interdependence. This can be termed quality or rele-
vance, depending on the conceptual perspective. However, it is fundamentally
linked to the fact that people value some social connections more highly than
others, and that some connections are more frequently utilized than others.
Understanding these statements from earlier theories can provide useful
guidelines for Internet startups.
131
However, it is not suggested that targeting lead users would not be a worthwhile investment of
marketing efforts; simply that, in the context of network effects, their increment to the network utility
is proportional to the increase of adding probability of matches, not the size of their network.
142
4.6 Monetization dilemma
All companies at some point must start generating revenue to
remain viable. (Lincoln Murphy)
4.6.1 Definition and exhibits
The monetization dilemma occurs when a startup needs to decide whether to
offer its platform for free at the loss of business viability, or charge for the
access and/or usage at the loss of users’ willingness to join. In other words,
willingness to join (WTJ ) and willingness to pay (WTP) are in conflict.
The following table presents exhibits of the dilemma.
Table 18 Exhibits of the monetization dilemma
Exhibit
[1] "This post attempts to summarize the [startup’s] story: how we got to be the most heavily used
browser synchronization service in the world and yet still find ourselves pulling the plug."
(Agulnick 2010).
[2] "For four years we have offered the synchronization service for no charge, predicated on the
hypothesis that a business model would emerge to support the free service. With that investment
thesis thwarted, there is no way to pay expenses, primarily salary and hosting costs. Without the
resources to keep the service going, we must shut it down." (Agulnick 2010).
[3] "The thesis of our business model […] was that there was a need for video producers and
content owners to make money from their videos, and that they could do that by charging their
audience. We found both sides of that equation didn’t really work. […] Video producers are
afraid of charging for content, because they don’t think people will pay. And they’re largely
right. Consumers still don’t like paying for stuff, period." (Diaz 2010).
[4] "[E]ven if enough people wanted the product, the business model around it is something which
we haven’t been able to figure out. We have the product’s version 2.0 sitting ready […] but we
do not see a clear exit yet, so are hesitant to launch it. Being blogged about major tech blogs
[…] we already got that love. If we stayed out in the market more – we’d probably get more
‘love’. But ‘love’ can only keep the servers humming for so long
." (Anonymous founder).
[5] "The experience has made me ask myself almost every time I see a cool web app – ‘OK, but how
will it make money?’, and if it can’t, then it would not be more than a short-lived dream for its
founders and backers." (Anonymous founder).
[6] "I felt like getting into the monetization stage was going to be long and difficult. And it was one
of those businesses where I liked the idea, but I didn’t think about monetization before I started,
because it was kind of a sexy idea, for me at least. And, I got some traction. I ended up with a
few thousand subscribers in a few weeks with the help of some larger companies that were
helping me out at the time. And, I kind of realized that to make my first dollar was going to be a
long time away […]." (Warner 2012).
[7] "Despite having over 200 beta testers at launch, it proved difficult to convert them into
customers. My prices started at $10/month, and though in my eyes this was a bargain, my
product didn?t demonstrate enough value to enough of my market quickly enough to justify the
operational costs of the business and my personal expenses." (Newberry 2010).
143
As the monetization dilemma relates to generating revenue, it takes place
independently of the size of user base ([1], [2], [4], and [5]), regardless of how
substantial this is; therefore, it concerns even popular platforms, such as
Twitter
132
. Consequently, popularity among users does not automatically lead
into financial success, unless successful monetization occurs. Other key tenets
of this dilemma are that the willingness to pay (WTP) of users in online
platforms is low [3], monetization requires time and effort [6], and even low
prices may not be "low enough" to attain WTP [7].
The dilemma is based on two critical conditions, which here are termed the
payment and revenue hypotheses:
a. Payment hypothesis: If a startup offers a paid product, it acquires zero
or very few customers; the potential risk here beingillusion of free.
b. Revenue hypothesis: If the startup offers a free product, it earns zero or
very little revenue (i.e., problem of free).
Seemingly, the startup cannot win. The dilemma therefore addresses diffi-
culties of both direct monetization (i.e., impossible to gain users) and indirect
monetization (i.e., impossible to create business); the former being impossible
under the premise that users refuse to pay when charged for access or usage of
a platform, and the latter under the premise that the startup is unable to
execute a successful model of indirect monetization even after gaining users,
which leads to an unviable business in the long run
133
.
The following table displays a choice matrix of the dilemma.
Table 19 Monetization dilemma simplified
Paid product Free product
Result No users No revenue
Major underlying assumptions, in economic terms, include high substituta-
bility between products (i.e., competition), low switching cost
134
between
them, so that users can easily switch providers and therefore cannot be locked
in, for example, by ‘bait and switch’, and strict homogenous price sensitivity
132
Twitter’s monetization troubles are well known in the industry (see e.g., Wired 2008).
133
Although, in the short term, this problem can be removed through venture funding; the goal
being to “capture market, monetize later”. This strategy, however, depends on successful
implementation of the indirect monetization model, not assumed in the dilemma’s premises.
134
“A consumer faces a switching cost between sellers when an investment specific to his current
seller must be duplicated for a new seller” (Farrell & Klemperer 2007, 1977).
144
whereby all customers always choose the lowest price, so that users move
from paid to free. Further, it is assumed that differentiation has no impact on
WTP. If even one of these assumptions were incorrect, one of the hypotheses
would fail, and the dilemma would dissolve. Consider the assumptions: first,
substitutability refers to the possibility of replacing the startup’s product with
competing products, which might or might not be true, totally depending on
the product. If the startup has created something that no competitor can
replicate, its product cannot be easily substituted and it would then be fair to
counter the first hypothesis and assume that users would be willing to pay for
the product, assuming that the differentiating property is something they want;
mere differentiation is insufficient.
From this abstraction, the nature of differentiation can be deduced. Whether
or not there is differentiation is irrelevant unless its nature leads to positive
WTP. The payment hypothesis argues that WTP is zero, thus any condition
that negates this argument invalidates the dilemma. The premise of competi-
tion assumes an association between competition, differentiation, and WTP,
which are practical issues that the startup needs to consider in determining the
validity of the hypothesis in its case
135
.
Price sensitivity is a key assumption, and a simple game demonstrates its
meaning. Consider, for example, a game in which players can choose between
free and paid versions of a product. The free version is of inferior but suffi-
cient quality, whereas the paid version is of better quality although costly.
Both players are price sensitive; that is, price is more important to them than
quality.
Table 20 Price sensitive (both)
B
Free Paid
A Free 2, 2 3, 1
Paid 1, 3 1, 1
The paid version can be more desirable if price is not included. However,
because it is, the dreaded cost makes price sensitive users prefer the free over
the paid version. Both users want to avoid a situation in which they are paying
and the other one is free riding. The free rider’s payoff is 3, because the
135
This estimation is required ex ante as the startup needs to decide on the monetization model.
However, the decision is not necessarily irreversible, although moving from free to paid might be
more difficult thanvice versa.
145
paying party helps to keep the platform free. The dominant strategy for each
player is to move from paid to free (1 ? 2 and 1 ? 3). Hence, neither of them
will pay.
Now consider the reverse, when the other player (A) is not price sensitive,
but prefers the premium features of the paid version.
Table 21 Quality sensitive (A)
B
Free Paid
A Free 1, 2 1, 1
Paid 2, 3 2, 1
Because A is not sensitive to price but quality, and the paid version offers
better quality, A will always prefer the paid version, regardless of B’s choice.
This is similar to generators always preferring to produce content, except in a
standalone sense. Whereas generators derive benefit from the existence of
consumers (i.e., positive indirect network effects), paid users are indifferent to
free users. Free users, however, enjoy benefit from paid users, becausein the
long term they help to keep the platform free. Therefore, the free user’s payoff
is 3 when there are paid users. From A’s perspective, however, there is no free
rider problem as he/she is not price sensitive. He/she will not switch from paid
to free, and B will not switch from free to paid; thus, there is a stable Nash
equilibrium
136
. Consider, however, that the free users now have an incentive to
keep paid users on board. The startup can leverage this as a part of their UG
strategy. Moreover, it becomes amarketing problem for the startup to find us-
ers who are more quality- than price-sensitive, and a product-development
problem to provide such quality that satisfies them.
Second, low switching cost implies that any lock-in mechanisms are weak
and, if fees are introduced, users can easily switch between alternative plat-
forms. This invalidates ‘bait and switch’ type of solutions; for example, first
offering free product and then charging for it. Given price sensitivity and low
switching cost, the user would abandon the platform when fees are introduced.
In the case of online platforms, switching costs are typically reduced by high
interoperability (e.g., API access), advanced import/export functions offered
by platforms that exclude proprietary storage of data, and the relatively small
136
Neither player can improve, given the choice of the other player (Nash 1950).
146
learning curve of new platforms as Web services might follow the same con-
ventions (see Cappel & Huang 2007).
Third, price sensitivity implies that users choose the platform based on
price and will therefore prefer free platforms to paid ones, even if this means
sacrificing some quality or features, a type of behavior termedsatisficing (see
Simon 1956). Similarly, if a platform is initially free, to solve the cold start
problem, but later turns to paid, to solve the monetization problem, that is, ap-
plying the bait and switch strategy, users will exit the platform. Given the
number of substitutes, they will always have a fallback. Therefore, if the as-
sumption of price sensitivity is correct, the payment hypothesis is true. Price
sensitivity sets it so that users have low WTP. Note that these premises are
neutral with regard to multihoming (Armstrong 2006); users might multihome
or not, but the introduction of fees would prevent adoption.
Fourth, it must be noted that if a startup is successful in indirect monetiza-
tion, it will not fall into the problem of free and the dilemma will dissolve.
That is, indirect monetization is only a solution when it is successful; a priori,
this can be difficult to determine, which is why the author hypothesizes that
startup founders might be more likely to underestimate their ability to directly
monetize and overestimate their ability to indirectly monetize.
Moreover, the notion of competitiveness in the definition is important, but
only indirectly. As noted, competition only matters if the lack of it is not due
to a lack of demand; that is, in markets where there is no demand, there is no
market although, at times, startups plan to create new markets. Second, the
offsetting factor to competition if differentiation, by which firms are able to
overcome competition. However, differentiation only matters if the differenti-
ating factor is perceived as more beneficial by the user. Some startups are able
to articulate differentiation although in praxis the difference can be trivial to
end users.
Finally, the hypotheses are subject to false confirmation (see e.g.,
Chesbrough 2004), meaning that their truthfulness might be incorrect. Con-
sider a hypothesis that is rejected even if it is true
137
or, conversely, a hypothe-
sis that is accepted when it is false. This relationship between validation and
truth is depicted in the following table.
137
Depending on the ontological position, truth can be defined as an objective property of the nature
(e.g., a law of physics) or an interpretation of the ideal condition (i.e., “social law”).
147
Table 22 Truth and assumptions
Real i t y
Bel i ef s
True False
True True False positive
False False negative True
To elaborate, in some cases users might exhibit behavior that is non-price-
sensitive: behavior that exists both offline and online. For example, every day,
some users pay Spotify for access to music, Netflix for access to movies, and
Dropbox for virtual storage space
138
. In cases where there is a motive to pay
but the startups abandon the alternative based on the payment hypothesis, it is
subject to a special case of confirmation error (i.e., false positive; see Table
22) that the author terms illusion of free (see Chapter 6.4). Under this condi-
tion, the startup is choosing freefying by default as it makes the assumption
that, without facts and contrary to the truth value, users would not be willing
to pay for the use of product, content, or access to a platform
139
. Therefore,
there is a risk of false positive: Users are willing to pay for the product, con-
trary to the founder’s belief. This is quite a significant risk as a startup faces a
choice horizon that might include creating products for which a proportion of
customers would pay, which is an opportunity neglected in the case of the
false positive, and the choice horizon becomes constrained to free offerings.
Therefore, the false negative might not only harm monetization but lead to
neglecting the discovery of other types of product and market space, otherwise
termed missed opportunities.
As previously stated, the dilemma will dissolve if the startup is unable to
successfully implement the indirect monetization model, as Google achieved
by aggregating Web content. There are, however, empirical foundations (i.e.,
reported by failed founders) to believe that indirect monetization is not as
straightforward as many founders assume, which will be discussed in the fol-
lowing subchapter.
138
It must be noted, however, that none of these services incorporate UG as the content production
mechanism. Nevertheless, this is irrelevant with regard to the premise of WTP.
139
In brief, WTP is assumed to correlate with willingness to join. Were the reverse shown, there
would be grounds to discard the dilemma.
148
4.6.2 The literature
In addition to the scholarly literature, there has been substantial discussion
among startup practitioners on the sustainability of free models (see e.g.,
Murphy 2011; Chen 2009; Kopelman 2009). Therefore, both academicians
and practitioners acknowledge the problem. This subchapter will focus on the
conditions of monetization in two-sided markets or, as put by Evans (2003),
“internalizing the externalities”.
As noted, the problem of monetizing online offerings has been identified in
the literature. For example, Beuscart and Mellet (2009, 166) note that viable
businesses must be “built around free access to content and principal
services; economic strategies mostly depend upon the sites’ ability to monetize
their growing audiences”. Also in the platform literature, Rochet and Tirole
(2005, 6) state that “[m]anagers devote considerable time and resources to
figure out which side should bear the pricing burden, and commonly end up
making little money on one side […] and recouping their costs on the other
side.” In a similar vein, Teece (2010, 172) argues that “[w]ithout a well-
developed business model, innovators will fail to either deliver or to capture
value from their innovations. This is particularly true of Internet companies,
where the creation of revenue streams is often most perplexing because of
customer expectations that basic services should be free.”
The technological nature of the Internet, as a communication network, is
compatible with creating platform-type of businesses. Clearly, technology
provides an apt solution for coordination problems that are difficult to handle
via manual coordination and negotiation (see e.g., Argyres 1999), which in-
creases the feasibility of entry. Second, various open-source platforms com-
pete with commercial platforms (Mian et al. 2011). While open source as a
phenomenon generally increases social welfare (e.g., Lerner & Tirole 2004),
open-source platforms are perceived as substitutes to commercial platforms,
and therefore create pressure for free offerings. It is often assumed that the
marginal cost of distributing an information good is close to zero, or negligible
(e.g., Shapiro & Varian 1998; Niculescu & Wu 2013).
However, supporting users in a platform is not equivalent to this assump-
tion. Free users are associated with marketing cost (i.e., user acquisition), ser-
vicing costs (e.g., bandwidth), and support cost (i.e., customer service)
140
.
Therefore, although low, the marginal cost of free users is not zero (Murphy
2011). How close to zero it is depends on the cost structure of the individual
startup; in any case, subsidizing free users comes with a cost. Furthermore, the
140
Although it can be assumed that the ‘free’ property in a good makes an average user less
demanding compared to paying customers.
149
cost is directly proportional to adding users, which means that if there is an
exponential growth of users, the costs also grow exponentially. Hence, a suc-
cessful startup, in terms of user growth, typically requires external funding as
it needs to cover expenses from growing the user base without being able to
monetize it directly. This has been the case for most of the famous Internet
platforms: Facebook, Google, Twitter, and even PayPal (Mas & Radcliffe
2011). Consequently, basing the free pricing strategy on the information-
goods argument does not appear to be valid.
Nevertheless, there are properties that can validate it, which relate to the
nature of two-sided markets. Pricing in a two-sided market can deviate from
that in a one-sided market, in that equilibrium pricing might not be equal to
marginal cost when examining the sides in isolation (Evans 2003). In contrast,
it is a part of a larger equilibrium where both sides are concerned. This makes
subsidization, and making a loss in one side, a possible strategy
141
. In their
model, Parker and Van Alstyne (2005, 1494) show that “even in the absence
of competition, a firm can rationally invest in a product it intends to give away
into perpetuity.” It is assumed that the losses will be recovered either by taxing
the other side of the market, or raising prices after the platform reaches market
dominance. Theoretically, premium sales should cover, and exceed, the costs
of serving free users, so that the firm is profitable (Parker & Van Alstyne
2005). As the user base benefits from a positive feedback-loop, free users at-
tract paid users who, in turn, attract more free users, and so on. This charac-
teristic of the network effects makes it difficult to estimateex ante which price
level is correct; that is, how much subsidization the platform can provide and
still become profitable in the long run.
Despite conflicting perspectives, the argument for low cost of digital ser-
vices is strong (Parker & Van Alstyne 2005, 1503): “This strategy also takes
advantage of information’s near zero marginal cost property as it allows a
firm to subsidize an arbitrarily large market at a modest fixed cost.” Based on
this study, setting up the platform cannot be termed “modest fixed cost”.
However, in a similar vein, Fletcher (2007, 221) notes that “the prices charged
on one side of the market need not reflect the costs incurred to serve that side
of the market.” Finally, Evans (2002, 68) argues that “there is not necessary
relationship between price and marginal cost on either side of the market. In
fact, the price on one side of the market could be well above marginal cost
while the price on the other side of the market could be below marginal cost.”
However, this implies that the rule of marginal cost being zero for information
goods does not apply for pricing the premium product, as there is a
141
However, “[l]osing money initially to buy penetration can also be an important phenomenon in
one-sided networks” (Evans 2002, 57).
150
disconnection between the two sides. In effect, the startup needs to consider
both the acquisition and serving cost of the premium side and also that of the
free side. This is a central conclusion from the disconnection of sides (cf.
Evans, 2002
142
).
Therefore, employing the marginal cost theory of information goods is not
appropriate for two-sided online platforms, the price structure for which is
dual-sided, comprising 1) fixed development costs, 2) acquisition cost of free
users, and 3) service costs (e.g., bandwidth; customer support) of free users, 4)
acquisition cost of paid users, who convert instantly, 5) conversion cost of
paid users; (i.e., cost of converting activities, minus cost of acquiring free us-
ers), and 6) service cost of paid users. This price structure needs to be consid-
ered in pricing strategy, which, as demonstrated by the study’s sample, is un-
fortunately not always the case. Furthermore, in the case of subscription-based
charging, as is the case for many so-called SaaS startups (Xin & Levina 2008),
the pricing strategy needs to consider the expected lifetime revenue of the av-
erage customer, resulting from the lifetime and service level chosen
143
, time
required to convert free users to paid customers, the percentage of this conver-
sion, and the average lifetime of a paid customer.
In the practitioner’s side, Anderson (2009) claims that “practically every-
thing Web technology touches” will end up as free for consumers, as the mar-
ginal costs are approaching zero and prices are approaching the marginal cost,
and as “there’s never been a more competitive market than the Internet”. Alt-
hough not a part of academic debate, this perspective has proved popular
among startup practitioners and managers, even reaching some kind of para-
digmatic status
144
. However, a counter-argument to this claim can also be
found, and it comes from Bekkelund (2011, 16) through a simple but powerful
connection to Hal Varian’s earlier contribution:
"Based on the arguments in Varian (1995) it is likely that the
prediction of free products on the Internet in Anderson (2009)
only yields for purely competitive markets, i.e. when there are
“several” producers of an identical commodity. On the other
hand, it is not necessarily true for markets with monopolistic
competition, i.e. when there are several somewhat different
products, some of which are close substitutes."
142
“Indeed, a key feature of these [two-sided] markets is that, because the product jointly benefits
two parties, there is no basis for separating benefits or costs.” (Evans 2002, 9).
143
Online startups often apply so-called ‘tiered pricing’, by which the customer chooses a service
level. For example, level A gives Y features, and a paid level B gives Y+n features. As customers
frequently switch between the service levels, the average lifetime value of a customer is a mixture of
A and B; generally lower than price level B times the customer’s lifetime.
144
The future will show if this is a temporary phenomenon, or a true change in business logic.
151
Indeed, to assume free is to assume commoditization. Although claimed by
Anderson (2009), and followed by many startups, there is no definitive proof
of the sustainability of the model, even more importantly of its necessity, if
WTP and its association to differentiation strategies can be shown in the real
world. In fact, it is observable that consumers are in fact paying for online
content
145
. It is therefore not categorically true, at this point in time at least,
that all industries converge to free.
The connection made by Bekkelund (2011) is therefore crucial as it implies
that differentiation is a potential strategy against the monetization dilemma. In
fact, Anderson’s (2009) assumption of strict price elasticity abandons the use
of marketing as a differentiating factor in both 1) communicating products and
2) acquiring information on the needs of customers to create differentiated
products. Such a fallacy occurs when marketing is not perceived to contribute
to price elasticity. As economists posit, advertising has the potential to create
“artificial product differentiation” and change tastes (i.e., preferences), and, as
a consequence, advertised products face less elastic demand, associated in the-
ory with higher prices (Beuscart & Mellet 2009).
However, a static set of assumptions such as these can be countered by ar-
guing that regardless of differentiation through marketing or product diversifi-
cation, rivals are able to compete away these benefits rather quickly. Arroyo-
Barrigüete et al. (2010, 644) argue that “[winner-takes-all] does not mean,
however, that competition is scarce; it is quite the opposite, in fact: the com-
petition can be very intense until a company succeeds in establishing its tech-
nology as the dominant one.” Indeed, the inclusion of competitive dynamics is
required; some aspects relating to this are further elaborated in Chapter 4.7.
Moreover, if offering a free product, the user base cannot be substituted in-
definitely, so the problem culminates in decisions such as “which side to
charge?” and “how much?” If the startup is able to create network value be-
tween two sides in the market, it should also internalize that value (Rochet &
Tirole 2003). Therefore, depending on the platform firm’s perspective, its
pricing strategy has to, or can, consider network effects; that is, the indirect
benefits derived by parties from interacting with one another.
Inherently, online markets seem to provide a fruitful ground for free offer-
ing, while making monetization difficult. According to Luchetta (2012), these
can include 1) low technical and financial entry barriers, and 2) strong
network externalities. While the latter premise is subject to specificities, such
as markets and user behavior (e.g., multihoming), entry to online platform
markets is inarguably easier than, for example, offering a shopping mall
145
Refer to the annual reports of Zynga and Spotify for anecdotal but valid evidence on discarding a
categorical rule.
152
platform. While this might increasehalf-hearted efforts and thereby lead to a
naturally high mortality rate
146
, it can also erode serious competitors’ price
levels. In particular, whether or not there are network effects does not, strictly
speaking, matter.
Consider a founder who believes there are network effects to be achieved in
a given market, and therefore sets pricesgratis to obtain a critical mass before
rivals. Whether users multihome or not, or how strongly network effects actu-
ally affect adoption, thus plays no role in the outcome; the startup will make
free offerings. In a sense, network effects can be red herrings, and so the con-
cept is a double-edged sword for a founder who fails to understand its impli-
cations.
It remains true, however, that theoretical support for critical mass is strong
and it is widely regarded as the reason for subsidizing users: “one way to do
this to obtain a critical mass of users on one side of the market by giving them
the service for free or even paying them to take it” (Evans 2002, 50). At the
same time, the fundamental notion is to refer to two-sided markets; however,
the strategy only makes sense if the startup is able to recoup loss leader in-
vestments on the paying side. Rochet and Tirole (2005, 3) imply that the pur-
pose of the two-sided market, in a commercial sense, is to charge total prices
that cannot be negotiated away by participants: “The platforms’ fine design of
the structure of variable and fixed charges is relevant only if the two sides do
not negotiate away the corresponding usage and membership externalities.”
Indeed, if the parties were able to negotiate the externalities, the market would
be feasible. If introducing fees to either side would lead to a collapse, this
would represent a “natural cause of death” because failure is a proper response
to a lack of demand
147
.
We can therefore differentiate ‘true’ or genuine demand from superficial
demand, which is present only when offered free. This is similar to the case of
giving away free beer. Such a business would obviously be successful on the
first day, but if the next day or a few days later it introduces fees and none of
the customers return, there might not be demand at the set price
148
. Applied in
the context of online startups, if the price is zero and still there is no
146
Given that Internet startups might require little sunk investments apart from learning, there is
also a lowexit barrier. In fact, exit is a positive strategy if the learning accumulated can be redeployed
in another startup that is expected to succeed better.
147
This remark relates to long-term scenarios; in the short term, firms might apply penetration
pricing and other loss-leading tactics. However, in the long run, demand will determine the venture’s
viability.
148
In other words, reducing price to zero can introduce pseudo-demand, which can be eliminated by
any price rise.
153
attendance, this might imply there is no demand at any price
149
. However, a
much more plausible explanation, based on the author’s analysis, is that the
startup simply lacks awareness to generate such feedback that would enable it
to develop the platform in the correct direction, eventually gaining legitimacy.
In any event, it is imperative that adoption is preceded by both price
considerations and awareness (i.e., marketing).
The literature lends support to the idea that a lack of demand results in dis-
appearing two-sided markets: “an important characteristic of two-sided mar-
kets is that the demand on each side vanishes if there is no demand on the
other-regardless of what the price is” (Evans 2002, 50). At the same time, su-
perficial demand is not considered as it is demand for the product, albeit in the
“free beer” sense, and would vanish if fees were introduced. Inarguably, if
both sides of the platform are unwilling to pay, it cannot be viable unless it is
non-profit, public, or venture-funded.
In a similar vein, Tajirian (2005, 1) recognizes the efficiency of platforms
in coordinating exchange, and therefore their newness, but still posits the plat-
form should be able to extract rents from its services:
"[A platform] is more efficient in facilitating the exchange coor-
dination than a bilateral relationship between buyers and
sellers. Nevertheless, for the existence of an economically viable
market, the marketplace must to be able to derive economic
profits from facilitating the coordination by appropriately
charging each side of an exchange."
It might also be the case that there is demand, which the startup is unable to
capture. This would be the case when the platform is too open and relies on
users’ goodwill to compensate it for its services. Given rationality, if users
gain advantage (e.g., financial, time, or effort) in bypassing the platform at the
transaction stage, while utilizing its services at the match-making stage, they
will take this opportunity. Therefore, platforms that are unable to capture rents
become easy free-riding targets. As noted by Roson (2005), the interaction
between users might not always be perfectly observed, or it might only be a
part of the interaction in the platform, while continuing elsewhere
150
. For
example, match-making services such as dating sites only benefit while people
149
If there is lack of demand at any price, the lack of demand can be said to be genuine and there
might be no need in the market for that product. In startup language, such a condition can be termed
“vitamin syndrome”, whereas highly demanded products would be “pain killers”.
150
Consider the author’s personal experience of utilizing a freelancer platform to run an auction in
the platform, and then contracting the developer outside the platform. In this case, the platform was
not bypassed, as they charged a listing fee, but had their fee been commission on the transaction, the
author would have been tempted to free-ride.
154
are searching for partners; after a match is made, users no longer need to uti-
lize the service
151
.
Paradoxically, the more efficient the platform is in providing matches, the
less it earns if it is unable to capture rents. Moreover, even when it captures
rents based on transactions, a relationship between users might develop that
will bypass the platform in their future interaction of a similar kind (Rochet &
Tirole 2005, 13): “Buyers and suppliers may find each other and trade once
on a B2B exchange, and then bypass the exchange altogether for future
trade.” Note that this observation is similar to the example given earlier of
‘ActivityGifts vs. Gidsy’. In brief, private knowledge can become an issue.
However, offering some slack on this condition can be appropriate if other UG
effects on average compensate for the lost taxable interaction. Such would be
the case when users still propagate the platform, give feedback, or assist other
users; compensating for opportunistic behavior. In any case, it can be seen that
the choices relating to a platform’s openness (Eisenmann et al. 2009) can
influence its ability to appropriate coordination services.
Moreover, there can be psychological aspects defending the free strategy;
namely, introducing fees, any fees, might result in a disproportionate negative
effect on users’ willingness to adopt (WTA). Shampanier, Mazar, and Ariely
(2007), for example, argue that the benefits of a free product are more highly
appreciated than paid products, regardless of their difference in quality. This
lends support to the idea of satisficing (Simon 1956), so that users fallback on
free products if they provide at least a somewhat satisfactory solution to their
problem (i.e., “get the job done”
152
), despite their inferior performance. This
would indicate considerable friction in adopting paid platforms if substitutes
are available, even if inferior. However, Pauwels and Weiss (2008) study a
successful transitionfrom free to fee, and conclude that a charge can be made
for content, even when the theoretical reference price is zero.
However, even if not appreciated by users, fees might be indirectly im-
portant. As such, pricing can influence the quality of the user base that, in turn,
influences the benefits users derive from the platforms. Thus, prices can have
an indirect effect on increasing network effects, and pricing is an important
connection as an adverse selection control mechanism (Akerlof 1970;
Cennamo & Santalo 2013; Dushnitsky & Klueter 2011). Therefore, pricing
can be regarded as a governance mechanism that filters out low-quality par-
ticipants from the platform. Removing low-quality interaction that influences
151
In some cases, this poses a moral hazard for the platform. For example, Hagiu and J ullien (2011)
discuss a price-comparison site’s incentives to divert searchers toward more profitable products.
152
Consider a mobile app for note-taking versus pen and paper. Clearly, they are not competitors in
the application marketplace. However, from a user’s perspective, they can be substitutes. In this sense,
substitutes and indirect competition can be similar (see Chen, Esteban, & Shum 2008).
155
perceived network effects is critical, and can increase user satisfaction,
loyalty, and finally, the basis for rent capturing. Prices can also signify quality
to users; consider the abundance of free services, and the problem of
determining quality in the absence of prices
153
.
In particular, low-quality users might generate negative cross-group exter-
nalities that would harm the other group of users. This can occur regardless of
which side is being charged. For example, too much low-quality advertising
on Facebook might turn users to different social platforms, whereas low-qual-
ity visitors resulting from Facebook advertising might turn advertisers to other
platforms such as Google’s AdWords
154
. Therefore, the lack of pricing (i.e.,
two-way free access) can lead to a situation of adverse selection, which is det-
rimental to the survival of the platform. There are two important issues here:
first, in a freemium model, is there a spillover from free users to paid users so
that the former might, under some circumstances, create negative externalities
for the latter (e.g., congestion; use of support resources)? If this is the case, the
freemium model risks adverse selection, as paid users become annoyed by the
presence of free users.
Second, price is not the only mechanism with which to prevent adverse se-
lection problems. For example, authentication through identity can reduce
spam caused by anonymity. This was discovered by the popular technology
blog TechCrunch; after introducing obligatory authentication through reveal-
ing identity when logging-in to Facebook, abusive comments decreased
(Burns & Blesener 2013). The theory being that, in the presence of social
penalties, people avoid abusive behavior when their identity is revealed,
whereas anonymity enables such behavior. Some support for this notion can
be found in Lea, Spears, and de Groot (2001). However, the latter is limited to
the inherently low quality of maliciousness, whereas quality problems can
arise regardless of malicious intent (see the earlier discussion in Chapter 4.5).
Sides can also be considered as separate markets so that, for example, there
are both developer and consumer markets. As such, the competitive dynamics
can result in interesting findings. For example, Chakravorti and Roson (2006)
consider a situation in which there are two competing platforms, A andB, that
compete over two market sides, x and y. They show that a price decrease by
platformA in one side of the market (x), apart from competition, will lead to a
153
Note that with the lack of prices, the startup omits revenue from its measures of success. If
growth of revenue is replaced by growth of user base, the startup replaces customers with users as a
proxy for success, and therefore commits to a fallacy of false popularity (see Chapter 3.4 for users and
customers). Briefly, as users do not reveal a product’s economic viability, or genuine demand, their
use as a decision criterion can give incorrect information on where to allocate resources.
154
This example is not only hypothetical. In the online advertising industry, it is commonly
acknowledged that Facebook ads are of lower quality than search advertisements on Google. As a
result, Google’s advertising revenue was ten times larger than Facebook’s in 2011.
156
price increase in the opposite side
. This is the basic notion of subsidization
(Rochet & Tirole, 2003). However, in a competitive setting, platform B will
respond by reverse action; that is, by increasing the price of x to capture that
market, and lowering it for y to regain losses. This is shown in Table 23,
which depicts the platforms’ strategic choices.
Table 23 Strategic pricing (adapted from Chakravorti & Roson 2006)
Platform A Platform B
Side x Lower prices Raise prices
Side y Raise prices Lower prices
Consequently, lowering the price for side x will gain relative advantage
over the other platform in that segment, and vice versa, and acquiring market
share there will attract more users to sidey, which can be taxed based on the
network value provided by side x. Note that this assumes sidex is willing to
pay; it might be that introducing fees offset the perceived network value.
The major contribution of the literature is that the two sides of the market
are interconnected, and that the startup might produce a loss on one side. This
principle is paraphrased by Parker & Van Alstyne (2005, 1498):
"If the increment to profit on one complementary good exceeds
the lost profit on the other good, then a discount or even subsidy
becomes profit maximizing. Free-goods markets can therefore
exist whenever the profit-maximizing price of zero or less gener-
ates cross-market network externality benefits greater than in-
tramarket losses."
However, in terms of monetization, this is not much help. Farrell and
Klemperer (2007, 2020) provide a more useful approach: “It is efficient to
subsidize a marginal adopter for whom the cost of service exceeds his private
willingness to pay, but exceeds it by less than the increase in other adopters’
value.” This effectively implies, transferred to the monetization context, that
subsidization can be a profitable strategy while the price paid by premium us-
ers exceeds the overall subsidization cost for free users, including acquisition,
serving, and support. Equally, the same applies for as long as advertising reve-
nue, when monetizing through adverts, exceeds the subsidies.
Finally, it is relevant to draw an analogy to dotcom failures as some argu-
ments presented here exhibit similar features to those made at the time. Essen-
tially, in the dotcom era (ca. 1999-2001), two-sided markets were known as
electronic marketplaces. Most of the literature addresses the failure of B2B
marketplaces with, arguably, a similar conclusion to that regarding the B2C
157
context, given that the platforms have similar dynamics. Some of the failure
factors included, for example, 1) lack of quality indication, so that buyers
could not distinguish between reputable and non-reputable sellers, 2) exces-
sive competition on price among platform supply-side participants, 3) brand
dilution, and 4) existing industry relationship (Evans 2009a). Describing the
hype at the time, Evans (ibid., 115) states:
"Various researchers forecasted that B2Bs would come to ac-
count for a large fraction of commerce. Goldman Sachs pre-
dicted in 2000 that B2B e-commerce transactions would equal
$4.5 trillion worldwide by 2005. […] Entrepreneurs and venture
capitalists poured into this new industry. Between 1995 and
2001 there were more than 1,500 B2B sites.[…] Most of them
collapsed in the early 2000s as investors realized that they did
not have a viable business model and as the expected buyers and
sellers failed to turn up.” (Present author’s emphasis)
Therefore, unrealistic business expectations of the new economy are posited
as a reason for the demise of the dotcom era. The argument, based on an anal-
ogy between dotcom e-marketplaces, which arguably displayed similar char-
acteristics such as multi-sidedness, and modern online platforms is that free
offerings should be monetized, otherwise modern startups will share a fate
similar to dotcoms.
However, if the company’s goal is not sustainability but acquisition, the
strategic choice of freefying can be better understood. It can be seen that the
platform sacrifices profitability, even in the long term, as an attempt to raise
interest from investors and larger companies, and then to monetize the, albeit
hyped, interest. If this is the goal then profit is secondary to liquidity, as noted
by Brunn, J ensen, and Skovgaard (2002).
The analogy to dotcoms is somewhat alarming as the key tenets for failure,
based on both the sample and the theoretical survey, still very much surround
online markets. It has been argued that the nature of information goods, with a
low marginal distribution cost, or the nature of two-sided markets are insuffi-
cient for free offeringsper se, and require a realistic plan to monetize. It seems
that the literature shows mixed approaches, and cannot debunk or positively
confirm the premises made in the definition of the dilemma. Therefore, it is
fair to argue that the dilemma endures quite a multi-faceted treatment. Next, a
potential solution, the freemium model, is considered.
158
4.6.3 Solution: Freemium
When one side is subsidized (refer to Subchapter 4.4.3), the startup is forced to
find a party that is willing to pay for the demand-side’s use of the product,
which is termed indirect monetization. When the monetization model is free-
mium, the startup has one user base that is split into free and paid users, and
the model is direct monetization, with price discrimination. Free users are of-
fered the basic service while paid customers receive extra features, quota, or
support (Pujol, 2010). Freemium is a widespread model in the context of Web
startups, as noted by, for example, Niculescu and Wu (2010) and Teece
(2010). Therefore, it is suitable to consider freemium a potential solution to
the monetization dilemma
155
.
According to Wilson (2006), freemium is defined as follows:
"Give your service away for free, possibly ad supported but
maybe not, acquire a lot of customers very efficiently through
word of mouth, referral networks, organic search marketing,
etc., then offer premium priced value added services or an en-
hanced version of your service to your customer base."
The definition can differ based on the type of platform. For example,
Riggins (2003, 70) considers content platforms: “What these information pro-
viders are essentially doing is degrading their information product to create a
free version of the good that satisfies low-type consumers, but holding back
enough content so that high-type consumers are not entirely satisfied and,
therefore, are willing to pay for the fee-based site.” Riggins divides users into
low- and high types based on their WTP; however, he does not consider that
users can move from one side to the other (i.e., downgrade or upgrade). Free-
mium can also be regarded as second-degree price discrimination or version-
ing (Varian 1983), whereby users are given a choice between low-quality and
high-quality products
156
. Note that the equivalents are free and paid product in
the freemium setting, and that the quality does not refer to “bad” quality but,
for example, that the other product has less features. The low quality still has
to be sufficiently substantial to invite adoption, as discussed previously.
Beuscart and Mellet (2009) suggests that although Internet platforms
categorically give free access, they are able to monetize through four means:
1) advertising (e.g., Facebook), 2) freemium (e.g., Evernote), 3) transaction
fees (e.g., eBay), and 4) donations (e.g., Wikipedia). Donations apply to
155
In general, any solution that either directly or indirectly increases WTP (i.e., converting from
free to paid users) is effective.
156
In contrast, first-degree price discrimination occurs when the startup is able to identify WTP, and
therefore targets users with precise products. In second-degree discrimination, users are presented
with both options, and they can self-select (Riggins 2003).
159
non-profit projects rather than commercial ventures, and are not considered
here
157
. Advertising is a form of indirect monetization; thus, it effectively
circumvents WTP. As WTP is made irrelevant, the problem can be solved by
finding advertisers that are willing to pay for access to users. However,
advertising is not generally considered a good option, unless the platform
generates a substantial mass of traffic; therefore, it is a winner’s choice in a
winner-takes-all market.
Riggins (2003, 81) notes that “sponsored sites have struggled to find a
profitable business model based on advertising revenues.” The performance of
advertising is also criticized by Beuscart and Mellet (2009, 165), who describe
advertising revenues of Web 2.0 companies as “weak and disappointing, espe-
cially related to their audience”
158
. Advertisers often seek economies of scale
that an early-stage platform is unable to provide and, due to transaction costs,
it does not make sense for them to contract a large number of weak plat-
forms.
159
A negative position is also taken by Clemons (2009) who concludes
that advertising will eventually fail as the primary business model because it is
distractive, consumers do not trust it, and its informative capability is being
replaced by recommendation platforms.
Additionally, transaction fees are a form of direct monetization and there-
fore beyond the assumptions of the monetization dilemma. If transaction fees
were applied successfully, there would be no problem with monetization,
given sufficient liquidity; however, contrary to the hypotheses, this is not the
case in the monetization dilemma
160
. In contrast, freemium aims to overcome
WTP by a form of second-degree price discrimination. This differs from ‘bait
and switch’ in that a startup does not normally change the price levels after
users join, but expects usage to grow naturally and, therefore, free users to
convert to paid users (i.e., customers)
161
. As a solution, freemium relies on its
ability to change negative WTP into positive WTP.
As noted, freemium is not the same as subsidization, in which, according to
the two-sided markets theory, it is assumed that two sides interact. In
157
There are initiatives (e.g., a startup termedFlattr) that propose voluntary micro-payments (i.e., a
form of donations) in exchange for the consumption of content. However, they have not reached the
mainstream at this point in time.
158
Google is a notable exception; its revenue, mostly from advertising, amounted to $50Bn in 2012
(Google Inc. Announces Fourth Quarter and Fiscal Year 2012 Results 2013).
159
However, any type of platform can access the advertising market though online advertising
networks (see Salminen 2010), which largely reduce transaction costs for both parties in finding,
negotiating, and monitoring performance of their counterparty. In exchange, advertising platforms
take a commission based on some revenue sharing principles; typically, the publisher, that is the
connecting platform, retains the majority of click-based revenues.
160
Moreover, transaction as a term only applies to exchange platforms.
161
In contrast, it might increase features over time; however, the free version typically remains an
option.
160
freemium, there is no such assumption; instead, users form a priori one
homogenous group, and then self-select into ‘free’ and ‘premium’ groups.
Even if this results in two groups of users (i.e., free and paid users), these
groups derive no immediate benefit from each other’s presence
162
. Although
the goal of growing the user base can be shared by a two-sided market in
general and a startup applying a freemium model, the difference is that free
subsidization in a two-sided market aims to provide positive network effects
for another side of the market, whereas freemium attracts free users who can
later be converted to paid customers. However, in a one-sided market,
freemium can be regarded as a type of subsidy. As such, it is similar to
differentiation
163
and versioning
164
.
There is support for the idea of converting free users to paying customers.
Traditionally, marketers have employed sampling to penetrate a market, so
that giving free samples converts users into loyal customers (e.g., Milgrom &
Roberts 1986). Peitz and Waelbroeck (2006, 907) argue that, as a result of
sampling, consumers are willing to pay more “because the match between
product characteristics and buyers’ tastes is improved”. This is especially true
for Web services that are similar experience goods, so that consumers need to
try before they buy because quality is difficult to determine prior to testing
(Shapiro, 1983). As noted by Niculescu and Wu (2013, 2), “y trying (sam-
pling) the product or part of it before committing to any purchase, consumers
could learn more about the quality and other attributes (such as performance,
functionality, interface, and features) of the software, capabilities of related
modules, compatibility issues, hardware requirements, etc.” However, in con-
trast to physical goods, with which sampling is limited due to replication and
distribution costs (Niculescu & Wu 2013), freemium benefits from infor-
mation goods properties (see Subchapter 4.6.2), and therefore scales much bet-
ter.
Moreover, there can be positive spillover effects relating to free users. For
example, Oestreicher-Singer and Zalmanson (2009) found that social features
built alongside content interaction increases the propensity to convert to a free
user, so that the more active users of social features were also more likely to
convert. This suggests a relationship between content interaction and social
interaction; that is, spillover effects between them. Potentially, particular types
of platform might benefit from building structures that support other types of
interaction; for example, exchange platforms compatible with sharing and
content platforms that enable exchange. In a similar vein, Albuquerque et al.
162
However, free users can derive long-term benefits, as shown in Chapter 4.4.1.
163
Creating features that make the product special in the eyes of users.
164
Creating several versions of the product.
161
(2012, 408) found that free users’ content creation activities led to increased
profits of the user-generated platform they studied, whereby “free marketing
activities and referrals bring in about 50% of the sales of the platform, and we
suggest that [the company] should provide additional incentives to content
creators to increase their referral behavior.” Although, in general, this study
advises against overly optimistic expectations concerning the role of peer
marketing, they seem to prevail in some circumstances.
Platform theorists have employed the concept of differentiation to explain
the coexistence of several platforms in a given market. For example,
Tanriverdi and Lee (2008, 382) note that “heterogeneity in customer prefer-
ences allows differentiation, limits market tipping, and leads to the coexistence
of multiple [OS] platforms.” It is seen that users might require different feature
sets. This enables platforms to attract different types of user within the same
markets, which relates to the assumption that while network effects hold
within a group, they might not generalize (see Subchapter 4.5.1). Cennamo
and Santalo (2013) refer to distinctive positioning in inter-platform competi-
tion, which means that the differentiating features and target markets depend
on the choices of other platforms. Over-differentiation can result in a niche
trap, whereby the mass-market provider is ultimately also able to capture the
niche users (cf. tipping; Farrell & Klemperer 2007). Applied to freemium, the
startup can create features that are distinct from other platforms while main-
taining the same minimum requirements set by the competition. Such an ap-
proach aims to simultaneously maintain differentiation while appealing to us-
ers’ standard expectations.
However, there are some limitations to freemium. First, Oestreicher-Singer
and Zalmanson (2009, 2) note that “conversion rates vary and are often very
low, and firms continue to seek effective strategies for converting consumers
‘from free to fee’.” Indeed, many authors note the ‘expectancy of free’ by
customers (see Subchapter 4.6.2), thus complicating reversion to paid offer-
ings. There are also varying accounts on how easy it is to convert free users to
paid customers; Murphy (2011) notes that industry averages are approximately
three to five percent of conversion rates. This study will not consider the tacti-
cal methods of conversion optimization; however, based on the author’s expe-
rience, it can be noted that a substantial amount of time and effort is spent by
startups on optimizing their conversion rates. In particular, Pauwels and Weiss
(2008) highlight, among other factors, the importance of 1) the correct price
point, given the competition, 2) accumulating a substantial base of free users
before attempting conversion-enhancing actions, and 3) properly executed
marketing communications.
Second, based on an extensive sample of users of the freemium-based web-
site Last.fm that is classified as a content platform, Oestreicher-Singer and
162
Zalmanson (2009, 19) found that that it took 652 days on average for a free
user to convert to a paid user. They concluded that “the typical subscription
decision […] requires deep familiarity with the website and its features.”
Therefore, even when the conversion might take place from free to paid, the
process can be long and require consistent persuasion (i.e., marketing) by the
startup. Then, the startup needs to consider supporting costs for the free users
during their free period (Subchapter 4.6.2), to find which user types are more
likely to convert, and to create potential tactics to facilitate earlier rather than
later conversion.
Third, Bakos and Katsamakas (2008) examine the optimal platform design
structure and conclude that a platform would be advantaged by focusing on
one side, and then charging that side. By applying this logic to freemium, this
would mean that the startup should serve paid users better than free users. Alt-
hough this is achieved with premium features that offer higher quality (i.e.,
more features), the free version needs to be able to solve the cold start prob-
lem. Under freemium, platform design issues therefore concentrate on choos-
ing the appropriate structure for product variations and ‘tiering’ (Semenzin,
Meulendijks, Seele, Wagner, & Brinkkemper 2012). Such design choices re-
quire a tradeoff, whereby the platform needs to determine what is sufficient to
include in the free version to attract free users while keeping paid users satis-
fied.
This is known as thecannibalization problem (e.g., Riggins 2003), which
has been studied extensively in the extant literature. Generally, a firm facing
the problem must balance its allocations so that customers do not have an in-
centive to fall back to their second-best choice (see Subchapter 4.6.1). How-
ever, not offering a free version might leave the cold start problem unresolved
(Subchapter 4.4.3). As offering a free version to some users is a strong form of
subsidization and creates the cannibalization problem, the startup needs to bal-
ance this with sufficient investments to paid users, or risk “spillovers from the
intermediary’s investments in the other side of the network” (Bakos &
Katsamakas 2008, 192). The spillovers would be, for example, overly gener-
ous features or usage quotas, depending on the type of premium constraints
enforced, which would reduce the incentive for paid users to stay on the paid
user side or free users converting to paid customers, when this would other-
wise be required by the stricter conditions.
Pujol (2010, 2) refers to the cannibalization problem as the reflective
competition dilemma: “Feature differentiation can be challenging as it
requires tradeoffs between growing the free user base and generating reve-
nues, [i.e.,] the reflective competition dilemma.” Riggins (2003, 70) describes
the problem as follows: “For two-tier sites […], the challenge is to provide
enough free content to keep users coming back in order to increase banner ad
163
revenues, but, at the same time, limit the free content such that high-end users
will still be willing to pay to access the premium services and information.”
Moreover, Riggins (2003) assumes that advertising is a sort of control mecha-
nism, so that it can be utilized by a platform to increase inconvenience up to
the point at which users are willing to convert. In a pure freemium strategy,
this option does not exist. The problem is also perceived by practitioners
(Chen 2009): “the key is to create the right mix of features to segment out the
people who are willing to pay, but without alienating the users who make up
your free audience.” For the sake of clarity, the problem can be conceptualized
in this study as the feature definition problem in the specific context of the
freemium business model
165
. This conceptualization captures the freemium
business model as opposed to, for example, advertising, and refers to measures
that need to be considered by the startup when designing its offering.
Bekkelund (2011, 16) argues the use of “observable characteristics, such as
memberships in particular social or demographic groups; or unobservable
characteristics, such as the quality of the choice the consumer purchases”.
These are needed because the WTP is not known to the startup in second-de-
gree price discrimination (Riggins 2003). Bekkelund (2011) notes that free-
mium enables startups to experiment with pricing plans to discover users’ true
WTP, which is compatible with Ries’ (2010) proposal. Consequently, by
identifying common characteristics of users who are willing to pay, the startup
might be able to move to first-degree price discrimination, in which it would
directly offer discriminatory pricing plans based on WTP (Laffont, Rey, &
Tirole 1998).
In other words, a potential move for startups is to find a proper niche to tar-
get. In saturated markets, it is generally more challenging to discover needs
that other platforms have not satisfied (cf. Parrish, Cassill, & Oxenham 2006).
To speculate whether, and to what extent, online markets have become satu-
rated or not goes beyond the scope of this study; however, generally, given the
low cost of experimenting with different platform models, many mass markets
tend to be difficult for challengers. Nevertheless, by employing different fac-
tors as differentiation criteria (e.g., geographic location; language; highly spe-
cific interests), startups might be able to create niche platforms that are able to
reach a critical mass for both self-propagation and active use.
Teece (2010, 178) mentions that it is also possible to apply a hybrid
strategy and puts forward Flickr as an example: “Flickr’s multiple revenue
stream business model involves collecting subscription fees, charging
165
The problem relates to two conflicting incentives: 1) when giving free content, paid users might
be willing to downgrade, and 2), when only offering paid content, users who would convert after a
trial period will not adopt the platform.
164
advertisers for contextual advertising, and receiving sponsorship and revenue-
sharing fees from partnerships with retail chains and complementary photo
service companies.” Several methods can therefore be applied alongside
freemium. As also mentioned by Riggins (2003), the platform can utilize
negative network effects associated with advertising to drive up the number of
users’ willingness to convert. However, such a strategy is risky because users
might also switch instead of converting. Thus, utilizingadvertising as a threat
might not be an effective tactic.
In sum, freemium is a way to split users into free and paid users. It is simi-
lar to, although distinct from, bait and switch. The startup does not expect us-
ers to pay for the initial adoption but, as their usage grows, they are offered
extra paid service. As such, the functionality of the method as a solution is
directly linked to the proportion of free and paid users. If the paid user base is
sufficiently significant to sustain free users and satisfy the startup’s financial
goals, the solution is successful. However, more research is needed to under-
stand the antecedents to conditions in which users can be made to convert.
This relates to emerging studies on conversion optimization (see e.g.,
J ankowski 2013; Paden 2011; Soonsawad 2013). For example, Pauwels and
Weiss (2008) document that their case company employed e-mail marketing
and price promotions to upsell content subscriptions to their base of free users.
4.6.4 Discussion
By offering access and usage of its platform for free (i.e., freefying), a startup
is able to attract users, but is unable to monetize (i.e., attract paid customers).
If monetized, users opt for free substitutes. Therefore, should a startup aim for
free users or charge for its product? How valuable are “users” actually? How
to capitalize on popularity? These are questions that arise from this dilemma.
In essence, the literature shows two camps. The first argues that free models
are fundamentally different from the old rules of doing business (i.e., a new
economy), whereas the other argues for “business laws” such as revenue, via-
bility, and sustainability. The strong analogy to dotcoms cannot be ignored in
this debate as there is a risk of history repeating itself.
By applying the freemium model to a product startup (e.g., one selling a
uniform product to all customers), it can be transformed into a two-sided
setting (see Chapter 6.1). This is because the startup’s user base can now be
divided into two groups, which is a condition for platforms, of free users (i.e.,
users) and paid users (i.e., customers). The question of the analysis then be-
comes: are there network effects between these groups? At first sight, the an-
swer is “No”, as the product qualities do not change for paid users regardless
165
of the number or quality of free users, or vice versa. However, it can be argued
that the free users gain long-term benefits from paid users as this guarantees
their free usage. Ultimately, if the startup is unable to attract a sufficient
number of paid users, it will close down, and both free and paid users will
lose. If free users are aware of this, they might become motivated to promote
the platform to their peers in the hope of converting some to paid users.
However, although free users gain benefit from the existence of paid users,
it is unclear whether paid users benefit from the existence of free users. If not,
the network effect is asymmetric: one side benefits more than the other. In
fact, this is not necessarily uncommon in online business. Consider advertisers
and free users (i.e., consumers of content); advertisers, in theory
166
, benefit
from the presence of users, and would not engage with a platform without
them. However, the reverse does not apply, and users would in most cases
happily frequent the platform without advertisers
167
. This condition, as previ-
ously stated, is a negative indirect network effect, and it is not clear whether it
exists between free and paid users (e.g., through congestion). While it is often
assumed that the marginal cost of distributing a digital product is “close to
zero”, if the startup invests its resources in acquiring free users, the overall
user acquisition cost is transferred to prices that paid users will ultimately
pay
168
. In other words, there is a potential free rider problem.
This is, in fact, well understood in the platform literature as it is perceived
fait accompli that one side is being subsidized by charging the other. To the
author’s knowledge, this only becomes an issue in two cases: first, when as-
suming ideal user generation (UG); in other words, when the startup expects a
positive non-economic contribution from free users, such as peer marketing.
According to the study’s analysis, there are startups that make this assumption
at least implicitly. The second case, in which free riding is problematic, is
when paid users are tempted to join free riders, and leave the cost to a de-
creasing group of paid users, or when they avoid conversion in the first place.
166
It is important to note here that the relationship between the number of users and advertisers’
increased utility (i.e., the network effect relationship) is not necessarily linear. From an advertiser’s
perspective, much attention is paid to whether users 1) are from the proper target group, whether
defined by demographics of interests, and 2) that they are actively processing advertisements or, more
preferably, clicking them. Furthermore, while clicks are relatively easy to track, banner blindness
(Benway & Lane 1999) complicates the processing of online advertising, and therefore can be a major
obstacle in the relationship of user base versus its worth to well-informed advertisers.
167
We can, therefore, argue for an implicit contract between the platform and its free users: users
accept advertisers in exchange for free content.
168
Consider, for example, that the startup pays 100 money units to acquire 100 users, of which 5
convert (i.e., shift from free to paid user). Assuming that acquisition costs are evenly distributed (i.e.,
acquiring each user costs 1 money unit), and that acquisition costs are transferred to prices, which they
will be in the long run according to economic rationality, the 100 money units will thus be added to
the total cost of the product that will be paid by the 5 converted users; therefore, each will pay 19
more money units than without free users: (100-5)/5=19.
166
This type of strategic behavior can occur, for example, when users reduce their
usage on purpose to avoid quotas; that is, where premium features are linked
to consumption
169
. In sum, freemium can have negative spillover effects on
conversion that restrict its use as a solution to the monetization dilemma.
Clearly, freemium and advertising can be combined so that free users, and
not paid users, are shown advertisements. As advertising can represent a nega-
tive indirect network effect for free users, some would convert to paid users to
avoid advertisements. In theory, this would provide the startup with two bene-
fits: 1) revenue from advertisers, and 2) a higher conversion rate from ‘free to
paid’ than otherwise would have been the case. The problem with advertising,
by applying the platform literature, is simply that the startup needs to be a
critical mass of users, impressions, or clicks to attract advertisers; otherwise,
there are no network effects. This critical mass can be considerably larger than
some startups assume, as those in the advertising side of the market are only
willing to reduce their transaction costs by dealing with the largest platforms.
Moreover, advertising can result in switching instead of conversion if users are
ad-sensitive.
A more advanced tactic is not to employ indirect monetization at all but, in-
stead, leverage the actions of free users in a frictional non-user-generated
manner, to increase the conversion rate from free to paid users. This is identi-
cal to creating positive direct network effects. For example, the startup can
gather the usage of free users as “templates” and offer them to paid users as
complements. Therefore, the network benefits will only be available to pre-
mium side users, which would be an incentive for free users to convert to paid
customers
170
. Because the generation of such content is ‘frictionless’, it is free
from the free rider dilemma depicted earlier.
If freefying (i.e., offering a free access and usage) is applied as an ex ante
solution for the cold start dilemma in an attempt to increase the adoption rate,
turning from freemium to paid access is an ex post solution to the
monetization dilemma. For example, some content portals have begun
introducing pay walls for access to content (Spulber 2010). It is still too early
to judge the results of these attempts; nevertheless, the publishing industry is
facing strong pressure to monetize free content. This highlights the dynamic
interaction, or vicious cycle, between dilemmas; that is, if fees are introduced,
solving the cold start dilemma might become more difficult, whereas if
169
Note that the definition of the freemium model applied here is based on versioning and tiered
pricing (i.e., where free is the base level) not, for example, on time limitations. Although some authors
include this as a variation of freemium, trial software stems from desktop software and does not offer
a permanent free version as offered by the pure freemium model.
170
In other words, the free side would be neutral to network effects but the standalone benefit would
be the same, regardless of how many users are in the paid side, which is an interesting juxtaposition
whereby the product simultaneously exhibits and does not exhibit network effects.
167
freefying is applied, monetization becomes more difficult. As in most cases,
the startup is seemingly forced to make a tradeoff between the strategies.
However, opting for a paid monetization model might serve as astress test
for quick failure. If no users are willing to pay for a product, even after be-
coming aware of it, perhaps there is no true demand. Offering for free is
equivalent to offering free beer; if it succeeds and the idea is later to charge for
beer, there is no guarantee that the business will be sustainable once fees are
introduced. Of course, this does not apply to indirect business models, alt-
hough these might face the issues discussed previously in this chapter. In ad-
dition, freefying makes it impossible for the startup to compensate its content
creators. Although this is not a major concern for a process of UG with mo-
tives deviating from economic rationality, it can become an issue in special
cases. For example, one startup in the sample aimed to market journalists’
writings, essentially relying on them to write for free, thereby commoditizing
the labor they were paid to perform in traditional industry.
Competition with free can become hazardous in an industry setting with
low distribution and entry costs, such as many online markets, because en-
trants are 1) many, causing increased pressure for subsidization across verti-
cals, and 2) able to subsidize one side on the expectation of an emerging se-
cond sideex post (cf. finding a business model), and because this is facilitated
by their cost structure. Eventually, this strategy can lead to arace to the bot-
tom, whereby all platforms subsidize some sides by lowering prices, resulting
in all markets becoming subsidized. In the medium term, the startup might be
able to cover the loss of subsidization through venture capital, in the hope of
either finding a sidey to tax, if all of its users are initially x, or converting us-
ers from the subsidized side to the paid side, if they apply the freemium
model. The situation will be resolved when entrants are unable to recoup their
losses and exit both markets, which will enable the remaining platforms to
adjust their prices upward and become profitable.
The conditions for a race to the bottom include theex ante undefined side y,
if sidex is subsidized, andvice versa, and also the allowance period of expec-
tations (e.g., hype) that enables new entrants to consume previously profitable
sides (i.e., markets). As the required initial investments are generally low in
the Internet, venture capitalists might favor disruption by sponsoring ventures
that offer free substitutes to previously paid product categories to gain quick
traction, and then sell these ventures based on the expected value of their user
base.
The monetization dilemma is connected to the cold start dilemma; a startup
is unable to subsidize users or pay for content if it employs a free business
model. Hence, external funding is needed. However, startups without external
funding might simply neglect paid actions, such as marketing, that would
168
facilitate solving the cold start or lonely user dilemma and thus, are unable to
test whether the product would gain a critical mass
171
. This can lead to failure
without even getting a chance. Thus, the solution of free models can lead to
far-reaching consequences with regard to other startup dilemmas.
Regardless of the increased interest in monetization problems, the conclu-
sions seem to be mixed. One side of the argument seems todefend the use and
viability of free models in platform markets, while the other side contests it.
The two-sided nature of the market, and network effects associated with it, are
critical to this discussion. Pricing decisions (i.e., level and structure) in two-
sided markets are distinct from those in one-sided market considerations.
Therefore, subsidizing free users can be defended as a strategy, as rents can, in
theory, be extracted from the other side with higher WTP. However, the mar-
ginal distribution cost (e.g., nearly zero in information goods) is too loose a
definition, as it misses this subsidization cost, and also, typically, the cost of
supporting free users.
Essentially, a startup can suffer from hidden information if the platform is
too open and enables parties to negotiate and transact on their own, which is a
particular risk for exchange platforms that often need to retain control over
transactions. For other platform types, interaction beyond the platform (i.e.,
bypassing the platform) is not a major issue unless users are transacting (e.g.,
dating sites). Social platforms and content platforms, which often apply an
indirect monetization model, need to consider the negative indirect network
effect imposed by advertising, although theory shows that the effect can be in
both directions, and also its particular challenges in providing sufficient re-
turns for long-term business viability. In particular, advertising is appropriate
for platforms with strong dominance, so that they can create a sufficient num-
ber of impressions and clicks to attract advertisers. Although the problem can
be alleviated by delegating negotiation and coordination to an advertising plat-
form, the risk of returns remaining low exists for this monetization model.
4.7 Remora’s curse
4.7.1 Definition and exhibits
The remora is a type of fish that attaches itself to a larger fish
like a shark or even a boat. It rides along with its host and feeds
on whatever comes by. The remora can also detach from its host,
swim on its own, and survive. (Don Dodge)
171
The classic marketing maxim: if customers don’t know about the product, how can they buy it?
169
As established in Chapter 4.4, startups can face greater than expected difficul-
ties in achieving a sufficient degree of user-generated content (UGC); “suffi-
cient” being enough to launch a self-sustaining process of content replication,
or a critical mass. To overcome this hurdle, some startups opt for theremora
strategy, which is to join an existing platform to gain access to its predominant
user base or content, and in this way solve the cold start dilemma. In practice,
this might mean developing applications on top of existing platforms, such as
Facebook, Google, or Twitter, and leveraging their application programming
interfaces (API)
172
and user bases; essentially, gaining access to network ef-
fects without generating a critical mass. The solution might appear solid in
theory, and there are several cases in which it has worked well (e.g., acquisi-
tion
173
or direct monetization
174
); however, our sample of failed startups also
showed its limitations (for exhibits, see Table 25). The purpose of this chapter
is to analyze these limitations.
The dilemma of a remora’s curse takes place when a platform entrant needs
to decide whether to integrate a critical functionality relating to distribution,
marketing, or monetization to a predominant platform at the cost of losing
power in those areas, or to develop an independent solution at the cost of los-
ing access to the platform host’s pre-existing user base, content, distribution,
monetization system, or any other asset to which the integration would grant
access.
Consequently, remora’s curse addresses the choice of either developing a
product on top of existing platform (i.e., become a ‘remora’) or not (i.e., start
an independent platform); the former gives access to a pre-existing user base
or content while the latter requires that the user base or content be created sep-
arately without the “kick-off” provided by the host platform. In both choices,
the startup pays a tradeoff cost, as depicted in Table 24.
172
API, in the case of Web applications, enables access between applications; in platforms terms,
interoperability.
173
Instagram, for example, was built to be compatible with Facebook, and was acquired by
Facebook for $1Bn in 2012 (Constine & Kutler 2012).
174
Zynga, for example, charges the user for virtual goods sold on the Facebook platform, and
generated revenue of $1.2Bn in 2012 (Zynga Inc. 2013).
170
Table 24 Remora’s choice
Join Not join
Tradeoff Lose power over tech-
nology, marketing, and
monetization
Lose access to the pre-existing user
base or content (i.e., cold star di-
lemma)
Therefore, by joining an existing platform
175
as a supply-side participant, a
startup gains access to an existing user base or content but increases its de-
pendency on the platform owner, thereby in effect trading off 1) technology
power, 2) marketing power, and 3) monetization power to a) the distribution
function and b) the marketing function, which are delegated to the host plat-
form
176
. Technology power implies that the host influences the startup’s
technology choices, and the startup incurs initial integration costs and, when-
ever the host platform’s specification changes, continuous adaptation costs.
This can be regarded as a form of asset specificity, as discussed in the litera-
ture subchapter. Losing control over marketing and monetization refers, re-
spectively, to the inability to differentiate via marketing, as the platform poses
marketing restrictions, and the inability to choose a monetization model as this
is imposed by the host. In effect, the startup will also forego customer relation-
ships because it is the platform owner that retains customer information
177
.
The host has more information on the users but it restricts sharing it due to 1)
privacy concerns and 2) the competitive value of information. Capturing value
is another conflict: while in the platform, a remora can never reach an outcome
by which its revenue supersedes that of the host. This is theoretically impossi-
ble when the host imposes a revenue sharing scheme and blocks all other
means of monetization
178
.
Exhibits of remora’s curse are presented in the following table.
175
Such as Facebook, Google, and Twitter, or iOS, Android, and Windows mobile.
176
Distribution is delegated as the startup’s platform is accessed through the parent platform’s
interface. For the same reason, marketing is expected to be self-organizing as users will find the
startup’s product inside the host platform.
177
This was a major concern for record labels considering distribution with iTunes, as Apple would
hold customer information (i.e., customer relationship); eventually, Steve J obs was able to convince
them of the mutual benefit (Isaacson 2011).
178
The revenue of a remora grows proportionally to the host’s share; so that a/b àk(a/b), in which
k =growth factor, a =host’s share, b =remora’s share; given that a >b anda +b =1.
171
Table 25 Exhibits of remora’s curse
Example
[1] "We exposed ourselves to a huge single point of failure called Facebook. I’ve ranted for years
about how bad an idea it is for startups to be mobile-carrier dependent. In retrospect, there is
no difference between Verizon Wireless and Facebook in this context. To succeed in that kind of
environment requires any number of resources." (Rafer 2009).
[2] "The killing blow was when Facebook changed its app platform to make things less spammy,
and thus less viral. We were toast." (Parr 2011a).
[3] "Predictably and reasonably, Facebook acted in their own interest rather than ours. Their
Summer 2008 redesign supported Facebook’s goals elegantly but hurt our publishers and us in
ways that became clear just weeks after we’d raised another ~$2M." (Rafer 2009).
[4] "We were doing some time-consuming processing on gathered data so there wasn’t a big time
buffer we could use. With each downtime (the website worked but with no actual data it didn’t
make much sense to use it anyway) we had to wait until a backlog was cleared. Chances were
good that by this time we had another issue to deal with – a bug in the code, on of auction plat-
form’s changing the structure of their data, a simple hardware malfunction, or running out of
disk space." (Brodzinski 2009).
[5] "[Facebook] wasn’t a perfect fit for the Nouncer services, but it still fit in with the overall strat-
egy and philosophy. It also looked like an easy thing to do with a big marketing potential. The
result was JabAbout, a Facebook application using the social graph to propagate short mes-
sages by following the friends-of-friends paths. JabAbout failed to build a user base and was
eventually shutdown." (Hammer 2008).
[6] "Mint’s dependence on Yodlee apparently suppressed their acquisition interest among compa-
nies that knew Yodlee well (such as Microsoft, Yahoo, and Google); since we had developed our
own technology for aggregation, we didn't have that particular problem, and in fact had some
acquisition interest simply for the aggregator we’d built." (Hedlund 2009).
The exhibits demonstrate remora's curse from several angles. First, atech-
nology lock-in [1] indicates a situation in which continuous investments from
the startup are required to keep its product up-to-date according to the tech-
nological specifications of the platform owner. This might limit the available
technologies to some extent while increasing dependence on the host’s tech-
nological choices. If the choices are not optimal for the startup’s product, this
will reduce its competitiveness. Further, changing functionality [2] requires
the startup to react and organize its product development according to that of
the platform owner, and it pays adaptation costs [4].
The bigger issue, however, is the lack of control with regard to the user
base. At any time, the platform owner can restrict or deny the startup’s access
to users, justified as a change of service terms [3] or platform design [2].
Losing access to users may also occur due to a technical breakdown. In
consequence, the solution to the chicken-and-egg problem dissolves [5].
Moreover, the platform sets rules for marketing over which a startup has
little control. For example, the platform might give additional visibility to
particular products and not others, thereby distributing competitive advantage.
A startup can have little control over its visibility in the platform as it cannot
172
influence the rules
179
, and in general advertising is not allowed
180
. Overall,
these limitations may reduce investors' willingness to invest in a startup [6].
As noted, adaptation costs arise when the startup is dependent on the plat-
form as a source of data. First, it has to build the product so as to be compati-
ble with the platform. Second, it has to account for changes that might easily
break the flow of data and, therefore, its own product. Third, the platform
owner can restrict access to data, rendering the product useless. Coordination
problems of this kind, therefore, relate to the functionality of the product, and
apply especially to startups following the aggregator content model
181
whereby, in theory, the startup’s product integrates into several host platforms
to fetch data. This solves the cold start problem well as the fetched content
will enable demand-side benefits; for example, the more websites indexed, the
better the search engine, all else being equal.
By aggregating data from several websites, the startup might gain an in
praxis a solution to the cold start problem. However, at the same time, it be-
comes dependent on these data sources; any change in which necessitates an
adaptive response or the startup’s platform loses its ability to function
182
. The
more aggregated platforms (i.e., data sources), the higher the risk for coordi-
nation problems; however, the less the dependency on individual sources, as
they become expendable in a large selection, and the more changes by plat-
form owners, the higher the risk of coordination problems. Further, the startup
is forced to constantly monitor the health of the third party data source. Note
that aggregation is a special case when joining a platform; its purpose is not to
acquire users directly (as in: host platform ? startup’s platform), but to pro-
vide benefit for, often, existing users by offering them content from other
sources, or to utilize the content indirectly through social interaction spillovers
or search-engine indexing, which can lead to website traffic.
179
However, if the platform is fair and the rules are transparent, the startup is able to increase its
position by adapting to them and outperforming competitors. Equally in this case, it is not affecting
the marketing variables set by the platform owner, but only adapting to them.
180
To compensate for the lack of marketing tools provided by platform owners, some developers
have created tools for peer marketing. In them, applications exchange users on a ‘give one, receive
one’ basis; the revenue comes from selling a small portion of slots to advertisers.
181
Assume that all websites in the world suddenly deny Google’s access to their content; the search
engine would instantly become worthless. As Google provides indirect network benefits (i.e., a large
number of searchers) this is unlikely to happen. Further, Google is inherently hedging its risk by
diversifying the aggregation to billions of sites; therefore, its dependence on an individual host
approaches zero.
182
A real-time service loses matching ability; a static platform becomes outdated.
173
Consider two “degrees” of integration:
· Full integration: building the product inside the host platform (i.e.,
turning to a full complement).
· Selective integration: accessing the host platform’s functions and
user base but retaining, for example, distribution and marketing
183
.
These types of service are sometimes termed ‘mashups’.
Due to its definition, remora’s curse applies to both degrees of integration.
The severity of dependence might be less in full integration as user base and
marketing freedom is retained. However, if the access provided by the host
platform is critical for the functionality of the remora platform, as is assumed
in the definition, the dependency is also critical.
Moreover, it is assumed that most users find products, including those of-
fered by the startup, within the platform. That is, the remora retains beyond-
platform marketing capabilities, although they are mostly irrelevant when dis-
tribution is delegated to the host. For example, currently, leaderboards and
rankings are controlled by the host in most online platforms However, if this
assumption was denied and the startup was able to successfully market so that
users connect to the platform to find the product, the marketing dependence
would be broken. This is not, however, a solution to the dilemma as the plat-
form owner retains control of technology, distribution
184
, and monetization. If
revenue sharing works in favor of a startupin praxis, this does not remove the
fact that, in theory, the host can change the terms; although, while there is
competition for complements, a choice such as that would most likely result in
inter-platform competition.
While any of the above functions are considered critical, removing them
partially from the host’s control does not solve the dilemma. However, partial
integration can be sufficient in solving the cold start dilemma; more precisely,
the startup might be able to draw users from its host to an extent whereby it
obtains a critical mass. Even if the host then exercises its power, this is not
detrimental to the startup as it has already gained a critical mass and is now
self-sustained.
The risks associated with delegation are presented in the following table.
183
In selective integration, delegated functions can be arbitrary based on functionalities offered by
the host and the startup’s strategy.
184
Even in selective or partial integration, whereby distribution would not be delegated, the problem
will persist while the host controls any of the critical functions.
174
Table 26 Risks of delegation
Delegated function Risk
Technology Technology lock-in
Marketing Favoritism
Monetization Unequal revenue sharing / no revenue
sharing
Distribution Breakdowns, changing terms
In a typical setting, the remora’s expected benefit of joining relates to dis-
tribution. In aggregation, the product is distributed outside the platform, there-
fore with distribution and marketing costs, whereas the platform brings, in
theory, delegation benefits. However, this matching, from the perspective of
any startup other than the category leader, is not automatic, and herein lays the
fallacy of believing that marketing investments are not required. In other
words, intra-platform competition exists even in the presence of network ef-
fects, and due to the host’s incentives to promote the strong remora at the ex-
pense of weak remoras, participating in a platform as opposed to being inde-
pendent can in fact become detrimental; that is, the required cost for differen-
tiation exceeds coordination benefits provided by the platform, which is easily
perceived when understanding that fair treatment is not a profit-maximizing
strategy of the host. Rather, it benefits from favoritism; particular killer apps
bring much more revenue, and are much more difficult to replace, than the
long tail of complements
185
.
Consider, for example, a simple game with two players: remora and the
host. Two versions will be presented: first, a version in which the remora is
weak, meaning that the host does not believe it will sell. In the second, the
remora is strong in the sense that the host believes in it and will give it addi-
tional marketing support (i.e., exposure). This is a sequential game with three
turns: first, the remora decides whether to join or not; second, the host will
either sell its product or not
186
; and third, the remora will decide to stay or
leave.
The players make investments which they might lose, and gain benefits
which they might keep. Sales are recurring (i.e., third round) and parties
185
Consider App Store with millions of applications. The existence of this many complements is
beneficial to the platform owner and also the end user, given that his search cost of having so much
choice is not paramount, which is another reason for the platform owner to apply favoritism.
However, the majority of developers are disadvantaged as their offering cannot be easily discovered
(Salminen & Teixeira 2013).
186
This simplification equals the remora’s expected benefits described earlier; that is, acquiring
users or content.
175
engage in revenue sharing. Network effects are assumed, as the following
figure explains.
Figure 13 Weak remora
In the first stage, the payoff is expected benefits. As the remora will avoid
marketing investments, such as advertising and hiring a marketing manager, it
has a positive payoff. The product as a stand-alone would have some intrinsic
value, but less than when combined with the host platform’s assets (i.e., ex-
pected network effects). If the host makes sales, each party’s payoff increases
in proportion to that of the other party (i.e., revenue sharing
187
).
Not selling a weak remora’s product gives a higher payoff to the host as it
can keep the incremental network value without extra effort; comparatively, it
incurs an opportunity cost of not selling the strong complement, which is why
the payoff for not selling is higher than for selling. However, the host gets a
positive payoff for the remora joining as the remora provides an increment to
its complement base
188
(i.e., marginal network effect).
If the remora defects, it will lose its platform-specific investment. It will
also need to redeploy its product and compensate for loss of marketing dele-
gation, which is similar to the hold-up problem. However, it is assumed that
the remora can recover some learning effects by redeploying the product either
to independence or to another platform. Its departure will cause the host to
lose the incremental value. If it stays, it incurs no additional cost, but can also
resort to multihoming, which is not considered in the game.
At this point, keeping the remora will not produce additional gains for the
host as it does not expect the remora to sell but to provide perceptible value.
However, losing the remora would mean the loss of its incremental value.
187
For example, Apple shares revenue with its App Store developers using a70/30 ratio, in favor of
developers (Gans 2012).
188
According to the indirect network effects assumption, the complement base, as a whole, provides
a sales argument for the demand-side users.
Remora
Remora
Host
Join
Not
Makesales
Not
Leave
Stay
(1, 0)
(2, 2)
(-2, -1)
(0, 1)
176
Figure 14 Strong remora
By joining, the strong remora gets the same expected delegation benefits as
the weak one. At this point, it will only provide the incremental network util-
ity. Not joining will also produce similar effects as in the case of a weak rem-
ora.
In this case, the host can make high sales and is incentivized to sell. If
changing, the host would lose both the incremental network effect and the
sales effect. The strong remora would lose its platform-specific investment
and sales effect. Thus, both parties have an incentive to continue
collaboration
189
. Here the demand-side user base (i.e., indirect network effects)
becomes important for the strong remora; while, due to lack of exposure,
theoretical network effects are important for the weak remora. Strong remoras
actually realize high payoffs from participation.
Essentially, expected network effects are crucial with regard to the failure
of a weak remora. While it provides an actual marginal increase as a network
effect for the demand-side that the host can monetize, the weak remora gains
nothing in return. For the weak remora, if it is a possible strategy, becoming
strong before joining the host platform would provide a potentially better way
of investing its resources than joining as a weak player. From the host’s per-
spective, because payoffs are similar in the first step, it would need to distin-
guish between strong and weak remoras (i.e., “cherries and lemons”).
For the host, intra-platform competition is often desirable; startups
represent supply-side complements that increase demand-side utility
190
. For
startups, the reverse can apply: the greater the competition, the more difficult
it is to acquire users or customers, and the remora’s marketing delegation
advantage dissolves. The more the startup commits relationship-specific
189
Alternatively, the strong remora might consider multihoming to several platforms, which is not
considered in this game.
190
The logic is such that greater selection increases customer benefit, a standard assumption of
indirect network effects.
Remora
Remora
Host
Join
Not
Makesales
Not
Leave
Stay
(1, 0)
(3, 3)
(-4, -3)
(3, 3)
177
investments to the platform, the higher the degree of lock-in. In addition, moti-
vation to join a platform might arise from the expectation that user acquisition
is less costly within than outside the platform. However, when intra-platform
competition is high, this is less likely to be the case because other startups and
established firms compete over the same users. The competition can, in fact,
lead to an outcome whereby user acquisition is equally, or more, costly than
outside the platform
191
. As a result, the perceived marketing benefits relating
to customer acquisition can dissolve, as demonstrated in Table 25 [5].
This implies that even if there is a potential market, and network effects ap-
ply so that the increase in end users is due theoretically to the startup, the
startup is forced to compete within the platform. Therefore, these types of
network effect are here referred to as ‘theoretical network effects’, which are
theoretical (i.e., potential) as they do not realize under high intra-platform
competitionunless the startup is a category leader. In other words, the network
effects are not shared equally; some participants enjoy them, while others,
perhaps the majority, depending on the competition, do not. Therefore, net-
work effects that do not take place in the real-world setting are worthless to
the startup, and it gains no advantage in joining a platform with strong plat-
form effects compared to the situation of starting a platform without a critical
mass
192
.
Furthermore, the remora strategy is distinct from utilizing a platform as a
traditional marketing channel because of the integration of one or many criti-
cal functions into the host platform. For example, a user cannot access the
startup’s platform in a specific platform without first joining the host platform
(i.e., full integration), or the user might not access it beyond the host platform
if API access is not available (i.e., partial integration). If the host platform is a
monopoly, then joining it might give the remora access to some monopoly
benefits
193
. In contrast, when there is effective inter-platform competition
whereby users are distributed between several competing platforms, it makes
sense for the startup to follow this pattern by diversifying. While a subset of
users will treat platforms as mutually exclusive and choose one among them,
another subset will adopt several competing platforms simultaneously,
regardless of interoperability. The platform literature respectively refers this to
191
Such a situation is exacerbated when the platform owner reduces diffusion subsidies within the
platform, thereby increasing friction between the startup and potential users; for example, when
Facebook reduces visibility of application invites or organic post visibility in user streams.
192
Theoretically, the start-up gains a diminishingly small advantage compared to a pure cold start;
although, the more competitive the market becomes, the more the start-up becomes a “long tail”
provider. In brief, such a market exhibits winner-takes-all dynamics; however, not due to network
effects but to favoritism and user preferences.
193
Such as user adoption, so that in the absence of alternatives, the host platform keeps growing the
number of users.
178
singlehoming and multihoming (see Subchapter 4.7.3). By multihoming, in the
supply side, the startup can gain access to both multihoming and single-hom-
ing users, given that the host platform does not require exclusivity. In contrast,
choosing one host platform (i.e., single-homing) excludes users who single-
home to a different platform than that chosen by the startup.
As the author has argued, under some circumstances, the expected benefits
of the remora model do not materialize. If there is a reason for the startup to
believe soex ante, the dilemma dissolves as there is no rational reason to join.
However, under no conditions will the potential power of the host be negated,
regardless of whether it is enforced or not. The sole relaxation of the di-
lemma’s validity from this side would be when integration only touches non-
critical functions, but this is not in accordance with the definition presented
here. The host choosing not to exercise its power is not a relaxation because,
although it leads to a favorable position for the remorain praxis, the benefits
are not stable as there is uncertainty concerning the host changing its strategy.
Another case is take it all, when the host is lazy in exercising its power in
any of the critical dimensions. In such cases, the remora can in effect trans-
form into a leech, gaining users while retaining all benefits. However, again
note that the dilemma in effect persists as, at any time, the host can change the
rules of the game. Twitter is a well-known industry example of a “lazy host”
that grants free access, does not enforce revenue sharing, and is built as a very
open communication platform with low lock-ins to the website. For example,
Facebook has implemented strong lock-ins because users have to log in to its
interface for each interaction, and are shown advertising; Twitter can be ac-
cessed from anywhere without realizing additional revenue
194
. Nevertheless,
Twitter has also been known to break the rules of sound business logic in other
areas, mainly monetization. In general, platforms expect reciprocity; even if
not charging their complements, they expect them to provide indirect network
effects that can be monetized according to their monetization model.
However, the loss of user base must be discussed; more precisely, the defi-
nition is ‘users’ not ‘customers’. In other words, we return to the issue of ‘user
versus customer’. It is therefore possible to argue against the premises of the
problem by stating that users are only desirable if they can be converted to
revenue, because there is an implicit assumption that the startup wants users.
In fact, this becomes an issue when the platform owner is possessive about the
opportunities to monetize; for example, as is the case with Apple, but currently
not Facebook, restricting available monetization methods. As such, assuming
194
However, the behavior of platform complements is not “opportunistic”, as they have no other
choice. In opportunistic behavior, the startup chooses a strategy, among other strategies, which
maximizes its profit at the platform host’s cost. By this logic, Twitter is a “non-profit” platform, and
should be excluded from commercial analyses.
179
indirect monetization is not possible, free users would not be worth the
startup’s efforts as the platform can internalize all complement benefits relat-
ing to monetization; that is, there is no revenue sharing.
In sum, remora’s curse addresses situations in which platform participants
are at the mercy of the platform owner, which often aims to control revenue
sharing within the platform; for example, Apple’s App Store dictates the reve-
nue sharing terms for developers. However, the reverse might occur if the plat-
form is open; that is, it enables free access to its data and does not control
monetization. This case is clearly demonstrated by Twitter: for a long time,
third party service providers tapped into tweets generated by Twitter without
contributing anything in return; applying the animal analogy, the remoras had
become leeches. These included services to monitor tweets, set alerts, and
manage tweet streams. Counter-examples such as these do not remove the ex-
istence of the dilemma because Twitter can willingly exercise its power, which
it has begun to do (Nickinson 2013).
The main risks of falling victim to the platform owner’s strategic behavior
can be attributed to platform design (i.e., rules, terms, and specifications) and
unpredictable changes, in which central functionality is altered with no influ-
ence from the participants. Although joining an existing platform, or becoming
a ‘remora’, might appear to be alow hanging fruit, or an easy solution to the
cold start problem, the startup should be cautious about the potential hazards.
As established in the introduction to this chapter, startups, as is the case with
most organizations, are obliged to trade off strategic alternatives. Therefore,
joining a platform is not a trivial matter as it can lead to strong lock-in effects
and might be irreversible, especially considering the startup constraint of lim-
ited runway (i.e., depletion of time and resources).
Nevertheless, benefits of joining an existing platform probably exist in
some form. It can be assumed that the advantages are strongest when the
product category or industry is unfamiliar to potential customers, and therefore
requires strong persuasion, market education, and heavy investments in pro-
motional activities. However, each platform has its own competitive dynamics
that might not always be fair and which can, in fact, lead to complete dissolu-
tion of the expected benefits. Although it might appear to be a good strategy
for solving cold start problems, joining a platform does not automatically se-
cure more customers due to intra-platform competition and the above-men-
tioned dynamics (Table 25). The platform owner’s aim is, in most cases, to
encourage competition among participants within its platform. An exception to
this goal exists when protecting category leaders (i.e., killer apps) due to their
higher benefit to the platform. In such cases, new entrants will have difficulty
because incumbents are protected by the platform owner, for example, through
180
dominant ranks in application listings, thereby increasing the risk of awinner
take all outcome.
4.7.2 The literature
This subchapter will examine the remora strategy from the perspective of the
platform literature, which refers to remora-type structures as complements
(Farrell & Klemperer 2007; Rochet & Tirole 2003); as such, becoming a rem-
ora is to become a complement. The problems derive from not having owner-
ship of the platform, while the benefits originate from the existence of the host
platform (i.e., coordination effects) and its end users (i.e., network effects).
The large number of participants in the other side of the market increases at-
tractiveness to join, whereas the platform’s specialized coordination abilities
increase matching to a point at which transaction costs of finding, negotiating,
and monitoring the other side of the interaction can considerably decrease.
The platform owner’s tendency to exercise power, and also the risks relat-
ing to the remora’s position have been extensively discussed in both the plat-
form and economic literatures; transaction cost-related concepts are especially
applicable. In general, despite the fact that Internet startups did not even exist
at the time of its invention, the classic hold-up problem (see Klein 1998) ad-
dresses this type of issue at a general level, although not necessarily from the
same perspective
195
.
More specifically, Hagiu and Yoffie (2009) identify three hold-up risks: 1)
the host raising prices after becoming successful, 2) vertical integration into
the remora’s business, and 3) losing the ability to differentiate. Hagiu and
Yoffie (ibid.) give respective examples: 1) after reaching dominance with
Windows, Microsoft raised OEM licensing prices; 2) Google has been bun-
dling applications into its core offerings; and 3) Toys ‘R’ Us was unable to
differentiate against small players in Amazon’s marketplace. In general, these
risks are compatible with the concerns voiced by the founders (see Subchapter
4.7.1).
The platform owner becoming the startup’s direct competitor is another
risk. Due to asymmetric information in favor of the platform owner, it is able
to monitor each product and decide whether or not to provide a substitute.
Such examples have been documented in the industry (Honan 2012). How-
ever, as mentioned earlier, there are also documented success cases of
195
The hold-up problem requires 1) asset-specificity, 2) incomplete contracts, and 3) incentive to
“hold” (Klein 1998). In Web platforms, these arise if the complement cannot reuse its platform-
specific investments. In general, no contractual agreements protect the complement, and the host can
treat individual complements as expendable when they are large in number.
181
employing the remora strategy to rapidly acquire new customers (Campbell
2012), although, even in these cases, the power imbalance and, therefore, the
dilemma is present (Kelly 2009).
Remora is a strategy that aims to internalize externalities of a larger net-
work (i.e., envelopment). In platform markets, a remora relates to implications
of compatibility. As opposed to competing technologies, especially rival
standards, online platforms invite compatibility through their application pro-
gramming interfaces, or APIs (Evans et al. 2006). This behavior might be
different to that in other industries where “t is unlikely that the sponsor(s) of
a network with a large installed base will grant compatibility. Doing so en-
hances intra-network competition and […] provides very little benefit to the
system sponsor. Compatibility eliminates the installed base advantage of the
incumbent, reducing its market power and profits” (Church & Gandal 2004,
21). The reverse is argued here, based on different assumptions. Church and
Gandal (2004) imply that compatibility enables substitution and envelopment.
Here, this is not primarily considered by remora’s strategy because the host
can either prevent access or absorb remoras, and therefore counter remoras
that aim at becoming substitutes
196
.
Church and Gandal’s (ibid.) concern relates mainly to standards and tech-
nology. Once a standard or technology is open, the host cannot cancel the de-
cision as the technology has become public knowledge. Therefore, accepting
remoras can be perceived as reversible, while opening technology can be irre-
versible. The exception is when a startup performs aggregation as its content
model, and can envelop through content; as such, it can envelop the target if
the user is not motivated to visit the host website because information is given
by the remora
197
. However, even in the case of standards, inviting competition
can actually lead to a better outcome from the technology-holder’s perspec-
tive. As noted by Shy (2011, 131), “Sony did not use [open] strategy and as a
result it had to abandon its Betamax video technology in 1988 because it re-
fused to license it to competitors, thereby paving the way to VHS standards.”
In sum, it has been established that the host has an incentive to offer its plat-
form, and remoras have incentives to join it.
196
Furthermore, the risk of envelopment only applies to complements that are platforms. Most
complements (i.e., apps) in social platforms, for example, are stand-alone products that together
increase the utility of the platform. Envelopment would take place if popular apps were to move away
from the social platform, taking users with them. Such a coordinated move seems unlikely and, even
in this case, the user would most likely multihome given that, even without the complements, the
social platform offers intrinsic benefit while sufficient friends, who are not associated with apps,
remain.
197
Google is an example of a remora; it scans and indexes host sites, and then displays information
in search results. It utilizes content from other websites to monetize.
182
The purpose of the remora strategy is to gain benefit from an existing in-
stalled base, and therefore it is assumed that users’ switching cost is low,
thereby moving from the host platform to the remora platform
198
. Eisenmann
et al. (2011, 136) assert that “f users switch between rival providers of a
shared platform, they do not forfeit platform-specific investments in comple-
ments or in learning the platform’s rules.” The interface remains similar and
users remain in a “trusted” environment. For example, building an application
on top of Facebook does not require users to migrate from Facebook. They can
find the app through Facebook and utilize the familiar interface to access it
(i.e., low learning curve), and therefore the cost of adoption can be less.
Additionally, the remora can gain brand spillover effects (Olson 2008) from
the host platform’s enhanced adoption. The host’s strategy, in contrast, is to
prevent over-excessive brand spillovers that might compromise its platform
through abuse or dilution by the remora (ibid.). The platform owner also aims
to benefit from direct monetization while avoiding thecommodity trap; that is,
offering infrastructure without control of customer relationships. Those
joining the platform want to benefit from the owner’s reputation. This conflict
is shown in practice by, for example, Facebook’s rules prohibiting the use of
its supposed endorsement, and the willingness of competitive organizations to
associate with Facebook by utilizing its logo or other signaling devices (see
Facebook 2013).
Essentially, by committing to a platform, a remora makes relationship-spe-
cific investments, and will therefore be vulnerable to related problems: sunk
costs (i.e., technology development that cannot be redeployed), power abuse
(e.g., host changing the terms), the hold-up problem (i.e., difficulty of switch-
ing in the case of abuse), and even the free-rider problem whereby the plat-
form owner employs remoras to increase its popularity among end customers
while retaining all associated economic gains
199
. Cennamo and Santalo (2013)
note that “[h]igher sunk costs that are relationship specific imply […] a
higher exposure to hold-up problems.” Applied to startups, a learning curve
can emerge for platform-specific skills if the host platform’s technology dif-
fers from the startup team’s skill set.
Compatibility with the current team’s skills is an influential factor as a high
degree of compatibility requires little adaptation with regard to product devel-
opment. However, developing for a single platform (i.e., single-homing) might
become highly asset-specific (see ‘multihoming’ in Subchapter 4.7.3). When
198
In fact, there is no switching as the startup will become a complement not a substitute. We can
refer to this as the ‘conversion cost’, essentially implying the same propensity for a user to join the
startup’s platform.
199
For example, free apps increase the attractiveness of Apple’s App Store, although an Apple
device is needed to access them. Developers do not receive revenue from hardware sales.
183
joining the platform requires skills that can be redeployed in the case of exit,
asset-specificity through skills will not become an issue. In fact, many current
online platforms utilize open Web standards and programming technologies
(Zeldman & Marcotte 2009); therefore, learning them, although being a sunk
cost, does not lead to asset specificity.
Further problems, from the remora’s perspective, include substitution by
acquisition or rivalry (i.e., absorption through substitution). The former hurts
non-acquired competing startups while the latter is harmful for all firms in the
vertical entered by the platform. The platform owner can utilize its marketing
power to secure better positions within the platform for its own features or
those of the acquisition target, when acquired. It is in a far superior position
regarding download trends and other types of information than remoras, from
which this type of information can remain hidden. In the presence of
asymmetric information (i.e., host advantage) and delegation of marketing and
distribution, a natural condition of moral hazard arises (e.g., Pauly 1968). In
other words, by utilizing its power to exercisefavoritism for developer A, the
host will neglect the delegated tasks from developer B. In reality, this is a
common practice
200
, although platform owners tend to build it as amarketing
mechanism, so that the most popular applications receive the most prominent
positions in leader boards and category views
201
.
The issues of power and dependency have been widely discussed in the lit-
erature beyond two-sided markets theory. For example, Yli-Renko and
J anakiraman (2008, 134) argue that “resource interdependencies with other
organizations are viewed as constraints and restrictions; that is, being de-
pendent on an exchange partner means that the partner has increased bar-
gaining power. Therefore, to survive and succeed, firms should take action to
minimize threats to organizational autonomy and attempt to control the re-
sources needed by other organizations to make others more dependent on
themselves.” In the platform context, whether to depend on the platform’s re-
sources or become the platform on which others are dependent is precisely the
question; both include risks, hence the dilemma. Exchanging power for dele-
gation, becoming dependent on sunk costs, and the opportunity cost of “going
solo” are hazards of the remora strategy. In contrast, opting for an independent
launch in platform markets is problematic when the platform is incompatible
with an incumbent platform.
200
“Sorting applications on the basis of popularity, the platform sponsor can choose to own the
highest rank order items, as Microsoft has chosen to do for its operating system and game platforms”
(Eisenmann et al. 2009, 147).
201
The classic conundrum for the non-favored application, therefore, is: How to get visibility
without downloads, and how to get downloads without visibility? Hence, the need for a marketing
function re-emerges.
184
This is noted by Farrell and Klemperer (2007, 2045): “Switching costs and
network effects can work in tandem to discourage incompatible entry: switch-
ing costs discourage large-scale entry […] while network effects discourage
gradual, small-scale entry.” A large-scale entry can be ineffective because the
installed base is unwilling to become a new platform, whereas a small-scale
entry would initially provide the platform for a small network of users, alt-
hough their adoption is prevented by the critical mass in the incumbent plat-
form. Therefore, the startup ends up in the familiar double-bind: the cold start
dilemma, which also represents the tendency of going “back to square one”, as
discussed in Chapter 4.8.
Although strategic thinking influences the behavior of the platform owner,
it is not entirely sovereign in its use of power. Instead, it needs to consider 1)
inter-platform competition, and 2) quality. Consider the negative effects that a
large-scale exit by high-quality supply-side actors would have on the demand-
side as a result of power abuse. This might also lead to high-quality users, in
terms of their high willingness to pay (WTP), exiting the platform; the re-
mainder would be low-quality complements (e.g., apps) and low-quality users
(e.g., free users). This type of escalating chain of events, led by the exit of ac-
tors from one side, has been highlighted by Akerlof (1970) who argued that a
lemon’s market can arise if high-quality actors from one market-side abandon
a market, followed by high-quality actors in the corresponding side, leaving
only low-quality actors in each side.
The degree to which hosts utilize power varies to a great extent. Open plat-
forms, such as Linux in the operating system market, allow the greatest free-
dom, although often the least business support, whereas more closed platforms
(e.g., App Store) can include users with higher willingness to pay (Developer
Economics 2012). If the startup monetizes directly, the feasibility of joining a
platform relates to its user base’s WTP. If network effects are a factor in WTP,
opening the platform might increase aggregate WTP because interoperability
enables access to a larger user base (Eisenmann et al. 2009).
It is relevant to note that the remora’s achieved network will not compete
against that of the platform owner while the remora’s users continue to emerge
from the network; this is because its user base will always remain a subset of
the platform’s user base. In contrast, envelopment aims to take the users away
from the remora. This is thecomplement effect, which makes it feasible for the
platform owner to attract new remoras; in other words, new complements in-
vite new subsets
202
, and the entire network size expands.
202
Although the host network will grow and feed remoras, the reason it will grow is because of
remoras. Effectively, this is a solution for the cold start dilemma, as suggested in Subchapter 4.5.3.
185
The characteristic of social networks to initiate sub-networks (see e.g.,
Ganley & Lampe 2009) in fact gives support to the remora strategy. This is
because they become a powerful entry barrier; for example, consider an en-
trant who would like to create a social network for a particular niche, such as
dog lovers. The entrant will soon discover that the dominant platform most
likely includes a sub-group sharing this interest. If not, then the entrant can
begin such a sub-group and become a complement, which explains why the
diffusion effects are so strong in the social platform field. In addition to net-
work effects, complements, including user-generated groups and applications
created by developers, increase the platform’s benefits for existing and new
users. At the same time, sub-groups introduce an entry barrier for new social
platforms; therefore, the remora strategy becomes feasible, given that the plat-
form allows monetization. For example, in Facebook this is possible inde-
pendently, but Apple controls it in App Store. Although application developers
might enjoy monetization gains, users who create new social sub-groups are
not typically included in revenue sharing (cf. Facebook); this is not in conflict
with their motivation, which is more intrinsic than profit-driven motives.
Some authors argue that the Internet is characterized by winner-takes-all ef-
fects, which are more or less stable in the presence of strong network exter-
nalities. For example, Herings and Schinkel (2001, 25) state:
"The fabulous dynamics of the information and communication
technology sector makes monopoly positions to be temporary. As
soon as the speed of technological innovation diminishes, […] it
is nearly impossible to enter into a sector with strong network
externalities and one monopolist."
The first argument is easily acceptable as seen by quick transformations in a
few key Internet markets; for example, MySpace replacing Friendster, then
Facebook replacing MySpace as the most popular social network, and simi-
larly, Yahoo being replaced by Google in a relatively short period of time
(Gawer & Cusumano 2008). However, a major relaxation to the second argu-
ment of impossible entry, it can be argued, is when the market is suitable for
network externalities in general, not only in the case of the dominant com-
pany; that is, multihoming takes place (Mital & Sarkar 2011). Thus, if we con-
sider multihoming behavior, which is customers willing to join several com-
peting platforms, the argument of impossible entry disappears. This is because
network externalities are not mutually exclusive and can be utilized by many
companies in the market, given that they are able to provide benefits that in-
terest users. If we were to apply a third assumption, namely interoperability,
the initial argument would become even weaker. In an environment of strong
interoperability between platforms (e.g., through APIs), there is less incentive
to remain a proprietary user, or provider, of a single platform, given that the
186
users are active in taking advantage of this feature, and that supply-side par-
ticipants can monetize within the platform. In fact, this can be seen in the an-
ecdotal evidence of people subscribing to several social networks and porting
contacts between them, and also developers creating products for several com-
peting platforms.
4.7.3 Solution: Diversification
A common strategy to reduce dependence on a single platform is diversifica-
tion to several host platforms, similar to multihoming in the platform literature
(Rochet & Tirole 2003; Armstrong 2006). In this strategy, the startup utilizes
several host platforms instead of only one. Salminen and Teixeira (2013) sug-
gest that developers should multihome to avoid being trapped in a single ap-
plication marketplace. Hagiu and Yoffie (2009) provide the example that firms
can advertise on both Google and Yahoo platforms; that is, employ them as
marketing channels to drive search traffic. This idea is now developed further
through a concept calledselective integration that, along with associated strat-
egies, is defined below:
· Selective integration: a strategy of choosing which parts or functions
of a platform are integrated with a host platform.
· Content envelopment: using aggregation from one or several host
platforms to gather a critical mass of content, after which UG negate
any dependence.
· Value envelopment: changing the monetization model when passing
users from the host to the remora.
First, diversification takes place when the remora sources content or users
through aggregation from several hosts. In aggregation, the remora feeds from
several sources; as such, the host might either not necessarily be aware of the
remora’s existence
203
or welcome it
204
. Aggregation reduces dependence from
a single source. A simple rule for dependence of the remora on the host can be
given to illustrate diversification effects through aggregation:
203
As in the case of the auction platform startup in the sample that aggregated results from several
sites. However, lack of awareness might not necessarily help, as it effectively prevents cooperation. If
hosts are cooperative, hiding from them achieves smaller payoffs.
204
As in the case of Google, whereby all websites want to be included regardless of the fact that
Google monetizes their content, the benefits of getting free traffic overcome this nuisance.
187
D =1/x, in whichD =Dependence, x =number of hosts
As x approaches infinity, D approaches zero, assuming equal performance
acrossx. In reality, we observe this effect, for example, through search engines
that aggregate the content of billions of websites; their dependence on one site
is diminishingly small, whereas a developer’s dependence on Facebook or
Twitter, given that the number of available hosts is much smaller, is naturally
bigger. A good example is Google: because it indexes billions of websites, its
dependence on a single site, no matter how big the size of the site, remains
very low; therefore, Google as a remora has the power advantage.
Second, as part of diversification, a startup might opt for selective integra-
tion, a type of mixed strategy that would take place when the startup partially
leverages one or many host platforms; for example, as a source of content or
users, while maintaining, for instance, monetization alternatives in its own
platform. This is the case when the monetization model changes in transition
from host to remora, so that:
Host ? indirect monetization
Remora ? direct monetization, without revenue sharing
The transformation of a monetization model can be referred to as value
envelopment. An example is when the users of a free platform become paid
users of a remora (e.g., Zynga as the remora and Facebook as the host). The
platform owner applies an indirect monetization model (e.g., advertising),
while the remora applies a direct monetization model (e.g., selling virtual
goods), without the host being part of revenue sharing. In another setting, the
host monetizes by distributing its complements (e.g., free apps), and then ap-
plies a more or less generous scheme of revenue sharing.
The reverse can also occur, whereby monetization is delegated to the host.
This is the case when the remora joins an online advertising network, such as
Google AdSense. The agent will then resell the advertising inventory
205
, and
the startup is able to capitalize on the aggregated content. A special case is
termedarbitrage, in which the platform simultaneously buys cheap clicks (i.e.,
visitors) from the network and sells more expensive clicks in return, profiting
from the price difference (Gunawardana, Meek, & Biggs 2008).
Third, content envelopment takes place when aggregation occurs for a lim-
ited time: that is, sufficient to obtain a critical mass, after which UG effects
begin to take place. In this option, the startup employs technology to aggregate
205
In practice, AdSense employs an algorithm based on, for example, semantic matching of content
and keywords, and advertisers’ placement preferences (see Salminen 2010).
188
content to solve the cold start problem, but then relies on user generation
(UG). Aggregation is a means of employing technology to retrieve content.
The goal of employing aggregation
206
as a content model is to solve the cold
start dilemma because the host provides the content. Additionally, when ag-
gregation is employed in relation to diversification, it has the potential to solve
the remora's curse by reducing dependence on a single host. After this, specific
problems relate to monetization and active use, and also, in some cases, the
access towalled garden systems, which are non-accessible by Web crawlers.
Bi-directionality and selectivity under the selective integration require
closer examination. First, both sourcing and spreading content back to content
or social platforms is possible. For example, a startup called AirBnB famously
applied this tactic by spreading its product listings to another much larger plat-
form. Through such efforts, the startup can reach potential users in relevant
verticals; however, the host might perceive programmatic solutions in a nega-
tive light, and block access, as was the case with AirBnB and its target. Man-
ual efforts, however, scale relatively poorly.
Second, technical solutions can be created to facilitate users’ interaction
with the content; a typical example being social media buttons that enable
sharing with various social networks. Because social platforms are dependent
on fresh and interesting content, its provision generates organic traffic for the
platform without the need of integrating its product to the host. Mital and
Sarkar (2011) mention two examples of mutual benefit among platforms:
Facebook and YouTube, which enable the sharing of videos on the social plat-
forms and, simultaneously, increase views as more people click to view the
videos. This is a form of symbiosis between content and social features.
Moreover, when platforms are differentiated, they can “share” users even
when outsiders consider them competitors in the same market. Mital and
Sarkar (ibid.) put forward the example of Facebook and LinkedIn; both are
social platforms, but the creation of connections is complementary as the for-
mer is specialized in private (i.e., strong) ties whereas the latter is for profes-
sional (i.e., weak) ties. Consequently, these features explain why platforms do
not expect exclusivity.
Third, the startup might aim to create embedded platforms (i.e., a platform
within a platform); for example, a dating application in both Facebook and
Google Plus. In this strategy, the payoff results from a spillover effect; some
fraction of the overall user base of the host platform is expected to convert to
206
A technological means to elicit information from various sources, such as public websites and
databases.
189
users of the embedded platform. In addition, brand effects can carry over when
a presence is established in a host platform
207
.
In one of the post-mortems, selective integration (i.e., sourcing data but not
customers) was employed to explain how a competitor was able to solve tech-
nical problems more rapidly and, therefore, produce more benefit to custom-
ers
208
. This suggests that it is possible to join a platform to source information
while retaining control over customer relationships. A potential application of
this solution is to keep the platform’s core technology proprietary while
reaching into online marketing channels; for example, by applying search
engine optimization (Berman & Katona 2012), social media marketing, and
inbound marketing
209
. Although successful application of these tactics might
not be easy to achieve, dominant online platforms are currently easy to access;
for example, search engines (i.e., content platforms) index all websites
210
, and
social media sites enable creation of open communities and fan pages.
From the host’s perspective, remoras are complements. Should they become
substitutes, the host’s attitude might quickly change to being overtly hostile.
As a consequence, within-platform integration sets specific boundaries to plat-
form design; the more active the platform owner is in defending its interests,
the less maneuvering space remoras generally have
211
. Diversification is facili-
tated by inter-platform competition and interoperability through APIs. Host
platforms are open to invite remoras because complements add demand-side
utility (i.e., indirect network effects) that the platform can tax. It is seemingly
a win-win situation, although exclusivity to one platform reduces space for a
remora’s strategic maneuvering, and might be demanded by hosts under some
circumstances (cf. Armstrong & Wright 2007). Aggarwal and Yu (2012) note
that interoperability can be utilized to replicate the host platform’s network
effects due to the fact that the remora accesses the same user base as the host.
They give the example of Google Social Circles that suggests to a new user
207
Consider the popular game Mafia Wars which started independently but become known, and
after integration to Facebook multiplied its user base.
208
“That one mistake (not using or replacing Yodlee [platform] before Mint had a chance to launch
on Yodlee) was probably enough to kill Wesabe alone. […] Everything I’ve mentioned […] are great,
rational reasons to pursue what we pursued. But none of them matter if […] a shorter-term
alternative is available.” (Hedlund 2010).
209
SEO aims to increase one’s search engine ranking, whereas inbound marketing aims to reach a
strong presence in communities relating to the startup’s industry (Halligan & Shah 2009).
210
For example, Google scans websites that provide their content, and in exchange receive free
traffic, also known as organic as opposed to paid traffic. In this model, Google is a remora that
employs aggregation to tap into content platforms to retrieve sites in which its user base is interested.
Any platform that has content which it allows Google to index is also a remora that receives the
organic traffic.
211
Consider Microsoft that declined to support Intel’s hardware solution due to its proprietary
software which could have risked Microsoft’s hegemony over the platform (Evans et al. 2006).
190
the option of replicating his or her existing social network structure, retrieved
by accessing Facebook.
There are also some downsides to diversification. Most importantly, the
remora faces additional costs of accommodating different platforms. As noted
by Hagiu and Yoffie (2009), downsides include “the extra engineering, mar-
keting, and support required to play with several MSPs [multisided plat-
forms].” The integration costs depend on the scope and depth of integration. In
marketing integration, the costs are relatively low, whereas building separate
functionality for different platforms is costly. This limits the effectiveness of
the solution, especially for startups subject to resource constraints. In general,
the startup’s ability to maintain a portfolio of different technologies in various
platforms involves resource constraints relating to team skills, time-to-market,
and finances. Finally, as a technical solution to social problems (e.g., content
sharing and user-driven user acquisition) aggregation has severe limitations
because these problems cannot be solved programmatically. In other words,
aggregation as a means to collect and treat data can be highly beneficial, but it
might not necessarily be sufficient to kick-off user interaction with the content
that is necessary to harvest UG effects.
In sum, aggregation from many sources is more effective in reducing the
host’s power than multihoming when there are a limited number of choices
(i.e., the market is oligopolistic). If, for example, there is an oligopoly in social
platforms, the dependence is larger due to fewer substitutes; however, assum-
ing that these platforms compete, they have a strong incentive to provide at-
tractive terms to complements, and thus dependence on fewer hosts can lead to
better bargaining power for remoras. As such, intense inter-platform competi-
tion is likely to curb hosts’ opportunistic behavior.
4.7.4 Discussion
In platform markets, remora’s curse can be regarded as a tradeoff between cre-
ating a platform and joining an existing one
212
, whereby the opportunity cost
of creating a platform is foregoing the customer base of an existing platform,
and the opportunity cost of joining a platform is the loss of power relating to
technology, monetization, and marketing choices, which are dictated by the
platform owner at its convenience. For example, the platform owner might
decide to favor some startups over others in terms of visibility within the plat-
form. It might also study them and decide to offer a substitute, sometimes
212
A platform startup will effectively build “a platform within a platform”, although it inherits the
rules of its parent platform, and therefore remora’s curse applies.
191
termed ‘absorption strategy’. The remora tradeoff is therefore a strategic
choice; instead of creating a platform, a startup decides to join one, thereby, in
theory, eliminating the cold start problem because the platform functions as a
channel for distributing the service and acquiring new users. More precisely,
the strategic choice of building on an existing platform grants benefits such as
saved development time, cost, and direct access to an existing user base and
content. Some of these benefits can become a source of competitive ad-
vantage. In other words, the choice to engage or not is also influenced by a
startup’s competitive strategy.
If the platform is “fair”, it does not take advantage of its power advantage.
However, the state of fairness is not necessarily stable, and eventually the plat-
form might behave strategically; for example, it would prevent a startup from
gaining such excessive rents that the platform itself would become dependent
on the startup. In other words, although it can be achieved, success in a plat-
form market is capped for remoras. Finally, although not explicitly mentioned
by informants, platform-reliant startups might be involved in a race to the
bottom if the platform they have chosen is replaced by another platform
213
.
Although platform dependence is embodied in remora’s curse, the benefits
might counterbalance this risk, at least in the medium term, which seems to be
suggested by some anecdotal evidence, including success cases and what if
statements
214
. By not leveraging a pre-existing platform, a startup faces other
dilemmas, mainly monetization and cold start dilemmas, in addition to its
variation of the lonely user dilemma. The benefits of the platform relate to
customer acquisition and monetization
215
that, for both, the platform offers a
potential channel with a critical mass. In other environments, monetization can
be more challenging and customer acquisition requires seemingly more ef-
forts
216
. Further, the platform can provide competitive advantages over rivals
if it is exclusive instead of inclusive; when it is not, a remora faces intra-plat-
form competition.
Therefore, if the startup applies a platform business model, it can become a
competing platform itself by choosing the independent option, or it might turn
into aplatform within a platform by choosing to join an existing platform. In
the latter case, however, its power is restricted in the same manner as if it
213
For example, within a couple of years, Facebook replaced MySpace as the most popular social
network, growing from zero to 600 million users in a few years (Hartung 2011).
214
“So how would I do things differently today? […] I would wait until [the] location is all clean
and dandy with the carriers and build on top of that.” (Bragiel 2008).
215
The host platform’s coordination can increase WTP by making it attractive and easy; often
termed ‘attributes’ of Apple’s App Store, which collects credit card numbers upon registration and
enables one-click purchases.
216
Although within the platform, the competition is typically comparable to that outside the
platform ecosystem; that is, the startup cannot escape the need for differentiation.
192
applied a non-platform business model. In other words, the strategic choice of
integrating the product on top of a platform leads to a degree of platform de-
pendence that, in turn, increases competition against other firms within the
platform
217
, and also puts the startup at risk of the platform’s strategic deci-
sions, which naturally deviate from the startup’s goals in some aspects. This
can result in changing, for example, the rules (i.e., terms of service), technol-
ogy, design, or behavior of the platform, increasing API cost or restricting ac-
cess, introducing a directly competitive feature (i.e., substitute), thereby mak-
ing the startup’s product redundant, or acquiring a competitor and then favor-
ing it within the platform, for example, by providing it with more visibility. As
there are some vivid examples of this strategic behavior, it has been widely
discussed in the startup community. Further, building a proprietary technology
to integrate into the platform imposes the opportunity cost of developing an
independent product, although in some cases a startup might resort to diversi-
fication to solve the dilemma
218
.
However, the sample demonstrates that the multihoming process is not au-
tomatic, even if transference logic is assumed. There is a need for more re-
search to identify the conditions for successful transference of business logic
across contexts. This study, however, demonstrates that such a problem exists
and, consequently, a working business model in one context (e.g., location)
might not generalize, or at least requires “starting from zero”, as the new mar-
ket is not part of the same network.
Especially, technologically oriented founders, who are sometimes charac-
terized by a lack of interest in marketing (see Roberts 1990), might perceive it
as feasible to delegate marketing and distribution functions to the platform
owner. However, they might fail to acknowledge that, within-platform, they
arealso subject to competition, and so the need for differentiation by, for ex-
ample, marketing, will not dissolve. In addition, marketing is needed even
when externalizing user acquisition for the platform owner, due to the plat-
form owner’s incentives to favor high-performing remoras
219
.
217
Assuming the platform is open and there are little or no barriers to entry.
218
Diversification can be regarded as developing a separate individual product for an independent
marketing channel (i.e., website) or developing for other platforms. Such a diversification reduces the
risk of lock-in, but does not eliminate remora’s curse because the startup is nevertheless investing its
scarce resources into the chosen platform(s).
219
As the platform owner’s revenues are directly proportional to participants in the platform, it
ensures most prominence to participants that generate most revenue. In the case of indirect
monetization, the revenue can be substituted with impression, clicks, or other types of economic
value. The notable limitation is that the incentive does not always lead to action, which is the case of
the “fair” platform owner, although the stability of such a state always represents a risk for the startup
as it has no means of influencing the platform owner’s strategic choices.
193
4.8 Summary and discussion on dilemmas
The idea for dilemmas was born from reading the material. Initially, it was
discovered that founders identified and named them in their post-mortem sto-
ries. For example, one founder mentioned a “cold start problem”, which was
then also discovered in other cases, although not by the same name.
Table 27 Applicability of dilemmas across platform types
Cold start Lonely user Monetization Remora
Content platform x x x
Social platform x x x x
Exchange platform x x x x
The cold start dilemma is applicable to all platform types considered in this
study; a particular number or amount (i.e. critical mass) of users or content is
needed to evoke willingness to join a platform, whether the platform is based
on content, social, or exchange interaction. Both the cold start and the lonely
user dilemmas are chicken-and-egg problems, and relate, respectively, to con-
tent platforms and social platforms, the latter requiring an active user base or
content to generate growth through network effects. In addition, marketplaces
(i.e., exchange platforms) face liquidity needs. It is negligible whether
liquidity in their context is understood as content (e.g., product listings) or us-
ers (i.e., buyers and sellers).
The cold start dilemma relates to content platforms with interaction such as
content creation and consumption, and also transactions in the context of ex-
change platforms, whereas the lonely user dilemma relates to social platforms
with interaction such as joining the platform; typically, users register or oth-
erwise subscribe as followers. Both, however, aim at user generation (UG)
effects, so that users’ actions lead to a desired response from other users, such
as content contribution, sharing, and invitations
220
.
Moreover, the cold start dilemma can be defined as a problem of one-sided
content platform, when users are homogeneous, or a two-sided problem, when
users are divided into consumers and contributors of content. Similarly, the
lonely user problem can be a problem of similar side critical mass (i.e., friends
or acquaintances are required to join and actively utilize the platform), or a
220
If the users are classified as one group, it is termed a one-sided platform. If they are classified as
two complementing groups, it is termed a two-sided platform. If they are classified as three or more
groups, it is termed a multisided platform.
194
two-sided problem (e.g., men and women finding each other in a dating web-
site). The only platform type that is categorically two-sided is the exchange
platform, which always requires different sides (i.e., buyers or sellers) for in-
teraction to take place.
In terms of implications, it is important to distinguish pure content plat-
forms from social platforms because contributing content can be regarded as
more demanding than engaging in social interaction; thus, different types of
incentive might be required. Then again, for exchange it is important to build
liquidity; a good volume of both sellers and buyers, so that goods are sold at
appropriate prices. The incentives of the platform owner and traders are usu-
ally well aligned as the rewards of exchange platforms tend to be tied to the
volume of transactions taking place in the platform
221
. Finally, social effects
are associated with UG; users, for example, upload videos on YouTube for
others to watch, not primarily to gain economic benefit
222
.
The monetization dilemma and remora’s curse are applicable to all platform
startups; the company needs to be financed which requires direct or indirect
monetization, that is, charging the user for access and/or usage or charging a
third party, most typically advertisers. In a similar vein, it depends on the
user/content acquisition strategy whether the remora model is applied and
therefore applicable, which is possible in all platform types: content platforms
can attempt to source content, social platforms users, and exchange platforms
product listings.
It is typical that attempts to solve one dilemma result in the discovery of
another. This principle is demonstrated in the following figure.
221
However, this is not always the case; eBay takes a commission but the Finnish auction site
Huuto.net only charges for premium services while also monetizing by offering advertising space.
222
Although YouTube offers a partnership program for the most popular content providers.
195
Figure 15 Dilemmas and associated problems
If solving the cold start problem or lonely user problem by offering a free
product [1], the startup faces the monetization dilemma [3]. Therefore, even
successfully building the user base does not guarantee business viability. This
is due to the fundamental difference between a customer and a user; the
former brings in revenue, while the latter brings a cost that needs to be
covered by indirect monetization. There is a discrepancy between the growth
of the user base and growth of revenue that is a consequence of indirect
business models not being perfectly elastic to the growth of user base. This is
implied, for example, in Goldfarb's (2003) model, based on the assumption
Cold start dilemma
Freefying Remora
Remora’s
curse
Monetization
dilemma
r
u
n
w
a
y
Paid product
Freemium
Feature
definition
problem
Problem of active
use
sol ved
Problem of
quality variance
Problem of
free
Illusion of
free
1 2
3
4
6
7
8
11
1
10
9
13
14
17
Lonely user dilemma
Transferability
problem Real-time problem
16 15
196
that users do not provide the revenue directly but it comes from advertisers
223
.
It then follows that users are not worthless while also not being as valuable as
many startup founders would like to think. Based on the author’s analysis,
seeking customers, even at the risk of “scaring away” users who are unwilling
to pay, seems a more recommendable strategy
224
. As a minimum, the startup
should look for ways to diversify indirect monetization instead of being de-
pendent on advertising. Further, the lack of consideration for business viability
also concerns the platform literature; for example, Evans and Schmalensee
(2010, 5) noted “we do not address whether a platform that attains a critical
mass would in fact be profitable; this would require the explicit consideration
of costs and other revenue.”
To solve the cold start dilemma, the startup is tempted to join a platform
with a pre-existing user base [2], anticipating that the barrier for users to join
is lower when they have already committed to the host platform. This comes at
the cost of giving away power (i.e., remora’s curse [3]). Whereas remora’s
curse addresses managing a relationship with the platform owner, the cold
start dilemma relates to becoming the platform owner. Particular problems of a
remora include platform dependence and potential hold-ups. Realization of
remora’s curse, that is, the host platform cutting access [6] to users or content,
will in effect lead the startup back to the cold start problem, but only given
that it has failed to reach a critical mass.
When the startup solves the monetization problem through the freemium
model [7], it is left with a problem of feature definition [8]. In other words,
giving away too many features leads to low conversion from paid to free user,
whereas giving too little away leads to lack of adoption in the first place. An-
other option is paid product [9], although this can lead to a similar problem of
lack of adoption (i.e., cold start). Note that there are two different states for
users’ WTP: positive and negative. For negative WTP, paid products always
result in defection, and there is a problem with free [10] because the startup is
forced to subsidize. However, if WTP is positive, then the startup risks an
illusion of free [11] in which it offers a free product, even though the users
would have been willing to pay.
With regard to users, different problems arise before and after they join a
platform, so that:
223
The assumption can be extended by arguing that the advertising market has its own dynamics,
which means that users in one website are not interchangeable with those in another with regard to
their advertising value. For example, consider Friendster that, when selling advertising space to
American companies, noticed their visitors mostly comprised Filipino consumers (see Chafkin 2007).
224
This line of thinking is based on the idea that not all startups can become category leaders (e.g.,
Facebook) that are able to accumulate hundreds of millions of page views per day, and thus attract
advertisers’ interest.
197
Before joining ? cold start dilemma
After joining ? lonely user dilemma, problem of active use & quality
variance
Theproblem of active use [13] implies that even after solving the cold start
problem, the startup is at risk of losing the achieved critical mass if the users
become inactive. This can cause ‘negative tipping’, which is essentially the
reverse of exponential growth. The problem of quality variance [14] will, in
effect, require the startup to introduce either manual or automatic monitoring
mechanisms. In the ideal user generation (UG) model, it is assumed that the
user base is self-controlling; thus, the platform offers tools such as a reporting
function and recruits some active members as moderators of quality. However,
even if the users are active in keeping misconduct in check, the problem arises
when the low-quality content is not malicious but otherwise not interesting to
other users. For example, consider the case of an indie music portal that failed
due to low-quality bands (Hagiu & Wright 2013). It seems reasonable to as-
sume that, in some cases, the startup needs to incur monitoring and interven-
tion costs to assure that the user-generated content matches the interest of
other users
225
.
Coincidentally, the runway [17] keeps depleting while the startup deter-
mines the problems. If founders are unaware of platform-specific issues, as
many of them were in the sample, it will take them some time to understand
the problem, and then some more time to think of potential solutions. Then,
they might run into additional problems as displayed in Figure 17. In contrast,
by being aware of potential risks, the startup is ablea priori to prepare a range
of solutions for multiple dilemmas at the same time.
Furthermore, relying on UG aggravates the cold start dilemma. Instead of
in-house production or syndication through partners to acquire customers and
content, the startup expects users to play this role. When the process fails, the
startup can find itself looking for “plan B”. However, at this stage, it might be
too late, as exemplified by one startup’s story:
"We modified our technology to be a very flexible and scalable
platform from which we could launch any type of application, for
any client, in any industry. We thought we could position our
solution as helping brands create a comprehensive distributed
touch point strategy by complementing their presences on Face-
book and Twitter with a presence on IM [instant messaging].
The plan was to partner with marketing agencies as well as sell
225
For example, refer to YouTube’s tactics of getting video material from attractive women by
posting on Craigslist (Evans 2009a, 113).
198
directly to clients similar to the approach taken by providers of
custom branded widgets, Facebook apps, and mobile apps. This
strategy eventually produced some great results but it was a case
of too little, too late. When we finally decided to pivot we had al-
ready spent most of the capital raised in our seed round."
The end of the runway signifies failure. In the absence of financial buffers,
the runway might not provide a sufficiently long period of time to solve the
problems. In contrast, venture funding, although providing resources, can lock
in some choices, which prevents a later adaptation (i.e., pivot). Furthermore,
venture funding can impose a situation of “go big or go home”, which might
negate the apparent freedom afforded by the funding
226
.
Finally, there are two specific problems associated with the lonely user di-
lemma in Figure 15. First, thetransferability problem [15], which implies that
a critical mass is not automatically transferable from one context (e.g., loca-
tion, niche market, or demography) to another context (e.g., another city or
user demography). Second, thereal-time problem [16], which implies that, in
particular circumstances, the emergence of a match between parties of a two-
sided platform (i.e., network effects) is dependent on time. An empty chat
room is an example: no matter how many users have registered, if none are
present, their value at timet is zero for the only user.
This also marks how the cold start and lonely user dilemmas differ: content
is static while social interaction is dynamic
227
. Registration does not guarantee
content production (e.g., becoming an active user) and content production
does not necessitate registration or other type of subscription. Therefore, the
root of these two motivational problems differs. Simply put, it is assumed that
users do not generate content for exactly the same reasons that they join a
social network, although there might be an overlap. More precisely, their
behavior can involve spillover effects, as implied in Chapter 3.3.
In sum, this chapter has shown empirical grounding to the chicken-and-egg
problem presented in the platform literature. More importantly, the study has
shown that the problem 1) can take specific forms (i.e., cold start and lonely
user) based on the type of coordination required (e.g., timeliness), and 2) is not
isolated, although some of its potential solutions applied by the failed plat-
forms startups are associated with further dilemmas; for example, the moneti-
zation dilemma and remora’s curse. This is an important finding as most of the
226
This was conceptualized as “Peter Pan’s dilemma”, although is not discussed thoroughly in the
study (see Chapter 4.1).
227
However, content can have different modes of freshness. A good treatment to the topic is given
by Kim and Tse (2011) who study knowledge-sharing markets and argue that there is both knowledge
that expires rapidly and knowledge that remains valid for a long time; although, while the content is
static in both cases, its benefit to the user is dynamic. For example, consider yesterday’s news that is
not so valuable today.
199
literature considers the chicken-and-egg problem in isolation. It is argued here
that potential solutions can aggravate the platform startup’s problems in the
big picture; for example, by denying monetization or making it dependent on
the host platform’s strategic choices. Hence, solving the cold start problem can
come at a significant cost, and thus 3) potential solutions need to be consid-
ered in terms of their impact on cascading dilemmas and problems.
201
5 SOLVING THE DILEMMAS
5.1 Introduction
As described in the method chapter, after several rounds of GT analysis, the
researcher reverted to the data, and coded 1) “what if’” statements from
founders; that is, what they would have done differently; and 2) the attempts
expressed in the post-mortems to solve the problem when it had been identi-
fied. This process was accompanied by interviewing six startup founders. The
proposed solutions are synthesized with the platform literature, and their
strengths and weaknesses discussed. In addition, separate solutions that arose
from the literature are analyzed in terms of their appropriateness. Finally, a
summary is presented.
The solutions here do not relate to pricing, subsidies, or integration into a
larger platform (i.e., remora), as these solutions and their strengths and weak-
nesses have already been discussed in the previous chapters. It is also note-
worthy to mention that in most solutions, sides of the platform are treated sep-
arately. Essentially, if growing each side separately from one another is taken
as a goal, the chicken-and-egg problem transforms into classic marketing
problems: "How to acquire customers?" and "How to build awareness?". This
vastly expands the scope of solutions as an array of marketing tactics (e.g.,
promotion, personal selling, and various means of digital marketing) becomes
available.
Despite this premise, a startup is forced to consider both market sides (i.e.,
sets of customers) to generate any action on the platform as their interdepend-
ence remains, regardless of the applied user-acquisition methods. However,
some observed solutions are now discussed.
5.2 Solutions
5.2.1 Exhibits
The solutions discussed in the following are formulated and given names
based on the post-hoc analysis. Exhibits of these solutions are presented in
Table 28.
202
Table 28 Exhibits from post-hoc analysis
ID Insight
[A] "One of the unseen benefits of the new system was that it enabled us to anonymize, extract, and
aggregate bookmark data. So we dove into that and started looking at what products we might
be able to deliver powered by the “corpus” of what would soon be 100 million bookmarks."
(Agulnick 2010).
"This was no mean engineering feat. We had a very, very large and complex back-end. And
even with this, the quality of the data coming through to the end-user was just not that good.
Too much spam, still. Duplicate posts. Sometimes mis-categorized. Difficulty applying our
reputation algorithms. Not good." (Ehrenberg 2008).
[C] "[The startup] employed a group of talented journalists and community representatives who
sought out and interacted constantly with members of each of our communities to encourage
them to participate." (Potts 2007).
[D] "The main failure of [the startup] was marketing. Dev and I came from PayPal, a strongly viral
product at a company almost hostile to marketing. Our efforts in SEO [search engine optimiza-
tion], SEM [search engine marketing], virality, platforms, PR, and partnerships weren’t terri-
ble, but drawing users to a live event requires constant, skillful work." (Goldenson 2009).
[E] "Like creating content, I no longer think marketing is something smart novices can figure out
part-time. As the Web gets super-saturated, marketing is the difference-maker, and it’s too deep
a skill to leave to amateurs […] Next time we’ll raise enough to hire a marketing expert early."
(Goldenson 2009).
[F] "We struck upon the idea that if we had fifty journalists, and they each cross-promoted each
other to their social networks, then over time we would get more and more people to read each
other’s content." (Biggar 2010).
[G] "Because we were basically calling on friends of friends who ran events to be our customers, we
didn’t learn what event organizers in general wanted or how to acquire them as customers in a
scalable way with the 'private social network product'." (J ohnson & Fraser 2010).
[H] "We could have and should have used the proceeds of the convertible note to get out from under
Facebook’s thumb rather to invest further in the Facebook platform." (Rafer 2009).
"Since the service was our child we were reluctant to make a decision about closing it faster
and limit losses. We’ve been tricking ourselves thinking that everything would be fine while we
couldn’t get the application back to work properly." (Brodzinski 2009).
[J] "[The startup] was designed as one community, but it really was a network of unaffiliated com-
munities. […]. The site was not optimized for that. We should have had more tools for assign-
ment creators to tie their contests to their existing communities." (Powazek 2008).
[K] "It’s very, very difficult to start from scratch in a community and get to critical mass without
help. For a variety of reasons that made sense at the time, [we] chose not to go the media part-
ner route. But as newspapers and broadcasters have become more savvy in the past few months
about their need for hyper-local efforts, it makes more sense for hyper-local entrepreneurs to
hook up with media partners […]." (Potts 2007).
[L] "Figure out the difference between a website, a service, a product, an application and a plat-
form. You need to figure out which one you’re building because what users do with each one
and how you make money is very different. If you answer all of the above, you’ve got a problem
because the answer determines why people use what you’re offering, and it says that your focus
is scattershot. The difference between these is another post entirely, and one I’m probably not
qualified to write yet." (Hemrajani 2010).
[M] "To serve investors and the entire ecosystem who we heard from, we launched CB Insights, a
subscription platform that offers faster, friendlier, comprehensive intelligence about private
companies. It was built after talking to customers this time […]. To serve entrepreneurs, we
remade ChubbyBrain as a place to leverage our data in ways that would benefit them. Our first
tool is the free Funding Discovery Engine (FDE), which emerged directly from the question we
repeatedly got from entrepreneurs to the old ChubbyBrain […]." (The Chubby Team 2010).
203
[N] "I had to deal with, while building BricaBox, why we weren’t modifying an existing Open
Source solution, like WordPress MU. We were a CMS [content management system] at heart,
after all. Next time, I’ll give more consideration to building off and participating in existing
Open Source project." (Westheimer 2008).
[O] "Another reason PlayCafe’s complexity hurt us is that developing good content and technology
simultaneously required too much time. We tried to make each deep and stable — important, we
thought, given our live nature — but we were too slow to iterate in a novelty- and entertain-
ment-based business." (Goldenson 2009).
[P] "The easiest way to avoid chicken-and-egg problems is simply to have a product that is useful
on its own. Del.icio.us, for example – it’s just a bookmark manager that happens to be more
useful as more people use it." (Tang 2008).
[Q] "The way to do it seems to be to make sure you're passionate enough about your own product to
use it yourself (like submitting your own stories to Reddit, or how we all created 3-4 sockpup-
pets at in Asphere and had conversations with ourselves on the forums) and to get out there and
put it in front of lots of other people. It’s that last point where we failed: we just kinda built the
product, launched it, and let it die." (Tang 2008).
The following sections address the treatment of the exhibits and interviews
in alphabetical order.
5.2.2 Advertising
Advertising emerged as a topic in the interviews. Especially in platforms that
serve consumers (i.e., side A, demand) and companies (i.e., side B, supply), it
was perceived as important that the platform owner conduct marketing to at-
tract consumers. Often, founders simplify marketing to mean advertising. Alt-
hough marketing comprises much more than advertising (e.g., Hunt 2002), it
is discussed here as a possible solution.
In general, there are two approaches to advertising:
Strategy A: mass media advertising
Strategy B: targeted, niche advertising
While the former is generally considered too expensive for startups, the
latter seems a more feasible tactic. In fact, both were mentioned by the inter-
viewed founders. One of the founders explained that their local market in Af-
rica has quite low mass media advertising costs, and that if they were able to
acquire funding, they would employ it to drive adoption through mass media.
Another founder mentioned that they only employ targeted, low-cost adver-
tising. Another interviewed founder mentioned, early in the interview, "all
morning we have done Facebook advertising" and revealed that it is the most
cost-effective marketing channel for them. This was interesting, as most
startups placed Google before Facebook.
204
In the discussion, it came apparent that marketing "super platforms" exist
that enable almost anyone connected on the Internet to be reached. In particu-
lar, two alternatives were discussed, Facebook and Google:
· Facebook: enables targeting social network users, based on their
demographic information and preferences (i.e., likes), with organic
and paid messages.
· Google: enables targeting searchers, based on, for example, keywords
used and location, with organic website content and paid ads.
Characteristic to the low-cost advertising approach are testing with small
budgets, carefully calculating the cost of conversion, and attempting to find
keywords or demographic niches that are more likely to attract users to the
platform.
In general, the marketing platforms require very little startup capital, and
enable freedom in managing the budget and also highly advanced targeting
functionalities:
Google AdWords ? search intent
Facebook Ads ? demographics and preferences (i.e., likes)
Somewhat consistent with the earlier division, these two often mentioned
marketing channels are distinct in terms of motive of usage (Google ?
content, and Facebook ? social interaction).
The main limitations of niche advertising are: 1) user acquisition through
linear growth; that is, the more users are wanted, the more it will cost. Often,
this results in issues, as the user acquisition cost is higher than the immediate
or lifetime revenue of the user, which is particularly relevant when the plat-
form offers free access and usage. In such a case, there is incompatibility: paid
user acquisition and free offerings can easily tilt the finances of a company
into a critical state; 2) there are natural limits to the size of a niche: if, for ex-
ample, a niche is based on a particular interest in a particular location, the
growth potential remains limited
228
; and 3) platforms are expected to accumu-
late new users as interaction becomes self-sustainable, and therefore advertis-
ing can be perceived as a kick-off or temporary solution, not one that is sus-
tainable or structural.
As a consequence, the following proposition can be formed:
228
Although this feature is a consequence of small market size, it is nevertheless a limitation.
205
Proposition: Advertising is a successful solution to the cold start dilemma
if it leads to exponential, not linear, growth of user base and interaction.
Other limitations of advertising are resources, know-how, and return on in-
vestment. Startups tend to lack marketing skills and budgets. For example, the
interviewed startup which mentioned Facebook as a cost-effective method of
user acquisition is in close partnership with an advertising agency that pro-
vides them with marketing services on an on-demand basis. Based on other
investigated cases, this is a rare luxury among startups. Return on investment,
also termed ROAS (i.e., return on ad spend), measures how well advertising
investments generated revenue. As noted by one of the interviewed founders,
if paid user acquisition is applied, the platform's revenue potential must be
high
229
.
5.2.3 Aggregation
The tactic applied in [A] was to aggregate data from various sources into a
single "corpus" of content that would be valuable to the platform’s users. Alt-
hough this tactic seems to have some potential, the example of [A] vividly
shows the linkage between the cold start and monetization dilemmas: that
solving the cold start dilemma by aggregating data leaves the question “how to
monetize this data?” unanswered.
First, it is not self-evident that the data, or content, per se are valuable to
users and, second, that they would be willing to pay for accessing it. Further,
as described in , the quality of the aggregated data can become problem-
atic. According to the ideal user generation (UG) model, deploying users to
“clean” the data is a potential solution, although this can be problematic given
that users might want high-quality data but might not necessarily want to pro-
duce/edit it. Thus, the cold start dilemma is effectively not solved by aggrega-
tion unless users consider the provided content suitable for their needs.
5.2.4 Community
If theoretical network effects do not materialize, the ideal UG model fails and
the startup will be in trouble. Consider the case reported in [C]: for the
purpose of kick-off, or community building, the approach seems logical and
sound. However, it simultaneously restricted the startup’s opportunity for
229
CAC (customer acquisition cost) 50) was larger
than the original sample, it was possible to gather useful industry insight that
enhanced the findings.
The discussions with founders outside the sample extended over a period of
more than three years, and involved active participation in startup events in
Finland, Sweden, and the United States. During these events, the researcher
conversed with founders and aimed to develop his strategic thinking, and also
an understanding on the circumstances and rationale of startup decision-mak-
ing.
264
Fourth, comparison with the extant literature helps to confirm findings of
inductive studies (Miles & Huberman 1994). According to Eisenhardt (1989,
544), based on potential conflicts “readers may assume that the results are
incorrect (a challenge to internal validity), or if correct, are idiosyncratic to
the specific cases of the study (a challenge to generalizability).” There were
no major conflicts with the constructs arising from the platform literature; in
fact, the strategic dilemmas elicited from post-mortems can be connected
rather easily to the platform literature. It was found that the literature
confirmed and deepened the analysis by presenting several strategic solutions.
In contrast, the empirical analysis sheds light on the dilemmas’ relationships in
the context of online platforms. The two approaches complement one another.
Fifth, the stories are public and can be accessed by anyone. Other research
might confirm or refute the conclusions made in this study, and “readers may
apply their own standards” (Eisenhardt 1989, 544). Hidden information can be
regarded as a major obstacle in assessing the credibility of research (Sommer
& Sommer 1992). Therefore, public data are an advantage for evaluating the
credibility of this study. General principles of scientific inquiries include repli-
cation (Easley, Madden, & Dunn 2000). For this purpose, and to demonstrate
the logic applied in the analysis, we have provided a coding paradigm that lists
the codes and their meaning. Thus, another researcher can internalize the cod-
ing structure and repeat the study in question. The coding guide can be found
in Appendix 1.
6.5.3 Success with theory
As the study’s purpose is to create a substantive theory, the evaluation of GT
should also consider the extent to which this attempt is successful. Kempster
and Parry (2011) note that a constant concern in GT research has been the ina-
bility to raise the abstraction level, while remaining “stuck” in description.
Although it cannot be stated that strategic problems as a core category explain
all similarities and variation (Kan & Parry 2004), the author maintains that,
according to his interpretations, these problems represent a remarkable pattern
in the data and, combined with the insight of the platform literature, arguably
influence the success or failure of any given platform startup that is compati-
ble with the online-specific typology presented in Chapter 3.
According to Charmaz (1990, 1164), “a theory explicates a phenomena,
specifies concepts which categorize the relevant phenomena, explains rela-
tionships between concepts and provides a framework for making predic-
tions.” Making exact predictions through our framework requires skill. How-
ever, the purpose is indeed to provide such characterization, or a dilemma
265
roadmap. The extent to which founders are able to understand the results of
this study is difficult to know. However, in verbal discussions with them,
identifying these specific dilemmas has intuitively resonated with the 50-or-so
founders with whom the author has conversed.
According to Wagner et al. (2010), an “adequate use” of GT would identify
which approach (i.e., Glaserian or Straussian) the researcher has followed,
mention the specific GT techniques employed, and generate real theory as op-
posed to case descriptions. This study was positioned in the internal GT debate
(see Chapter 2.4), explicated its use of GT techniques (Chapter 2.4), and
aimed to transcend description by conceptualizing “lasting” dilemmas. How-
ever, as a substantive theory, the theory presented here is limited in analytical
generalizability. It depends on the reader to determine its usefulness. Argua-
bly, the study offers more insight for scholars and practitioners familiar with
the strategic problems than for others. This is simply the nature of substantive
theory, the credibility of which is a “joint venture” of the researcher and the
reader (Glaser & Strauss 1965).
According to Strauss and Corbin (1994), a theory is “ready” when there are
no more novel possibilities; that is, a point of theoretical saturation has been
reached. Theoretical saturation occurs when there are no new variations in
terms of codes and their explanation is relevant to the central construct, and
when the imaginable settings and configurations relating to the theory,
emerging from and applied to the data at hand, have been exhausted. A corol-
lary to this method is, however, that GT is never truly ready as, by constant
comparison, it can always be expanded (Glaser 1978). Grounded theory gives
a representation of reality, but this representation is not meant to be decisive as
reality and contexts keep changing
259
.
6.5.4 Saturation
Theoretical saturation can be deemed a criterion for GT studies, although its
existence or absence is difficult to verify (Gasson 2003). Accordingly, data
collection (i.e., theoretical sampling) must continue until saturation. Strauss
and Corbin (1994) refer to this as “category saturation”, implying that the core
category and its subcategories need to be exhaustively covered. Simultane-
ously, theoretical sampling is a technique of verification applied by the re-
searcher (Suddaby 2006). Saturation emerges when there are no more
259
Theory is never absolute because it can never be completely proven; it can falsified (i.e., refuted)
or verified to some extent (Hunt 2002). However, a theory has an indirect relationship to facts that, in
turn, have an indirect relationship to what is termed reality (Meyling 1997).
266
surprises that challenge the emerging coding system (Finch 2002). Suddaby
(2006, 639) states that saturation is signaled by “repetition of information and
confirmation of existing conceptual categories”, which he notes depends not
only on the empirical context but also on the researcher’s experience and ex-
pertise; in grounded theory terms, theoretical sensitivity (Glaser 1978).
Reaching saturation in this study was signaled by three indices. First, the re-
searcher found patterns; several instances of each dilemma were found across
different post-mortems, and thus the data were perceived to be sufficient (refer
to exhibits in Chapter 4). Second, discussion with founders beyond the sample
yielded no significant new insight that would have challenged or increased
understanding beyond the initial findings. Third, familiarization with industry
circumstances supported the existence of the dilemmas but did not yield addi-
tional, platform-specific dilemmas, which was a criterion in the theoretical
integration phase. Although the author evaluates that the findings relating to
the dilemmas are saturated, he does not make the same claim for the solutions
which require more research to be comprehensive. In fact, it seems that more
strategies and tactics to solve the issues emerge constantly; therefore, it is ar-
guable whether a definitely comprehensive description of them is even possi-
ble. Moreover, theory-wise, the biases should be integrated to the dilemmas,
as they seem to increase explanatory power concerning why the founders were
unable to solve the dilemmas. In this sense, the theoretical claims made in this
study do not establish a saturated whole, or a "ready" theory, that cannot be
expanded by further studies. Thus, further studies are required for the above-
mentioned purposes.
A part of the reason for employing multiple criteria to assess research qual-
ity stems from different scientific paradigms and philosophical stances (Kuhn
1970). This study represents acritical realist approach, thereby assuming that
the analyzed material reflects reality (Kempster & Parry 2011). While it has
been argued that classic GT follows a positivist agenda in “discovering” the-
ory (Gasson 2003), Glaser in his later works (see e.g., Glaser 2004 & 2008)
seems to belittle the risk of researcher’s bias in finding this theory
260
. In this
study, it has been deemed important to recognize the fallibility of the human
condition. After all, asystematic distortion in either the data or their interpre-
tation would lead to unhelpful conclusions and be contrary to the purpose of
GT, as embedded in Glaser’s (1971) criterion of “theory that works”.
260
Glaser’s logic is twofold: if an informant’s account is biased, either the bias becomes a social
process to examine (e.g., impression management) or it is irrelevant because it nevertheless influences
the informant’s actions. However, Glaser fails to explain how the researcher concludes whether a
piece of data is biased or not, and how this influences GT’s ability to account for mechanisms of
reality.
267
For example, consider the existence of asystematic bias in the post-mortem
stories. While a relativist might include this as an interpretation of the world
and accept it at face value, it is assumed in this study that a systematic bias
would destroy the applicability of the results. If all interpretations are false,
then correcting for the problems will not work because the problems were de-
fined correctly in the first instance; that is, the proposed mechanism is not
faulty. In other words, there would be some other unidentified mechanism(s)
that account for the actualization of real strategic dilemmas.
Therefore, potential biases need to be considered; not only those arising
from the data but also the researcher, given that the analysis of this study is in
fact an interpretation of interpretations of reality. Wagner et al. (2010) observe
that some researchers have employed Glaser and Strauss’ (1967) unwilling-
ness to address biases as an excuse for their own ignorance, but contend that
excluding discussion on biases is not the correct way to approach credibility. It
is important to acknowledge that interpretation is inherent in GT, although this
does not reduce the credibility of its results. Interpretations influence and
shape reality, and therefore understanding them might lead to results that can
be applied to action or employed to predict actions of others. As Glaser and
Strauss (1965, 9) state: “Not infrequently people successfully stake their
money, reputations and even lives as well as the fate of others upon their in-
terpretations.” There is potentially an unlimited number of biases in any type
of research with human respondents (Tourangeau 1984). The author has tried
to identify the major ones relating to this study, which will be discussed next
in terms of risks relating to data, method, and the researcher.
6.5.5 Risks relating to data
6.5.5.1 Ulterior motives
The “truthfulness” of data is linked in many ways to the motives for ex-
pounding them. Table 34 illustrates founders’ reasons for sharing their stories.
268
Table 34 Reasons for writing post-mortems
Explicit reasons for story Example
Reflection of what happened /
Avoidance of repeating mis-
takes
"The purpose of this postmortem is to thoroughly reflect on what
went wrong, so I, and perhaps others, will not make the same mis-
takes again." (Nowak 2010).
Inspiring other founders, practi-
cal usefulness for other startups
"A year from now this story will either be a testament to our meth-
odology or an embarrassing reminder of all the mistakes we made.
Either way, the hope is that it avoids the polish of hindsight and
will be not only inspirational, but methodically practical to some-
one considering quitting their job." (Lance & Snider 2006).
Therapeutic purpose, addresses
emotions relating to failure
"In the last five years, writing about my failures has been the best
possible therapy […] I could have managed for myself." (Feld
2006).
Introspection, making sense of
the failure experience
"This post-mortem will serve to get things off my chest, organize
my thoughts, get the most out of the experience, and share my ex-
perience with others." (Diaz 2010).
Responding to questions and
third-party interpretations
"I’m […] writing this to be able to point to a single, detailed,
lengthy answer to the inevitable questions I’ll be getting from
friends and colleagues about what happened with [my startup].
Now people can read to their heart’s content." (Diaz 2010).
Stories are generally told for specific audiences. In the case of startup post-
mortems, founders tend to address the stories to other founders (see Table 34).
In this sense, we can refer to knowledge transfer; founders are interested in
helping others to avoid repeating their mistakes. Indeed, one of the most fre-
quent explicit motives to write a post-mortem is to help other founders avoid
common mistakes. The second motive is psychological, and might serve a
therapeutic purpose: by telling their stories, founders are able to reflect on
failure, a stressful experience with which to come to terms. Reflecting on past
failures was also perceived as learning for future startup projects; some found-
ers might encourage potential founders to create a startup.
In addition to reported reasons, there might be implicit reasons for story-
telling. This study can only speculate on the founders’ true motives, which
remain hidden. For example, a founder might engage in strategic behavior
through storytelling, which can hinder the story’s trustworthiness. Social ef-
fects can also take place, as noted by a practitioner: “it’s in nobody’s best in-
terest to call attention to their own bad decisions, and it’s even less wise to
poke fun at the bad decisions of your co-workers, who may be a vital part of
the personal network that will keep you alive after the startup explodes.”
Therefore, founders might “soften” their own part in failures by omitting some
information. Moreover, the story can be written in such a way that is intended
to maintain a professional profile, for example, to impress investors or future
employers (Bansal & Clelland 2004).
269
These points are not necessarily detrimental to the credibility of a story;
secondary motives can underlie narratives, and are only problematic if the
story in question is distorted. For example, the ulterior motive of ‘appearing
experienced’ does not make the story less credible if the facts and interpreta-
tions within it are otherwise objective. In other words, the person making a
claim is to be separated from the claim (cf. argumentum ad hominem).
6.5.5.2 Self-serving bias
Several authors have recognized the risk of self-serving bias in entrepreneurial
studies. The self-serving motives risk distorting the trustworthiness of a story
by omitting personal mistakes and assigning the blame to external as opposed
to internal reasons. For example, Lussier (1996) found in his study that only
five percent of the entrepreneurs surveyed identified poor management as a
failure factor; thereby implying a tendency to blame external factors. Beaver
and J ennings (2005) advise against employing surveys to find truthful ac-
counts for failure as people are more likely to give self-serving responses and
less likely to admit personal fault. Following the attribution theory, Zacharakis
(1999) argues that individuals are more likely to attribute their own failures to
external causes (e.g., recession), and failures of others to internal causes (e.g.,
poor management skills).
In contrast to Lussier's (1996) findings, Mantere, Aula, Schildt, and Vaara
(2013) found that entrepreneurs in their case companies were ready to accept
blame, and attributed less of it to their subordinates than that attributed by
subordinates to the entrepreneurs, indicating a low self-serving bias. However,
they stress that cognitive and emotional processing relating to the failure expe-
rience influence failure narratives; namely, grief recovery and self-justification
(Mantere et al. 2013). There is little that can be done to control such effects.
Nevertheless, a proxy measure of self-attribution is employed in this study.
The measure is constructed so that if a founder in his/her post-mortem explic-
itly attributes failure, at least in part, to his own actions, this is considered self-
attribution.
Exhibits based on this analysis are presented in the following table.
270
Table 35 Examples of self-attribution
Exhibit
[1] "I’d also like to thank our venture-capital investors […] who took a big risk on an unproven
concept and then took a large financial loss when we were unable to successfully execute on
that concept." (Potts 2007).
[2]
"I have a tremendous amount of respect towards everyone that I’ve worked with on this en-
deavor and do not wish to even hint at a “should've”/“would’ve” discussion. What’s done is
done. There is no way to go back. And ultimately it failed under my watch, and that is mine to
bear." (Yaghmour 2010).
[3]
"As co-founder and CEO of [the startup], the buck stops with me and no one else." (Rafer
2009).
[4]
"For one, we stuck with the wrong strategy for too long. I think this was partly because it was
hard to admit the idea wasn’t as good as I originally thought or that we couldn’t make it work.
If we had been honest with ourselves earlier on we may have been able to pivot sooner and
have enough capital left to properly execute the new strategy. I believe the biggest mistake I
made as CEO of [the startup] was failing to pivot sooner." (Nowak 2010).
[5]
"A final point that should be made is that this is not an attempt to blame anyone. The
journalists aren’t to blame: we didn’t make a sufficiently good product for them. The developer
isn’t to blame; we tried to hire someone for a startup role who had no interest in startups. No,
the only people to blame is us, and more specifically me, since I was at the helm when it all
went down." (Biggar 2010).
[6]
"When [my co-founder] had to leave the company due to a family illness, I took over as CEO
and led the company without a formal peer for the final two years. All that adds up to me hav-
ing absolutely no one to blame for [the startup’s] failure but myself, and as a result I can’t now
nor could ever be dispassionate in thinking about what happened." (Hedlund 2010).
Self-attribution refers to a founder explicitly attributing failure to self; that
is, taking responsibility rather than blaming external conditions. To be coded,
the founder had to explicitly indicate his/her or the team’s shortcoming in ex-
plaining the failure. There is a variation in interpretation as some founders
sought to attribute reasons to external factors more than others. Self-attribution
was coded in the material, and 41% of cases included references to self-attrib-
ution. Note that this does not indicate that the founder was not attributing
blame to self, only whether or not it was explicitly stated in the post-mortem.
Therefore, it is concluded that, in general, the risk of self-justification con-
siderably altering the stories is negligible. It is important to note that post-
mortems have been made public on the Internet; thus, an intentional “twisting
of facts” would risk the founders losing face and credibility (Krumpal 2011).
Founders are likely to be aware of this and increase their level of candor. The
final interpretation is thus in line with Mantere et al. (2003). As such, founders
in our sample can generally be regarded as candid in their accounts.
271
6.5.5.3 Recall bias
Apart from apparent candidness, there are other types of bias that relate to rec-
ollection and interpretation. Elliot (2005) describes recall bias as forgetting
past events or details in them, leading to deterioration of the data quality.
261
In this study, recall bias, or forgetting important points, might be less rele-
vant because the bias concerns details, not a gestalt (i.e., an organized whole
that is perceived as more than the sum of its parts) of problems experienced
(see Maitlis 2005). It is not likely that the gestalt (i.e., the whole story) would
have been falsely remembered. It is more a problem of interpretation than
memory if a specific dilemma was incorrect.
In the case of this study, all narratives were written within a year of failure.
Although this does not remove the risk of recall bias, the longer the delay in
reporting the event, the greater the likelihood of confusion (Coughlin 1990). A
year might be regarded as too short a period to forget critical details, although
sufficient for distancing one from the immediacy of failure, or grief recovery
(Shepherd & Kuratko 2009). This might, in fact, improve the quality of post-
mortems as the founders have had time to reflect, and are perhaps better able
to place their story into a “bigger picture”.
6.5.5.4 Unintentional false attribution
Moreover, false attribution can be a consequence of sense-making even with-
out being self-serving or memory-distorted (Mantere et al. 2013). This is when
a founder is unable to find the correct reasons, interprets facts incorrectly, or
there is simply no prominent explanation other than the failure was a combi-
nation of many events, some unforeseen and random. In other words, it is a
type of sense-making bias.
The fact that reports were voluntary (i.e., unpaid, not commanded) supports
this perspective as, clearly, founders wanted to share some experiences they
genuinely believed were of interest to others. If interpretations are internally
consistent, there is no reason why “amateur theories” should be perceived as
meaningless before closer examination. In fact, economics has also suffered
from time to time with regard to questioning inaccurate rationality assump-
tions (Kanazawa 1998; Nagel 1963).
However, the sense-making styledoes differ among founders. For instance,
the founder of a location-based startup was highly capable in conceptualizing
specific problems, which indicates good analytical skills, whereas other
261
In psychology, this is defined assystematic error in reproducing past events (Coughlin 1990).
272
founders were less capable of doing so or this capability was not visible in
their stories. Inherently, this results in some incommensurability of data. This
is why superficial stories were omitted from the sample; their analysis was not
sufficiently rich to provide grounds for proper theorization. In our case, the
hindsight perspective is useful as it enables reflection by founders. This prop-
erty was particularly useful in the post-hoc analysis, when the data were re-
analyzed for “what if” thoughts and proposals for solutions. After experienc-
ing failure, founders tend to be more aware of the errors committed during the
startup phase, whereas insights might be more anecdotal if the founders were
interviewed during the experience (Schoenberger 1991). When founders share
their a posteriori insights, they become a rich and valuable research material.
6.5.5.5 Communality
Another risk might be communality between the stories, so that they are not
independent in interpreting the chain of events. More precisely, if the stories
were written after first reading other stories, these other stories might have
influenced how the founders explained their own failures, potentially leading
to a systematic bias in interpretation. It is very difficult to account for such a
bias, as this would necessitate determining whether or not founders read each
other’s stories. There are indications of this behavior occurring; for example,
one founder wrote (Diaz 2010):
"There are many post-mortems from failed startups out there,
mainly because there are a lot of failed startups, and the people
that start them tend to be very introspective and public about
their successes and failures. I’m no different."
For example, data can be biased by the influence of other post-mortem sto-
ries or thought-leaders, so that founders identify challenges selectively, based
on what others have previously noted. As a consequence, they might miss
some points that they otherwise would have noticed. If peer influence is a
source of sharing motivation, the risk is that stories become contaminated by
other stories, so that interpretation follows earlier interpretation. Undoubtedly,
some communality or patterns in the data relate to this bias. For example,
founders employ vocabulary such as “iteration”, “pivots”, and “minimum via-
ble product” to refer to their failures. Some of these terms originate from fa-
mous startup thinkers who have their own approaches to failure. Thus, the
thinking of experts might have influenced the founders’ interpretations. How-
ever, if founders perceive these expert explanations as accurate, they should be
273
treated with a degree of plausibility insofar as many of the experts are also
entrepreneurs
262
. Moreover, the vocabulary utilized by Internet standards has
evolved as a combination of, for example, many startups, experts, and inves-
tors. The collective “slang” therefore cannot be regarded as a bias, but a way
of communicating within a community (Mazrui 1995).
However, communality is only a problem if an interpretation would other-
wise have been different because, due to a desire to adhere to other stories, the
founder has changed his/her interpretation. A positive sign of the lack of this
bias is included in Appendix 3, which represents support for two opposing
perspectives on whether or not to launch early. If the founders were subject to
conformity bias, conflicting perspectives would be less likely to arise. Based
on a strong dichotomy regarding early launch, and dissimilarities of details
presented in the dilemma exhibits, a systematic conformity bias (Moscovici &
Faucheux 1972) is not perceived to be a major issue.
6.5.5.6 Anchoring bias
Anchoring bias can take place if founders are overly focused on one aspect of
the story. When anchoring, individuals focus on one value, or piece of infor-
mation, over other values or information (Bunn 1975). According to Elliott
(2005), the narrative story format is prone to simplifications; that is, empha-
sizing particular information over other information. For startups this might
include overemphasizing one aspect of failure, while neglecting others. When
these other aspects are judged critical by some objective measure, the resulting
account would be distorted.
In other words, founders might highlight one specific point as the over-
arching explanation for the failure, while ignoring other aspects. For example,
they might not have a perfect understanding on the platform model, and there-
fore might not properly interpret the dynamics leading to failure. However, at
the same time, their interpretation might be more valuable as they are not
“contaminated” by platform theory, and utilize their own terminology and
experience to explicate critical aspects. Furthermore, founders are very
knowledgeable on their cases; thus, even if they highlight particular aspects,
they are more likely to have a more comprehensive picture than outsiders.
In assessing the seriousness of this bias, one has to acknowledge that em-
ploying GT, or conducting any research, requires simplification of the
surrounding world’s complexities. In this case, the stories generally included
262
For example, Paul Graham, Eric Ries, and Steven Blank are thought-leaders whose terminology
is widely applied by startup founders in their post-mortems.
274
not only one overarching explanation for failure but several of them, thereby
reducing the impression of a strong anchoring bias.
6.5.5.7 Different interpretations
Founders, customers, investors, and other stakeholder groups can have differ-
ent opinions on what actually happened. However, this study is only based on
founders’ interpretations. Table 36 presents some examples of interpretations,
collected from comments on post-mortem stories, which differ from those
found in the post-mortems.
Table 36 Examples of different interpretations
Source Interpretation
Comments in
Bragiel (2008)
"I checked [the startup] twice […], when they launched and again in a year.
There were zero improvements in that year! Still their blog was filled with crap
like 'look at our brand new and cool office'. I posted a comment asking what's up
and reporting some bugs. The comment was quickly deleted and nobody ever got
back to me. That day I knew they were dead."
"I was probably what you would call an early Chicago adopter of [the startup],
and I thought it was a great idea, and still do. With each succeeding version of
the software, however, the interface seemed to get weirder and weirder until I
couldn’t figure it out anymore."
Comments in
Goldenson (2009)
"As it pertains to critical mass, we (if you remember, The Legion Team) always
found it very frustrating that there were only a certain number of slots available
for each team. In my opinion, it was this limitation that created the bottleneck. "
"Of course, scalability becomes an issue; but it's a critical one to place at the top
of the priority list. Each week the 'ringleader' would have to go back and decide
who was getting invites that day, and it ultimately limited participation. Why?
Because we were all already part of the same community. We were all already on
the same team. We wanted to play with each other, not against each other. It was
us against the world, albeit if only 20 at a time. "
"But with the inability for all of us to come back night after night, interest waned,
and the lure of the Internet equivalent of A.D.D. [attention deficit disorder] took
hold. ('Oh, look! A chicken!')"
"Your words here are nothing short of courageous, and I have the utmost respect
for them. This was just meant to shed some insight from a user perspective on
why it was hard for us to grow as a […] community."
"The way this guy characterizes this is grossly misleading. [The startup] was a
great idea, but extremely poorly executed. They spent $600/$900k of investors’
money. [...] The quality of their video was easily eclipsed by an $800 trip to best
buy as far as technology goes. They'd often have laziness issues on cam, where
the host would just plain forget to have put new/charge the batteries in their mic,
causing long delays. Why the talent was left to do anything requiring responsi-
bility past showing up is befuddling."
275
It seems reasonable to assume that the founders were unable to fully under-
stand the customer experience (i.e., that of outsiders), or perspectives of other
third parties; as noted by Elliot (2005), narratives concern giving meaning to
personal experiences, not those of others. However, the data might be biased
because they only contains single-person interpretations, excluding the per-
spectives of co-founders, customers, investors, and other stakeholders. This
bias could be reduced by interviewing other parties, such as co-founders, cus-
tomers, and even competitors, who “view the focal phenomena from diverse
perspectives” (Eisenhardt & Graebner 2007, 28). Stories were not tested in
this “360 degrees” perspective, because the truthfulness of particular cases
was not perceived as a problem due to storytelling motives expressed by the
founders (see Table 34). Thus, this study regards founders as a reliable source
for identifying strategic challenges, despite the coexistence of alternative inter-
pretations of failure reasons.
6.5.5.8 Reverse survivorship bias
Finally, there might be a risk of a reverse survivorship bias because all cases
were failures. If we consider the existence of survivorship bias (Brown et al.
1992) as only relating to successful cases, “failure bias” is the reverse by only
relating to stories of failure. Technically, the end result might be the same, a
skewed realization of the world. Therefore, it is important to keep in mind that
the context of the observed dilemmas was failure, and failed startups. Poten-
tially, successful startups do not observe the same challenges at all, although
the author finds this difficult to believe. Alternatively, successful startups
might have overcome these dilemmas and encountered even more problems at
later stages, which remain undiscovered in this study. However, given the fact
that there were many analyzed cases, and especially that the focus is on early-
stage online platforms, implicitly assuming that overcoming these dilemmas
leads to later stages where the strategic problems are different, and purpose-
fully not considered in this study, the reverse survivorship bias cannot be re-
garded as a major risk for accomplishing the aim of the study.
Moreover, based on discussions with successful founders and also those
with undecided outcomes, it was discovered that founders beyond the sample
recognized the same issues identified in this study. This leads the author to
believe that the same dilemmas exist more widely in the context of online
startups.
276
6.5.5.9 Final assessment of data
The post-mortems included in this dissertation were written by the founders
(i.e., entrepreneurs) of failed companies, offering a useful insight on self-re-
flected reasons for failure. In other words, the primary analyst is the founder.
The researcher is a secondary analyst who synthesizes stories, aims to find
similarities and differences in them, and increases the level of abstraction
(Goulding 2005).
Many founders seemed talented at conceptualizing their problems; for ex-
ample, the concept of ‘cold start’ is taken from a founder’s terminology. How-
ever, as post-mortems are subjective interpretations of reality, their reliability
needs to be judged case by case and, when necessary, their trustworthiness
questioned.
If the stories are not truth but interpretation, how does this statement affect
their value? First, it is clear that, in human sciences, interpretations of reality
vary according to the observer or storyteller. However, this premise excludes
neither their usefulness nor their credibility as a potential explanation of
events. In fact, theories are similar in that they arepotential explanations, not
observed facts. Therefore, employing interpretations, given that they are
treated critically in the analysis, is a natural method for theory formulation
(Glaser & Strauss 1967).
Therefore, it seems that the correct way to approach post-mortems is to
consider them interpretations, not objective truth. According to Charmaz
(1990), GT is highly compatible with interpretation. For example, consider a
hypothetical startup failure and the research task of determining why it failed.
An accountant might argue that it failed because of a lack of revenue (i.e., a
correct statement); a marketer because they were unable to convert sufficient
customers (i.e., a second correct statement); and the CEO because it was a
recession when they started, and potential customers lacked the money to buy
(i.e., a third correct statement). Essentially, GT is compatible with various
epistemological stances and does not argue that interpretation means no relia-
ble information can be acquired (Charmaz 1990). According to Glaser (2002),
it is more important in GT to argue for usefulness than for accuracy.
Verifying the trustworthiness of the stories is hindered by the researcher’s
non-participation, which would have enabled a deeper understanding on a par-
ticular story (Poon, Swatman, & Swatman 1999). Lacking a better measure,
stories were included that appeared candid, fulfilled the formal criteria, and
contained sufficient detail to be analyzed. However, as noted, they might be
subject to several biases, which have been discussed in the previous
277
subchapters. Essentially, the post-mortems are founders’ interpretations, not
objective or factual statements of the “truth”
263
in a positivistic sense. That
being stated, it is not plausible, considering the often altruistic motives for
writing them (see Table 34, p. 268), that founders would deliberately mislead
readers. Rather, it is plausible that informants were not inventing strategic
business problems but reporting what they actually perceived.
Moreover, some founders seemed aware of biases and expressed a desire to
prevent them affecting the truthfulness of their story. For example, one
founder wrote (Ehrenberg 2008):
"These observations aren’t worth much. But the interpersonal
dynamics, the issues of organizational structure, the need to
change strategy in light of new information, the relationship with
key investors, all of these are very instructive. I will endeavor to
be as honest and candid as possible."
Statements of this type, along with the self-attribution analysis and explicit
motives for story writing, and also the general requirements of GT for data,
lead the research to believe that the material is of adequate quality for achiev-
ing the research purpose. Moreover, it can be concluded that they are suffi-
ciently precise to articulate the key issues of platform startups on the Internet.
6.5.6 Risks relating to method
Grounded theory has been employed in previous studies focusing on platforms
(e.g., Curchod & Neysen 2009; Palka, Pousttchi, and Wiedemann 2009;
Mantere et al. 2013). Curchod and Neysen (2009) employed GT to identify
perceived positive and negative network effects by users of eBay, an exchange
platform. Palka et al. (2009) applied it to identify users’ motives to share viral
messages in a mobile platform. Mantere et al. (2013) analyzed interviews to
build narrative attributions of entrepreneurial failure. Generally, GT is re-
garded as providing a robust method for analyzing qualitative data (Corbin &
Strauss 1990); in particular, employing a systematic approach to build theo-
retical constructs, while remaining rooted in informants’ experiences. Most of
the platform literature remains analytical as it is derived from economics
(Birke 2008; Shy 2011). Qualitative inquiries are therefore necessary to vali-
date and complement analytical models as they provide a high level of depth
(Salomon 1991).
263
Given that failure is a relative concept (Watson & Everett 1999), and its interpretation varies
among the venture’s participants, it is not clear that one truth exists. In contrast, it is more likely that
each party has their own interpretation, and the failure outcome is a combination of events and
conditions.
278
In general, purposive, or judgmental, sampling is regarded as appropriate
when the research seeks to create new concepts or theory, as opposed to test-
ing hypotheses with statistical inference (Eisenhardt & Graebner 2007).
Considering the purpose of this study, selective sampling is not perceived as
an issue. Moreover, secondary data pose no problem for GT. Goulding (2005)
mentions secondary data as one format in her methodological overview on
GT. According to Glaser (2004), “all is data” and the researcher should decide
on the most appropriate data for the study. Post-mortem stories were not only
publicly available, they also enabled a substantial and varietal sample of Web
2.0 startups to be quickly amassed. Variety was perceived to be useful for the
purpose of constant comparison, whereas having multiple informants on one
case would have enabled a thicker and perhaps more accurate description.
However, according to Glaser (2004), accuracy of description is not a major
concern in GT.
The methodology’s influence is a particular risk, meaning that results origi-
nate from the use of a particular method. As GT aims to find relationships
between constructs (Strauss & Corbin 1994), finding a relationship between
individual strategic dilemmas (i.e., categories) is a natural result of applying
the method. However, credibility is not threatened for two reasons. First, there
are connections between the problems in the real world and they are revealed
as grounded in the data; thus, they are not concocted. Second, the connections
were filtered by selective coding, with some being discarded. Applying judg-
ment emphasizes the most important connections; thus the method does not
allow a pre-determined model to emerge.
Charmaz (1990) discusses the merits and challenges of grounded theory in
more detail. In general, Eisenhardt (1989) notes that the resulting theory might
be narrow and limited to a specific context, as opposed to “a grand theory”. As
previously established, grand theory was not the purpose of this study, but ra-
ther a substantive one. In general, the chosen method was seen to provide rigor
in the analytical process, which was perceived to be vague and complicated
before learning the key principles of GT. Therefore, the researcher feels the
method was appropriate and useful. For a description of how grounded theory
was applied in this study, refer to Chapter 2.4.
6.5.7 Risks relating to researcher
In general, the nature of qualitative analysis puts strong emphasis on the re-
searcher’s judgment (Seale 1999). This is also a limitation of this study,
because the data include self-assessment a priori, and are interpreted by the
researcher a posteriori; in other words, interpreting interpretations.
279
In particular, researcher bias becomes an issue in GT through the interpre-
tation mechanism by which the researcher elicits meanings from the data
(Partington 2000). For example, during data collection, the researcher can
influence informants’ responses. This is a particular concern with the inter-
view method. Elliott (2005, 11) elaborates on this concern:
"[W]ithin certain contexts the narrator may be influenced by im-
agined or possible future audiences […]. The very fact that the
conversation is being recorded suggests that it will at least be
listened to at some future time and may also be transcribed and
parts of it translated into a written text."
In this study, stories were collected from the Internet and, considering they
were published in personal blogs targeting other startup founders, it is unlikely
that founders expected them to become part of academic scrutiny. This elimi-
nates the possibility that the researcher might have influenced the reports. The
researcher did not influence in any way the writing of post-mortems, and
therefore data collection was completely unobtrusive from the researcher’s
side.
Theoretical sampling can risk confirmation bias if one purposefully seeks
only positive answers (Nickerson 1998). It was borne in mind that feedback
might not only confirm but also conflict with the researcher’s assumptions.
Hence, it can be argued that contextualization is a double-edged sword: by
increasing domain-specific information, understanding is increased on the
conditions in which startups make strategic decisions. At the same time, con-
textualization can risk going native so that informants’ sense-making is
adopted as truth (Marker 1998); for example, the researcher might start to
overemphasize and/or neglect facts in accordance with founders’ biases. Also,
Charmaz (1990) mentions the risk of going native. This might occur when the
researcher is so immersed in the informants’ reality that he/she loses objectiv-
ity, becomes naïve, and accepts informants’ conclusions at face value. As the
author was not actually a member in any of the observed or analyzed startups,
this risk can be deemed low. Partial immersion, in fact, is beneficial to under-
standing the context; that is, specific circumstances and conditions which in-
fluence actions such as strategic choices. This is implied in theoretical sensi-
tivity, denoting that the researcher is not a tabula rasa but does have a per-
sonal, professional, and theoretical perspective on the studied phenomenon
(Strauss & Corbin 1994).
Regarding analysis, the researcher might be biased in his interpretation;
even if no harmful
264
bias exists, some degree of bias is inherent in interpreta-
tive qualitative analysis (J ohnson 1997). It is important to note that GT is an
264
Harmful as in deliberately excluding facts that would contest the theory under development.
280
interpretative method (Gasson 2003), and therefore particular risks can arise
regarding misinterpretation. However, interpretation is associated with theo-
retical sensitivity (Glaser 1978; Strauss & Corbin 1994). Relating to this
study, the author read a large amount of related information on online business
to increase his understanding on the topic, including, for example, startup-fo-
cused blogs, online community discussions, and books, which is likely to in-
fluence his understanding (J ones & Alony 2011), although hopefully in a way
that leads to more useful conclusions.
Finally, Glaser (2004), who places great faith on a researcher’s ability to
employ data in the correct way, states that:
"GT discovers and conceptualizes the latent patterns of what is
going on. It is always relevant. If a GT is accused as being inter-
pretive, which is probably meaningless, it is a very relevant in-
terpretation."
In other words, Glaser consistently approaches constructivism in his un-
derlying logic that descriptive accuracy is not as relevant as the theory’s abil-
ity to influence action. The way in which subjects perceive the world in turn
influences their action, and so perceptions/interpretation have the potential to
become self-fulfilling prophecies (Merton 1948; see also Charmaz 1990). This
perspective is adopted insofar as to state that whether or not the interpretation
is perfectly correct in description, its implications point the reader to the cor-
rect direction.
In GT, verification is not generally performed by others but by the re-
searcher through engaging in constant comparison of subjects (Glaser &
Strauss 1967) and utilizing his/her curiosity in theoretical sampling, or ac-
quiring new slices of data to challenge the emerging theory. Glaser and
Strauss (1965) place a great deal of trust in the researcher, and their message is
that researchers should trust their findings, achieved through hard work and
immersion in the context of the research subjects. Even if two analysts rarely
reach exactly the same conclusions when analyzing a topic independently
(Glaser & Strauss 1967), both can agree on each other’s works.
Although this might seem paradoxical, it only highlights the non-linear cre-
ative process associated with GT, leading to individuals emphasizing different
angles. For example, this study could have focused on founders’ decision-
making biases, or on the ideal UG model. According to the author, both would
have explained the problematic features of the failed startups. It was simply
the author’s judgment call to focus on strategic dilemmas, which can be re-
garded as no more right or wrong than alternative calls. What can be further
stated on this, however, is that the author believes the few identified dilemmas
can be employed better to explain platform startups’ failures than utilizing a
281
large number of generic failure factors that seem to be randomly scattered in
startups; that is, in the context of platform startups, their relevance is higher.
6.5.8 Generalizability
A question typically posed for any type of research is “how generalizable is
it?” The issue is tackled in this subchapter.
The highest level of abstraction in GT is termed aformal theory, which re-
lates to a phenomenon in general, whereas a substantive theory relates to a
phenomenon in a given context (Glaser 1971). For example, a theory on busi-
ness failure would be considered a formal theory (i.e., general focus), whereas
a theory on startup failure would be a substantive failure (i.e., contextual fo-
cus). To these, Glaser (1971) adds “grand theories” that can cover almost all
types of situation and exist across phenomena; for example, explaining failure
in any type of context could constitute such a theory.
According to Glaser and Strauss (1965), a theory should be judged on its
intended generalizability. In some cases, substantive theory can be regarded as
being equivalent to interim theory (Glaser 1971), or a step in the process to
formal theory. However, it can also be perceived as an independent entity, and
therefore its merits should be judged by considering the context for which it
was devised. It is noteworthy that, in this case, the researcher relies onanalyti-
cal or logical generalization when applying the theoretical conclusions to
other units in the same context (Collingridge & Gantt 2008). This process
should not be confused with statistical inference, that is, generalization from
sample to population, as it is a statistical technique for determining applicabil-
ity (Wagner et al. 2010).
Corbin and Strauss (1990) relate generalizability to a study’s reproducibil-
ity, which resembles the transference criteria discussed earlier. They (ibid.,
250) note that “probably no theory that deals with a […] phenomenon is actu-
ally reproducible insofar as finding new situations or other situations whose
conditions exactly match those of the original study, though many conditions
may be similar.” Consequently, Strauss and Corbin (1994) postulate that there
are spheres of applicability, ranging from an individual case to a local com-
munity to an international setting, and so on, towards a global pattern. In a
similar vein, Urquehart et al. (2010, 364) note that “as the researcher moves
up the level of abstraction, the range and scope of the theory increases.”
By applying these ideas, it can be contemplated that there are spheres of
applicability also for this study, which are presented in the following figure.
282
Figure 18 Spheres of applicability
Therefore, the findings of this study are generalizable, at a minimum, to the
four online platform types that tend to apply UG, at least implicitly, and indi-
rect monetization, including the freemium model in an attempt to internalize
externalities from interaction among platform participants.
Increasing the level of abstraction is generally perceived to increase a the-
ory’s applicability (Strauss & Corbin 1994). Moving away from an online
platform also means that more general theory literature becomes available,
although at the cost of losing empirical context (Glaser 1971). Generally, the
process of generalization aims to lose contextual factors and introduce a gen-
eral logic (i.e., formal theory) concerning why particular relationships hold
across many contexts. Increasing the abstraction therefore removes online-
specificities; for example, different motives of interaction manifested in plat-
form types, or commonly applied UG and indirect monetization.
This way of examining the generalizability implies that strategic problems
do not always exist; that is, if their existence depends on specific conditions,
when some contextual conditions are removed from the picture, the dilemmas
can no longer be identified. In particular, the problems addressed here require
online-specificities, or they might not emerge. For example, if the platform
charges an access fee and customers are willing to pay, there is no monetiza-
tion dilemma. Consistent with the perspective of critical realism (Kempster &
Parry 2011), the dilemmas are contingent upon the context.
This issue can be illustrated by attempting to increase abstraction on the
study’s main model, Figure 15 (p. 195), which shows the relationships be-
tween dilemmas. In Figure 19, the author followed Glaser’s (2008) advice to
abstract from time, place, and people, and only to apply general concepts. In
online
platforms
other firms
other platforms
other contexts
283
this study, a probable or most appropriate way is arguably towards business
failure in general, a formative theory on failure of sorts.
Figure 19 A tentative formal theory
The theory explains failure through strategic problems and biases. It argues
that there are strategic problems requiring solutions for the company to avoid
failure. If these problems are solved, there will be a derivative problem that
requires a solution, and so on. It is argued that solving the problems is pre-
ceded by their identification. This identification is potentially prevented by
bias; because of his or her biases, the founder or manager might be unable to
identify the strategic problems. If this happens, the company will fail. How-
ever, even if the problem is identified, its solutions might be associated with a
bias; the founder or manager is unlikely to think of a solution in the proper
way due to his or her bias. Again, the company will fail. Only by solving all of
the problems and their derivative problems can the company achieve its true
potential in the market. However, it is argued that the true potential can also
equal failure; for example, when there is no true demand.
While being logical, this approach highlights the problem of moving from
substantive to formal theory: it loses the context. Thus, there is now a general
description, but with two emergent issues. First, the theory still needs to be
applied to other contexts to determine how well it fits to them (Charmaz
1990). Second, by moving from the substantive context, the practical applica-
bility to online startups has, to a major extent, now been lost. Indeed, this
seems to contest the critical realist criteria, namely, practical adequacy and
plausibility (Kempster & Parry 2011), as there are no longer grounds to esti-
mate the contexts in which the theory works without reintroducing and study-
ing those contexts. As a consequence, practitioners are less likely to
Failure
Strategic
problem
Identification
yes no
Solution
Derivative
problem
Bias
no
yes
yes
True potential no
284
understand what this means in their context. Thus, generalizability at worst
seems to lead to a double bind of losing both context and relevance. If the
study accomplishes other evaluative criteria for GT, the added usefulness of
generalizing from substantive theory can be regarded as negligible.
This is somewhat in conflict with Corbin and Strauss (1990, 267) who ar-
gue that “[t]he more systematic and widespread the theoretical sampling, the
more conditions and variations that will be discovered, therefore the greater
the generalizability, precision, and predictive capacity of the theory.” It is ar-
gued here, based on the aforementioned logic, that while theoretical sampling
increases the generalizability of a study, it reduces precision, and that this
might decrease, not increase, the predictive capacity of the theory. This con-
flict seems also to apply when examining classic GT. In fact, Glaser (2004)
mentions that the general concepts apply in any domain where they exist, but
that they require modification by constant comparison. Essentially, if they
need to be modified when applied, then they do not apply at their general
level, and are thus not generalizable prior to being employed in action.
The Glaserian and Straussian approaches differ in their understanding on
reproducibility. While Corbin and Strauss (1990, 251) require that “given the
same theoretical perspective of the original researcher and following the same
general rules for data gathering and analysis, plus a similar set of conditions,
another investigator should be able to come up with the same general
scheme.” Glaser (2002) leans more towards variation in ability to conceptual-
ize, and perceives that some are more talented in this than others and that con-
ceptualization differs from description, which is a simpler cognitive process.
In fact, such an idea is implicit in classic GT through the concept of theoretical
sensitivity. If traits such as personal and professional experience (Strauss &
Corbin 1994) influence how well the researcher is able to elicit concepts and
theory from the data, it is unlikely that two persons who vary greatly on these
dimensions would reach exactly the same conclusions.
Corbin and Strauss (1990) fail to explain this discrepancy, unless their “the-
oretical perspective” means the exact same theoretical sensitivity. Given the
interpretative nature of GT, such a match can be considered realistic as two
identical snowflakes. In contrast, the origins of GT lie in the substantive the-
ory with its contextual score (i.e., usefulness): “The invalidation or adjustment
of a theory is only legitimate for those social worlds or structures to which it
is applicable” (Glaser & Strauss 1965, 10). In this study, the author contends
that it would be unlikely for another researcher reproducing the study to draw
exactly the same conclusions. However, in whatever ways they might explain
failure, other researchers would most probably rely on the properties of the
dilemmas. In other words, when reconciling perceptions between the imagi-
nary new research and this study, there should be no fundamental
285
disagreement. Thus, the spirit of the conclusions would remain intact, and the
author assesses that reproducing this study would yield results leading to
similar practical implications and usefulness.
As can be seen in the platform literature, strategic challenges vary by plat-
form type. For example, lessons from online platforms might not be applicable
to platform markets such as the newspaper/media industry, payment cards, or
operating systems examined in other studies (Rysman 2009). For example,
taking the general assumption of network effects prevent switching from the
standards discussion (Katz & Shapiro 1994) would be an overstatement, as
network effects are by definition not decisive in an environment with multiple
coexisting platforms and multihoming, which the Internet as a business envi-
ronment clearly is and proprietary standards are not.
In terms of improving the quality of the substantive theory, Figure 10 (p.
93), which displays the totality of emerging dilemmas and biases, is a strong
candidate as it explains much of the startup process leading to failure. In par-
ticular, biases and bounded rationality in decision-making have been acknowl-
edged since Simon (1956); including cognitive psychology and behavioral
game theory would increase the explanatory power of the substantive theory,
and potentially lead to discoveries that could be formalized and applied across
contexts. Indeed, including biases would improve the explanatory power of
why founders choose particular strategies that can afterwards be considered
destructive, and thus improve the work done here.
In the spirit of substantive theory, the results are limited to Internet startups,
or early-stage technology ventures in online business. Therefore, based on this
study, the results cannot readily be generalized to other types of technology
startup such as life science or clean-tech, other types of startup, platform com-
panies, or firms in general. These entities are likely to have different dynamics
that render the conclusions of this study inapplicable. However, the generali-
zability is extended beyond failed startups to all platform startups on the Inter-
net. There is no reason to believe that successful platform startups have not
faced these issues; indeed, they have solved them. Thus, further theoretical
sampling can be targeted at the strategies of successful platform startups.
Moreover, there are startups withindecisive outcomes and, without doubt, the
greatest share of startups encountered by the author in the various startup
events belong to this group. They can benefit from this study by identifying
potential strategic problems and solutions, and by adopting strategic platform
thinking.
Finally, the author would like to point out that two-sidedness is not an ex-
clusive feature of platforms; in contrast, as a perspective, it can be applied to
examine very many situations. As two-sided dynamics relate not only to plat-
forms, Rysman (2009) refers to two-sided strategies as opposed to two-sided
286
platforms. By further developing his idea, one can speak of platform logic, or
two-sided logic, that is applicable to contexts not normally understood as plat-
forms. For example, events: the better the speakers, the more tickets can be
sold. Relatedly, the higher the speaker fees, the higher the sales; given that the
quality of the speakers is proportional to his or her fee. Another example is
education: the better the teachers are in a given school, the better the caliber of
students wanting to apply. Moreover, two-sided logic applies to retailing,
which was considered distinctly separate from the platform model in Chapter
3.1. Clearly, the selection and variety of merchants affects how likely the end
customer is to buy, and thus two-sided dynamics are present. Internal market-
ing is another example of two-sided logic. The happier the employees, the
better they serve the customers, and thus the happier customers become; vice
versa, the more unhappy or moodier the customer, the worse the motivation of
the employees serving him/her, and so on
265
. The implication for firms is to
understand what each interactive side appreciates, and assess the degree of
critical mass needed to evoke the desired response (i.e., conversion).
6.5.9 Overall assessment of credibility
Regarding credibility, the nature of data means that they cannot be treated as
facts. Risks to credibility comprise informant biases (e.g., recall bias and re-
spondent bias) that can involve consciously or unconsciously ignoring relevant
aspects, researcher bias closely associated with interpretation, and the specific
methodological choices.
In general, a limited number of firms and respondents might result in a one-
sided perspective and prevent generalization in a statistical sense; although, if
judged as a qualitative case study, the number of cases here can considered in
line with such studies. Moreover, there are also benefits associated with the
approach taken in the study and, considering the research purpose, the issue of
“myopia” cannot be regarded as a major issue. First, the founders are highly
knowledgeable on the challenges they faced. Second, stories are independent
accounts, written in different places at different times and interpreting differ-
ent startups, which lends support to identified patterns in the real world. Third,
the researcher has confidence in knowing the startup reality sufficiently well
to make credible claims on it. Finally, the analysis suggests that the founders
were aware of their personal biases, and might even have made them explicit.
Overall, this leads the author to believe that the material and analytical
265
The mentioned chain effect relies on contagion of emotions, but the logic is two-sided as the
welfare of each group is linked to the other.
287
procedures are appropriate to draw credible conclusions. Regarding saturation,
the researcher does not expect that the conclusions would drastically change if
more post-mortems had been included in the study. This has been validated by
discussions with numerous founders beyond the initial sample.
Other strengths arise from the use of GT. By detailing specific techniques,
it enabled a novice PhD student to choose techniques that felt natural and
flexible but still guided the analytical process from start to finish. Grounded
theory is distinct from qualitative data analysis (QDA) methods due to its em-
phasis on theory, not “thick descriptions" or case studies. It also differs from
analytical modeling, which often makes strong assumptions to remain tracta-
ble; GT is not constrained by the same rigor, and can therefore expand to great
lengths while remaining rooted in the relevancy of the data. Truly under-
standing the strengths of the method motivates one to continue, and creates
confidence in one’s analytical abilities. Overall, the author feels confident in
recommending the grounded theory method to studies that aim at conceptual-
ization and theory development.
289
REFERENCES
Aarikka-Stenroos, L. (2011) Reference communication and third actors in the
initiation of business relationships. Turku School of Economics
Series A-6:2001, Turku.
Ackoff, R. (1969) Systems, organizations, and interdisciplinary research. In:
Systems thinking, ed. F. Emery, 330–347. Penguin Books, Balti-
more.
Agarwal, R., – Gort, M. (2001) First-Mover Advantage and the Speed of
Competitive Entry, 1887–1986. Journal of Law and Economics,
Vol. 44 (1), 161–177.
Aggarwal, C. C. – Yu, P. S. (2012) On the network effect in Web 2.0 applica-
tions. Electronic Commerce Research and Applications, Vol. 11
(2), 142–151.
Agulnick, T. (2010) Xmarks is shutting down. , retrieved J une 2012.
Akerlof, G. A. (1970) The Market for “Lemons”: Quality Uncertainty and the
Market Mechanism. The Quarterly Journal of Economics, Vol. 84
(3), 488–500.
Albuquerque, P. – Pavlidis, P. – Chatow, U. – Chen, K.-Y. – J amal, Z. (2012)
Evaluating promotional activities in an online two-sided market of
user-generated content. Marketing Science, Vol. 31 (3), 406–432.
Almeida, P. – Dokko, G. – Rosenkopf, L. (2003) Startup size and the mecha-
nisms of external learning: increasing opportunity and decreasing
ability? Research Policy, Vol. 32 (2), 301–315.
Anderson, C. (2009) Free: how today’s smartest businesses profit by giving
something for nothing. Hyperion, New York, NY.
Anderson, S. P. – Gabszewicz, J . J . (2006) The media and advertising: a tale
of two-sided markets. In: Handbook of the Economics of Art and
Culture, Volume 1, eds. K. Arrow – M. Intriligator, 567–614.
Elsevier B. V., Amsterdam.
Anding, M. – Hess, T. (2002) Online Content Syndication-A Critical Analysis
From The Perspective Of Transaction Cost Theory. In: Proceed-
ings of the 10
th
European Conference on Information Systems, ed.
S. Wrycza, 551–563.
290
Argyres, N. S. (1999) The impact of information technology on coordination:
Evidence from the B-2 “Stealth” bomber. Organization Science,
Vol. 10 (2), 162–180.
Arkes, H. R. – Blumer, C. (1985) The psychology of sunk cost. Organiza-
tional Behavior and Human Decision Processes, Vol. 35 (1), 124–
140.
Armstrong, M. (2006) Competition in Two-Sided Markets. RAND Journal of
Economics, Vol. 37 (3), 668–691.
Armstrong, M. – Wright, J . (2007) Two-sided markets, competitive bottle-
necks and exclusive contracts. Economic Theory, Vol. 32 (2),
353–380.
Arndt, J . (1978) How broad should the marketing concept be? Journal of Mar-
keting, Vol. 42 (1), 101–103.
Arroyo-Barrigüete, J . – Ernst, R. – López-Sánchez, J . – Orero-Giménez, A.
(2010) On the identification of critical mass in Internet-based ser-
vices subject to network effects. The Service Industries Journal,
Vol. 30 (5), 643–654.
Audretsch, D. – Acs, Z. (1994) New-firm startups, technology, and macro-
economic fluctuations. Small Business Economics, Vol. 6 (6),
439–449.
Axelrod, R. M. (2006) The evolution of cooperation. Basic Books, New York,
NY.
Azoulay, P. – Shane, S. (2001) Entrepreneurs, Contracts, and the Failure of
Young Firms. Management Science, Vol. 47 (3), 337–358.
Bakos, Y. (1998) The emerging role of electronic marketplaces on the Inter-
net. Communications of the ACM, Vol. 41 (8), 35–42.
Bakos, Y. – Katsamakas, E. (2008) Design and ownership of two-sided net-
works: Implications for Internet platforms. Journal of Manage-
ment Information Systems, Vol. 25 (2), 171–202.
Ball, L. – Mankiw, N. – Romer, D. – Akerlof, G. – Rose, A. – Yellen, J . –
Sims, C. (1988) The new Keynesian economics and the output-in-
flation trade-off. Brookings Papers on Economic Activity, Vol.
1988 (1), 1–82.
Banerjee, A. V. (1992) A simple model of herd behavior. The Quarterly Jour-
nal of Economics, Vol. 107 (3), 797–817.
Banerji, A. – Dutta, B. (2009) Local network externalities and market seg-
mentation. International Journal of Industrial Organization,
27(5), 605–614.
Banks, J . – Deuze, M. (2009) Co-creative labour. International Journal of
Cultural Studies, Vol. 12 (5), 419–431.
291
Bansal, P. – Clelland, I. (2004) Talking trash: Legitimacy, impression man-
agement, and unsystematic risk in the context of the natural envi-
ronment. Academy of Management Journal, Vol. 47 (1), 93–103.
Baum, J . A. C. – Mezias, S. J . (1992) Localized Competition and Organiza-
tional Failure in the Manhattan Hotel Industry, 1898-1990. Ad-
ministrative Science Quarterly, Vol. 37 (4), 580–604.
Bauman, Z. (2010) Hermeneutics and Social Science (Routledge Revivals):
Approaches to Understanding. Taylor & Francis, New York, NY.
Beaver, G. – J ennings, P. (2005) Competitive advantage and entrepreneurial
power: The dark side of entrepreneurship. Journal of Small Busi-
ness and Enterprise Development, Vol. 12 (1), 9–23.
Beck, R. – Beimborn, D. – Weitzel, T. – König, W. (2008) Network effects as
drivers of individual technology adoption: Analyzing adoption
and diffusion of mobile communication services. Information
Systems Frontiers, Vol. 10 (4), 415–429.
Bekkelund, K. (2011) Succeeding with freemium: Exploring why companies
have succeeded and failed with freemium. Innovation and Entre-
preneurship, Specialization Project 4350, Norwegian University
of Science and Technology, Trondheim. , retrieved J une 2013.
Belleflamme, P. – Toulemonde, E. (2004) B2B marketplaces: when it pays to
be small. Presented at the5th ZEW Conference: The Economics of
Information and Communication Technologies, Mannheim, Ger-
many.
Benkler, Y. (2002) Coase’s Penguin, or, Linux and “The Nature of the Firm.”
Yale Law Journal, Vol. 112 (3), 369–446.
Benway, J . P. – Lane, D. M. (1999) Banner blindness: What searching users
notice and do not notice on the World Wide Web. Rice University
Electronic Theses and Dissertations, Rice University, Houston,
TX.
Bergen, M. – Dutta, S. – Walker, O. C., J r. (1992). Agency relationships in
marketing: A review of the implications and applications of
agency and related theories. Journal of Marketing, Vol. 56 (J uly),
1–24.
Berman, R. – Katona, Z. (2012) The role of search engine optimization in
search marketing. Marketing Science, Vol. 32 (4), 644–651.
Berners-Lee, T. (2010) Long Live the Web: A Call for Continued Open
Standards and Neutrality. Scientific American, 26.11.2010.
Besen, S. – Farrell, J . (1994) Choosing How to Compete: Strategies and Tac-
tics in Standardization. Journal of Economic Perspectives, Vol. 8
(2), 117–131.
292
Beuscart, J . – Mellet, K. (2009) Business Models of the Web 2.0: Advertising
or the Tale of Two Stories. Communications and Strategies, Spe-
cial Issue (November 2008), 165-81
Biernacki, P. – Waldorf, D. (1981) Snowball Sampling: Problems and Tech-
niques of Chain Referral Sampling. Sociological Methods &
Research, Vol. 10 (2), 141–163.
Biggar, P. (2010) Why we shut NewsTilt down. , retrieved J une 2012.
Birke, D. (2008). The Economics of Networks - A Survey of the Empirical
Literature. Journal of Economic Surveys, Vol. 23 (4), 762–793.
Blank, S. (2010) Customer Development for Web Startups. Steve Blank.
, retrieved 2.7.2013.
Boudreau, K. (2010) Open platform strategies and innovation: Granting access
vs. devolving control. Management Science, Vol. 56 (10), 1849–
1872.
Boudreau, K. – Hagiu, A. (2009) Platform rules: multi-sided platforms as reg-
ulators. In: Platforms, Markets and Innovation, ed. A. Gawer,
169–191. Edward Elgar Publishing, Cheltenham.
Boudreau, K. – Lakhani, K. R. (2012). How to manage outside innovation.
MIT Sloan Management Review, Vol. 50 (4), 69–76.
Boyle, R. D. – Desai, H. B. (1991) Turnaround strategies for small firms.
Journal of Small Business Management, 29 (3), 33–42.
Bragiel, P. (2008) Anatomy Of A Failure: Lessons Learned.
, retrieved J une 2012.
Briscoe, B. – Odlyzko, A. – Tilly, B. (2006) Metcalfe’s law is wrong. IEEE
Spectrum. , retrieved J une 2013.
Brown, S. J . – Goetzmann, W. – Ibbotson, R. G. – Ross, S. A. (1992) Survi-
vorship bias in performance studies. Review of Financial Studies,
Vol. 5 (4), 553–580.
Brodzinski, P. (2009) Lessons Learned: Startup Failure Part 1.
, retrieved J une 2012.
Bruderl, J . – Schussler, R. (1990) Organizational mortality: The liabilities of
newness and adolescence. Administrative Science Quarterly, Vol.
35 (3), 530–547.
Brunn, P. – J ensen, M. – Skovgaard, J . (2002). e-Marketplaces:: Crafting A
Winning Strategy. European Management Journal, Vol. 20 (3),
286–298.
293
Bunn, D. W. (1975) Anchoring bias in the assessment of subjective probabil-
ity. Operational Research Quarterly, Vol. 26 (2), 449–454.
Burns, M. Blesener, E. (2013) Commenters, we want you back. TechCrunch.
,
retrieved 1.7.2013.
Cable, D. – Shane, S. (1997) A prisoner’s dilemma approach to entrepreneur-
venture capitalist relationships. Academy of Management Review,
22 (1), 142–176.
Caillaud, B. – J ullien, B. (2003) Chicken & egg: Competition among interme-
diation service providers. RAND Journal of Economics, Vol. 34
(2), 309–328.
Calvano, E. – J ullien, B. (2012) Issues in online advertising and competition
policy: A two sided markets perspective. In: Recent Advances in
the Analysis of Competition Policy and Regulation, eds. J . E.
Harrington – Y. Katsoulacos, 179–197. Edward Elgar Publishing,
Cheltenham.
Camerer, C. (2003) Behavioral Game Theory: Experiments in Strategic Inter-
action. Princeton University Press, Princeton, NJ .
Campbell, C. (2012) How much is Zynga paying for new gamers? IGN.
, retrieved 5.7.2013.
Cappel, J . J . – Huang, Z. (2007). A usability analysis of company websites.
Journal of Computer Information Systems, Vol. 48 (1), 117.
Casey, T. R. – Töyli, J . (2012) Dynamics of two-sided platform success and
failure: An analysis of public wireless local area access. Techno-
vation, Vol. 32 (12), 703–716.
Cennamo, C. – Santalo, J . (2013). Platform competition: Strategic trade-offs in
platform markets. Strategic Management Journal, Vol. 34 (11),
1331–1350.
Chafkin, M. (2007) How to kill a great idea! Inc. Magazine. , retrieved J une 2013.
Chakravorti, S. – Roson, R. (2006) Platform competition in two-sided mar-
kets: The case of payment networks. Review of Network Econom-
ics, Vol. 5 (1), 118–143.
Charmaz, K. (1990) “Discovering” chronic illness: Using grounded theory.
Social Science & Medicine, Vol. 30 (11), 1161–1172.
Charmaz, K. (2006) Constructing grounded theory: A practical guide through
qualitative analysis. SAGE Publications, London.
294
Chauhan, S. S. – Proth, J .-M. (2005). Analysis of a supply chain partnership
with revenue sharing. International Journal of Production Eco-
nomics, Vol. 97 (1), 44–51.
Chen, A. (2009) How to create a profitable Freemium startup. @andrewchen.
, retrieved
1.7.2013.
Chen, J . – Esteban, S. – Shum, M. (2008) How much competition is a second-
ary market? Texas A&M University Workshop. , retrieved J une
2013.
Chesbrough, H. (2004) Managing open innovation. Research-Technology
Management, Vol. 47 (1), 23–26.
Chesler, M. (1987) Professionals’ views of the “dangers” of self-help groups.
Center for Research on Social Organization Working Paper No.
345, Ann Arbor, MI.
ChubbyBrain (2011) Analyzing 32 startup failure post-mortems to find the 20
top reasons that startups fail. Chubby Brain.
, retrieved J une 2013.
Church, J . – Gandal, N. (1992) Network effects, software provision, and
standardization. Journal of Industrial Economics, Vol. 40 (1), 85–
103.
Church, J . – Gandal, N. (2004) Platform competition in telecommunications.
Centre for Economic Policy Research No. 4659, London.
Clements, M. T. (2004) Direct and indirect network effects: are they equiva-
lent? International Journal of Industrial Organization, Vol. 22
(5), 633–645.
Clements, M. T. – Ohashi, H. (2005) Indirect network effects and the product
cycle: video games in the U.S., 1994–2002. Journal of Industrial
Economics, Vol. 53 (4), 515–542.
Clemons, E. K. (2009) Business Models for Monetizing Internet Applications
and Web Sites: Experience, Theory, and Predictions. Journal of
Management Information Systems, Vol. 26 (2), 15–41.
Coase, R. H. (1937) The nature of the firm. Economica, Vol. 4 (16), 386–405.
Coch, L. – French, J . (1948) Overcoming resistance to change. Human Rela-
tions, Vol. 1 (4), 512–532.
Cochran, J . J . – Darrat, A. F. – Elkhal, K. (2006) On the bankruptcy of internet
companies: An empirical inquiry. Journal of Business Research,
Vol. 59 (10–11), 1193–1200.
295
Cohen, A. (2003) The perfect store: inside eBay. Back Bay Books; Boston,
MA.
Cohen, W. M. – Klepper, S. (1992) The tradeoff between firm size and diver-
sity in the pursuit of technological progress. Small Business Eco-
nomics, Vol. 4 (1), 1–14.
Collingridge, D. S. – Gantt, E. E. (2008) The quality of qualitative research.
American Journal of Medical Quality, Vol. 23 (5), 389–395.
Constine, J . – Kutler K.-M. (2012) Facebook buys Instagram for $1 billion,
turns budding rival into its standalone photo app. TechCrunch.
, retrieved 30.6.2013.
Corbin, J . – Strauss, A. (1990) Grounded theory research: Procedures, canons,
and evaluative criteria. Qualitative Sociology, Vol. 13 (1), 3–21.
Corbin, J . – Strauss, A. (2008) Basics of Qualitative Research: Techniques
and Procedures for Developing Grounded Theory. SAGE Publi-
cations, Los Angeles, CA.
Coughlin, S. S. (1990) Recall bias in epidemiologic studies. Journal of Clini-
cal Epidemiology, Vol. 43 (1), 87–91.
Cova, B. – Dalli, D. (2009) Working consumers: the next step in marketing
theory? Marketing Theory, Vol. 9 (3), 315–339.
Covey, N. (2010) Changing models a global perspective on paying for content
online. Nielsen Newswire. , retrieved J une 2013.
Creswell, J . W. (2008) Research Design: Qualitative, Quantitative, and Mixed
Methods Approaches (3rd ed.). SAGE Publications, Los Angeles,
CA.
Curchod, C. – Neysen, N. (2009) Disentangling Positive and Negative Exter-
nalities on Two-Sided Markets: the EBay Case. Université
catholique de Louvain Working Paper No. 09/03, Louvain-la-
Neuve.
Cusumano, M. (2010) Technology strategy and management: The evolution of
platform thinking. Communications of the ACM, Vol. 53 (1), 32–
34.
D’aveni, R. A. (1994) Hypercompetition: Managing the dynamics of strategic
maneuvering. The Free Press, New York, NY.
Dagger, D. – O’Connor, A. – Lawless, S. – Walsh, E. – Wade, V. P. (2007)
Service-oriented e-learning platforms: From monolithic systems
to flexible services. Internet Computing, IEEE, Vol. 11 (3), 28–
35.
296
Daugherty, T. – Eastin, M. S. – Bright, L. (2008). Exploring consumer moti-
vations for creating user-generated content. Journal of Interactive
Advertising, Vol. 8 (2), 1–24.
Davis, J . (2009) Revealing the research “hole”of early childhood education for
sustainability: A preliminary survey of the literature. Environ-
mental Education Research, Vol. 15 (2), 227–241.
Developer Economics 2012 (2012) VisionMobile. , retrieved J une 2013.
Diaz, A. (2010) YouCastr - A post-mortem. , retrieved J une 2012.
Diaz Ruiz, C. (2012) Theories of markets: Insights from marketing and the
sociology of markets. The Marketing Review, Vol. 12 (1), 61–77.
Dickens, Z. (2010) Start-up Post-mortem: Contrastream. , retrieved
J une 2012.
Dimitras, A. I. – Zanakis, S. H. – Zopounidis, C. (1996) A survey of business
failures with an emphasis on prediction methods and industrial
applications. European Journal of Operational Research, Vol. 90
(3), 487–513.
Dixon, C. (2010) The bowling pin strategy. , retrieved J une 2013.
Dow, S. – Kulkarni, A. – Bunge, B. – Nguyen, T. – Klemmer, S. – Hartmann,
B. (2011) Shepherding the crowd yields better work. Proceedings
of the ACM Conference on Human Factors in Computing Systems,
eds. D. Tan – G. Fitzpatrick – C. Gutwin –B. Begole – W. A.
Kellogg, 1669-1674. ACM, New York, NY.
Dushnitsky, G. – Klueter, T. (2011) Is There an eBay for Ideas? Insights from
Online Knowledge Marketplaces. European Management Review,
Vol. 8 (1), 17–32.
Dwyer, F. R. – Schurr, P. H. – Oh, S. (1987) Developing buyer-seller relation-
ships. J ournal of Marketing, Vol. 51 (2), 11–27.
Easley, R. W. – Madden, C. S. – Dunn, M. G. (2000) Conducting marketing
science: the role of replication in the research process. Journal of
Business Research, Vol. 48 (1), 83–92.
Economides, N. – Katsamakas, E. (2006) Two-sided competition of proprie-
tary vs. open source technology platforms and the implications for
the software industry. Management Science, Vol. 52 (7), 1057–
1071.
297
Edwards, T. (2013) Facebook targets 76 million fake users in war on bogus
accounts. Business Insider. , retrieved J une 2013.
Ehrenberg, R. (2008) Monitor110: A Post Mortem. , retrieved J une
2012.
Eisenhardt, K. (1989) Building Theories from Case Study Research. The
Academy of Management Review, Vol. 14 (4), 532–550.
Eisenhardt, K. – Graebner, M. E. (2007) Theory Building From Cases: Op-
portunities And Challenges. The Academy of Management Jour-
nal, Vol. 50 (1), 25–32.
Eisenmann, T. – Parker, G. – Van Alstyne, M. (2009) Opening platforms:
how, when and why? In: Platforms, Markets and Innovation, ed.
A. Gawer, 131–162. Edward Elgar Publishing, Cheltenham.
Eisenmann, T. – Parker, G. – Van Alstyne, M. W. (2006) Strategies for two-
sided markets. Harvard Business Review, Vol. 84 (10), 92–101.
Eisenmann, T. – Parker, G. – Van Alstyne, M. (2011) Platform envelopment.
Strategic Management Journal, Vol. 32 (12), 1270–1285.
Elliott, J . (2005) Using narrative in social research: Qualitative and quantita-
tive approaches. SAGE Publications, London.
Elliott, N. – Lazenbatt, A. (2005) How to recognise a “quality” grounded the-
ory research study. The Australian Journal of Advanced Nursing,
Vol. 22 (3), 48–52.
Ellison, N. B. (2007) Social network sites: Definition, history, and scholarship.
Journal of Computer-Mediated Communication, Vol. 13 (1), 210–
230.
Evans, D. S. (2002) The Antitrust Economics of Two-Sided Markets. SSRN
Working Paper No. 332022. Social Science Research Network,
Rochester, NY.
Evans, D. S. (2003) Some Empirical Aspects of Multi-sided Platform Indus-
tries. SSRN Working Paper No. 447981. Social Science Research
Network, Rochester, NY.
Evans, D. S. (2009a) How catalysts ignite: the economics of platform-based
start-ups. In: Platforms, Markets and Innovation, ed. A. Gawer,
99–130. Edward Elgar Publishing Cheltenham.
Evans, D. S. (2009b) Two-Sided Market Definition. SSRN Working Paper No.
1396751. Social Science Research Network, Rochester, NY.
Evans, D. S. – Hagiu, A. – Schmalensee, R. (2006) Invisible Engines: How
Software Platforms Drive Innovation and Transform Industries.
MIT Press, Cambridge, MA.
298
Evans, D. S. – Schmalensee, R. (2010) Failure to Launch: Critical Mass in
Platform Businesses. SSRN Working Paper 1353502. Social Sci-
ence Research Network, Rochester, NY
Facebook (2013) Advertising guidelines. Facebook. , retrieved J une 2013.
Farrell, J . – Klemperer, P. (2007) Coordination and lock-in: Competition with
switching costs and network effects. In: Handbook of industrial
organization, Volume 3, eds. M. Armstrong – R. Porter, 1967–
2072. Elsevier B. D., Amsterdam.
Farrell, J . – Saloner, G. (1985) Standardization, compatibility, and innovation.
The RAND Journal of Economics, Vol. 16 (1), 70–83.
Farrell, J . – Saloner, G. (1986) Installed Base and Compatibility: Innovation,
Product Preannouncements, and Predation. The American Eco-
nomic Review, Vol. 76 (5), 940–955
Fehr, E. – Kirchler, E. – Weichbold, A. – Gächter, S. (1998) When social
norms overpower competition: Gift exchange in experimental la-
bor markets. Journal of Labor Economics, Vol. 16 (2), 324–351.
Feld, B. (2006) Talking About Failure. , retrieved J une 2012.
Festinger, L. (1962) A theory of cognitive dissonance, Volume 2, Stanford
university press, Stanford, CA.
Finch, J . H. (2002) The role of grounded theory in developing economic the-
ory. Journal of Economic Methodology, Vol. 9 (2), 213–234.
Fischer, M. – Albers, S. – Wagner, N. – Frie, M. (2011) Practice prize win-
ner—dynamic marketing budget allocation across countries, prod-
ucts, and marketing activities. Marketing Science, Vol. 30 (4),
568–585.
Fletcher, A. (2007) Predatory pricing in two-sided markets: A brief comment.
Competition Policy International, Vol. 3 (1), 221–224.
Flynn, L. R. – Goldsmith, R. E. – Eastman, J . K. (1996) Opinion leaders and
opinion seekers: two new measurement scales. J ournal of the
Academy of Marketing Science, Vol. 24 (2), 137–147.
Freeman, J . – Carroll, G. R. – Hannan, M. T. (1983) The liability of newness:
Age dependence in organizational death rates. American Socio-
logical Review, Vol. 48 (5), 692–710.
Funk, J . (2006) Early competition in complex network industries: The case of
the mobile internet. Hitotshubashi University Institute of Innova-
tion Research Working Paper, Tokyo, J apan.
Galbreth, M. R. – March, S. T. – Scudder, G. D. – Shor, M. (2003) A game-
theoretic model of e-marketplace participation growth. Journal of
Management Information Systems, Vol. 22 (1), 295–319.
299
Ganley, D. – Lampe, C. (2009) The ties that bind: Social network principles in
online communities. Decision Support Systems, Vol. 47 (3), 266–
274.
Gans, J . S. (2012) Mobile application pricing. Information Economics and
Policy, Vol. 24 (1), 52–59.
Gaskill, L. R. – Van Auken, H. E. – Manning, R. A. (1993) A factor analytic
study of the perceived causes of small business failure. Journal of
Small Business Management, Vol. 31 (4), 18.
Gasson, S. (2003) Rigor in grounded theory research - an interpretive perspec-
tive on generating theory from qualitative field studies. In: Hand-
book for information systems research, eds. M. Whitman – A.
Woszczynski, 79–102. Idea Group Publishing, Hershey, PA.
Gawer, A. (2009) Platform dynamics and strategies: from products to services.
In: Platforms, Markets and Innovation, ed. A. Gawer, 45–76.
Edward Elgar Publishing, Cheltenham.
Gawer, A. – Cusumano, M. (2008) How Companies Become Platform Lead-
ers. MIT Sloan Management Review, Vol. 49 (2), 28–35.
Gawer, A. – Henderson, R. (2007) Platform owner entry and innovation in
complementary markets: Evidence from Intel. Journal of Eco-
nomics & Management Strategy, Vol. 16 (1), 1–34.
Gazé, P. – Vaubourg, A-G. (2011) Electronic platforms and two-sided mar-
kets: A side-switching analysis. The Journal of High Technology
Management Research, Vol. 22 (2), 158–165.
Gilder, G. (1993) Metcalf’s law and legacy. Forbes ASAP, 13.9.1993.
Gillin, P. (2009) The new influencers: A marketer’s guide to the new social
media. Linden Publishing, Freshno, CA.
Glaser, B. G. (1971) Status Passage. Transaction Publishers, Chicago, IL.
Glaser, B. G. (1978) Theoretical sensitivity: advances in the methodology of
grounded theory. Sociology Press, Mill Valley, CA.
Glaser, B. G. (1992) Basics of grounded theory analysis: Emergence vs forc-
ing. Sociology Press, Mill Valley, CA.
Glaser, B. G. (2004) “Naturalist inquiry” and grounded theory. Forum: Qual-
itative Social Research, Vol. 5 (1), 1–14.
Glaser, B. G. (2002) Conceptualization: On theory and theorizing using
grounded theory. International Journal of Qualitative Methods,
Vol. 1 (2), 23–38.
Glaser, B. G. – Holton, J . (2004) Remodeling grounded theory. Forum: Qual-
itative Social Research, Vol. 5 (2), 1–17.
Glaser, B. G. – Strauss, A. L. (1965) Discovery of substantive theory: A basic
strategy underlying qualitative research. American Behavioral
Scientist, Vol. 8 (6), 5–12.
300
Glaser, B. G. – Strauss, A. L. (1967) The discovery of grounded theory: strat-
egies for qualitative research. Transaction Publishers, Piscataway,
NJ .
Golder, P. N. – Tellis, G. J . (1993) Pioneer advantage: Marketing logic or
marketing legend? Journal of Marketing Research, Vol. 30 (2),
158–170.
Goldenson, M. (2009) 10 lessons from a failed startup.
, retrieved J une 2012.
Goldfarb, A. (2003) Advertising, profits, switching costs, and the internet. In:
Proceedings of the 5th international conference on electronic
commerce, eds. N. M. Sadeh – M. J . Dively – R. J . Kauffman – Y.
Labrou – O. Shehory – R. Telang – L. F. Cranor, 266–275. ACM
Press, New York, NY.
Google Inc. Announces Fourth Quarter and Fiscal Year 2012 Results (2013)
Google Investor Relations. , retrieved J une 2013.
Goulding, C. (2005) Grounded theory, ethnography and phenomenology: A
comparative analysis of three qualitative strategies for marketing
research. European Journal of Marketing, Vol. 39 (3/4), 294–308.
Grabner-Kräuter, S. – Kaluscha, E. A. (2003) Empirical research in on-line
trust: A review and critical assessment. International Journal of
Human-Computer Studies, Vol. 58 (6), 783–812.
Gulli, A. – Signorini, A. (2005) The indexable web is more than 11.5 billion
pages. In: Proceedings of WWW '05 Special interest tracks and
posters of the 14th international conference on World Wide Web,
eds. A. Ellis – T. Hagino – F. Douglis – P. Raghavan, 902–903.
ACM Press, New York, NY.
Gunawardana, A. – Meek, C. – Biggs, J . (2008) A quality-based auction for
search ad markets with aggregators. Paper presented at theACM
EC Workshop on Ad Auctions, Chicago, IL.
Hagiu, A. (2006) Pricing and commitment by two-sided platforms. The RAND
Journal of Economics, Vol. 37 (3), 720–737.
Hagiu, A. (2007) Merchant or two-sided platform? Review of Network Eco-
nomics, Vol. 6 (2), 1–19.
Hagiu, A. (2008) Platforms, pricing, commitment and variety in two-sided
markets. Doctoral dissertation, Princeton University, Princeton,
NJ .
Hagiu, A. (2009) Multi-sided platforms: From microfoundations to design and
expansion strategies. Harvard Business School Working Paper,
No. 09–115, Boston, MA.
301
Hagiu, A. and T. Eisenmann (2007) A Staged Solution to the Catch-22. Har-
vard Business Review, Vol. 85 (11).
Hagiu, A. – J ullien, B. (2011) Why do intermediaries divert search? The
RAND Journal of Economics, Vol. 42 (2), 337–362.
Hagiu, A. – Spulber, D. (2012) First-party content and coordination in two-
sided markets. Management Science, Vol. 59 (4), 933–949.
Hagiu, A. – Wright, J . (2011) Multi-sided platforms. Harvard Business School
Working Paper, No. 12-024, Boston, MA.
Hagiu, A. – Wright, J . (2013) Do you really want to be an eBay? Harvard
Business Review, Vol. 91 (3), 102-108.
Hagiu, A. – Yoffie, D. B. (2009) What’s your Google strategy? Harvard Busi-
ness Review, Vol. 87 (4), 74–81.
Halligan, B. – Shah, D. (2009) Inbound marketing: get found using Google,
social media, and blogs. J ohn Wiley & Sons, Hoboken, NJ .
Haltiwanger, J . – J armin, R. – Miranda, J . (2009) Business dynamics statistics
briefing: High growth and failure of young firms. Ewing Marion
Kauffman Foundation, Kansas City, MO.
Hammer, E. (2008) The Last AnNounce(r)ment. , retrieved J une 2012.
Han, K. S. – Noh, M. H. (1999) Critical failure factors that discourage the
growth of electronic commerce. International Journal of Elec-
tronic Commerce, Vol. 4 (2), 25–43.
Hartung, A. (2011) How Facebook beat MySpace. , retrieved 2.7.2013.
Heath, H. – Cowley, S. (2004) Developing a grounded theory approach: a
comparison of Glaser and Strauss. International Journal of Nurs-
ing Studies, Vol. 41 (2), 141–150.
Hedlund, M. (2010) Why Wesabe Lost to Mint. , retrieved J une 2012.
Hemrajani, K. (2010) Post-mortem. , retrieved J une 2012.
Hennig-Thurau, T. – Gwinner, K. P. – Walsh, G. – Gremler, D. D. (2004)
Electronic word-of-mouth via consumer-opinion platforms: What
motivates consumers to articulate themselves on the internet?
Journal of Interactive Marketing, Vol. 18 (1), 38–52.
Henshel, R. L. – J ohnston, W. (1987) The emergence of bandwagon effects: A
theory. The Sociological Quarterly, Vol. 28 (4), 493–511.
302
Herings, P. J . J . – Schinkel, M. P. (2001) World-Wide-Welfare: a micro-eco-
nomic analysis of the new economy. METEOR Working Paper,
No. RM/01/004, Maastricht research school of Economics of
Technology and Organizations, Maastricht.
Hermida, A. – Thurman, N. (2008) A clash of cultures: The integration of
user-generated content within professional journalistic frame-
works at British newspaper websites. Journalism Practice, Vol. 2
(3), 343–356.
Hill, S. (2006) Network-Based Marketing: Identifying Likely Adopters via
Consumer Networks. Statistical Science, Vol. 21 (2), 256–276.
Hinde, S. (2003) Spam: the evolution of a nuisance. Computers & Security,
Vol. 22 (6), 474–478.
Hinz, O. – Skiera, B. – Barrot, C. – Becker, J . U. (2011) Seeding Strategies for
Viral Marketing: An Empirical Comparison. Journal of Market-
ing, Vol. 75 (6), 55–71.
Hofer, C. W. (1980) Turnaround strategies. Journal of Business Strategy, Vol.
1 (1), 19–31.
Holton, J . (2010) The Coding Process and Its Challenges. The Grounded The-
ory Review - An International Journal, Vol. 9 (1), 21–40.
Honjo, Y. (2000) Business failure of new firms: an empirical analysis using a
multiplicative hazards model. International Journal of Industrial
Organization, Vol. 18 (4), 557–574.
Hopkins, C. C. (1968). Scientific advertising. New Line Publishing.
Horowitz, D. – Kamvar, S. D. (2010) The anatomy of a large-scale social
search engine. In: Proceedings of the 19th international confer-
ence on World Wide Web, eds. M. Rappa – P. J ones – J . Freire –S.
Chakraparti, 431–440. ACM Press, New York, NY.
Howe, J . (2006) The rise of crowdsourcing. Wired Magazine, Vol. 14 (6), 1–4.
Hsiao, R-L. (2003) Technology fears: distrust and cultural persistence in elec-
tronic marketplace adoption. The Journal of Strategic Information
Systems, Vol. 12 (3), 169–199.
Hsieh-Yee, I. (2001) Research on Web search behavior. Library & Infor-
mation Science Research, Vol. 23 (2), 167–185.
Huang, P. – Ceccagnoli, M. – Forman, C. – Wu, D. J . (2009). Participation in
a Platform Ecosystem: Appropriability, Competition and Access
to the Installed Base. Working paper 09-14, NET Institute. Avail-
able athttp://papers.ssrn.com/abstract=1480900.
Hunt, S. D. (2002) Foundations of Marketing Theory: Toward a General The-
ory of Marketing. M.E. Sharpe.
303
Hyrynsalmi, S. – Mäkilä, T. – J ärvi, A. – Suominen, A. – Seppänen, M. –
Knuutila, T. (2012) App store, marketplace, play! an analysis of
multi-homing in mobile software ecosystems. In: Proceedings of
the Fourth International Workshop on Software Ecosystems, eds.
S. J ansen – J . Bosch – C. Alves, 59–72. CEUR Workshop Pro-
ceedings, Cambridge, MA.
Iacovou, C. L. – Benbasat, I. – Dexter, A. S. (1995) Electronic data inter-
change and small organizations: adoption and impact of technol-
ogy. MIS Quarterly, Vol. 19 (4), 465–485.
Idu, A. – van de Zande, T. – J ansen, S. (2011) Multi-homing in the apple eco-
system: why and how developers target multiple apple app stores.
In: Proceedings of the International Conference on Management
of Emergent Digital EcoSystems, eds. W. I. Grosky – Y. Badr – R.
Chbeir, 122–128. ACM Press, New York, NY.
Darwell, B. (2013) Facebook’s frictionless sharing mistake. Inside Facebook,
22.1.2013. Insider Facebook. , re-
trieved J une 2013.
Honan, M. (2012) Twitter’s New API Rules Likely Spell the End of Third-
Party Clients. Wired. , retrieved J une 2013.
Isaacson, W. (2011) Steve Jobs. Simon & Schuster, New York, NY.
J anis, I. L. (1982) Groupthink. Houghton Mifflin, Boston, MA.
J ankowski, J . (2013) Increasing website conversions using content repetitions
with different levels of persuasion. In: Intelligent Information and
Database Systems, 5
th
Asian Conference, Proceedings, Part II,
eds. A. Selamat – N. T. Nguyen – H. Haron, 439–448. Springer,
Berlin.
J ansen, S. – Bloemendal, E. (2013) Defining app stores: The role of curated
marketplaces in software ecosystems. In: Software Business. From
Physical Products to Software Services and Solutions. Proceed-
ings of the 4
th
International Conference, ICSOB 2013, eds. G.
Herzwurm – T. Margaria, 195–206. Springer, Berlin.
J ohnson, R. B. (1997) Examining the validity structure of qualitative research.
Education, Vol. 118 (2), 282–292.
J ohnson, R. – Fraser, J . (2010) EventVue post-mortem. , retrieved J une 2012.
J ones, M. – Alony, I. (2011) Guiding the use of grounded theory in doctoral
studies – An example from the Australian film industry. Interna-
tional Journal of Doctoral Studies, Vol. 6, 95–114.
304
J ullien, B. (2005) Two-sided markets and electronic intermediaries. CESifo
Economic Studies, Vol. 51 (2-3), 233–260.
Katila, R. – Rosenberger, J . – Eisenhardt, K. (2008) Swimming with Sharks:
Technology Ventures, Defense Mechanisms and Corporate Rela-
tionships. Administrative Science Quarterly, Vol. 53 (2), 295–332.
Kaiser, U. – Wright, J . (2006) Price structure in two-sided markets: Evidence
from the magazine industry. International Journal of Industrial
Organization, Vol. 24 (1), 1–28.
Kan, M. – Parry, K. W. (2004) Identifying paradox: A grounded theory of
leadership in overcoming resistance to change. The Leadership
Quarterly, Vol. 15 (4), 467–491.
Kan, J . (2006) Actual lessons from Kiko. , retrieved J une 2012.
Kanazawa, S. (1998) In defense of unrealistic assumptions. Sociological The-
ory, Vol. 16 (2), 193–204.
Kane, T. (2010) The importance of startups in job creation and job destruction.
Kauffman Foundation Research Series: Firm Formation and Eco-
nomic Growth. Ewing Marion Kauffman Foundation, Kansas
City, MO.
Karjaluoto, E. (2009) Why your web startup will fail. , retrieved J une 2012.
Kärrberg, P. (2010) A two-sided market approach to value chain dynamics in
telecom services: A study lens for mobile platform innovation and
pricing strategies. In: Proceedings of the 2010 Ninth International
Conference on Mobile Business and 2010 Ninth Global Mobility
Roundtable, ed. J . E. Guerrero, 505–509. IEEE Computer Society,
Conference Publishing Services, Los Alamitos, CA.
Katz, M. L. – Shapiro, C. (1985) Network externalities, competition, and
compatibility. The American Economic Review, Vol. 75 (3), 424–
440.
Katz, M. L. – Shapiro, C. (1986) Technology adoption in the presence of net-
work externalities. Journal of Political Economy, Vol. 94 (4),
822–841.
Katz, M. L. – Shapiro, C. (1994) Systems competition and network effects.
The Journal of Economic Perspectives, Vol. 8 (2), 93–115.
Kaufman, B. E. (2007) The institutional economics of J ohn R. Commons:
complement and substitute for neoclassical economic theory. So-
cio-Economic Review, Vol. 5 (1), 3–45.
Kelly, T. (2009) Zynga and the End of the Beginning. Gamasutra.
, retrieved J une 2013.
305
Kempster, S. – Parry, K. W. (2011) Grounded theory and leadership research:
A critical realist perspective. The Leadership Quarterly, Vol. 22
(1), 106–120.
Kerin, R. A. – Varadarajan, P. R. – Peterson, R. A. (1992) First-mover ad-
vantage: A synthesis, conceptual framework, and research propo-
sitions. The Journal of Marketing, Vol. 56 (4), 33–52.
Kim, K., & Tse, E. (2011). Dynamic competition strategy for online
knowledge-sharing platforms. International Journal of Electronic
Commerce, Vol. 16 (1), 41–76.
Kirkpatrick, D. (2010) The Facebook effect: the inside story of the company
that is connecting the world. Simon & Schuster, New York, NY.
Kittur, A. – Nickerson, J . V. – Bernstein, M. – Gerber, E. – Shaw, A. – Zim-
merman, J . – Lease, M. – Horton, J . (2013) The future of crowd
work. In Proceedings of the 2013 conference on Computer sup-
ported cooperative work, eds. A. Bruckman – S. Counts – C.
Lampe – L. Terveen, 1301–1318. ACM Press, New York, NY.
Klein, B. (1998) Hold-up problem. The New Palgrave Dictionary of Econom-
ics and the Law, 2, 241.
Klein, D. A. A. (2012) The strategic management of intellectual capital.
Routledge, London.
Kopelman, J . (2009) Let’s just add in a little virality. Redeye VC.
, retrieved J une 2013.
Kotha, S. – Vadlamani, B. L. (1995) Assessing generic strategies: an empirical
investigation of two competing typologies in discrete manufac-
turing industries. Strategic Management Journal, Vol. 16 (1), 75–
83.
Kousha, K. – Thelwall, M. (2007) Google Scholar citations and Google
Web/URL citations: A multi-discipline exploratory analysis.
Journal of the American Society for Information Science and
Technology, Vol. 58 (7), 1055–1065.
Kretz, L. (2010) A Happy Failure Story: Lessons Learned From My First
Startup. , retrieved J une 2012.
Kristiansen, E. G. (1998) R&D in the presence of network externalities: Tim-
ing and compatibility. The RAND Journal of Economics, Vol. 29
(3), 531–547.
Krumpal, I. (2011) Determinants of social desirability bias in sensitive sur-
veys: a literature review. Quality & Quantity, Vol. 47 (4), 2025–
2047.
Kuhn, T. S. (1970). The structure of scientific revolutions. Chicago/London.
306
Köster, D. (1999) Wettbewerb in Netzproduktmärkten. Gabler Verlag:
Wiesbaden.
Laffont, J .-J . – Rey, P. – Tirole, J . (1998) Network competition: II. Price dis-
crimination. The RAND Journal of Economics, Vol. 29 (1), 38–56.
Lance, E. – Snider, D. (2006) Boompa.com: Post-mortem. , retrieved J une 2012.
Laudan, L. (1981) Science and hypothesis: Historical essays on scientific
methodology. Cambridge University Press; Cambridge.
Layard, P. R. G. – Layard, R. – Glaister, S. (1994) Cost-benefit analysis.
Cambridge University Press, Cambridge.
Lea, M. – Spears, R. – de Groot, D. (2001) Knowing me, knowing you: Ano-
nymity effects on social identity processes within groups. Person-
ality and Social Psychology Bulletin, Vol. 27 (5), 526–537.
Lee, R. S. (2013) Vertical integration and exclusivity in platform and two-
sided markets. American Economic Review, Vol. 103 (7), 2960–
3000.
Lee, S. C. – Brandyberry, A. A. (2003) The e-tailer’s dilemma. SIGMIS Data-
base, Vol. 34 (2), 10–22.
Lerner, J . – Tirole, J . (2004) The economics of technology sharing: Open
source and beyond. NBER Working Paper No. 10956, National
Bureau of Economic Research, Cambridge, MA.
Li, S. – Liu, Y. – Bandyopadhyay, S. (2010) Network effects in online two-
sided market platforms: A research note. Decision Support Sys-
tems, Vol. 49 (2), 245–249.
Li, Z. – Penard, T. (2013) The role of quantitative and qualitative network ef-
fects in b2b platform competition. Managerial and Decision Eco-
nomics, Vol. 35 (1), 1–19.
Libai, B. – Biyalogorsky, E. – Gerstner, E. (2003) Setting referral fees in af-
filiate marketing. Journal of Service Research, Vol. 5 (4), 303–
315.
Lieberman, M. B. (2005) Did first-mover advantage survive the dot-com
crash. Working Paper, Anderson Graduate School of Manage-
ment, UCLA, Los Angeles, CA.
Lieberman, M. B. – Montgomery, D. B. (1988) First-mover advantages. Stra-
tegic Management J ournal, Vol. 9 (S1), 41–58.
Lincoln, Y. S. – Guba, E. G. (1985) Naturalistic Inquiry (1st edition.). SAGE
Publications, London.
Locke, K. (1996) Rewriting the Discovery of Grounded Theory after 25
Years? Journal of Management Inquiry, Vol. 5 (3), 239–245.
307
Luchetta, G. (2012) Is the Google platform a two-sided market? SSRN
Working Paper No. 2048683. Social Science Research Network,
Rochester, NY.
Lussier, R. N. (1996) Reasons why small businesses fail: and how to avoid
failure. The Entrepreneurial Executive, Vol. 1 (2), 10–17.
Lyles, M. A. (1981) Formulating strategic problems: Empirical analysis and
model development. Strategic Management J ournal, Vol. 2 (1),
61–75.
Lyles, M. A. – Howard, T. (1985) Problem formulation and organizational
decision-making?: Biases and assumptions underlying alternative
models of strategic problem formulation. University of Illinois at
Urbana-Champaign, College of Commerce and Business Admin-
istration, Champaign, IL.
Lyles, M. A. – Howard, T. (1988) Strategic problem formulation: Biases and
assumptions embedded in alternative decision-making models.
Journal of Management Studies, Vol. 25 (2), 131–145.
Lyon, D. W. – Lumpkin, G. – Dess, G. G. (2000) Enhancing entrepreneurial
orientation research: Operationalizing and measuring a key strate-
gic decision making process. Journal of Management, Vol. 26 (5),
1055–1085.
Lyons, K. – Messinger, P. R. – Niu, R. H. – Stroulia, E. (2012) A tale of two
pricing systems for services. Information Systems and E-Business
Management, Vol. 10 (1), 19–42.
Maicas, J . P. – Polo, Y. – Sese, F. J . (2009) The role of (personal) network
effects and switching costs in determining mobile users’ choice.
Journal of Information Technology, Vol. 24 (2), 160–171.
Maitlis, S. (2005) The social processes of organizational sensemaking. Acad-
emy of Management Journal, Vol. 48 (1), 21–49.
Mangold, W. G. – Faulds, D. J . (2009) Social media: The new hybrid element
of the promotion mix. Business Horizons, Vol. 52 (4), 357–365.
Mantere, S. – Aula, P. – Schildt, H. – Vaara, E. (2013) Narrative attributions
of entrepreneurial failure. Journal of Business Venturing, Vol. 28
(4), 459–473.
Mäntymäki, M. (2011) Continuous use and purchasing behaviour in social
virtual worlds. Turku School of Economics Series A-3:2011,
Turku.
Marker, M. (1998) Going native in the academy: Choosing the exotic over the
critical. Anthropology & Education Quarterly, Vol. 29 (4), 473–
480.
308
Martin, P. Y. – Turner, B. A. (1986) Grounded theory and organizational re-
search. The Journal of Applied Behavioral Science, Vol. 22 (2),
141–157.
Maruyama, M. – Ohkita, K. (2011) Platform strategy of video game software
in J apan, 1984-1994. Managerial and Decision Economics, Vol.
32 (2), 105–118.
Mas, I. – Radcliffe, D. (2011) Scaling mobile money. Journal of Payments
Strategy & Systems, Vol. 5 (3), 298–315.
Mason, R. O. – Mitroff, I. I. (1981) Challenging strategic planning assump-
tions: Theory, cases, and techniques. J ohn Wiley & Sons,
Hoboken, NJ .
Mata, J . – Portugal, P. – Guimarães, P. (1995) The survival of new plants:
Start-up conditions and post-entry evolution. International Jour-
nal of Industrial Organization, Vol. 13 (4), 459–481.
Mathur, V. K. (1979) Some unresolved issues in the location theory of the
firm. Journal of Urban Economics, Vol. 6 (3), 299–318.
May, R. (2007) A Startup Idea Postmortem: Proof That Good Ideas Aren’t
Always Good Business. , retrieved J une 2012.
Mazrui, A. (1995) Slang and code switching: the case of Sheng in Kenya. Af-
rikanistiche Arbeitspapiere, 42, 168–179.
McCarthy, P. – Plantholt, B. – Riordan, S. (1981) Success versus survival?: the
dilemma of high technology firms. Massachusetts Institute of
Technology, Cambridge, MA.
McGhee, G. – Marland, G. R. – Atkinson, J . (2007) Grounded theory research:
literature reviewing and reflexivity. Journal of Advanced Nursing,
Vol. 60 (3), 334–342.
McGrady, D. (2008) 7 Reasons Why My Social Music Site Never Took Off.
, retrieved J une 2012.
McIntyre, D. (2011) Network intensity: Assessing the strategic value of in-
stalled base. Journal of Business Strategies, Vol. 28 (2), 177–195.
McIntyre, D. – Subramaniam, M. (2009) Strategy in network industries: a re-
view and research agenda. Journal of Management, Vol. 35 (6),
1494–517.
Mears, P. (1966) Discussion of financial ratios as predictors of failure. Journal
of Accounting Research, Vol. 4, 119–122.
Melin, L. (1985) Strategies in managing turnaround. Long Range Planning,
Vol. 18 (1), 80–86.
309
Merton, R. K. (1948) The self-fulfilling prophecy. The Antioch Review, Vol. 8
(2), 193–210.
Meyling, H. (1997) How to change students ‘conceptions of the epistemology
of science. Science & Education, Vol. 6 (4), 397–416.
Mian, S. Q. – Teixeira, J . – Koskivaara, E. (2011) Open-source software im-
plications in the competitive mobile platforms market. In: Build-
ing the e-World Ecosystem, eds. T. Skersys – R. Butleris – L.
Nemuraite – R. Suomi, 110–128. Springer, Berlin.
Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An ex-
panded sourcebook. Sage, Thousand Oaks, CA.
Milgrom, P. – Roberts, J . (1986) Price and advertising signals of product
quality. The Journal of Political Economy, Vol. 94 (4), 796–821.
Miller, D. (1977) Common syndromes of business failure. Business Horizons,
Vol. 20 (6), 43–53.
Mital, M. – Sarkar, S. (2011) Multihoming behavior of users in social net-
working web sites: a theoretical model. Information Technology &
People, Vol. 24 (4), 378–392.
Moh, T.-S. – Murmann, A. J . (2010) Can you judge a man by his friends? –
Enhancing spammer detection on the Twitter microblogging plat-
form using friends and followers. In: Information Systems, Tech-
nology and Management; Proceedings of the 4th International
Conference, ICISTM 2010, Bangkok, Thailand, eds. S. K. Prasad
– H. M. Vin – S. Sahni – M. P. J aiswal – B. Thipakorn, 210–220.
Springer, Berlin.
Moore, G. A. (1991) Crossing the chasm. Harper Business, New York, NY.
Moscovici, S. – Faucheux, C. (1972) Social influence, conformity bias, and
the study of active minorities. In: Advances in experimental social
psychology, Vol. 6, ed. L. Berkowitz, 149–202. Academic Press,
London.
Mungamuru, B. – Weis, S. (2008) Competition and fraud in online advertising
markets. In: Financial Cryptography and Data Security, Revised
Selected Papers from the 12th International Conference, FC
2008, Cozumel, Mexico, ed. G. Tsudik, 187–191. Springer, Berlin.
Murphy, L. (2011) Classical freemium doesn’t exist at scale. Saas growth
strategies. , retrieved J une 2013.
Nagel, E. (1963) Assumptions in economic theory. The American Economic
Review, Vol. 53 (2), 211–219.
Nash, J . (1950) Equilibrium points in n-person games. Proceedings of the Na-
tional Academy of Sciences, Vol. 36 (1), 48–49.
310
Neilsen, E. H. – Rao, M. V. H. (1987) The strategy-legitimacy nexus: A thick
description. Academy of Management Review, Vol. 12 (3), 523–
533.
Newberry, M. (2010) Don't Drink Your Own Kool-Aid , retrieved J une 2012.
Nickerson, R. S. (1998) Confirmation bias: a ubiquitous phenomenon in many
guises. Review of General Psychology, Vol. 2 (2), 175–220.
Nickinson, P. (2013) Two more casualities of Twitter’s power play. Android
Central. , retrieved 5.7.2013.
Niculescu, M. – Wu, D. J. (2013) Economics of free under perpetual licensing:
Implications for the software industry. Information Systems Re-
search, Vol. 25 (1), 173–199
Nielsen, J . (1998) Web pages must live forever. Nielsen Norman Group.
, retrieved J une 2013.
Noe, T. – Parker, G. (2005) Winner take all: Competition, strategy, and the
structure of returns in the internet economy. Journal of Economics
& Management Strategy, Vol. 14 (1), 141–164.
Nowak, K. (2010) Getting Caught Mid-Pivot: Lessons from the Imercive
Postmortem. , retrieved J une
2012.
O’Reilly, T. (2005) Web 2.0. , retrieved J une 2013.
Ochoa, X. – Duval, E. (2008) Quantitative analysis of user-generated content
on the web. In: Proceedings of First International Workshop Un-
derstanding Web Evolution (WebEvolve '08), eds. D. De Roure
and W. Hall, 19–26, Beijing, China.
Odlyzko, A. – Tilly, B. (2005) A refutation of Metcalfe’s Law and a better
estimate for the value of networks and network interconnections.
, retrieved J une 2013.
Oestreicher-Singer, G. – Zalmanson, L. (2009). Paying for content or paying
for community? The effect of social involvement on subscribing
to media web sites. ISIC 2009 Proceedings, eds. H. Chen – S.
Slaughter. AIS.
Olson, E. L. (2008) The implications of platform sharing on brand value.
Journal of Product & Brand Management, Vol. 17 (4), 244–253.
311
Ordanini, A. – Micelli, S. – Di Maria, E. (2004) Failure and success of B-to-B
exchange business models: A contingent analysis of their perfor-
mance. European Management Journal, Vol. 22 (3), 281–289.
Oreg, S. (2003) Resistance to change: developing an individual differences
measure. Journal of Applied Psychology, Vol. 88 (4), 680–693.
Orton, J . D. (1997) From inductive to iterative grounded theory: Zipping the
gap between process theory and process data. Scandinavian Jour-
nal of Management, Vol. 13 (4), 419–438.
Oxford Dictionary. (2013) dilemma. , retrieved J une 2013.
Packham, G. A. (2002) Competitive advantage and growth: the challenge for
small firms. International Journal of Management and Decision
Making, Vol. 3 (2), 165–179.
Paden, J . (2011) Why conversion rate optimization is guaranteed to improve
SEO results. Digital Relevance. , retrieved J une 2013.
Palka, W. – Pousttchi, K. – Wiedemann, D. G. (2009) Mobile word-of-mouth
– A grounded theory of mobile viral marketing. Journal of Infor-
mation Technology, Vol. 24 (2), 172–185.
Pandit, N. R. (1996) The creation of theory: A recent application of the
grounded theory method. The Qualitative Report, Vol. 2 (4), 1–
14.
Park, N. K. – Mezias, J . M. – Song, J . (2004) A Resource-based view of stra-
tegic alliances and firm value in the electronic marketplace. Jour-
nal of Management, Vol. 30 (1), 7–27.
Parker, G. G. – Van Alstyne, M. W. (2005) Two-sided network effects: A the-
ory of information product design. Management Science, Vol. 51
(10), 1494–1504.
Parker, G. – Van Alstyne, M. (2008) Innovation, openness, and platform con-
trol. SSRN Working Paper No. 1079712. Social Science Research
Network, Rochester, NY.
Parker, G. – Van Alstyne, M. (2010) Two-sided network effects: A theory of
information product design. Management Science, Vol. 51 (10),
1494–1504.
Parr, B. (2011a) What's it like to work at a failing startup? , retrieved J une
2012.
Parr, B. (2011b) Twitter acquires TweetDeck. Mashable. , retrieved J une 2013.
312
Parrish, E. D. – Cassill, N. L. – Oxenham, W. (2006) Niche market strategy in
the textile and apparel industry. Journal of Fashion Marketing
and Management, Vol. 10 (4), 420–432.
Partington, D. (2000) Building grounded theories of management action. Brit-
ish Journal of Management, Vol. 11 (2), 91–102.
Pauly, M. V. (1968) The economics of moral hazard: comment. The American
Economic Review, Vol. 58 (3), 531–537.
Pauwels, K. – Weiss, A. (2008) Moving from free to fee: How online firms
market to change their business model successfully. Journal of
Marketing, Vol. 72 (3), 14–31.
Pawson, R. – Tilley, N. (2009) Realist evaluation. In: Evidence-based Practice
– Modernising the Knowledge Base of Social Work? Eds. H.-U.
Otto – A. Polutta – H. Ziegler, 151–184. Verlag Barbara Budrich,
Germany.
Peitz, M. – Waelbroeck, P. (2006) Why the music industry may gain from free
downloading – the role of sampling. International Journal of In-
dustrial Organization, Vol. 24 (5), 907–913.
Pempek, T. A. – Yermolayeva, Y. A. – Calvert, S. L. (2009) College students’
social networking experiences on Facebook. Journal of Applied
Developmental Psychology, Vol. 30 (3), 227–238.
Perry, C. – J ensen, O. (2001) Approaches to combining induction and deduc-
tion in one research study. In: Australian and New Zealand Mar-
keting Academy Conference, Massey University, Auckland, New
Zealand. , retrieved J une 2013.
Piezunka, H. (2011) Technological platforms; An assessment of the primary
types of technological platforms, their strategic issues and their
linkages to organizational theory. Journal Für Betriebswirtschaft,
Vol. 61 (2-3), 179–226.
Pitkänen, L. – Salminen, J . (2012) Crowdsourcing research to mobile consum-
ers? Emerging themes on videographic data collection. Paper pre-
sented at the Planetary Scientific Research Centre Conference,
Dubai, United Arab Emirates.
Poon, S. – Swatman, P. – Swatman, P. P. (1999) A longitudinal study of ex-
pectations in small business internet commerce. International
Journal of Electronic Commerce, 3 (3), 21–33.
Porter, C. E. (2004) A typology of virtual communities: A multi-disciplinary
foundation for future research. Journal of Computer-Mediated
Communication, Vol. 10 (1).
313
Potts, M. (2007) Backfence: Lessons Learned.http://recoveringjournalist.
typepad.com/recovering_journalist/2007/07/backfence-lesso.html,
retrieved J une 2012.
Powazek, D. (2008) Pixish Closing October 31. , retrieved J une 2012.
Pujol, N. (2010) Freemium: Attributes of an emerging business model. SSRN
Working Paper No. 1718663. Social Science Research Network,
Rochester, NY.
Rafer, S. (2009) Couldery Shouldery. , retrieved J une 2012.
Raivio, Y. – Luukkainen, S. (2011) Mobile networks as a two-sided platform -
Case Open Telco. Journal of Theoretical and Applied Electronic
Commerce Research, Vol. 6 (2), 77–89.
Rao, R. S. – Chandy, R. K. – Prabhu, J . C. (2008) The fruits of legitimacy:
Why some new ventures gain more from innovation than others.
Journal of Marketing, Vol. 72 (4), 58–75.
Rappa, M. (2013) Business models on the Web. , retrieved J une 2013.
Razi, M. A. – Tarn, J . M. – Siddiqui, F. A. (2004) Exploring the failure and
success of DotComs. Information Management & Computer Secu-
rity, Vol. 12 (3), 228–244.
Reeves, C. A. – Bednar, D. A. (1994) Defining quality: alternatives and impli-
cations. Academy of Management Review, Vol. 19 (3), 419–445.
Reisinger, M. (2004) Two-sided markets with negative externalities. Discus-
sion Papers in Economics 2004-27, Ludwig-Maximilians-Univer-
sität, Munich.
Reisinger, M. (2012) Platform competition for advertisers and users in media
markets. International Journal of Industrial Organization, Vol. 30
(2), 243–252.
Ren, S. – Park, J . – van der Schaar, M. (2011) Profit maximization on user-
generated content platforms. In: 49th Annual Allerton Conference
on Communication, Control, and Computing, 135–142. IEEE.
Ries, E. (2008) Three decisions to make on virtual goods. Startup lessons
learned. , retrieved J une 2013.
Ries, E. (2010) Stop lying on stage. Startup lessons learned. , re-
trieved J une 2013.
Ries, E. (2011) The Lean Startup. Penguin Books Ltd, London.
314
Riggins, F. J . (2003) Market segmentation and information development costs
in a two-tiered fee-based and sponsorship-based web site. Journal
of Management Information Systems, Vol. 19 (3), 69–86.
Riordan, M. H. – Williamson, O. E. (1985) Asset specificity and economic
organization. International Journal of Industrial Organization,
Vol. 3 (4), 365–378.
Roberts, E. B. (1990) Evolving toward product and market-orientation: the
early years of technology-based firms. Journal of Product Inno-
vation Management, Vol. 7 (4), 274–287.
Robinson, W. T. – Fornell, C. – Sullivan, M. (1992) Are market pioneers in-
trinsically stronger than later entrants? Strategic Management
Journal, Vol. 13 (8), 609–624.
Rochet, J .-C. – Tirole, J . (2003) Platform competition in two-sided markets.
Journal of the European Economic Association, Vol. 1 (4), 990–
1029.
Rochet, J .-C. – Tirole, J. (2005) Two-sided markets?: A progress report. IDEI
Working Paper No. 275, Institut d’Économie Industrielle, Tou-
louse.
Rogers, E. M. (1995) Diffusion of Innovations. Simon and Schuster, New
York, NY.
Rohlfs, J . (1974) A theory of interdependent demand for a communications
service. The Bell Journal of Economics and Management Science,
Vol. 5 (1), 16–37.
Roseman, S. (2010) Out with the Old business in with the New.
, retrieved J une 2012.
Roson, R. (2005) Two-sided markets: A tentative survey. Review of Network
Economics, Vol. 4 (2), 142–160.
Rumbo, J . D. (2002) Consumer resistance in a world of advertising clutter:
The case of Adbusters. Psychology & Marketing, Vol. 19 (2),
127–148.
Rysman, M. (2004) Competition between networks: A study of the market for
yellow pages. The Review of Economic Studies, Vol. 71 (2), 483–
512.
Rysman, M. (2007) An empirical analysis of payment card usage. The Journal
of Industrial Economics, Vol. 55 (1), 1–36.
Rysman, M. (2009) The economics of two-sided markets. Journal of Eco-
nomic Perspectives, Vol. 23 (3), 125–143.
Saha, R. – Mantena, R. – Tilson, V. (2012) Essays on economics of technol-
ogy-enabled intermediaries. University of Rochester, Rochester,
NY.
315
Salminen, J . (2010) Power of Google: A study on online advertising exchange.
LAP Lambert Academic Publishing, Saarbrücken, Germany.
Salminen, J . – Hytönen, A. (2012) Viral coefficient – Unveiling the Holy Grail
of online marketing. Paper presented at the Conference of Theo-
retical and Applied Aspects of Cybernetics, Kiev, Ukraine.
Salminen, J . – Teixeira, J . (2013) Fool’s gold? Developer dilemmas in a
closed mobile application market platform. Presented at the ICEC
2013 The 15th International Conference on Electronic Commerce,
Turku.
Salomon, G. (1991) Transcending the qualitative-quantitative debate: The an-
alytic and systemic approaches to educational research. Educa-
tional Researcher, Vol. 20 (6), 10–18.
Samanta, S. K. (2009) Impact of the utility and cost on price in a two sided
market, SSRN Working Paper No. 1339620. Social Science Re-
search Network, Rochester, NY.
Sawhney, M. S. (1998) Leveraged high-variety strategies: From portfolio
thinking to platform thinking. Journal of the Academy of Market-
ing Science, Vol. 26 (1), 54–61.
Sawhney, M. – Verona, G. – Prandelli, E. (2005) Collaborating to create: The
Internet as a platform for customer engagement in product inno-
vation. Journal of Interactive Marketing, Vol. 19 (4), 4–17.
Schilling, M. (2009) Protecting or diffusing a technology platform: tradeoffs
in appropriability, network externalities, and architectural control.
In: Platforms, Markets and Innovation, ed. A. Gawer. Edward El-
gar Publishing, Cheltenham.
Schlosser, A. E. – White, T. B. – Lloyd, S. M. (2006) Converting web site
visitors into buyers: How web site investment increases consumer
trusting beliefs and online purchase intentions. Journal of Mar-
keting, Vol. 70 (2), 133–148.
Schmalensee, R. – Evans, D. S. (2007) Industrial organization of markets with
two-sided platforms. Competition Policy International, Vol. 3 (1).
Schmidt, C. (2004) The Analysis of Semi-structured Interviews. In: A Com-
panion to Qualitative Research, eds. U. Flick – E. von Kardoff –
I. Steinke, 253–258. SAGE Publications, London.
Schnettler, S. (2009) A structured overview of 50 years of small-world re-
search. Social Networks, Vol. 31 (3), 165–178.
Schoenberger, E. (1991) The corporate interview as a research method in eco-
nomic geography. The Professional Geographer, Vol. 43 (2),
180–189.
316
Schumpeter, J . A. (1961) The theory of economic development: An inquiry
into profits, capital, credit, interest, and the business cycle, Vol-
ume 55. Transaction Books, New Brunswick, NJ .
Schwenk, C. R. (1984) Cognitive simplification processes in strategic deci-
sion-making. Strategic Management Journal, Vol. 5 (2), 111–128.
Seale, C. (1999) Quality in qualitative research. Qualitative Inquiry, Vol. 5
(4), 465–478.
Semenzin, D. – Meulendijks, E. – Seele, W. – Wagner, C. – Brinkkemper, S.
(2012) Differentiation in freemium: Where does the line lie? In:
Software Business; Proceedings of the Third International Con-
ference, ICSOB 2012, Cambridge, M. A. Cusumano – B. Iyer – N.
Venkatraman, 291–296. Springer, Berlin.
Shampanier, K. – Mazar, N. – Ariely, D. (2007) Zero as a special price: The
true value of free products. Marketing Science, Vol. 26 (6), 742–
757.
Shapiro, C. (1983) Optimal pricing of experience goods. The Bell Journal of
Economics, Vol. 14 (2), 497–507.
Shapiro, C. – Varian, H. (1998) Information rules: A Strategic guide to the
economy. Harvard Business School Press, Boston, MA.
Sharma, S. – Mahajan, V. (1980) Early Warning Indicators of Business Fail-
ure. Journal of Marketing, Vol. 44 (4), 80.
Shaughnessy, H. (2012) Facebook’s 1 billion users: Why the sky is still the
limit. Forbes. , retrieved J une 2013.
Shepherd, D. A. – Kuratko, D. F. (2009) The death of an innovative project:
How grief recovery enhances learning. Business Horizons, Vol. 52
(5), 451–458.
Sheridan, J . (2011) Free as in beer and free as in speech: Public sector infor-
mation after 450 days of open government. Australian Law Li-
brarian, Vol. 19 (4), 251–255.
Shy, O. (2011) A short survey of network economics. Review of Industrial
Organization, Vol. 38 (2), 119–149.
Simon, H. A. (1956) Rational choice and the structure of the environment.
Psychological Review, Vol. 63 (2), 129–138.
Singh, J . V. – Tucker, D. J . – House, R. J . (1986) Organizational legitimacy
and the liability of newness. Administrative Science Quarterly,
Vol. 31 (2), 171–193.
Smith, D. – Menon, S. – Sivakumar, K. (2005) Online peer and editorial rec-
ommendations, trust, and choice in virtual markets. Journal of In-
teractive Marketing, Vol. 19 (3), 15–37.
317
Smith, T. (2008) Transmutable Postmortem. , retrieved J une
2012.
Sommer, R. – Sommer, B. B. (1992) A practical guide to behavioral research.
Oxford University Press, New York, NY.
Soonsawad, P. (2013) Developing a new model for conversion rate optimiza-
tion: A case study. International Journal of Business and Man-
agement, Vol. 8 (10), 41–51.
Spulber, D. F. (2010) Solving the circular conundrum: communication and
coordination in internet markets. Northwestern University Law
Review, Vol. 104 (2), 537–592.
Standing, S. – Standing, C. – Love, P. E. D. (2010) A review of research on e-
marketplaces 1997–2008. Decision Support Systems, Vol. 49 (1),
41–51.
Stanoevska-Slabeva, K. (2002) Toward a community-oriented design of Inter-
net platforms. International Journal of Electronic Commerce,
Vol. 6 (3), 71–95.
Summary of Startup Genome Report Extra: Premature Scaling (2011) Startup
Compass. , retrieved December 2011.
Stinchcombe, A. L. (1965) Social structure and organizations. In: Handbook of
Organizations, ed. J . G. March, 153–193. Rand McNally, Chi-
cago, IL.
Strauss, A. – Corbin, J . (1994) Grounded theory methodology. In: Handbook
of qualitative research, eds. N. K. Denzin – Y. S. Lincoln, 273–
285. Sage, Thousand Oaks, CA.
Stvilia, B. – Twidale, M. B. – Smith, L. C. – Gasser, L. (2008) Information
quality work organization in Wikipedia. Journal of the American
Society for Information Science and Technology, Vol. 59 (6),
983–1001.
Suarez, F. F. (2005) Network effects revisited: the role of strong ties in tech-
nology selection. Academy of Management Journal, Vol. 48 (4),
710–720.
Suddaby, R. (2006) From the editors: What grounded theory is not. Academy
of Management Journal, Vol. 49 (4), 633–642.
Sun, M. – Tse, E. (2007) When does the winner take all in two-sided markets?
Review of Network Economics, Vol. 6 (1), 1–25.
Susarla, A. – Oh, J .-H. – Tan, Y. (2012). Social networks and the diffusion of
user-generated content: evidence from YouTube. Information
Systems Research, Vol. 23 (1), 23–41.
318
Tajirian, A. (2005) Multiple viable domain name marketplaces can co-exist.
DomainMart.
, retrieved J une 2013.
Tang, J . (2008) Postmortem. , retrieved J une 2012.
Tanriverdi, H. – Lee, C.-H. (2008. Within-industry diversification and firm
performance in the presence of network externalities: Evidence
from the software industry. Academy of Management Journal,
Vol. 51 (2), 381–397.
Teece, D. J. (2010) Business models, business strategy and innovation. Long
Range Planning, Vol. 43 (2), 172–194.
Teece, D. J. (2011) Dynamic capabilities: A framework for profitability. Ivey
Business Journal, Vol. 18 (7), 509–533.
The Chubby Team (2010) How to Develop a Product that Nobody Wants: The
Story of ChubbyBrain. , retrieved J une 2012.
Teixeira, J . (2012) Open-source technologies realizing social networks: A
multiple descriptive case-study. In: Proceedings of the 8th IFIP
WG 2.13 International Conference, OSS 2012, Hammamet, Tuni-
sia, September 10-13, 2012, IFIP Advances in Information and
Communication Technology 378/2012, eds. I. Hammouda – B.
Lundell – T. Mikkonen – W. Scacchi, 250–255. Springer, Berlin.
The Entrepreneurial Enlightenment (2012) Startup Compass ,
retrieved J une 2013.
Thelwall, M. (2008) Social networks, gender, and friending: An analysis of
MySpace member profiles. Journal of the American Society for
Information Science and Technology, Vol. 59 (8), 1321–1330.
Thompson, S. A. – Sinha, R. K. (2008) Brand communities and new product
adoption: the influence and limits of oppositional loyalty. Journal
of Marketing, Vol. 72 (6), 65–80.
Tourangeau, R. (1984) Cognitive science and survey methods. In: Cognitive
aspects of survey methodology: Building a bridge between disci-
plines, eds. T. B. J abine – M. L. Straf – J . M. Tanur – R.
Tourangeau, 73–100. National Academy Press, Washington, D.C.
319
Trusov, M. – Bucklin, R. E. – Pauwels, K. H. (2008) Effects of word-of-mouth
versus traditional marketing: Findings from an internet social
networking site. Robert H. Smith School of Business Research
Paper No. RHS 06-065, Robert H. Smith School of Business,
University of Maryland, College Park, MD.
Tversky, A. – Kahneman, D. (1974) J udgment under uncertainty: Heuristics
and biases. Science, Vol. 185 (4157), 1124–1131.
Urquhart, C. – Lehmann, H. – Myers, M. D. (2010) Putting the “theory” back
into grounded theory: guidelines for grounded theory studies in in-
formation systems. Information Systems Journal, Vol. 20 (4),
357–381.
Uyar, A. (2009) Investigation of the accuracy of search engine hit counts.
Journal of Information Science, Vol. 35 (4), 469–480.
Valente, T. W. (1995) Network models of the diffusion of innovations (Quan-
titative Methods in Communication). Hampton Press, New York,
NY.
Van Huyck, J . B. – Battalio, R. C. – Beil, R. O. (1990) Tacit coordination
games, strategic uncertainty, and coordination failure. The Ameri-
can Economic Review, Vol. 80 (1), 234–248.
Varian, H. R. (2003) Innovation, components and complements. MISQ Special
Issue Workshop, University of California, Berkeley, CA.
Venkatesh, V. – Davis, F. D. (2000) A theoretical extension of the technology
acceptance model: four longitudinal field studies. Management
Science, Vol. 46 (2), 186–204.
Viégas, F. B. – Donath, J . (2004) Social network visualization: Can we go be-
yond the graph. Paper presented at the Workshop on Social Net-
works (CSCW'04) , Chicago, IL.
Vesterinen, V. (2010) Cross-Promotion Network Applifier Hit 55 Million Us-
ers In 4 Months. Arctic Startup. , retrieved 4.7.2013.
Wagner, S. M. – Lukassen, P. – Mahlendorf, M. (2010) Misused and missed
use – Grounded theory and objective hermeneutics as methods for
research in industrial marketing. Industrial Marketing Manage-
ment, Vol. 39 (1), 5–15.
Wang, S. – Zheng, S. – Xu, L. – Li, D. – Meng, H. (2008) A literature review
of electronic marketplace research: Themes, theories and an inte-
grative framework. Information Systems Frontiers, Vol. 10 (5),
555–571.
320
Warner, A. (2009) Founder Of Fit Fuel On What You Can Learn From His
Company’s Closure – With Luke Burgis. , retrieved J une 2012.
Warner, A. (2012) Maids In Black: An Entrepreneurial Dilemma (Tech Or
Non-Tech?) – with Rohan Gilkes. , retrieved J une 2012.
Wasserman, N. (2013) The Founder’s Dilemmas: Anticipating and Avoiding
the Pitfalls That Can Sink a Startup. Princeton University Press.
Wasserman, T. – West, C. M. (2006) The business of open source. Paper pre-
sented at the International Conference on Open Source Systems,
Como, Italy.
Watson, J . – Everett, J . (1993) Defining small business failure. International
Small Business Journal, Vol. 11 (3), 35–48.
Watson, J . – Everett, J . (1999) Small business failure rates: Choice of defini-
tion and industry effects. International Small Business Journal,
Vol. 17 (2), 31–47.
Webster, F. E. (1969). New product adoption in industrial markets: a frame-
work for analysis. The Journal of Marketing, Vol. 33 (3), 35–39.
Weitzel, T. – Beimborn, D. – König, W. (2006) A Unified Economic Model of
Standard Diffusion: The Impact of Standardization Cost, Network
Effects, and Network Topology. MIS Quarterly, Vol. 30 (1), 489–
514.
Westheimer, N. (2008) Bricabox: Goodbye world! , retrieved J une 2012.
Westland, J . C. (2010). Critical mass and willingness to pay for social net-
works. Electronic Commerce Research and Applications, Vol. 9
(1), 6–19.
Wilcox, J . W. (1971) A simple theory of financial ratios as predictors of fail-
ure. Journal of Accounting Research, Vol. 9 (2), 389–395.
Williamson, O. E. (1981) The economics of organization: The transaction cost
approach. The American Journal of Sociology, Vol. 87 (3), 548–
577.
Wilson, F. (2006) The Freemium Business Model. , retrieved J une 2013.
Wilson, M. (2006) Bubble 2.0 begins to deflate – 9 lessons. , retrieved J une 2012.
Wiltbank, R. – Dew, N. – Read, S. – Sarasvathy, S. D. (2006) What to do
next? The case for non-predictive strategy. Strategic Management
Journal, Vol. 27 (10), 981–998.
321
Wired (2008) Twitter’s business model? Well, ummmm... Wired.
, retrieved 5.7.2013.
Wright, J . (2004) One-sided logic in two-sided markets. Review of Network
Economics, Vol. 3 (1), 1–21.
Wu, R. – Lin, M. (2012) Seller Diversity on a Two-Sided Platform. Research
Collection, School Of Information Systems Working Paper,
Singapore Management University, Singapore.
Wunsch-Vincent, S. – Vickery, G. (2007) Participative web: user-created
content. OECD Directorate for Science, Technology and Industry.
Xia, M. – Huang, Y. – Duan, W. – Whinston, A. B. (2007) To continue shar-
ing or not to continue sharing? – An empirical analysis of user de-
cision in peer-to-peer sharing networks. Information Systems Re-
search, Vol. 23 (1), 247–259.
Xin, M. – Levina, N. (2008) Software-as-a-service model: Elaborating client-
side adoption factors. In: Proceedings of the 29th International
Conference on Information Systems, eds. R. Boland – M.
Limayem – B. Pentland, Paris, France.
Yelle, L. E. (1979) The learning curve: Historical review and comprehensive
survey. Decision Sciences, Vol. 10 (2), 302–328.
Yaghmour, K. (2010) Failing Forward. , retrieved J une 2012.
Yli-Renko, H – J anakiraman, R. (2008) How Customer Portfolio Affects New
Product Development in Technology-Based Entrepreneurial
Firms. Journal of Marketing, Vol. 72 (September), 131–148.
Yoo, C. (2010) Innovations in the Internet’s architecture that challenge the
status quo. Journal of Telecommunications and High Technology
Law, Vol. 8 (1), 79–99.
Zacharakis A, M. G. D. (1999) Differing perceptions of new venture failure: a
matched exploratory study of venture capitalists and entrepre-
neurs. Journal of Small Business Management, Vol. 37 (3), 1–14.
Zeithaml, V. A. (1988) Consumer perceptions of price, quality, and value: a
means-end model and synthesis of evidence. The Journal of Mar-
keting, Vol. 52 (3), 2–22.
Zeldman, J . – Marcotte, E. (2009). Designing with web standards. New Rid-
ers, Berkeley, CA.
Zhang, X. – Zhu, F. (2011) Group Size and Incentives to Contribute: A Natu-
ral Experiment at Chinese Wikipedia. American Economic Re-
view, Vol. 101 (4), 1601–1615.
322
Zhao, Y. – Zhu, Q. (2012) Evaluation on crowdsourcing research: Current
status and future direction. Information Systems Frontiers, April
2012, 1–18.
Zheng, H. – Li, D. – Hou, W. (2011) Task design, motivation, and participa-
tion in crowdsourcing contests. International Journal of Elec-
tronic Commerce, Vol. 15 (4), 57–88.
Zynga Inc. (2013) Zynga reports fourth quarter and full year 2012 financial
results. Nasdaq OMX GlobeNewswire. , retrieved 1.7.2013.
3
2
3
A
P
P
E
N
D
I
X
1
C
O
D
I
N
G
G
U
I
D
E
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
2
_
B
U
S
I
N
E
S
S
M
O
D
E
L
e
c
o
n
o
m
i
c
v
i
a
b
i
l
i
t
y
I
m
p
l
y
i
n
g
l
a
c
k
o
f
e
c
o
n
o
m
i
c
v
i
a
b
i
l
i
t
y
.
T
h
e
l
a
r
g
e
r
l
e
s
s
o
n
h
e
r
e
i
s
t
h
a
t
b
u
s
i
n
e
s
s
i
s
u
l
t
i
m
a
t
e
l
y
d
r
i
v
e
n
b
y
e
c
o
n
o
m
i
c
s
,
a
n
d
e
c
o
n
o
m
i
c
s
a
r
e
d
r
i
v
e
n
b
y
i
n
c
e
n
t
i
v
e
s
.
E
n
t
r
e
p
r
e
n
e
u
r
s
,
a
n
d
i
n
p
a
r
t
i
c
u
l
a
r
w
e
b
e
n
-
t
r
e
p
r
e
n
e
u
r
s
,
a
r
e
o
f
t
e
n
d
r
i
v
e
n
b
y
a
m
y
o
p
i
a
t
h
a
t
s
a
y
s
"
i
f
I
l
i
k
e
i
t
,
s
o
w
i
l
l
e
v
e
r
y
o
n
e
e
l
s
e
.
"
a
g
g
r
e
g
a
t
i
o
n
S
t
r
a
t
e
g
y
t
o
s
o
u
r
c
e
d
a
t
a
f
r
o
m
o
t
h
e
r
w
e
b
s
i
t
e
s
F
i
r
s
t
a
n
d
f
o
r
e
m
o
s
t
,
w
e
’
l
l
u
s
e
t
e
c
h
n
o
l
o
g
y
w
e
’
v
e
b
u
i
l
t
t
o
a
g
g
r
e
g
a
t
e
i
n
f
o
r
m
a
t
i
o
n
a
b
o
u
t
s
t
a
r
t
u
p
s
.
e
x
t
e
n
d
e
d
f
i
r
m
D
e
l
e
g
a
t
i
n
g
t
a
s
k
s
t
o
u
s
e
r
s
.
n
e
t
w
o
r
k
e
f
f
e
c
t
s
R
e
f
e
r
r
i
n
g
t
o
n
e
t
w
o
r
k
e
f
f
e
c
t
s
o
r
n
e
t
w
o
r
k
v
a
l
u
e
.
W
e
s
t
r
u
c
k
u
p
o
n
t
h
e
i
d
e
a
t
h
a
t
i
f
w
e
h
a
d
f
i
f
t
y
j
o
u
r
n
a
l
i
s
t
s
,
a
n
d
t
h
e
y
e
a
c
h
c
r
o
s
s
-
p
r
o
m
o
t
e
d
e
a
c
h
o
t
h
e
r
t
o
t
h
e
i
r
s
o
c
i
a
l
n
e
t
w
o
r
k
s
,
t
h
e
n
o
v
e
r
t
i
m
e
w
e
w
o
u
l
d
g
e
t
m
o
r
e
a
n
d
m
o
r
e
p
e
o
p
l
e
t
o
r
e
a
d
e
a
c
h
o
t
h
e
r
’
s
c
o
n
t
e
n
t
.
p
l
a
t
f
o
r
m
F
o
u
n
d
e
r
s
d
e
s
c
r
i
b
i
n
g
p
l
a
t
f
o
r
m
b
u
s
i
n
e
s
s
m
o
d
e
l
.
F
o
r
t
h
o
s
e
u
n
f
a
m
i
l
i
a
r
w
i
t
h
i
m
e
r
c
i
v
e
,
w
e
w
e
r
e
a
s
o
c
i
a
l
m
e
d
i
a
m
a
r
k
e
t
i
n
g
c
o
m
p
a
n
y
t
h
a
t
p
r
o
v
i
d
e
d
b
r
a
n
d
e
d
i
n
s
t
a
n
t
m
e
s
s
a
g
i
n
g
a
p
p
l
i
c
a
t
i
o
n
s
a
s
n
e
w
t
o
u
c
h
p
o
i
n
t
s
f
o
r
c
o
n
s
u
m
e
r
e
n
g
a
g
e
m
e
n
t
.
u
s
e
r
g
e
n
e
r
a
t
i
o
n
c
o
m
m
u
n
i
t
y
s
t
r
a
t
e
g
y
F
o
u
n
d
e
r
t
a
l
k
s
o
n
a
c
o
m
m
u
n
i
t
y
o
f
u
s
e
r
s
.
T
h
e
a
u
d
i
e
n
c
e
t
o
o
k
r
e
s
p
o
n
s
i
b
i
l
i
t
y
f
o
r
w
h
a
t
w
e
n
t
o
n
a
t
e
a
c
h
l
o
c
a
l
[
s
t
a
r
t
u
p
]
s
i
t
e
a
n
d
d
e
b
a
t
e
d
l
o
c
a
l
i
s
s
u
e
s
i
n
a
c
i
v
i
l
m
a
n
n
e
r
b
e
c
a
u
s
e
i
t
w
a
s
a
b
o
u
t
t
h
e
i
r
c
o
m
m
u
-
n
i
t
y
.
A
n
d
a
s
i
n
t
h
e
i
r
p
h
y
s
i
c
a
l
c
o
m
m
u
n
i
t
y
,
t
h
e
y
w
e
r
e
p
r
o
u
d
o
f
i
t
a
n
d
t
o
o
k
c
a
r
e
o
f
i
t
.
u
g
c
,
c
a
r
e
f
u
l
f
o
r
n
o
t
t
a
k
i
n
g
a
d
v
a
n
t
a
g
e
E
x
p
l
o
i
t
i
n
g
u
s
e
r
s
w
i
l
l
r
e
s
u
l
t
i
n
r
e
s
i
s
t
a
n
c
e
.
S
i
n
c
e
w
e
w
e
r
e
n
’
t
m
a
k
i
n
g
a
n
y
m
o
n
e
y
f
o
r
t
h
e
m
,
a
n
d
i
t
a
p
p
e
a
r
e
d
t
o
t
h
e
m
–
c
o
r
-
r
e
c
t
l
y
–
t
h
a
t
n
o
o
n
e
w
a
s
e
v
e
n
r
e
a
d
i
n
g
t
h
e
i
r
c
o
n
t
e
n
t
,
t
h
e
r
e
w
a
s
n
o
e
a
r
t
h
l
y
r
e
a
s
o
n
t
h
e
y
w
o
u
l
d
k
e
e
p
w
r
i
t
i
n
g
“
f
o
r
u
s
”
.
s
e
a
r
c
h
e
x
t
e
r
n
a
l
i
t
i
e
s
F
o
u
n
d
e
r
s
t
a
l
k
s
o
n
s
e
a
r
c
h
e
n
g
i
n
e
b
e
n
e
f
i
t
s
.
F
o
r
y
e
a
r
s
t
h
e
r
e
w
e
r
e
c
e
r
t
a
i
n
q
u
e
r
i
e
s
a
b
o
u
t
c
e
m
e
n
t
p
r
o
d
u
c
t
i
o
n
a
n
d
u
n
e
m
p
l
o
y
-
m
e
n
t
w
h
e
r
e
w
e
w
e
r
e
t
h
e
f
i
r
s
t
o
r
s
e
c
o
n
d
l
i
n
k
s
.
T
h
e
c
o
r
o
l
l
a
r
y
t
o
t
h
i
s
r
u
l
e
i
s
g
o
o
d
323 323
3
2
4
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
g
o
o
g
l
e
f
o
o
[
w
e
b
s
i
t
e
t
r
a
f
f
i
c
]
w
o
n
’
t
s
a
v
e
y
o
u
.
Y
o
u
n
e
e
d
t
h
a
t
t
r
a
f
f
i
c
t
o
t
r
a
n
s
l
a
t
e
i
n
t
o
a
c
o
m
m
u
n
i
t
y
.
A
v
i
s
i
t
i
s
n
o
t
a
n
i
n
t
e
r
e
s
t
i
n
g
s
t
a
t
i
s
t
i
c
e
s
p
e
c
i
a
l
l
y
i
n
a
b
u
s
i
n
e
s
s
t
h
a
t
r
e
q
u
i
r
e
s
t
h
e
c
o
m
m
u
n
i
t
y
t
o
p
r
o
d
u
c
e
c
o
n
t
e
n
t
.
u
s
e
r
g
e
n
e
r
a
t
i
o
n
,
i
n
c
e
n
t
i
v
e
s
I
n
c
e
n
t
i
v
e
s
t
o
m
o
t
i
v
a
t
e
c
r
o
w
p
a
r
t
i
c
i
p
a
t
i
o
n
.
A
n
d
w
h
e
n
w
e
o
f
f
e
r
e
d
c
o
f
f
e
e
c
a
r
d
s
i
n
e
x
c
h
a
n
g
e
f
o
r
p
o
s
t
s
,
[
s
t
a
r
t
u
p
]
u
s
e
r
s
a
c
t
u
-
a
l
l
y
r
e
b
e
l
l
e
d
.
T
h
e
y
t
h
o
u
g
h
t
w
e
w
e
r
e
m
a
k
i
n
g
t
h
e
s
i
t
e
t
o
o
c
o
m
m
e
r
c
i
a
l
,
e
v
e
n
t
h
o
u
g
h
a
d
s
d
i
d
n
'
t
s
e
e
m
t
o
b
o
t
h
e
r
t
h
e
m
a
t
a
l
l
[
…
]
b
e
c
a
u
s
e
l
o
c
a
l
a
d
s
a
r
e
l
o
c
a
l
i
n
f
o
r
m
a
t
i
o
n
.
v
i
r
a
l
s
t
r
a
t
e
g
y
I
d
e
a
l
p
r
o
d
u
c
t
d
i
f
f
u
s
i
o
n
e
i
t
h
e
r
t
h
r
o
u
g
h
“
b
u
i
l
t
-
i
n
”
v
i
r
a
l
f
e
a
t
u
r
e
s
o
r
f
r
i
c
t
i
o
n
l
e
s
s
s
h
a
r
i
n
g
.
A
n
e
x
c
e
p
t
i
o
n
i
s
i
n
h
e
r
e
n
t
l
y
v
i
r
a
l
i
d
e
a
s
,
e
s
p
e
c
i
a
l
l
y
o
n
e
-
t
o
-
m
a
n
y
v
i
r
a
l
i
t
y
,
w
h
e
r
e
n
o
r
m
a
l
u
s
e
o
f
y
o
u
r
p
r
o
d
u
c
t
r
e
a
c
h
e
s
n
e
w
u
s
e
r
s
,
n
o
t
“
w
o
r
d
-
o
f
-
m
o
u
t
h
”
v
i
r
a
l
t
h
a
t
r
e
q
u
i
r
e
s
u
s
e
r
s
t
o
a
d
v
o
c
a
t
e
y
o
u
.
W
i
t
h
i
n
h
e
r
e
n
t
v
i
r
a
l
i
t
y
,
a
b
a
r
e
l
y
a
d
e
q
u
a
t
e
p
r
o
d
-
u
c
t
m
i
g
h
t
s
u
f
f
i
c
e
,
t
h
o
u
g
h
e
v
e
n
t
h
e
n
m
a
r
k
e
t
i
n
g
s
h
o
u
l
d
a
c
c
e
l
e
r
a
t
e
g
r
o
w
t
h
.
2
_
F
U
N
D
I
N
G
[
r
e
l
a
t
e
s
t
o
P
e
t
e
r
P
a
n
’
s
d
i
l
e
m
m
a
]
m
o
n
e
y
e
x
p
e
n
d
i
t
u
r
e
O
v
e
r
l
y
f
a
s
t
s
p
e
n
d
i
n
g
o
f
f
u
n
d
s
.
[
T
h
e
s
t
a
r
t
u
p
]
w
a
s
f
e
d
-
b
u
t
-
a
n
e
m
i
c
a
n
d
s
l
o
w
l
y
r
o
a
s
t
i
n
g
i
t
s
c
a
s
h
.
D
o
i
t
a
g
a
i
n
a
n
d
I
’
l
l
c
o
n
c
e
n
t
r
a
t
e
t
h
o
s
e
c
o
s
t
s
a
t
t
h
e
b
e
g
i
n
n
i
n
g
.
W
e
w
o
u
l
d
h
a
v
e
h
a
d
t
w
i
c
e
t
h
e
p
r
o
d
u
c
t
i
n
2
/
3
o
f
t
h
e
t
i
m
e
.
m
o
n
e
y
,
e
x
c
e
s
s
G
e
t
t
i
n
g
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
S
o
t
h
e
n
e
t
e
f
f
e
c
t
w
a
s
t
h
a
t
i
t
e
n
a
b
l
e
d
t
o
r
a
i
s
e
a
l
l
t
h
i
s
m
o
n
e
y
t
h
a
t
k
e
p
t
u
s
f
a
r
f
r
o
m
t
h
e
c
u
s
t
o
m
e
r
.
T
r
u
t
h
b
e
t
o
l
d
,
w
e
w
e
r
e
p
r
o
b
a
b
l
y
a
f
r
a
i
d
o
f
c
u
s
t
o
m
e
r
s
a
t
t
h
i
s
p
o
i
n
t
b
e
c
a
u
s
e
w
e
d
i
d
n
’
t
w
a
n
t
t
o
d
i
s
a
p
p
o
i
n
t
t
h
e
m
o
r
l
o
o
k
b
a
d
.
m
o
n
e
y
,
i
r
r
e
l
e
v
a
n
t
F
o
u
n
d
e
r
i
n
d
i
c
a
t
e
s
f
u
n
d
i
n
g
i
s
n
o
t
c
r
i
t
i
c
a
l
.
U
n
l
e
s
s
[
t
h
e
s
t
a
r
t
u
p
]
r
e
a
c
h
e
d
a
n
e
x
p
a
n
s
i
o
n
s
t
a
g
e
,
w
h
e
r
e
t
h
e
s
a
l
e
s
m
o
d
e
l
w
a
s
k
n
o
w
n
a
n
d
r
e
p
e
a
t
a
b
l
e
,
I
’
m
e
v
e
r
-
m
o
r
e
c
o
n
v
i
n
c
e
d
t
h
a
t
r
a
i
s
i
n
g
i
n
s
t
i
t
u
t
i
o
n
a
l
c
a
p
i
-
t
a
l
w
o
u
l
d
n
o
t
h
a
v
e
i
n
c
r
e
a
s
e
d
o
u
r
c
h
a
n
c
e
s
o
f
c
r
e
a
t
i
n
g
v
a
l
u
e
f
o
r
t
h
e
e
x
i
s
t
i
n
g
s
h
a
r
e
h
o
l
d
e
r
s
.
T
h
e
c
o
m
p
a
n
y
w
o
u
l
d
h
a
v
e
s
u
r
v
i
v
e
d
l
o
n
g
e
r
,
b
u
t
t
h
a
t
’
s
n
o
t
t
h
e
g
o
a
l
.
m
o
n
e
y
,
l
a
c
k
L
a
c
k
o
f
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
W
h
e
n
y
o
u
g
o
i
t
a
l
o
n
e
,
y
o
u
’
v
e
g
o
t
t
h
e
f
r
e
e
d
o
m
t
o
d
o
w
h
a
t
e
v
e
r
y
o
u
w
a
n
t
,
b
u
t
i
t
’
s
h
a
r
d
t
o
s
t
a
y
f
o
c
u
s
e
d
w
h
e
n
t
h
e
p
r
o
j
e
c
t
i
s
a
l
w
a
y
s
c
o
m
p
e
t
i
n
g
f
o
r
y
o
u
r
a
t
t
e
n
t
i
o
n
w
i
t
h
o
t
h
e
r
,
u
s
u
a
l
l
y
p
a
y
i
n
g
,
w
o
r
k
.
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
,
m
o
n
e
y
M
o
n
e
y
w
a
s
i
n
d
i
c
a
t
e
d
a
s
a
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
.
T
h
e
s
i
n
g
l
e
b
i
g
g
e
s
t
r
e
a
s
o
n
w
e
a
r
e
c
l
o
s
i
n
g
d
o
w
n
(
a
c
o
m
m
o
n
o
n
e
)
i
s
r
u
n
n
i
n
g
o
u
t
o
f
c
a
s
h
.
D
e
s
p
i
t
e
p
u
t
t
i
n
g
t
h
e
c
o
m
p
a
n
y
i
n
a
n
E
X
T
R
E
M
E
L
Y
l
e
a
n
p
o
s
i
t
i
o
n
,
g
e
n
e
r
-
a
t
i
n
g
r
e
v
e
n
u
e
,
a
n
d
h
o
l
d
i
n
g
o
u
t
a
s
l
o
n
g
a
s
w
e
c
o
u
l
d
,
w
e
d
i
d
n
'
t
h
a
v
e
t
h
e
c
a
s
h
t
o
324 324
3
2
5
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
k
e
e
p
g
o
i
n
g
.
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
,
s
p
e
n
t
t
i
m
e
o
n
f
u
n
d
i
n
g
S
p
e
n
d
i
n
g
t
i
m
e
l
o
o
k
i
n
g
f
o
r
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
W
e
s
p
e
n
t
6
m
o
n
t
h
s
r
a
i
s
i
n
g
o
u
r
a
n
g
e
l
r
o
u
n
d
,
a
n
d
f
i
n
a
l
l
y
c
l
o
s
e
d
i
t
a
f
t
e
r
l
a
u
n
c
h
i
n
g
o
u
r
p
u
b
l
i
c
b
e
t
a
i
n
F
e
b
r
u
a
r
y
2
0
0
8
.
W
e
r
a
i
s
e
d
l
e
s
s
t
h
a
n
w
e
s
h
o
u
l
d
h
a
v
e
(
C
h
r
i
s
D
i
x
o
n
n
o
t
e
s
t
h
a
t
t
h
e
w
o
r
s
t
t
h
i
n
g
a
s
e
e
d
-
s
t
a
g
e
c
o
m
p
a
n
y
c
a
n
d
o
i
s
r
a
i
s
e
t
o
o
l
i
t
t
l
e
m
o
n
e
y
a
n
d
o
n
l
y
r
e
a
c
h
p
a
r
t
w
a
y
t
o
a
m
i
l
e
s
t
o
n
e
,
w
h
i
c
h
i
s
g
r
e
a
t
a
d
v
i
c
e
)
.
W
e
s
p
e
n
t
3
m
o
n
t
h
s
g
e
t
t
i
n
g
r
e
a
d
y
f
o
r
r
a
i
s
i
n
g
o
u
r
n
e
x
t
r
o
u
n
d
,
t
h
e
n
6
m
o
n
t
h
s
b
e
a
t
i
n
g
o
u
r
h
e
a
d
a
g
a
i
n
s
t
a
w
a
l
l
t
r
y
i
n
g
t
o
r
a
i
s
e
m
o
n
e
y
d
u
r
i
n
g
t
h
e
f
i
n
a
n
c
i
a
l
c
o
l
l
a
p
s
e
.
2
_
M
A
R
K
E
T
b
a
r
r
i
e
r
s
t
o
a
d
o
p
t
i
o
n
R
e
a
s
o
n
s
w
h
y
u
s
e
r
s
w
o
u
l
d
n
o
t
a
d
o
p
t
t
h
e
p
l
a
t
f
o
r
m
.
A
f
t
e
r
e
v
e
n
t
u
a
l
l
y
d
i
s
c
o
v
e
r
i
n
g
t
h
a
t
s
e
t
u
p
a
n
d
c
o
n
f
i
g
u
r
a
t
i
o
n
w
a
s
o
u
r
p
r
i
m
a
r
y
a
d
o
p
t
i
o
n
p
r
o
b
l
e
m
(
a
n
d
a
f
t
e
r
t
r
y
i
n
g
a
n
d
f
a
i
l
i
n
g
t
o
i
m
p
l
e
m
e
n
t
v
a
r
i
o
u
s
s
t
r
a
t
e
g
i
e
s
t
o
m
a
k
e
i
t
s
i
m
p
l
e
a
n
d
e
a
s
y
)
,
w
e
t
r
i
e
d
t
o
m
o
v
e
t
o
t
h
e
o
t
h
e
r
e
n
d
o
f
t
h
e
s
p
e
c
t
r
u
m
.
c
h
a
n
g
i
n
g
b
e
h
a
v
i
o
r
P
e
r
c
e
i
v
e
d
d
i
f
f
i
c
u
l
t
y
o
f
c
h
a
n
g
i
n
g
b
e
h
a
v
i
o
r
.
I
b
e
g
a
n
t
o
r
e
a
l
i
z
e
t
h
a
t
e
x
p
e
r
t
s
i
n
m
o
s
t
f
i
e
l
d
s
a
r
e
u
s
u
a
l
l
y
p
r
e
t
t
y
b
u
s
y
p
e
o
p
l
e
,
a
n
d
w
i
t
h
t
h
e
e
x
c
e
p
t
i
o
n
o
f
t
e
c
h
n
i
c
a
l
f
o
l
k
s
(
w
h
o
s
p
e
n
d
t
r
e
m
e
n
d
o
u
s
a
m
o
u
n
t
s
o
f
t
i
m
e
i
n
f
r
o
n
t
o
f
c
o
m
p
u
t
e
r
s
)
,
m
o
s
t
o
t
h
e
r
p
e
o
p
l
e
d
o
n
'
t
l
i
k
e
s
i
t
t
i
n
g
i
n
f
r
o
n
t
o
f
o
n
e
,
a
n
d
p
r
o
b
a
b
l
y
w
o
n
'
t
a
n
s
w
e
r
q
u
e
s
t
i
o
n
s
f
o
r
a
f
e
w
m
e
a
s
l
y
b
u
c
k
s
.
c
o
m
p
e
t
i
t
i
o
n
W
h
y
c
o
m
p
e
t
i
n
g
p
l
a
t
f
o
r
m
s
a
r
e
s
u
p
e
-
r
i
o
r
/
i
n
f
e
r
i
o
r
.
W
e
w
e
r
e
o
f
f
e
r
i
n
g
i
n
f
o
r
m
a
t
i
o
n
o
n
g
r
e
a
t
a
l
b
u
m
s
a
n
d
c
o
m
m
u
n
i
t
y
v
o
t
i
n
g
.
B
u
t
o
t
h
e
r
s
i
t
e
s
l
i
k
e
L
a
s
t
.
f
m
a
n
d
H
y
p
e
M
a
c
h
i
n
e
w
e
r
e
o
f
f
e
r
i
n
g
t
h
e
a
c
t
u
a
l
m
u
s
i
c
.
T
h
a
t
w
a
s
a
c
o
m
p
e
t
i
t
i
v
e
a
d
v
a
n
t
a
g
e
t
h
a
t
’
s
h
a
r
d
t
o
b
e
a
t
a
n
d
w
e
l
a
c
k
e
d
a
s
i
g
n
i
f
i
c
a
n
t
u
s
e
r
b
a
s
e
t
o
c
o
n
v
i
n
c
e
e
n
o
u
g
h
p
e
o
p
l
e
l
a
c
k
o
f
h
i
s
t
o
r
y
,
t
r
u
s
t
R
e
p
o
r
t
e
d
c
a
u
s
e
f
o
r
m
i
s
t
r
u
s
t
b
y
c
u
s
t
o
m
e
r
s
.
I
n
c
o
m
m
u
n
i
t
y
-
g
e
n
e
r
a
t
e
d
m
e
d
i
a
,
t
r
u
s
t
i
s
e
v
e
r
y
t
h
i
n
g
.
W
h
e
n
y
o
u
a
s
k
f
o
r
s
u
b
m
i
s
-
s
i
o
n
s
,
c
o
n
t
r
i
b
u
t
o
r
s
g
o
t
h
r
o
u
g
h
a
n
i
n
s
t
a
n
t
i
n
t
e
r
n
a
l
c
a
l
c
u
l
a
t
i
o
n
:
“
D
o
I
t
r
u
s
t
t
h
e
s
e
p
e
o
p
l
e
w
i
t
h
m
y
w
o
r
k
?
”
W
h
e
n
y
o
u
r
s
i
t
e
i
s
b
r
a
n
d
n
e
w
,
y
o
u
’
v
e
g
o
t
n
o
r
e
c
o
r
d
t
o
r
e
l
y
o
n
.
A
n
d
w
i
t
h
m
o
r
e
s
h
a
d
y
“
u
s
e
r
-
g
e
n
e
r
a
t
e
d
c
o
n
t
e
n
t
”
s
c
h
e
m
e
s
p
o
p
p
i
n
g
u
p
e
v
e
r
y
d
a
y
,
p
e
o
p
l
e
h
a
v
e
t
h
e
i
r
d
e
f
e
n
s
e
s
u
p
(
a
s
w
e
l
l
t
h
e
y
s
h
o
u
l
d
)
.
m
a
r
k
e
t
,
s
m
a
l
l
m
a
r
k
e
t
s
y
n
-
d
r
o
m
e
C
o
n
c
l
u
s
i
o
n
o
f
t
h
e
f
o
u
n
d
e
r
t
h
a
t
t
h
e
m
a
r
k
e
t
i
s
t
o
o
s
m
a
l
l
v
i
s
-
à
-
v
i
s
a
m
b
i
t
i
o
n
s
.
G
i
v
e
n
t
h
e
i
n
d
u
s
t
r
y
t
r
e
n
d
s
w
e
w
e
r
e
s
e
e
i
n
g
,
w
e
j
u
s
t
d
i
d
n
'
t
s
e
e
a
l
i
g
h
t
a
t
t
h
e
e
n
d
o
f
t
h
e
t
u
n
n
e
l
w
o
r
t
h
s
u
r
v
i
v
i
n
g
f
o
r
.
T
h
i
s
m
a
r
k
e
t
m
a
y
g
e
t
s
l
i
g
h
t
l
y
b
e
t
t
e
r
,
b
u
t
i
t
'
s
n
o
t
g
o
i
n
g
t
o
b
e
a
b
i
g
c
o
m
p
a
n
y
.
2
_
M
A
R
K
E
T
I
N
G
b
r
a
n
d
i
n
g
B
r
a
n
d
i
n
g
t
h
e
p
l
a
t
f
o
r
m
.
O
u
r
n
a
m
e
,
C
h
u
b
b
y
B
r
a
i
n
,
i
s
m
e
m
o
r
a
b
l
e
,
a
b
i
t
p
l
a
y
f
u
l
a
n
d
a
t
o
u
c
h
i
r
r
e
v
e
r
e
n
t
.
325 325
3
2
6
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
W
e
,
h
o
w
e
v
e
r
,
g
e
n
e
r
a
l
l
y
p
u
t
o
u
t
p
e
r
s
o
n
a
l
i
t
y
-
l
e
s
s
,
f
a
c
t
-
b
a
s
e
d
a
n
a
l
y
s
e
s
t
o
s
h
o
w
h
o
w
g
o
o
d
o
u
r
d
a
t
a
w
a
s
(
t
h
e
r
e
w
e
r
e
s
o
m
e
e
x
c
e
p
t
i
o
n
s
)
.
I
n
s
t
e
a
d
o
f
m
a
k
i
n
g
o
u
r
n
a
m
e
a
n
d
c
o
n
t
e
n
t
/
a
p
p
r
o
a
c
h
c
o
n
s
i
s
t
e
n
t
a
n
d
h
e
n
c
e
a
n
a
s
s
e
t
,
o
u
r
b
r
a
n
d
i
n
g
m
u
l
l
e
t
c
o
n
f
u
s
e
d
p
e
o
p
l
e
.
c
o
m
m
u
n
i
c
a
t
i
n
g
p
r
o
d
u
c
t
,
p
o
s
i
-
t
i
o
n
i
n
g
P
o
s
i
t
i
o
n
i
n
g
t
h
e
p
l
a
t
f
o
r
m
a
g
a
i
n
s
t
c
o
m
p
e
t
i
-
t
i
o
n
.
W
e
d
i
d
n
’
t
d
e
s
c
r
i
b
e
w
h
a
t
w
e
w
a
n
t
e
d
t
o
d
o
c
l
e
a
r
l
y
e
n
o
u
g
h
.
W
h
e
n
I
t
o
l
d
p
e
o
p
l
e
t
h
e
i
d
e
a
i
n
p
e
r
s
o
n
,
t
h
e
y
a
l
w
a
y
s
r
e
a
l
l
y
l
i
k
e
d
i
t
.
B
u
t
w
h
e
n
t
h
e
y
c
a
m
e
t
o
t
h
e
s
i
t
e
,
t
h
e
y
d
i
d
n
’
t
g
e
t
i
t
.
h
i
g
h
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
c
o
s
t
E
x
p
r
e
s
s
e
d
c
o
n
c
e
r
n
o
v
e
r
c
o
s
t
l
y
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
.
W
e
n
e
v
e
r
s
e
r
i
o
u
s
l
y
f
i
g
u
r
e
d
o
u
t
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
,
a
n
d
h
a
d
t
r
o
u
b
l
e
g
r
o
w
i
n
g
t
h
r
o
u
g
h
o
u
t
o
u
r
v
a
r
i
o
u
s
p
i
v
o
t
s
.
C
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
i
s
h
a
r
d
a
n
d
m
o
r
e
e
x
p
e
n
-
s
i
v
e
t
h
a
n
m
o
s
t
p
e
o
p
l
e
r
e
a
l
i
z
e
w
h
e
n
s
t
a
r
t
i
n
g
a
c
o
m
p
a
n
y
.
p
e
r
s
o
n
a
l
s
e
l
l
i
n
g
P
e
r
s
o
n
a
l
s
e
l
l
i
n
g
a
s
a
m
e
a
n
s
t
o
d
r
i
v
e
p
l
a
t
-
f
o
r
m
p
a
r
t
i
c
i
p
a
t
i
o
n
.
T
h
e
p
l
a
n
w
a
s
t
o
p
a
r
t
n
e
r
w
i
t
h
m
a
r
k
e
t
i
n
g
a
g
e
n
c
i
e
s
a
s
w
e
l
l
a
s
s
e
l
l
d
i
r
e
c
t
l
y
t
o
c
l
i
-
e
n
t
s
s
i
m
i
l
a
r
t
o
t
h
e
a
p
p
r
o
a
c
h
t
a
k
e
n
b
y
p
r
o
v
i
d
e
r
s
o
f
c
u
s
t
o
m
b
r
a
n
d
e
d
w
i
d
g
e
t
s
,
F
a
c
e
b
o
o
k
a
p
p
s
,
a
n
d
m
o
b
i
l
e
a
p
p
s
.
T
h
i
s
s
t
r
a
t
e
g
y
e
v
e
n
t
u
a
l
l
y
p
r
o
d
u
c
e
d
s
o
m
e
g
r
e
a
t
r
e
s
u
l
t
s
b
u
t
i
t
w
a
s
a
c
a
s
e
o
f
t
o
o
l
i
t
t
l
e
,
t
o
o
l
a
t
e
.
2
_
P
R
O
D
U
C
T
t
e
c
h
n
o
l
o
g
y
c
h
o
i
c
e
H
o
w
t
e
c
h
n
o
l
o
g
y
c
h
o
i
c
e
a
f
f
e
c
t
e
d
p
l
a
t
f
o
r
m
d
e
v
e
l
o
p
m
e
n
t
.
I
r
e
a
s
o
n
e
d
t
h
a
t
t
h
e
p
l
a
t
f
o
r
m
w
a
s
t
h
e
c
o
r
e
o
f
o
u
r
t
e
c
h
n
o
l
o
g
y
,
a
n
d
w
e
w
e
r
e
a
t
e
c
h
n
o
l
o
g
y
c
o
m
p
a
n
y
,
a
n
d
s
m
a
r
t
t
e
c
h
n
o
l
o
g
y
c
o
m
p
a
n
i
e
s
n
e
e
d
e
d
t
h
e
f
l
e
x
i
b
i
l
i
t
y
t
h
a
t
c
o
m
e
s
f
r
o
m
w
r
i
t
i
n
g
t
h
e
c
o
r
e
o
f
t
h
e
i
r
p
l
a
t
f
o
r
m
t
h
e
m
s
e
l
v
e
s
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
i
s
c
o
u
l
d
o
n
l
y
b
e
c
o
n
s
i
d
e
r
e
d
p
r
e
m
a
t
u
r
e
o
p
t
i
m
i
z
a
t
i
o
n
.
v
a
l
i
d
a
t
i
n
g
i
d
e
a
,
t
e
c
h
n
o
l
o
g
y
E
n
s
u
r
i
n
g
o
r
n
o
t
e
n
s
u
r
i
n
g
t
h
a
t
t
h
e
r
e
i
s
d
e
-
m
a
n
d
f
o
r
t
h
e
p
r
o
d
u
c
t
.
A
n
o
t
h
e
r
c
h
o
i
c
e
w
a
s
t
h
e
d
e
c
i
s
i
o
n
t
o
s
t
a
r
t
w
i
t
h
a
r
c
a
d
e
g
a
m
e
s
,
w
h
i
c
h
w
a
s
d
o
n
e
v
e
r
y
e
a
r
l
y
o
n
(
e
v
e
n
b
e
f
o
r
e
I
j
o
i
n
e
d
,
I
t
h
i
n
k
)
w
i
t
h
o
u
t
m
u
c
h
t
h
o
u
g
h
t
a
t
a
l
l
.
I
t
w
a
s
j
u
s
t
t
h
e
f
i
r
s
t
t
h
i
n
g
t
h
a
t
c
a
m
e
t
o
m
i
n
d
w
h
e
n
w
e
t
h
o
u
g
h
t
o
f
c
a
s
u
a
l
g
a
m
i
n
g
,
s
o
w
e
s
a
i
d
"
L
e
t
'
s
s
t
a
r
t
w
i
t
h
t
h
a
t
,
a
n
d
t
h
e
n
e
x
p
a
n
d
i
n
t
o
o
t
h
e
r
a
r
c
h
e
t
y
p
e
s
w
h
e
n
w
e
h
a
v
e
m
o
r
e
e
x
p
e
r
i
e
n
c
e
.
"
l
a
u
n
c
h
r
e
l
a
t
e
s
t
o
p
i
o
n
e
e
r
’
s
d
i
l
e
m
m
a
f
o
l
l
o
w
e
r
s
a
d
v
a
n
t
a
g
e
s
E
x
a
m
p
l
e
s
o
f
l
a
t
e
-
m
o
v
e
r
’
s
b
e
n
e
f
i
t
s
.
W
e
s
a
b
e
l
a
u
n
c
h
e
d
a
b
o
u
t
1
0
m
o
n
t
h
s
b
e
f
o
r
e
M
i
n
t
.
M
o
r
e
t
h
e
s
h
a
m
e
t
h
a
t
w
e
d
i
d
n
'
t
c
a
p
i
t
a
l
i
z
e
o
n
t
h
a
t
e
a
r
l
y
l
e
a
d
.
T
h
e
r
e
'
s
a
l
o
t
t
o
b
e
s
a
i
d
f
o
r
n
o
t
r
u
s
h
i
n
g
t
o
m
a
r
k
e
t
,
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
i
s
t
a
k
e
s
t
h
e
f
i
r
s
t
e
n
t
r
a
n
t
s
m
a
k
e
.
S
h
i
p
p
i
n
g
a
"
m
i
n
i
m
u
m
v
i
a
b
l
e
p
r
o
d
u
c
t
"
i
m
m
e
d
i
a
t
e
l
y
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
a
r
k
e
t
d
i
r
e
c
t
l
y
m
a
k
e
s
g
o
o
d
s
e
n
s
e
t
o
m
e
,
b
u
t
e
n
g
a
g
i
n
g
w
i
t
h
a
n
d
s
u
p
p
o
r
t
i
n
g
u
s
e
r
s
i
s
a
n
y
t
h
i
n
g
b
u
t
f
r
e
e
.
326 326
3
2
7
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
l
a
u
n
c
h
,
b
e
n
e
f
i
t
s
o
f
e
a
r
l
y
E
x
a
m
p
l
e
s
o
f
e
a
r
l
y
-
m
o
v
e
r
’
s
b
e
n
e
f
i
t
s
.
W
e
t
a
l
k
e
d
a
b
o
u
t
“
r
e
l
e
a
s
e
e
a
r
l
y
/
r
e
l
e
a
s
e
o
f
t
e
n
,
”
b
u
t
w
e
r
e
s
c
a
r
e
d
o
f
l
o
o
k
i
n
g
l
i
k
e
i
d
i
o
t
s
i
n
f
r
o
n
t
o
f
m
a
j
o
r
W
a
l
l
S
t
r
e
e
t
a
n
d
h
e
d
g
e
f
u
n
d
c
l
i
e
n
t
s
.
I
s
i
t
b
e
t
t
e
r
t
o
w
a
i
t
a
b
i
t
b
e
f
o
r
e
r
e
l
e
a
s
i
n
g
t
o
h
a
v
e
a
m
o
r
e
c
o
m
p
e
l
l
i
n
g
p
r
o
d
u
c
t
o
r
t
o
b
e
g
i
n
g
e
t
t
i
n
g
f
e
e
d
-
b
a
c
k
o
n
a
l
e
s
s
i
m
p
r
e
s
s
i
v
e
o
f
f
e
r
i
n
g
?
W
e
c
h
o
s
e
#
1
;
i
n
r
e
t
r
o
s
p
e
c
t
I
t
h
i
n
k
w
e
s
h
o
u
l
d
h
a
v
e
c
h
o
s
e
n
#
2
.
B
y
c
h
o
o
s
i
n
g
t
o
w
a
i
t
w
e
l
o
s
t
o
u
r
i
n
t
i
m
a
c
y
w
i
t
h
t
h
e
c
u
s
t
o
m
e
r
(
M
i
s
t
a
k
e
#
5
a
g
a
i
n
)
,
f
a
l
l
i
n
g
i
n
t
o
t
h
e
c
l
a
s
s
i
c
(
a
s
a
“
g
r
e
e
n
”
e
n
t
r
e
p
r
e
n
e
u
r
I
d
i
d
n
’
t
k
n
o
w
t
h
i
s
,
b
u
t
a
s
a
s
e
a
s
o
n
e
d
f
o
u
r
-
y
e
a
r
v
e
n
t
u
r
e
i
n
v
e
s
t
o
r
I
k
n
o
w
t
h
i
s
n
o
w
)
t
r
a
p
o
f
p
u
r
s
u
i
n
g
a
“
s
c
i
e
n
c
e
p
r
o
j
e
c
t
,
”
n
o
t
b
u
i
l
d
i
n
g
a
c
o
m
m
e
r
c
i
a
l
l
y
s
a
l
a
b
l
e
p
r
o
d
u
c
t
.
l
a
u
n
c
h
,
d
e
l
a
y
e
d
E
x
a
m
p
l
e
s
o
f
l
a
t
e
-
l
a
u
n
c
h
p
r
o
b
l
e
m
s
.
A
t
s
o
m
e
p
o
i
n
t
,
w
e
r
e
a
c
h
e
d
a
n
a
l
m
o
s
t
g
o
-
l
i
v
e
p
r
o
d
u
c
t
.
B
u
t
a
f
t
e
r
s
h
o
w
i
n
g
i
t
t
o
a
f
e
w
V
C
s
,
w
e
r
e
a
l
i
z
e
d
w
e
m
i
s
s
e
d
s
o
m
e
f
e
a
t
u
r
e
s
t
o
m
a
k
e
i
t
t
r
u
l
y
s
o
c
i
a
l
.
I
n
s
t
e
a
d
o
f
r
e
l
e
a
s
i
n
g
t
h
e
b
e
t
a
,
w
e
d
e
c
i
d
e
d
t
o
c
r
e
a
t
e
t
h
e
v
2
o
f
t
h
e
p
r
o
d
u
c
t
.
W
r
o
n
g
.
W
e
w
i
l
l
n
e
v
e
r
f
i
n
i
s
h
t
h
i
s
v
2
(
l
a
c
k
o
f
m
o
n
e
y
,
s
l
o
w
l
y
d
y
i
n
g
m
o
t
i
v
a
t
i
o
n
)
.
l
a
u
n
c
h
,
d
i
s
a
d
v
a
n
t
a
g
e
o
f
e
a
r
l
y
E
x
a
m
p
l
e
s
o
f
e
a
r
l
y
-
l
a
u
n
c
h
p
r
o
b
l
e
m
s
.
A
f
t
e
r
o
u
r
f
u
n
d
i
n
g
,
i
n
s
t
e
a
d
o
f
i
n
v
e
s
t
i
n
g
i
n
t
h
e
p
r
o
d
u
c
t
f
e
a
t
u
r
e
s
t
o
d
e
l
i
v
e
r
o
n
E
v
e
n
t
V
u
e
d
i
r
e
c
t
l
y
d
r
i
v
i
n
g
p
e
o
p
l
e
t
o
e
v
e
n
t
s
(
h
i
n
d
s
i
g
h
t
:
m
i
s
t
a
k
e
)
,
w
e
i
m
m
e
d
i
a
t
e
l
y
w
e
n
t
o
u
t
a
n
d
t
r
i
e
d
t
o
s
e
l
l
t
h
e
s
o
c
i
a
l
n
e
t
w
o
r
k
t
o
o
l
t
o
c
o
n
f
e
r
e
n
c
e
o
r
g
a
n
i
z
e
r
s
.
2
_
T
E
A
M
h
o
b
b
y
,
h
a
l
f
-
h
e
a
r
t
e
d
e
f
f
o
r
t
I
n
d
i
c
a
t
i
o
n
o
f
n
o
n
-
s
e
r
i
o
u
s
b
u
s
i
n
e
s
s
.
I
t
t
u
r
n
e
d
s
o
m
e
t
h
i
n
g
I
l
i
k
e
d
d
o
i
n
g
i
n
t
o
a
c
h
o
r
e
m
a
i
n
l
y
b
e
c
a
u
s
e
a
t
t
h
e
s
a
m
e
t
i
m
e
I
w
a
s
b
u
s
y
m
a
r
k
e
t
i
n
g
t
h
e
s
i
t
e
,
r
e
d
e
s
i
g
n
i
n
g
i
t
,
a
n
d
a
t
t
e
n
d
i
n
g
c
l
a
s
s
e
s
.
(
I
w
a
s
i
n
c
o
l
l
e
g
e
f
o
r
b
u
s
i
n
e
s
s
f
o
r
1
8
h
o
u
r
s
o
f
t
h
e
w
e
e
k
)
.
i
n
t
e
r
n
a
l
i
z
a
t
i
o
n
I
n
d
i
c
a
t
e
s
p
r
o
f
o
u
n
d
u
n
d
e
r
s
t
a
n
d
i
n
g
o
f
b
u
s
i
-
n
e
s
s
l
o
g
i
c
.
B
u
t
n
o
m
a
t
t
e
r
h
o
w
m
a
n
y
t
i
m
e
s
y
o
u
r
e
a
d
o
r
h
e
a
r
c
e
r
t
a
i
n
l
e
s
s
o
n
s
,
y
o
u
h
a
v
e
t
o
d
o
i
t
y
o
u
r
s
e
l
f
t
o
r
e
a
l
l
y
u
n
d
e
r
s
t
a
n
d
.
l
a
c
k
o
f
f
o
c
u
s
M
e
n
t
i
o
n
e
d
l
a
c
k
o
f
f
o
c
u
s
(
r
e
l
a
t
e
s
t
o
p
i
v
o
t
d
i
l
e
m
m
a
)
.
W
e
m
a
d
e
t
h
e
p
r
o
d
u
c
t
m
u
c
h
t
o
o
b
r
o
a
d
s
u
c
h
t
h
a
t
i
t
d
i
d
a
l
i
t
t
l
e
b
i
t
o
f
e
v
e
r
y
t
h
i
n
g
.
U
n
f
o
r
t
u
n
a
t
e
l
y
,
i
t
d
i
d
n
’
t
d
o
a
n
y
t
h
i
n
g
e
x
c
e
p
t
i
o
n
a
l
l
y
w
e
l
l
.
A
s
a
n
e
x
a
m
p
l
e
,
w
h
e
n
w
e
u
s
e
d
i
t
t
o
r
e
p
l
a
c
e
o
u
r
p
h
p
B
B
m
e
s
s
a
g
e
b
o
a
r
d
i
n
s
t
a
n
c
e
,
o
u
r
c
o
m
m
u
n
i
t
y
w
a
s
i
n
a
n
u
p
r
o
a
r
b
e
c
a
u
s
e
i
t
l
a
c
k
e
d
m
a
n
y
o
f
s
p
e
c
i
a
l
i
z
e
d
f
e
a
t
u
r
e
s
a
m
a
t
u
r
e
f
o
r
u
m
o
f
f
e
r
s
(
e
.
g
.
n
o
t
i
f
i
c
a
t
i
o
n
s
a
n
d
h
a
n
d
l
i
n
g
c
o
d
e
e
x
a
m
p
l
e
s
i
n
a
c
o
m
m
e
n
t
)
.
l
e
a
r
n
i
n
g
g
a
i
n
s
M
e
n
t
i
o
n
e
d
l
e
a
r
n
i
n
g
b
e
n
e
f
i
t
s
d
u
e
t
o
s
t
a
r
t
u
p
(
r
e
l
a
t
e
s
t
o
p
l
a
t
f
o
r
m
s
p
e
c
i
f
i
c
i
t
y
)
.
A
s
a
m
u
s
i
c
s
i
t
e
i
t
h
a
s
h
a
s
n
’
t
e
x
a
c
t
l
y
“
f
a
i
l
e
d
”
a
t
t
h
e
m
o
m
e
n
t
,
i
t
s
t
i
l
l
p
u
l
l
s
i
n
a
r
o
u
n
d
3
0
0
0
p
e
o
p
l
e
a
m
o
n
t
h
m
a
i
n
l
y
f
r
o
m
s
e
a
r
c
h
e
n
g
i
n
e
s
.
I
n
o
w
c
o
n
s
i
d
e
r
i
t
a
h
o
b
b
y
s
i
t
e
a
n
d
I
’
m
l
o
o
k
i
n
g
f
o
r
w
a
r
d
t
o
a
p
p
l
y
i
n
g
w
h
a
t
I
l
e
a
r
n
e
d
w
i
t
h
m
y
n
e
x
t
s
t
a
r
t
-
u
p
,
I
n
t
e
g
r
a
t
e
.
327 327
3
2
8
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
m
a
r
k
e
t
m
e
m
o
r
y
R
e
f
e
r
r
i
n
g
t
o
s
i
m
u
l
t
a
n
e
o
u
s
s
i
m
i
l
a
r
s
t
r
a
t
e
g
i
c
d
e
c
i
s
i
o
n
s
t
h
a
t
a
r
e
b
a
s
e
d
o
n
c
o
m
m
o
n
i
n
f
o
r
-
m
a
t
i
o
n
(
r
e
l
a
t
e
s
t
o
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
)
.
A
s
T
w
i
t
t
e
r
r
e
c
e
i
v
e
d
m
o
r
e
a
t
t
e
n
t
i
o
n
,
c
l
o
n
e
s
s
h
o
w
e
d
u
p
e
v
e
r
y
w
h
e
r
e
.
M
y
d
e
c
i
s
i
o
n
w
a
s
n
o
t
t
o
c
o
m
p
e
t
e
w
i
t
h
a
n
e
w
a
n
d
a
l
r
e
a
d
y
c
r
o
w
d
e
d
s
p
a
c
e
,
b
u
t
t
o
f
i
n
d
a
d
i
f
f
e
r
-
e
n
t
a
n
g
l
e
.
I
d
e
c
i
d
e
d
t
o
f
o
c
u
s
o
n
s
c
a
l
a
b
l
e
t
e
c
h
n
o
l
o
g
y
,
m
a
k
i
n
g
a
b
e
t
t
h
a
t
o
n
c
e
T
w
i
t
t
e
r
a
n
d
o
t
h
e
r
s
w
i
l
l
g
e
t
p
o
p
u
l
a
r
,
t
h
e
r
e
w
i
l
l
b
e
a
r
e
a
l
n
e
e
d
f
o
r
t
h
a
t
k
i
n
d
o
f
t
e
c
h
n
o
l
o
g
y
b
y
o
t
h
e
r
p
l
a
y
e
r
s
.
m
o
t
i
v
e
s
f
o
r
s
h
a
r
i
n
g
,
w
h
y
t
e
l
l
s
t
o
r
y
M
o
t
i
v
e
f
o
r
w
r
i
t
i
n
g
p
o
s
t
-
m
o
r
t
e
m
.
I
r
e
a
d
p
o
s
t
s
o
n
H
a
c
k
e
r
N
e
w
s
a
b
o
u
t
“
W
h
y
m
y
s
t
a
r
t
-
u
p
f
a
i
l
e
d
”
a
n
d
f
o
u
n
d
t
h
e
m
i
n
t
e
r
e
s
t
i
n
g
b
e
c
a
u
s
e
i
t
’
s
a
r
a
w
s
t
a
r
t
-
u
p
s
t
o
r
y
n
o
t
t
h
e
f
a
i
r
y
t
a
l
e
s
t
h
a
t
m
a
g
a
z
i
n
e
s
l
i
k
e
t
o
w
r
i
t
e
a
b
o
u
t
.
S
o
h
e
r
e
i
s
m
y
s
h
o
t
a
t
i
t
.
s
e
c
o
n
d
o
p
i
n
i
o
n
,
i
n
t
e
r
p
r
e
t
a
t
i
o
n
O
t
h
e
r
c
o
m
m
e
n
t
s
o
n
t
h
e
s
t
a
r
t
u
p
i
n
a
d
d
i
t
i
o
n
t
o
t
h
o
s
e
o
f
t
h
e
f
o
u
n
d
e
r
.
I
w
a
s
p
r
o
b
a
b
l
y
w
h
a
t
y
o
u
w
o
u
l
d
c
a
l
l
a
n
e
a
r
l
y
C
h
i
c
a
g
o
a
d
o
p
t
e
r
o
f
M
e
e
t
r
o
,
a
n
d
I
t
h
o
u
g
h
t
i
t
w
a
s
a
g
r
e
a
t
i
d
e
a
,
a
n
d
s
t
i
l
l
d
o
.
W
i
t
h
e
a
c
h
s
u
c
c
e
e
d
i
n
g
v
e
r
s
i
o
n
o
f
t
h
e
s
o
f
t
w
a
r
e
,
h
o
w
e
v
e
r
,
t
h
e
i
n
t
e
r
f
a
c
e
s
e
e
m
e
d
t
o
g
e
t
w
e
i
r
d
e
r
a
n
d
w
e
i
r
d
e
r
u
n
t
i
l
I
c
o
u
l
d
n
’
t
f
i
g
u
r
e
i
t
o
u
t
a
n
y
m
o
r
e
.
s
e
l
f
-
a
t
t
r
i
b
u
t
i
o
n
F
o
u
n
d
e
r
a
s
s
i
g
n
i
n
g
b
l
a
m
e
o
f
f
a
i
l
u
r
e
t
o
s
e
l
f
.
A
s
c
o
-
f
o
u
n
d
e
r
a
n
d
C
E
O
o
f
L
o
o
k
e
r
y
,
t
h
e
b
u
c
k
s
t
o
p
s
w
i
t
h
m
e
a
n
d
n
o
o
n
e
e
l
s
e
.
s
t
a
r
t
u
p
f
a
t
i
g
u
e
S
e
n
t
i
m
e
n
t
o
f
f
r
u
s
t
r
a
t
i
o
n
o
v
e
r
m
a
n
y
h
a
r
d
-
s
h
i
p
s
(
r
e
l
a
t
e
s
t
o
‘
r
u
n
w
a
y
’
)
W
e
c
o
u
l
d
h
a
v
e
g
o
n
e
a
b
o
u
t
t
r
y
i
n
g
t
o
f
i
x
M
e
e
t
r
o
b
u
t
t
h
e
t
e
a
m
w
a
s
j
u
s
t
r
e
a
d
y
t
o
m
o
v
e
o
n
.
3
_
T
H
E
O
R
Y
A
N
D
C
O
N
-
S
T
R
U
C
T
S
[
t
h
i
s
s
e
c
t
i
o
n
i
n
c
l
u
d
e
s
d
i
l
e
m
m
a
s
]
d
i
l
e
m
m
a
s
c
o
l
d
s
t
a
r
t
I
n
a
b
i
l
i
t
y
t
o
g
e
t
c
o
n
t
e
n
t
w
i
t
h
o
u
t
u
s
e
r
s
.
U
n
d
e
r
e
s
t
i
m
a
t
e
d
t
h
e
“
C
o
l
d
S
t
a
r
t
”
p
r
o
b
l
e
m
,
I
r
e
a
d
t
h
i
s
a
r
t
i
c
l
e
b
y
B
o
k
a
d
o
S
o
c
i
a
l
D
e
s
i
g
n
w
h
i
c
h
t
a
l
k
s
a
b
o
u
t
a
b
i
g
i
s
s
u
e
y
o
u
f
a
c
e
w
i
t
h
a
s
o
c
i
a
l
s
i
t
e
,
e
s
p
e
c
i
a
l
l
y
w
h
e
n
i
t
r
e
l
i
e
s
o
n
u
s
e
r
-
g
e
n
e
r
a
t
e
d
c
o
n
t
e
n
t
.
T
h
e
v
a
l
u
e
y
o
u
p
r
o
v
i
d
e
t
o
y
o
u
r
u
s
e
r
s
c
e
n
t
e
r
s
a
r
o
u
n
d
t
h
e
c
o
n
t
e
n
t
o
n
t
h
e
s
i
t
e
,
s
o
t
o
b
u
i
l
d
a
u
s
e
r
-
b
a
s
e
y
o
u
n
e
e
d
a
l
o
t
o
f
c
o
n
t
e
n
t
c
r
e
a
t
e
d
b
y
t
h
e
f
i
r
s
t
u
s
e
r
s
t
o
k
i
c
k
-
o
f
f
t
h
e
c
o
m
m
u
n
i
t
y
.
l
o
n
e
l
y
u
s
e
r
I
n
a
b
i
l
i
t
y
t
o
g
e
t
u
s
e
r
s
w
i
t
h
o
u
t
o
t
h
e
r
u
s
e
r
s
.
[
I
]
f
s
o
m
e
o
n
e
w
a
s
n
’
t
o
n
l
i
n
e
w
h
e
n
y
o
u
w
e
r
e
o
n
l
i
n
e
,
t
h
e
y
w
e
r
e
n
o
g
o
o
d
t
o
y
o
u
.
W
h
i
l
e
t
h
e
r
e
a
l
-
t
i
m
e
c
h
a
t
a
s
p
e
c
t
o
f
t
h
e
a
p
p
l
i
c
a
t
i
o
n
m
a
d
e
f
o
r
s
o
m
e
r
e
a
l
l
y
s
e
r
e
n
-
d
i
p
i
t
o
u
s
m
e
e
t
i
n
g
s
,
i
t
a
l
s
o
m
a
d
e
i
t
h
a
r
d
e
r
f
o
r
p
e
o
p
l
e
t
o
g
a
u
g
e
t
h
e
a
c
t
i
v
i
t
y
o
f
t
h
e
i
r
c
o
m
m
u
n
i
t
i
e
s
,
e
s
p
e
c
i
a
l
l
y
i
f
t
h
e
y
l
o
g
g
e
d
i
n
a
t
o
d
d
h
o
u
r
s
,
p
e
o
p
l
e
w
e
r
e
s
e
t
a
s
a
w
a
y
,
e
t
c
.
m
o
n
e
t
i
z
a
t
i
o
n
I
n
a
b
i
l
i
t
y
t
o
c
h
a
r
g
e
m
o
n
e
y
a
n
d
g
e
t
u
s
e
r
s
.
F
o
r
f
o
u
r
y
e
a
r
s
w
e
h
a
v
e
o
f
f
e
r
e
d
t
h
e
s
y
n
c
h
r
o
n
i
z
a
t
i
o
n
s
e
r
v
i
c
e
f
o
r
n
o
c
h
a
r
g
e
,
328 328
3
2
9
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
p
r
e
d
i
c
a
t
e
d
o
n
t
h
e
h
y
p
o
t
h
e
s
i
s
t
h
a
t
a
b
u
s
i
n
e
s
s
m
o
d
e
l
w
o
u
l
d
e
m
e
r
g
e
t
o
s
u
p
p
o
r
t
t
h
e
f
r
e
e
s
e
r
v
i
c
e
.
W
i
t
h
t
h
a
t
i
n
v
e
s
t
m
e
n
t
t
h
e
s
i
s
t
h
w
a
r
t
e
d
,
t
h
e
r
e
i
s
n
o
w
a
y
t
o
p
a
y
e
x
-
p
e
n
s
e
s
,
p
r
i
m
a
r
i
l
y
s
a
l
a
r
y
a
n
d
h
o
s
t
i
n
g
c
o
s
t
s
.
W
i
t
h
o
u
t
t
h
e
r
e
s
o
u
r
c
e
s
t
o
k
e
e
p
t
h
e
s
e
r
v
i
c
e
g
o
i
n
g
,
w
e
m
u
s
t
s
h
u
t
i
t
d
o
w
n
.
p
e
t
e
r
p
a
n
I
n
a
b
i
l
i
t
y
t
o
g
r
o
w
w
i
t
h
o
u
t
f
u
n
d
i
n
g
.
W
h
e
n
y
o
u
’
v
e
g
o
t
m
o
n
e
y
,
y
o
u
c
a
n
t
a
k
e
m
o
r
e
t
i
m
e
t
o
d
o
i
t
r
i
g
h
t
,
b
u
t
y
o
u
’
v
e
g
o
t
h
i
g
h
e
r
e
x
p
e
c
t
a
t
i
o
n
s
f
o
r
r
e
t
u
r
n
s
,
a
n
d
a
w
h
o
l
e
h
o
s
t
o
f
o
t
h
e
r
c
o
m
p
l
i
c
a
t
i
o
n
s
.
W
h
e
n
y
o
u
g
o
i
t
a
l
o
n
e
,
y
o
u
’
v
e
g
o
t
t
h
e
f
r
e
e
d
o
m
t
o
d
o
w
h
a
t
e
v
e
r
y
o
u
w
a
n
t
,
b
u
t
i
t
’
s
h
a
r
d
t
o
s
t
a
y
f
o
c
u
s
e
d
w
h
e
n
t
h
e
p
r
o
j
e
c
t
i
s
a
l
w
a
y
s
c
o
m
p
e
t
i
n
g
f
o
r
y
o
u
r
a
t
t
e
n
t
i
o
n
w
i
t
h
o
t
h
e
r
,
u
s
u
a
l
l
y
p
a
y
i
n
g
,
w
o
r
k
.
p
i
v
o
t
I
m
p
o
s
s
i
b
l
e
t
o
c
h
a
n
g
e
t
h
e
c
o
u
r
s
e
o
f
b
u
s
i
n
e
s
s
b
u
t
a
l
s
o
t
o
k
e
e
p
i
t
.
[
T
]
h
e
m
o
s
t
s
i
g
n
i
f
i
c
a
n
t
c
h
a
n
g
e
,
w
h
i
c
h
I
d
i
d
n
o
t
a
p
p
r
e
c
i
a
t
e
u
n
t
i
l
i
t
w
a
s
t
o
o
l
a
t
e
,
w
a
s
t
h
e
f
u
n
d
a
m
e
n
t
a
l
s
h
i
f
t
i
n
t
h
e
n
a
t
u
r
e
o
f
t
h
e
s
t
a
r
t
u
p
.
I
m
o
v
e
d
f
r
o
m
b
u
i
l
d
i
n
g
a
c
o
n
s
u
m
e
r
-
f
a
c
i
n
g
w
e
b
s
i
t
e
t
o
a
c
o
m
b
i
n
a
t
i
o
n
o
f
i
n
f
r
a
s
t
r
u
c
t
u
r
e
p
r
o
v
i
d
e
r
a
n
d
s
o
f
t
-
w
a
r
e
d
e
v
e
l
o
p
e
r
w
h
i
c
h
m
a
d
e
t
h
e
o
v
e
r
h
e
a
d
c
o
s
t
m
u
c
h
h
i
g
h
e
r
,
t
i
m
e
t
o
m
a
r
k
e
t
l
o
n
g
e
r
,
a
n
d
r
e
s
o
u
r
c
e
s
m
u
c
h
m
o
r
e
d
i
f
f
i
c
u
l
t
t
o
f
i
n
d
.
r
e
m
o
r
a
s
c
u
r
s
e
J
o
i
n
i
n
g
a
p
l
a
t
f
o
r
m
r
e
t
u
r
n
s
u
s
e
r
s
b
u
t
g
i
v
e
s
a
w
a
y
p
o
w
e
r
.
W
e
e
x
p
o
s
e
d
o
u
r
s
e
l
v
e
s
t
o
a
h
u
g
e
s
i
n
g
l
e
p
o
i
n
t
o
f
f
a
i
l
u
r
e
c
a
l
l
e
d
F
a
c
e
b
o
o
k
.
I
’
v
e
r
a
n
t
e
d
f
o
r
y
e
a
r
s
a
b
o
u
t
h
o
w
b
a
d
a
n
i
d
e
a
i
t
i
s
f
o
r
s
t
a
r
t
u
p
s
t
o
b
e
m
o
b
i
l
e
-
c
a
r
r
i
e
r
d
e
p
e
n
d
e
n
t
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
e
r
e
i
s
n
o
d
i
f
f
e
r
e
n
c
e
b
e
t
w
e
e
n
V
e
r
i
z
o
n
W
i
r
e
l
e
s
s
a
n
d
F
a
c
e
b
o
o
k
i
n
t
h
i
s
c
o
n
t
e
x
t
.
i
n
d
u
s
t
r
y
-
s
p
e
c
i
f
i
c
p
r
o
b
l
e
m
A
p
r
o
b
l
e
m
t
h
a
t
i
s
s
p
e
c
i
f
i
c
t
o
a
m
a
r
k
e
t
o
r
i
n
d
u
s
t
r
y
.
I
k
n
o
w
t
h
e
r
e
’
s
a
p
l
a
c
e
f
o
r
a
w
i
s
d
o
m
o
f
c
r
o
w
d
s
a
p
p
r
o
a
c
h
i
n
t
h
e
p
u
b
l
i
s
h
i
n
g
w
o
r
l
d
,
b
u
t
p
u
b
l
i
s
h
e
r
s
a
n
d
a
r
t
i
s
t
s
a
r
e
s
t
i
l
l
f
i
g
u
r
i
n
g
i
t
o
u
t
.
F
o
r
n
o
w
,
i
t
s
e
e
m
s
l
i
k
e
a
d
i
r
e
c
t
s
u
b
m
i
s
s
i
o
n
m
o
d
e
l
i
s
w
h
a
t
p
e
o
p
l
e
a
r
e
m
o
s
t
c
o
m
f
o
r
t
a
b
l
e
w
i
t
h
.
p
r
o
b
l
e
m
o
f
a
c
t
i
v
e
u
s
e
K
e
e
p
i
n
g
u
s
e
r
s
a
c
t
i
v
e
a
f
t
e
r
r
e
g
i
s
t
e
r
i
n
g
.
L
o
o
k
c
a
r
e
f
u
l
l
y
a
t
m
o
s
t
h
y
p
e
r
-
l
o
c
a
l
s
i
t
e
s
a
n
d
s
e
e
j
u
s
t
h
o
w
m
u
c
h
p
o
s
t
i
n
g
i
s
r
e
a
l
l
y
b
e
i
n
g
d
o
n
e
,
e
s
p
e
c
i
a
l
l
y
b
y
m
e
m
b
e
r
s
o
f
t
h
e
c
o
m
m
u
n
i
t
y
a
s
o
p
p
o
s
e
d
t
o
t
h
e
s
i
t
e
s
'
o
p
e
r
a
t
o
r
s
.
A
n
y
b
o
d
y
w
h
o
h
a
s
r
u
n
a
h
y
p
e
r
-
l
o
c
a
l
s
i
t
e
w
i
l
l
t
e
l
l
y
o
u
t
h
a
t
i
t
t
a
k
e
s
a
c
o
u
p
l
e
o
f
y
e
a
r
s
j
u
s
t
t
o
g
e
t
t
o
a
p
o
i
n
t
w
h
e
r
e
y
o
u
'
v
e
t
r
u
l
y
g
o
t
a
v
i
b
r
a
n
t
o
n
l
i
n
e
c
o
m
m
u
n
i
t
y
.
i
l
l
u
s
i
o
n
s
,
f
a
l
l
a
c
i
e
s
a
n
d
b
i
a
s
e
s
b
u
i
l
d
i
t
a
n
d
t
h
e
y
w
i
l
l
c
o
m
e
E
x
p
e
c
t
i
n
g
t
h
e
p
r
o
d
u
c
t
t
o
“
s
e
l
l
i
t
s
e
l
f
”
.
W
e
f
e
l
l
i
n
t
o
t
h
e
“
b
u
i
l
d
i
t
a
n
d
t
h
e
y
w
i
l
l
”
c
o
m
e
s
c
h
o
o
l
o
f
t
h
o
u
g
h
t
(
a
l
t
h
o
u
g
h
e
v
e
n
w
h
e
n
t
h
e
y
c
a
m
e
,
w
e
s
t
i
l
l
w
e
r
e
n
’
t
i
n
g
o
o
d
s
h
a
p
e
)
.
U
s
e
r
s
d
i
d
n
’
t
r
e
v
i
e
w
b
e
c
a
u
s
e
t
h
e
r
e
w
a
s
n
o
e
n
l
i
g
h
t
e
n
e
d
s
e
l
f
-
i
n
t
e
r
e
s
t
f
o
r
t
h
e
m
t
o
d
o
s
o
.
N
o
b
o
d
y
w
a
n
t
e
d
t
o
e
d
i
t
o
u
r
d
a
t
a
f
o
r
t
h
e
s
a
m
e
r
e
a
s
o
n
.
P
e
o
p
l
e
s
u
b
m
i
t
t
e
d
c
o
m
p
a
n
i
e
s
t
o
o
u
r
d
a
t
a
b
a
s
e
329 329
3
3
0
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
m
a
i
n
l
y
l
o
o
k
i
n
g
f
o
r
a
l
i
n
k
.
d
o
g
f
o
o
d
b
l
i
n
d
n
e
s
s
R
e
f
u
s
i
n
g
t
o
a
d
m
i
t
f
l
a
w
s
i
n
t
h
e
p
r
o
d
u
c
t
.
T
h
e
o
r
i
g
i
n
a
l
C
h
u
b
b
y
B
r
a
i
n
w
a
s
a
h
a
p
p
y
m
a
r
r
i
a
g
e
o
f
o
u
r
i
n
t
e
r
e
s
t
s
w
r
a
p
p
e
d
i
n
c
u
s
t
o
m
e
r
i
n
d
i
f
f
e
r
e
n
c
e
.
W
e
n
e
v
e
r
t
a
l
k
e
d
t
o
u
s
e
r
s
/
c
u
s
t
o
m
e
r
s
t
o
s
e
e
i
f
t
h
i
s
w
o
u
l
d
b
e
o
f
i
n
t
e
r
e
s
t
t
o
t
h
e
m
.
A
n
d
s
o
c
u
s
t
o
m
e
r
s
u
l
t
i
m
a
t
e
l
y
t
o
l
d
u
s
t
h
r
o
u
g
h
t
h
e
i
r
i
n
d
i
f
-
f
e
r
e
n
c
e
t
h
a
t
t
h
e
y
d
i
d
n
’
t
c
a
r
e
a
b
o
u
t
o
u
r
i
d
e
a
.
c
o
n
f
i
r
m
a
t
i
o
n
b
i
a
s
S
e
e
k
i
n
g
p
o
s
i
t
i
v
e
f
e
e
d
b
a
c
k
a
n
d
i
g
n
o
r
i
n
g
n
e
g
a
t
i
v
e
o
n
e
.
M
a
n
w
a
s
I
s
t
u
p
i
d
.
W
h
e
n
p
e
o
p
l
e
a
s
k
e
d
t
o
s
i
g
n
u
p
w
i
t
h
o
u
t
i
t
,
I
t
o
l
d
t
h
e
m
n
o
.
W
h
e
n
t
h
e
p
e
o
p
l
e
w
h
o
d
i
d
s
i
g
n
u
p
w
e
r
e
w
o
r
r
i
e
d
a
b
o
u
t
t
h
i
n
g
s
b
e
i
n
g
p
o
s
t
e
d
t
o
t
h
e
i
r
w
a
l
l
s
,
I
d
i
d
n
’
t
u
n
d
e
r
s
t
a
n
d
t
h
e
p
r
o
b
l
e
m
.
W
h
e
n
r
e
a
d
e
r
s
s
a
i
d
t
h
e
y
w
o
u
l
d
n
’
t
s
i
g
n
u
p
t
o
c
o
m
m
e
n
t
,
I
t
h
o
u
g
h
t
i
t
w
a
s
j
u
s
t
a
s
m
a
l
l
m
i
n
o
r
i
t
y
.
p
e
r
f
e
c
t
p
r
o
d
u
c
t
f
a
l
l
a
c
y
R
e
f
u
s
a
l
t
o
l
a
u
n
c
h
u
n
t
i
l
t
h
e
p
r
o
d
u
c
t
i
s
p
e
r
-
f
e
c
t
(
w
h
i
c
h
i
s
t
h
e
n
n
e
v
e
r
i
s
)
.
A
n
d
t
h
e
s
e
b
a
d
b
e
h
a
v
i
o
r
s
w
e
r
e
r
e
i
n
f
o
r
c
e
d
b
y
a
n
u
n
p
l
a
n
n
e
d
e
v
e
n
t
t
h
a
t
s
h
a
r
p
l
y
i
m
p
a
c
t
e
d
o
u
r
p
s
y
c
h
e
:
b
e
i
n
g
o
n
t
h
e
f
r
o
n
t
p
a
g
e
o
f
t
h
e
F
i
n
a
n
c
i
a
l
T
i
m
e
s
.
I
t
i
s
h
a
r
d
t
o
c
a
l
l
i
t
a
m
i
s
t
a
k
e
s
i
n
c
e
w
e
d
i
d
n
’
t
s
e
e
k
t
o
g
e
t
s
u
c
h
e
x
p
o
s
u
r
e
,
b
u
t
I
p
u
t
i
t
d
o
w
n
a
s
M
i
s
t
a
k
e
#
3
.
T
o
b
e
h
o
n
e
s
t
,
t
h
i
s
s
i
n
g
l
e
f
a
c
t
w
a
s
a
v
e
r
y
m
e
a
n
i
n
g
f
u
l
f
a
c
t
o
r
i
n
o
u
r
f
a
i
l
u
r
e
.
I
t
r
a
i
s
e
d
t
h
e
l
e
v
e
l
o
f
e
x
p
e
c
t
a
t
i
o
n
s
s
o
h
i
g
h
t
h
a
t
i
t
m
a
d
e
u
s
r
e
l
u
c
t
a
n
t
t
o
r
e
l
e
a
s
e
a
n
y
t
h
i
n
g
t
h
a
t
w
a
s
n
’
t
e
a
r
t
h
-
s
h
a
t
t
e
r
i
n
g
.
i
l
l
u
s
i
o
n
o
f
f
r
e
e
E
x
p
e
c
t
i
n
g
a
p
r
i
o
r
i
t
h
a
t
c
u
s
t
o
m
e
r
s
w
o
u
l
d
n
o
t
p
a
y
.
V
i
d
e
o
p
r
o
d
u
c
e
r
s
a
r
e
a
f
r
a
i
d
o
f
c
h
a
r
g
i
n
g
f
o
r
c
o
n
t
e
n
t
,
b
e
c
a
u
s
e
t
h
e
y
d
o
n
'
t
t
h
i
n
k
p
e
o
p
l
e
w
i
l
l
p
a
y
.
A
n
d
t
h
e
y
'
r
e
l
a
r
g
e
l
y
r
i
g
h
t
.
C
o
n
s
u
m
e
r
s
s
t
i
l
l
d
o
n
'
t
l
i
k
e
p
a
y
i
n
g
f
o
r
s
t
u
f
f
,
p
e
r
i
o
d
.
i
t
w
o
n
’
t
h
a
p
p
e
n
t
o
m
e
T
e
n
d
e
n
c
y
t
o
e
x
p
e
c
t
t
h
a
t
r
i
s
k
w
i
l
l
n
o
t
r
e
a
l
i
z
e
i
n
o
w
n
s
t
a
r
t
u
p
.
W
e
s
a
w
t
h
a
t
t
h
i
s
d
i
d
n
’
t
y
e
t
e
x
i
s
t
i
n
a
n
e
a
s
y
t
o
u
s
e
f
o
r
m
a
t
.
W
e
s
a
w
o
t
h
e
r
s
t
h
a
t
h
a
d
t
r
i
e
d
a
n
d
w
e
f
i
g
u
r
e
d
t
h
e
y
h
a
d
f
a
i
l
e
d
b
e
c
a
u
s
e
t
h
e
y
d
i
d
a
b
a
d
j
o
b
a
n
d
w
e
c
o
u
l
d
d
o
b
e
t
t
e
r
.
T
h
e
l
e
s
s
o
n
t
o
l
e
a
r
n
h
e
r
e
i
s
I
f
t
h
e
r
e
a
r
e
b
u
r
n
i
n
g
h
u
s
k
s
o
f
c
a
r
s
a
l
o
n
g
t
h
e
r
o
a
d
w
a
y
,
y
o
u
s
h
o
u
l
d
p
r
o
b
a
b
l
y
a
s
s
u
m
e
t
h
a
t
d
r
i
v
i
n
g
t
h
e
r
e
i
s
d
a
n
g
e
r
-
o
u
s
.
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
E
m
p
l
o
y
i
n
g
f
a
m
o
u
s
s
u
c
c
e
s
s
s
t
o
r
i
e
s
a
s
a
m
o
d
e
l
f
o
r
d
e
c
i
s
i
o
n
s
r
e
g
a
r
d
i
n
g
o
n
e
’
s
o
w
n
s
t
a
r
t
u
p
.
[
W
]
e
t
h
o
u
g
h
t
“
l
e
t
’
s
t
a
k
e
e
l
e
m
e
n
t
s
f
r
o
m
s
i
t
e
s
w
e
l
i
k
e
a
n
d
t
w
e
a
k
t
h
e
m
”
a
n
d
w
e
’
l
l
g
e
t
t
h
e
s
a
m
e
m
a
g
i
c
a
l
e
f
f
e
c
t
s
o
n
o
u
r
s
i
t
e
t
h
a
t
t
h
e
y
’
v
e
g
o
t
t
e
n
.
W
r
o
n
g
.
F
e
a
t
u
r
e
s
d
o
n
’
t
w
o
r
k
i
n
a
v
a
c
u
u
m
.
T
h
e
y
w
o
r
k
b
e
c
a
u
s
e
y
o
u
t
a
k
e
t
i
m
e
t
o
u
n
d
e
r
s
t
a
n
d
y
o
u
r
c
u
s
t
o
m
e
r
a
n
d
t
h
e
n
b
u
i
l
d
f
e
a
t
u
r
e
s
t
o
a
c
c
o
m
m
o
d
a
t
e
t
h
e
m
.
c
o
p
y
b
u
s
i
n
e
s
s
m
o
d
e
l
t
o
n
e
w
c
o
n
t
e
x
t
E
x
p
l
i
c
i
t
a
m
b
i
t
i
o
n
t
o
c
o
p
y
e
l
e
m
e
n
t
s
o
f
a
n
-
o
t
h
e
r
b
u
s
i
n
e
s
s
m
o
d
e
l
i
n
t
o
a
n
e
w
i
n
d
u
s
t
r
y
.
T
h
e
i
d
e
a
i
t
s
e
l
f
f
o
r
C
o
n
t
r
a
s
t
r
e
a
m
w
a
s
n
’
t
t
o
o
i
n
n
o
v
a
t
i
v
e
o
r
o
r
i
g
i
n
a
l
.
I
t
w
a
s
i
n
-
s
p
i
r
e
d
b
y
D
i
g
g
.
c
o
m
w
h
i
c
h
h
a
d
a
p
p
l
i
e
d
t
h
e
s
a
m
e
m
o
d
e
l
t
o
n
e
w
s
a
n
d
a
r
t
i
c
l
e
s
o
n
t
h
e
W
e
b
.
I
s
t
i
l
l
b
e
l
i
e
v
e
a
p
p
l
y
i
n
g
t
h
i
s
m
o
d
e
l
t
o
m
u
s
i
c
i
s
i
n
t
e
r
e
s
t
i
n
g
a
n
d
u
s
e
f
u
l
,
b
u
t
t
h
e
r
e
w
e
r
e
s
o
m
a
n
y
o
t
h
e
r
m
e
-
t
o
o
D
i
g
g
s
i
t
e
s
.
330 330
3
3
1
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
s
u
n
k
c
o
s
t
f
a
l
l
a
c
y
R
e
f
u
s
i
n
g
t
o
m
a
k
e
c
h
a
n
g
e
s
d
u
e
t
o
t
i
m
e
s
p
e
n
t
.
S
i
n
c
e
t
h
e
s
e
r
v
i
c
e
w
a
s
o
u
r
c
h
i
l
d
w
e
w
e
r
e
r
e
l
u
c
t
a
n
t
t
o
m
a
k
e
a
d
e
c
i
s
i
o
n
a
b
o
u
t
c
l
o
s
i
n
g
i
t
f
a
s
t
e
r
a
n
d
l
i
m
i
t
l
o
s
s
e
s
.
W
e
’
v
e
b
e
e
n
t
r
i
c
k
i
n
g
o
u
r
s
e
l
v
e
s
t
h
i
n
k
i
n
g
t
h
a
t
e
v
e
r
y
t
h
i
n
g
w
o
u
l
d
b
e
f
i
n
e
w
h
i
l
e
w
e
c
o
u
l
d
n
’
t
g
e
t
t
h
e
a
p
p
l
i
c
a
t
i
o
n
b
a
c
k
t
o
w
o
r
k
p
r
o
p
e
r
l
y
.
t
e
c
h
n
o
l
o
g
y
b
i
a
s
T
e
n
d
e
n
c
y
t
o
b
e
l
i
e
v
e
t
h
a
t
t
e
c
h
n
o
l
o
g
y
w
i
l
l
s
o
l
v
e
b
u
s
i
n
e
s
s
p
r
o
b
l
e
m
s
.
I
n
s
t
e
a
d
o
f
h
a
v
i
n
g
p
r
o
d
u
c
t
m
a
n
a
g
e
m
e
n
t
a
s
t
h
e
a
d
v
o
c
a
t
e
f
o
r
t
h
e
c
u
s
t
o
m
e
r
a
n
d
t
h
e
p
r
o
d
u
c
t
e
v
a
n
g
e
l
i
s
t
,
w
e
h
a
d
t
e
c
h
n
o
l
o
g
y
r
u
n
n
i
n
g
t
h
e
s
h
o
w
i
n
a
v
a
c
u
u
m
.
H
u
g
e
m
i
s
t
a
k
e
.
T
h
i
s
a
l
l
o
w
e
d
u
s
t
o
p
e
r
p
e
t
u
a
t
e
t
h
e
s
c
i
e
n
c
e
p
r
o
j
e
c
t
f
o
r
m
u
c
h
,
m
u
c
h
l
o
n
g
e
r
t
h
a
n
w
e
s
h
o
u
l
d
h
a
v
e
.
r
a
c
e
t
o
b
o
t
t
o
m
R
e
l
a
t
e
s
t
o
c
o
m
p
e
t
i
t
i
v
e
p
r
e
s
s
u
r
e
t
o
o
f
f
e
r
a
f
r
e
e
p
r
o
d
u
c
t
.
W
e
a
l
s
o
c
o
n
s
i
d
e
r
e
d
r
e
f
o
c
u
s
i
n
g
X
m
a
r
k
s
a
s
a
f
r
e
e
m
i
u
m
s
y
n
c
b
u
s
i
n
e
s
s
,
b
u
t
t
h
e
p
r
o
s
p
e
c
t
s
t
h
e
r
e
a
r
e
g
r
i
m
t
o
o
:
w
i
t
h
t
h
e
e
m
e
r
g
e
n
c
e
o
f
c
o
m
p
e
t
e
n
t
s
y
n
c
f
e
a
t
u
r
e
s
b
u
i
l
t
i
n
t
o
M
o
z
i
l
l
a
F
i
r
e
f
o
x
a
n
d
G
o
o
g
l
e
C
h
r
o
m
e
,
i
t
’
s
h
a
r
d
t
o
s
e
e
u
s
e
r
s
p
a
y
i
n
g
f
o
r
a
s
e
r
v
i
c
e
t
h
a
t
t
h
e
y
c
a
n
n
o
w
g
e
t
f
o
r
f
r
e
e
.
r
u
n
w
a
y
I
n
d
i
c
a
t
e
s
t
i
m
e
r
u
n
n
i
n
g
o
u
t
.
B
i
g
c
o
r
p
o
r
a
t
i
o
n
s
c
a
n
a
f
f
o
r
d
t
o
s
p
e
n
d
a
b
i
l
l
i
o
n
d
o
l
l
a
r
s
a
n
d
w
a
i
t
1
0
y
e
a
r
s
,
t
h
o
u
g
h
,
w
h
i
l
e
s
t
a
r
t
u
p
s
u
s
u
a
l
l
y
c
a
n
o
n
l
y
a
f
f
o
r
d
a
f
e
w
t
e
n
s
o
f
t
h
o
u
s
a
n
d
s
a
n
d
a
b
o
u
t
6
m
o
n
t
h
s
.
Y
o
u
h
a
v
e
a
b
o
u
t
z
e
r
o
c
h
a
n
c
e
o
f
b
u
i
l
d
i
n
g
a
n
d
g
e
t
t
i
n
g
p
e
o
p
l
e
t
o
a
d
o
p
t
a
p
l
a
t
f
o
r
m
i
n
6
m
o
n
t
h
s
.
p
a
t
h
d
e
p
e
n
d
e
n
c
e
A
l
t
e
r
n
a
t
i
v
e
e
x
p
l
a
n
a
t
i
o
n
f
o
r
r
e
f
u
s
a
l
t
o
m
a
k
e
c
h
a
n
g
e
s
(
r
e
l
a
t
e
s
t
o
‘
s
u
n
k
c
o
s
t
f
a
l
l
a
c
y
’
)
.
T
h
e
i
n
i
t
i
a
l
i
d
e
a
d
o
e
s
c
h
a
n
g
e
,
a
n
d
i
t
'
s
a
l
m
o
s
t
c
e
r
t
a
i
n
l
y
w
r
o
n
g
.
T
h
e
t
h
i
n
g
i
s
,
t
h
e
i
n
i
t
i
a
l
i
d
e
a
d
e
t
e
r
m
i
n
e
s
h
o
w
t
h
e
i
n
i
t
i
a
l
i
d
e
a
w
i
l
l
c
h
a
n
g
e
,
w
h
i
c
h
i
s
c
r
u
c
i
a
l
t
o
a
l
l
t
h
e
e
x
e
c
u
t
i
o
n
t
h
a
t
f
o
l
l
o
w
s
i
t
.
A
n
d
t
h
e
d
e
v
i
l
i
s
o
f
t
e
n
i
n
t
h
e
d
e
t
a
i
l
s
:
l
i
t
t
l
e
d
e
c
i
s
i
o
n
s
m
a
d
e
e
a
r
l
y
o
n
c
a
n
c
r
u
c
i
a
l
l
y
i
m
p
a
c
t
b
i
g
d
e
c
i
s
i
o
n
s
m
a
d
e
l
a
t
e
r
.
t
r
a
n
s
f
e
r
e
n
c
e
p
r
o
b
l
e
m
R
e
a
l
i
z
a
t
i
o
n
t
h
a
t
f
a
m
o
u
s
e
x
a
m
p
l
e
s
d
o
n
o
t
a
p
p
l
y
i
n
a
l
l
c
a
s
e
s
(
r
e
l
a
t
e
s
t
o
‘
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
’
)
.
A
m
a
z
o
n
p
r
o
v
e
d
b
e
y
o
n
d
a
n
y
d
o
u
b
t
t
h
a
t
t
h
i
s
i
s
a
v
a
l
i
d
a
n
d
p
o
w
e
r
f
u
l
b
u
s
i
n
e
s
s
m
o
d
e
l
,
b
u
t
v
e
r
y
f
e
w
c
o
m
p
a
n
i
e
s
c
a
n
a
f
f
o
r
d
t
o
b
u
i
l
d
s
u
c
h
a
s
e
r
v
i
c
e
a
n
d
t
h
e
n
w
a
i
t
f
o
r
p
e
o
p
l
e
t
o
c
o
m
e
.
I
t
a
l
s
o
h
e
l
p
s
t
h
a
t
A
m
a
z
o
n
d
i
d
n
o
t
s
t
a
r
t
w
i
t
h
t
h
e
g
o
a
l
o
f
p
r
o
v
i
d
i
n
g
t
h
e
s
e
s
e
r
v
i
c
e
s
,
b
u
t
i
d
e
n
t
i
f
i
e
d
a
n
o
p
p
o
r
t
u
n
i
t
y
o
n
c
e
t
h
e
i
r
o
w
n
r
e
t
a
i
l
i
n
f
r
a
s
t
r
u
c
t
u
r
e
w
a
s
i
n
p
l
a
c
e
.
N
B
.
S
o
m
e
c
o
d
e
s
a
r
e
o
m
i
t
t
e
d
f
o
r
p
a
r
s
i
m
o
n
y
.
331 331
332
APPENDIX 2 IS THE COLD START DILEMMA
REALLY A DILEMMA?
Strictly speaking, any chicken-and-egg dilemma is not a strategic dilemma in
the sense of a contradictory decision-making situation, as it does not involve a
strategic choice, but a dependency. In its base form, it is a rather a question of
“how to attract the first party to the platform?” rather than “which one of the
two negative outcomes should I choose?” However, it is possible to formulate
it as a dilemma that will satisfy the definition, without assuming too much
context and therefore losing generality. This is achieved by introducing two
conflicting conditions:
1. If the startup provides the content, user generation effects will not
take place as the startup has provided the content.
2. If the startup does not provide the content, user generation effects
will not take place as there is no content.
Hence, the strategic action would be to create content or not, and both out-
comes would result in a “cold platform”.
Behind these conditions, there is the premise that users only create content
because other users have previously created content; for example, Wikipedia
only exists because it is Encyclopedia Britannica. Note that the problem was
framed so that there is a dependence on user generation (UG): that is, that us-
ers will take charge of content production in the long run. This excludes first-
party content platforms, such as Spotify that utilizes a music library provided
by the record labels, whereas an indie music platform needs to attract user-
generated music from indie artists. This restriction is based on the theoretical
model of ideal UG (see Chapter 3.4).
A second alternative would be to frame the dilemma in the question:
“Which party should the startup focus first?”, whereby focus on A would ne-
glect B which is required, and vice versa. This, however, would introduce
contextual factors unless strong assumptions were made concerning the feasi-
bility of parties, the applied monetization model, and also the propensity to
produce and generate content. For example, clearly the startup should focus on
those parties who are most likely to generate content; however, this is not a
strategic problem but, rather, one of identification (i.e., how they are to be
found).
Furthermore, the cold start dilemma, as defined here, is irrelevant if UG
does not lead to the virtuous cycle: What happens if user-generated content
does not lead to the creation of new content?
3
3
3
A
P
P
E
N
D
I
X
3
S
U
P
P
O
R
T
F
O
R
E
A
R
L
Y
A
N
D
L
A
T
E
L
A
U
N
C
H
E
S
T
o
o
e
a
r
l
y
T
o
o
l
a
t
e
1
.
"
C
u
t
e
h
a
c
k
s
c
a
n
c
o
s
t
y
o
u
t
i
m
e
.
T
a
k
e
t
h
e
t
i
m
e
t
o
d
o
t
h
i
n
g
s
r
i
g
h
t
f
r
o
m
t
h
e
b
e
g
i
n
n
i
n
g
.
"
(
W
i
l
s
o
n
2
0
0
6
)
.
"
T
h
e
b
i
g
g
e
s
t
f
a
l
l
o
u
t
o
f
b
u
i
l
d
i
n
g
o
u
r
p
l
a
t
f
o
r
m
o
u
r
s
e
l
v
e
s
w
a
s
t
h
a
t
w
e
c
o
u
l
d
n
’
t
b
u
i
l
d
q
u
i
c
k
l
y
e
n
o
u
g
h
.
W
h
e
n
y
o
u
r
o
l
l
y
o
u
r
o
w
n
i
n
f
r
a
s
t
r
u
c
t
u
r
e
,
e
v
e
r
y
t
h
i
n
g
t
a
k
e
s
t
i
m
e
,
m
o
r
e
t
i
m
e
t
h
a
n
y
o
u
c
a
n
a
f
f
o
r
d
.
A
n
d
w
e
h
a
d
p
r
o
m
i
s
e
d
t
h
e
j
o
u
r
n
a
l
i
s
t
s
t
h
a
t
w
e
w
o
u
l
d
v
e
r
y
q
u
i
c
k
l
y
b
u
i
l
d
a
l
a
r
g
e
l
i
s
t
o
f
f
e
a
t
u
r
e
s
,
n
o
n
e
o
f
w
h
i
c
h
w
e
r
e
p
r
o
d
u
c
e
d
n
e
a
r
l
y
q
u
i
c
k
l
y
e
n
o
u
g
h
.
T
h
i
s
w
a
s
t
h
e
m
a
j
o
r
c
a
u
s
e
o
f
d
i
s
i
l
l
u
s
i
o
n
m
e
n
t
–
w
e
o
v
e
r
-
p
r
o
m
i
s
e
d
a
n
d
u
n
d
e
r
-
d
e
l
i
v
e
r
e
d
[
…
]
.
"
(
B
i
g
g
a
r
2
0
1
0
)
.
2
.
"
T
h
e
r
e
’
s
a
l
o
t
t
o
b
e
s
a
i
d
f
o
r
n
o
t
r
u
s
h
i
n
g
t
o
m
a
r
k
e
t
,
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
i
s
t
a
k
e
s
t
h
e
f
i
r
s
t
e
n
t
r
a
n
t
s
m
a
k
e
.
S
h
i
p
p
i
n
g
a
'
m
i
n
i
m
u
m
v
i
a
b
l
e
p
r
o
d
u
c
t
'
i
m
m
e
d
i
a
t
e
l
y
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
a
r
k
e
t
d
i
r
e
c
t
l
y
m
a
k
e
s
g
o
o
d
s
e
n
s
e
t
o
m
e
,
b
u
t
e
n
g
a
g
i
n
g
w
i
t
h
a
n
d
s
u
p
p
o
r
t
i
n
g
u
s
e
r
s
i
s
a
n
y
t
h
i
n
g
b
u
t
f
r
e
e
.
O
b
s
e
r
-
v
a
t
i
o
n
c
a
n
b
e
c
h
e
a
p
e
r
.
M
i
n
t
(
a
n
d
s
o
m
e
o
t
h
e
r
s
)
d
i
d
w
e
l
l
b
y
s
e
e
i
n
g
w
h
e
r
e
w
e
s
c
r
e
w
e
d
u
p
,
a
n
d
w
a
i
t
i
n
g
t
o
l
a
u
n
c
h
u
n
t
i
l
t
h
e
y
h
a
d
a
b
e
t
t
e
r
a
p
p
r
o
a
c
h
.
"
(
H
e
d
l
u
n
d
2
0
1
0
)
.
"
E
v
e
n
t
h
o
u
g
h
I
w
a
s
v
e
r
y
c
a
r
e
f
u
l
a
b
o
u
t
s
c
o
p
e
a
n
d
o
u
r
f
o
c
u
s
o
n
p
r
o
d
u
c
t
,
I
u
n
d
e
r
e
s
-
t
i
m
a
t
e
d
t
h
e
t
i
m
e
i
t
w
o
u
l
d
t
a
k
e
t
o
g
e
t
t
o
a
V
1
a
n
d
t
h
e
n
i
n
j
e
c
t
i
t
i
n
t
o
t
h
e
m
a
r
k
e
t
.
"
(
S
m
i
t
h
2
0
0
8
)
.
3
.
"
A
f
t
e
r
o
u
r
f
u
n
d
i
n
g
,
i
n
s
t
e
a
d
o
f
i
n
v
e
s
t
i
n
g
i
n
t
h
e
p
r
o
d
u
c
t
f
e
a
t
u
r
e
s
t
o
d
e
l
i
v
e
r
o
n
E
v
e
n
t
V
u
e
d
i
r
e
c
t
l
y
d
r
i
v
i
n
g
p
e
o
p
l
e
t
o
e
v
e
n
t
s
(
h
i
n
d
s
i
g
h
t
:
m
i
s
t
a
k
e
)
,
w
e
i
m
m
e
d
i
a
t
e
l
y
w
e
n
t
o
u
t
a
n
d
t
r
i
e
d
t
o
s
e
l
l
t
h
e
s
o
c
i
a
l
n
e
t
w
o
r
k
t
o
o
l
t
o
c
o
n
f
e
r
-
e
n
c
e
o
r
g
a
n
i
z
e
r
s
.
"
(
J
o
h
n
s
o
n
&
F
r
a
s
e
r
2
0
1
0
)
.
"
W
e
t
r
i
e
d
t
o
b
u
i
l
d
t
h
e
u
l
t
i
m
a
t
e
A
J
A
X
c
a
l
e
n
d
a
r
a
l
l
a
t
o
n
c
e
.
I
t
t
o
o
k
a
l
o
n
g
t
i
m
e
.
W
e
c
o
u
l
d
h
a
v
e
d
o
n
e
i
t
p
i
e
c
e
b
y
p
i
e
c
e
.
"
(
K
a
n
2
0
0
6
)
.
4
.
"
[
T
]
r
i
e
d
t
o
b
u
i
l
d
a
s
a
l
e
s
e
f
f
o
r
t
t
o
o
e
a
r
l
y
,
w
i
t
h
t
o
o
w
e
a
k
o
f
a
p
r
o
d
u
c
t
a
f
t
e
r
i
n
i
t
i
a
l
f
i
n
a
n
c
i
n
g
[
…
]
.
"
(
J
o
h
n
s
o
n
&
F
r
a
s
e
r
2
0
1
0
)
.
"
T
o
o
m
u
c
h
m
o
n
e
y
i
s
l
i
k
e
t
o
o
m
u
c
h
t
i
m
e
;
w
o
r
k
e
x
p
a
n
d
s
t
o
f
i
l
l
t
h
e
t
i
m
e
a
l
l
o
t
t
e
d
,
a
n
d
w
a
y
s
t
o
s
p
e
n
d
m
o
n
e
y
m
u
l
t
i
p
l
y
w
h
e
n
a
b
u
n
d
a
n
t
f
i
n
a
n
c
i
a
l
r
e
s
o
u
r
c
e
s
a
r
e
a
v
a
i
l
a
b
l
e
.
B
y
b
e
i
n
g
s
i
m
p
l
y
t
o
o
g
o
o
d
a
t
r
a
i
s
i
n
g
m
o
n
e
y
,
i
t
e
n
a
b
l
e
d
u
s
t
o
p
e
r
p
e
t
u
a
t
e
p
o
o
r
o
r
g
a
n
i
z
a
t
i
o
n
a
l
s
t
r
u
c
t
u
r
e
a
n
d
s
u
b
o
p
t
i
m
a
l
s
t
r
a
t
e
g
i
c
d
e
c
i
s
i
o
n
s
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
5
.
"
W
e
l
a
u
n
c
h
e
d
t
o
o
f
a
s
t
.
I
’
m
a
b
i
g
b
e
l
i
e
v
e
r
i
n
l
a
u
n
c
h
f
a
s
t
,
g
e
t
f
e
e
d
b
a
c
k
,
m
a
k
e
c
h
a
n
g
e
s
.
W
e
l
a
u
n
c
h
e
d
f
a
s
t
,
g
o
t
m
o
r
e
f
e
e
d
b
a
c
k
t
h
a
n
w
e
c
o
u
l
d
h
a
n
-
d
l
e
,
a
n
d
f
a
i
l
e
d
t
o
m
a
k
e
c
h
a
n
g
e
s
.
Y
o
u
c
a
n
’
t
e
x
p
e
c
t
p
e
o
p
l
e
t
o
w
a
i
t
f
o
r
y
o
u
t
o
g
e
t
i
t
r
i
g
h
t
.
"
(
P
o
w
a
z
e
k
2
0
0
8
)
.
"
W
e
t
a
l
k
e
d
a
b
o
u
t
'
r
e
l
e
a
s
e
e
a
r
l
y
/
r
e
l
e
a
s
e
o
f
t
e
n
,
'
b
u
t
w
e
r
e
s
c
a
r
e
d
o
f
l
o
o
k
i
n
g
l
i
k
e
i
d
i
o
t
s
i
n
f
r
o
n
t
o
f
m
a
j
o
r
W
a
l
l
S
t
r
e
e
t
a
n
d
h
e
d
g
e
f
u
n
d
c
l
i
e
n
t
s
.
[
…
]
B
y
c
h
o
o
s
i
n
g
t
o
w
a
i
t
w
e
l
o
s
t
o
u
r
i
n
t
i
m
a
c
y
w
i
t
h
t
h
e
c
u
s
t
o
m
e
r
[
…
]
f
a
l
l
i
n
g
i
n
t
o
t
h
e
c
l
a
s
s
i
c
[
…
]
t
r
a
p
o
f
p
u
r
s
u
i
n
g
a
'
s
c
i
e
n
c
e
p
r
o
j
e
c
t
,
'
n
o
t
b
u
i
l
d
i
n
g
a
c
o
m
m
e
r
c
i
a
l
l
y
s
a
l
a
b
l
e
p
r
o
d
u
c
t
.
"
333 333
3
3
4
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
6
.
"
T
h
a
t
o
l
d
c
l
i
c
h
é
a
b
o
u
t
n
o
t
g
e
t
t
i
n
g
a
s
e
c
o
n
d
c
h
a
n
c
e
t
o
m
a
k
e
a
f
i
r
s
t
i
m
p
r
e
s
s
i
o
n
?
C
o
r
n
y
b
u
t
t
r
u
e
.
W
h
e
n
y
o
u
s
t
u
m
b
l
e
o
u
t
o
f
t
h
e
g
a
t
e
,
i
t
c
a
n
b
e
h
a
r
d
t
o
r
e
g
a
i
n
y
o
u
r
f
o
o
t
i
n
g
.
W
e
s
h
o
u
l
d
h
a
v
e
d
o
n
e
m
o
r
e
t
e
s
t
i
n
g
o
f
b
o
t
h
t
h
e
c
o
r
e
i
d
e
a
a
n
d
t
h
e
s
i
t
e
i
t
s
e
l
f
.
"
(
P
o
w
a
z
e
k
2
0
0
8
)
.
"
A
n
d
t
h
e
s
e
b
a
d
b
e
h
a
v
i
o
r
s
w
e
r
e
r
e
i
n
f
o
r
c
e
d
b
y
a
n
u
n
p
l
a
n
n
e
d
e
v
e
n
t
t
h
a
t
s
h
a
r
p
l
y
i
m
p
a
c
t
e
d
o
u
r
p
s
y
c
h
e
:
b
e
i
n
g
o
n
t
h
e
f
r
o
n
t
p
a
g
e
o
f
t
h
e
F
i
n
a
n
c
i
a
l
T
i
m
e
s
.
I
t
i
s
h
a
r
d
t
o
c
a
l
l
i
t
a
m
i
s
t
a
k
e
s
i
n
c
e
w
e
d
i
d
n
’
t
s
e
e
k
t
o
g
e
t
s
u
c
h
e
x
p
o
s
u
r
e
,
b
u
t
I
p
u
t
i
t
d
o
w
n
a
s
M
i
s
t
a
k
e
#
3
.
T
o
b
e
h
o
n
e
s
t
,
t
h
i
s
s
i
n
g
l
e
f
a
c
t
w
a
s
a
v
e
r
y
m
e
a
n
i
n
g
f
u
l
f
a
c
t
o
r
i
n
o
u
r
f
a
i
l
u
r
e
.
I
t
r
a
i
s
e
d
t
h
e
l
e
v
e
l
o
f
e
x
p
e
c
t
a
t
i
o
n
s
s
o
h
i
g
h
t
h
a
t
i
t
m
a
d
e
u
s
r
e
l
u
c
t
a
n
t
t
o
r
e
l
e
a
s
e
a
n
y
t
h
i
n
g
t
h
a
t
w
a
s
n
’
t
e
a
r
t
h
-
s
h
a
t
t
e
r
i
n
g
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
7
.
"
L
o
o
k
e
r
y
’
s
P
r
o
f
i
l
e
S
a
a
S
/
u
n
i
v
e
r
s
a
l
c
o
o
k
i
e
m
e
c
h
a
n
i
s
m
i
s
f
a
r
m
o
r
e
e
c
o
n
o
m
i
c
a
n
d
e
f
f
e
c
t
i
v
e
t
h
a
n
c
o
o
k
i
e
e
x
c
h
a
n
g
e
s
y
s
t
e
m
s
i
n
a
w
o
r
l
d
w
h
e
r
e
a
d
m
e
d
i
a
a
n
d
t
a
r
g
e
t
i
n
g
d
a
t
a
a
r
e
s
e
p
a
r
a
t
e
c
o
m
m
o
d
i
t
i
e
s
.
T
h
a
t
w
o
r
l
d
i
s
a
y
e
a
r
o
r
m
o
r
e
i
n
t
h
e
f
u
t
u
r
e
.
"
(
R
a
f
e
r
2
0
0
9
)
.
"
I
t
t
o
o
k
u
s
2
4
m
o
n
t
h
s
t
o
f
a
i
l
.
2
y
e
a
r
s
.
T
h
a
t
’
s
a
l
i
f
e
t
i
m
e
i
n
w
e
b
s
t
a
n
d
a
r
d
s
.
W
e
s
h
o
u
l
d
h
a
v
e
l
a
u
n
c
h
e
d
s
o
m
e
t
h
i
n
g
(
e
v
e
n
u
g
l
y
)
i
n
t
h
e
f
i
r
s
t
6
m
o
n
t
h
s
,
a
n
d
p
u
l
l
e
d
t
h
e
p
l
u
g
a
f
t
e
r
1
2
m
o
n
t
h
s
.
D
o
n
’
t
w
a
s
t
e
t
i
m
e
a
n
d
m
o
n
e
y
o
n
a
l
o
n
g
l
a
s
t
i
n
g
p
r
o
j
e
c
t
i
f
y
o
u
d
o
n
’
t
s
e
e
t
h
e
e
n
d
o
f
t
h
e
t
u
n
n
e
l
.
"
(
K
r
e
t
z
2
0
1
0
)
.
8
.
"
[
O
]
u
r
i
n
i
t
i
a
l
a
r
c
h
e
t
y
p
e
c
h
o
i
c
e
b
e
c
a
m
e
t
h
e
b
a
s
i
s
f
o
r
o
u
r
p
r
o
d
u
c
t
.
M
e
a
n
-
w
h
i
l
e
,
m
o
s
t
o
f
t
h
e
g
r
o
w
t
h
i
n
t
h
e
c
a
s
u
a
l
g
a
m
i
n
g
i
n
d
u
s
t
r
y
h
a
s
b
e
e
n
i
n
p
u
z
-
z
l
e
&
w
o
r
d
g
a
m
e
s
,
s
o
w
e
w
e
r
e
p
o
s
i
t
i
o
n
e
d
w
r
o
n
g
f
r
o
m
t
h
e
s
t
a
r
t
.
"
(
T
a
n
g
2
0
0
8
)
.
"
T
h
e
t
e
c
h
n
o
l
o
g
y
d
e
c
i
s
i
o
n
t
o
u
s
e
C
+
+
w
a
s
w
e
l
l
f
o
u
n
d
e
d
.
H
o
w
e
v
e
r
,
i
t
f
a
i
l
e
d
t
o
t
a
k
e
i
n
t
o
a
c
c
o
u
n
t
,
o
r
a
t
l
e
a
s
t
p
r
e
p
a
r
e
f
o
r
t
h
e
i
n
e
v
i
t
a
b
l
e
r
e
a
l
i
t
y
o
f
b
e
i
n
g
u
n
a
b
l
e
t
o
f
i
n
d
p
e
o
p
l
e
t
o
j
o
i
n
m
e
.
A
n
y
m
a
j
o
r
w
e
b
c
o
m
p
a
n
y
r
e
a
c
h
i
n
g
m
a
s
s
i
v
e
s
c
a
l
e
e
v
e
n
t
u
a
l
l
y
t
u
r
n
s
t
o
C
/
C
+
+
f
o
r
p
e
r
f
o
r
m
a
n
c
e
.
G
i
v
e
n
t
h
e
f
a
c
t
t
h
a
t
N
o
u
n
c
e
r
h
a
d
n
o
c
o
m
p
e
t
i
t
i
o
n
a
t
a
l
l
w
h
e
n
t
h
e
i
d
e
a
w
a
s
f
i
r
s
t
c
o
n
c
e
i
v
e
d
[
.
.
.
]
,
I
d
e
c
i
d
e
d
t
o
t
a
k
e
a
l
i
t
t
l
e
b
i
t
l
o
n
g
e
r
b
e
f
o
r
e
l
a
u
n
c
h
i
n
g
t
o
b
u
i
l
d
a
s
o
l
i
d
f
o
u
n
d
a
t
i
o
n
.
T
h
i
s
,
o
f
c
o
u
r
s
e
,
t
u
r
n
e
d
u
p
a
s
a
b
i
g
m
i
s
t
a
k
e
–
b
u
t
i
t
i
s
f
a
r
f
r
o
m
c
l
e
a
r
t
h
a
t
I
w
o
u
l
d
h
a
v
e
b
e
e
n
f
i
r
s
t
t
o
m
a
r
k
e
t
e
i
t
h
e
r
w
a
y
.
"
(
H
a
m
m
e
r
2
0
0
8
)
.
9
.
"
A
s
a
p
i
o
n
e
e
r
i
n
t
h
e
u
s
e
r
-
g
e
n
e
r
a
t
e
d
,
h
y
p
e
r
-
l
o
c
a
l
f
i
e
l
d
,
B
a
c
k
f
e
n
c
e
h
o
p
e
f
u
l
l
y
w
i
l
l
p
a
v
e
t
h
e
w
a
y
f
o
r
m
a
n
y
o
t
h
e
r
e
f
f
o
r
t
s
t
o
c
r
e
a
t
e
l
o
c
a
l
l
y
f
o
c
u
s
e
d
o
n
l
i
n
e
c
o
m
m
u
n
i
t
i
e
s
t
h
a
t
u
l
t
i
m
a
t
e
l
y
w
i
l
l
b
e
c
o
m
e
p
r
o
f
i
t
a
b
l
e
b
u
s
i
n
e
s
s
e
s
.
"
(
P
o
t
t
s
2
0
0
7
)
.
"
[
T
]
h
e
s
p
a
m
i
s
s
u
e
[
…
]
p
r
o
m
p
t
e
d
u
s
t
o
n
o
t
p
u
s
h
v
e
r
s
i
o
n
1
.
0
,
i
n
s
t
e
a
d
w
a
n
t
i
n
g
t
o
s
e
e
i
f
w
e
c
o
u
l
d
c
o
m
e
u
p
w
i
t
h
a
m
o
r
e
p
o
w
e
r
f
u
l
r
e
l
e
a
s
e
u
s
i
n
g
N
L
P
t
o
m
a
r
k
t
h
e
k
i
c
k
-
o
f
f
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
i
s
w
a
s
a
b
i
g
m
i
s
t
a
k
e
[
…
]
W
e
s
h
o
u
l
d
h
a
v
e
g
o
t
t
e
n
i
t
o
u
t
t
h
e
r
e
,
b
e
e
n
k
i
c
k
e
d
i
n
t
h
e
h
e
a
d
b
y
t
o
u
g
h
c
u
s
t
o
m
e
r
s
,
a
n
d
i
t
e
r
a
t
e
d
l
i
k
e
c
r
a
z
y
t
o
a
d
d
r
e
s
s
t
h
e
i
r
n
e
e
d
s
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
1
0
.
"
A
t
s
o
m
e
p
o
i
n
t
,
w
e
r
e
a
c
h
e
d
a
n
a
l
m
o
s
t
g
o
-
l
i
v
e
p
r
o
d
u
c
t
.
B
u
t
a
f
t
e
r
s
h
o
w
i
n
g
i
t
t
o
a
f
e
w
V
C
s
,
w
e
r
e
a
l
i
z
e
d
w
e
m
i
s
s
e
d
s
o
m
e
f
e
a
t
u
r
e
s
t
o
m
a
k
e
i
t
t
r
u
l
y
s
o
c
i
a
l
.
I
n
s
t
e
a
d
o
f
r
e
l
e
a
s
i
n
g
t
h
e
b
e
t
a
,
w
e
d
e
c
i
d
e
d
t
o
c
r
e
a
t
e
t
h
e
v
2
o
f
t
h
e
p
r
o
d
u
c
t
.
W
r
o
n
g
.
W
e
w
i
l
l
n
e
v
e
r
f
i
n
i
s
h
t
h
i
s
v
2
(
l
a
c
k
o
f
m
o
n
e
y
,
s
l
o
w
l
y
d
y
i
n
g
m
o
t
i
v
a
t
i
o
n
)
.
"
(
K
r
e
t
z
2
0
1
0
)
.
1
1
.
"
[
T
]
h
e
c
o
s
t
o
f
b
u
i
l
d
i
n
g
s
c
a
l
a
b
i
l
i
t
y
a
h
e
a
d
o
f
t
i
m
e
i
s
e
x
t
r
e
m
e
l
y
h
i
g
h
,
a
n
d
f
o
r
m
o
s
t
s
t
a
r
t
u
p
i
s
c
o
s
t
p
r
o
h
i
b
i
t
i
v
e
.
T
h
e
d
e
c
i
s
i
o
n
t
o
t
u
r
n
N
o
u
n
c
e
r
f
r
o
m
a
c
o
n
s
u
m
e
r
s
e
r
v
i
c
e
t
o
a
n
i
n
f
r
a
s
t
r
u
c
t
u
r
e
p
r
o
v
i
d
e
r
f
o
r
c
e
d
a
s
i
g
n
i
f
i
c
a
n
t
e
a
r
l
y
i
n
v
e
s
t
m
e
n
t
t
h
a
t
h
a
d
v
e
r
y
l
i
t
t
l
e
c
h
a
n
g
e
o
f
m
a
t
e
r
i
a
l
i
z
i
n
g
.
"
(
H
a
m
m
e
r
2
0
0
8
)
.
334 334
3
3
5
A
P
P
E
N
D
I
X
4
S
T
R
A
U
S
S
I
A
N
E
V
A
L
U
A
T
I
O
N
O
F
C
R
E
D
I
B
I
L
I
T
Y
C
r
i
t
e
r
i
a
(
C
o
r
b
i
n
&
S
t
r
a
u
s
s
1
9
9
0
)
S
e
l
f
-
e
v
a
l
u
a
t
i
o
n
R
e
s
e
a
r
c
h
p
r
o
c
e
s
s
C
1
:
H
o
w
w
a
s
t
h
e
o
r
i
g
i
n
a
l
s
a
m
p
l
e
s
e
l
e
c
t
e
d
?
T
h
e
s
a
m
p
l
e
w
a
s
s
e
l
e
c
t
e
d
b
a
s
e
d
o
n
i
t
s
d
e
g
r
e
e
o
f
i
n
t
e
r
e
s
t
;
p
o
s
t
-
m
o
r
t
e
m
s
t
o
r
i
e
s
s
e
e
m
e
d
l
i
k
e
a
r
i
c
h
s
o
u
r
c
e
o
f
i
n
f
o
r
m
a
t
i
o
n
.
C
2
:
W
h
a
t
m
a
j
o
r
c
a
t
e
g
o
r
i
e
s
e
m
e
r
g
e
d
?
I
n
i
t
i
a
l
l
y
,
t
h
e
r
e
s
e
a
r
c
h
w
a
s
g
u
i
d
e
d
b
y
t
h
e
i
n
t
e
n
t
t
o
i
d
e
n
t
i
f
y
f
a
i
l
u
r
e
f
a
c
t
o
r
s
.
E
m
e
r
g
i
n
g
c
a
t
e
g
o
r
i
e
s
f
o
l
l
o
w
e
d
t
h
e
f
u
n
c
t
i
o
n
s
w
i
t
h
w
h
i
c
h
t
h
e
s
e
f
a
c
t
o
r
s
w
e
r
e
a
s
s
o
c
i
a
t
e
d
;
f
o
r
e
x
a
m
p
l
e
,
M
a
r
k
e
t
i
n
g
,
P
r
o
d
u
c
t
,
a
n
d
F
i
n
a
n
c
e
.
C
3
:
W
h
a
t
w
e
r
e
s
o
m
e
o
f
t
h
e
e
v
e
n
t
s
,
i
n
c
i
d
e
n
t
s
,
a
c
t
i
o
n
s
,
a
n
d
s
o
o
n
t
h
a
t
,
a
s
i
n
d
i
c
a
t
o
r
s
,
p
o
i
n
t
e
d
t
o
s
o
m
e
o
f
t
h
e
s
e
m
a
j
o
r
c
a
t
e
g
o
-
r
i
e
s
?
T
h
e
f
o
u
n
d
e
r
s
’
s
e
n
s
e
-
m
a
k
i
n
g
i
m
p
l
i
e
d
t
h
e
r
e
a
r
e
m
a
n
y
r
e
a
s
o
n
s
f
o
r
f
a
i
l
u
r
e
;
t
h
e
g
r
o
u
p
i
n
g
s
e
e
m
e
d
t
o
n
a
t
u
r
a
l
l
y
e
v
o
l
v
e
i
n
t
o
t
h
e
s
e
f
u
n
c
t
i
o
n
a
l
a
r
e
a
s
.
C
4
:
O
n
t
h
e
b
a
s
i
s
o
f
w
h
i
c
h
c
a
t
e
g
o
r
i
e
s
d
i
d
t
h
e
o
r
e
t
i
c
a
l
s
a
m
p
l
i
n
g
p
r
o
c
e
e
d
?
T
h
a
t
i
s
,
h
o
w
d
i
d
t
h
e
o
r
e
t
i
c
a
l
f
o
r
m
u
l
a
t
i
o
n
s
g
u
i
d
e
s
o
m
e
o
f
t
h
e
d
a
t
a
c
o
l
l
e
c
t
i
o
n
?
T
h
i
s
w
a
s
b
a
s
e
d
o
n
t
h
e
c
o
r
e
c
a
t
e
g
o
r
y
,
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
.
T
h
e
y
w
e
r
e
p
r
e
s
e
n
t
e
d
t
o
d
i
f
f
e
r
e
n
t
f
o
u
n
d
e
r
s
o
n
v
a
r
i
o
u
s
o
c
c
a
s
i
o
n
s
.
T
h
e
s
e
w
e
r
e
n
o
t
i
n
t
e
r
v
i
e
w
s
,
b
u
t
c
a
s
u
a
l
s
t
a
r
t
u
p
e
v
e
n
t
s
.
A
f
t
e
r
t
h
e
t
h
e
o
r
e
t
i
c
a
l
s
a
m
p
l
i
n
g
w
a
s
d
o
n
e
,
h
o
w
r
e
p
r
e
s
e
n
t
a
t
i
v
e
d
i
d
t
h
e
s
e
c
a
t
e
g
o
r
i
e
s
p
r
o
v
e
t
o
b
e
?
T
h
e
r
e
w
e
r
e
n
o
i
n
d
i
c
e
s
o
f
n
e
w
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
p
e
r
t
a
i
n
i
n
g
t
o
t
h
e
s
p
e
c
i
f
i
c
t
y
p
e
o
f
f
i
r
m
s
i
n
w
h
i
c
h
t
h
i
s
s
t
u
d
y
w
a
s
i
n
t
e
r
e
s
t
e
d
.
C
5
:
W
h
a
t
w
e
r
e
s
o
m
e
o
f
t
h
e
h
y
p
o
t
h
e
s
e
s
p
e
r
t
a
i
n
i
n
g
t
o
c
o
n
c
e
p
-
t
u
a
l
r
e
l
a
t
i
o
n
s
,
t
h
a
t
i
s
,
a
m
o
n
g
c
a
t
e
g
o
r
i
e
s
,
a
n
d
o
n
w
h
a
t
g
r
o
u
n
d
s
w
e
r
e
t
h
e
y
f
o
r
m
u
l
a
t
e
d
a
n
d
t
e
s
t
e
d
?
F
o
r
e
x
a
m
p
l
e
,
t
h
e
r
e
l
a
t
i
o
n
s
h
i
p
b
e
t
w
e
e
n
t
h
e
c
o
l
d
s
t
a
r
t
a
n
d
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
s
.
A
f
t
e
r
a
x
i
a
l
c
o
d
i
n
g
,
i
t
s
e
e
m
e
d
c
l
e
a
r
t
h
a
t
s
t
a
r
t
u
p
s
w
h
i
c
h
t
r
y
t
o
c
o
m
p
e
n
s
a
t
e
a
l
a
c
k
o
f
u
s
e
r
s
b
y
f
r
e
e
p
r
o
d
u
c
t
n
o
t
o
n
l
y
f
a
i
l
t
o
s
o
l
v
e
i
t
,
b
u
t
r
u
n
i
n
t
o
a
d
d
i
t
i
o
n
a
l
t
r
o
u
b
l
e
.
T
h
r
o
u
g
h
c
o
m
p
a
r
i
s
o
n
w
i
t
h
o
t
h
e
r
c
a
s
e
s
,
t
h
i
s
s
e
e
m
e
d
t
o
b
e
a
p
a
t
t
e
r
n
i
n
t
h
e
m
a
t
e
r
i
a
l
.
C
6
:
W
e
r
e
t
h
e
r
e
i
n
s
t
a
n
c
e
s
w
h
e
n
h
y
p
o
t
h
e
s
e
s
d
i
d
n
o
t
h
o
l
d
u
p
a
g
a
i
n
s
t
w
h
a
t
w
a
s
a
c
t
u
a
l
l
y
s
e
e
n
?
H
o
w
w
e
r
e
t
h
e
s
e
d
i
s
c
r
e
p
a
n
-
c
i
e
s
a
c
c
o
u
n
t
e
d
f
o
r
?
H
o
w
d
i
d
t
h
e
y
a
f
f
e
c
t
t
h
e
h
y
p
o
t
h
e
s
e
s
?
N
o
n
e
c
o
m
e
t
o
m
i
n
d
;
t
h
i
s
a
s
p
e
c
t
o
f
a
n
a
l
y
s
i
s
c
o
u
l
d
a
r
g
u
a
b
l
y
h
a
v
e
b
e
e
n
i
m
p
r
o
v
e
d
b
y
m
o
r
e
a
c
-
t
i
v
e
l
y
d
e
v
e
l
o
p
i
n
g
h
y
p
o
t
h
e
s
e
s
.
C
7
:
H
o
w
a
n
d
w
h
y
w
a
s
t
h
e
c
o
r
e
c
a
t
e
g
o
r
y
s
e
l
e
c
t
e
d
?
W
a
s
t
h
i
s
s
e
l
e
c
t
i
o
n
s
u
d
d
e
n
o
r
g
r
a
d
u
a
l
,
d
i
f
f
i
c
u
l
t
o
r
e
a
s
y
?
O
n
w
h
a
t
g
r
o
u
n
d
s
w
e
r
e
t
h
e
f
i
n
a
l
a
n
a
l
y
t
i
c
a
l
d
e
c
i
s
i
o
n
s
m
a
d
e
?
T
h
e
s
e
l
e
c
t
i
o
n
w
a
s
g
r
a
d
u
a
l
,
b
a
s
e
d
o
n
s
e
v
e
r
a
l
c
o
d
i
n
g
p
h
a
s
e
s
:
f
a
i
l
u
r
e
à
f
a
i
l
u
r
e
f
a
c
t
o
r
s
à
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
a
n
d
b
i
a
s
e
s
à
d
i
l
e
m
m
a
s
.
O
u
t
o
f
a
l
l
t
h
e
c
h
o
i
c
e
s
,
d
i
l
e
m
m
a
s
s
e
e
m
e
d
t
o
m
o
s
t
i
n
f
l
u
e
n
c
e
f
a
i
l
u
r
e
.
T
h
e
s
e
l
e
c
t
i
o
n
w
a
s
n
o
t
e
a
s
y
o
r
d
i
f
f
i
c
u
l
t
,
i
t
f
e
l
t
n
a
t
u
r
a
l
a
f
t
e
r
r
e
a
d
i
n
g
t
h
e
m
a
t
e
r
i
a
l
m
a
n
y
t
i
m
e
s
.
T
h
e
c
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
b
y
f
o
u
n
d
e
r
s
a
l
s
o
p
o
i
n
t
e
d
t
o
t
h
i
s
d
i
r
e
c
t
i
o
n
.
335 335
3
3
6
E
m
p
i
r
i
c
a
l
g
r
o
u
n
d
i
n
g
o
f
f
i
n
d
i
n
g
s
C
1
:
A
r
e
c
o
n
c
e
p
t
s
g
e
n
e
r
a
t
e
d
?
T
h
e
s
t
u
d
y
g
e
n
e
r
a
t
e
s
n
e
w
c
o
n
s
t
r
u
c
t
s
,
s
u
c
h
a
s
t
h
e
c
o
l
d
s
t
a
r
t
d
i
l
e
m
m
a
,
t
h
e
l
o
n
e
l
y
u
s
e
r
d
i
l
e
m
m
a
,
r
e
m
o
r
a
’
s
c
u
r
s
e
,
a
n
d
t
h
e
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
.
N
e
w
c
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
s
a
l
s
o
r
e
l
a
t
e
t
o
o
n
l
i
n
e
p
l
a
t
f
o
r
m
t
y
p
e
s
a
n
d
t
h
e
‘
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
’
.
T
h
e
t
e
r
m
i
n
o
l
o
g
y
i
s
p
a
r
t
i
a
l
l
y
d
e
r
i
v
e
d
f
r
o
m
t
h
e
f
o
u
n
d
e
r
s
a
n
d
i
n
d
u
s
t
r
y
,
t
h
e
r
e
f
o
r
e
b
e
i
n
g
g
r
o
u
n
d
e
d
.
C
2
:
A
r
e
t
h
e
c
o
n
c
e
p
t
s
s
y
s
t
e
m
a
t
i
c
a
l
l
y
r
e
l
a
t
e
d
?
T
h
e
c
o
n
c
e
p
t
s
a
r
e
r
e
l
a
t
e
d
a
s
a
n
t
e
c
e
d
e
n
t
s
(
e
.
g
.
,
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
)
,
o
u
t
c
o
m
e
s
(
e
.
g
.
,
c
o
l
d
s
t
a
r
t
d
i
l
e
m
m
a
)
a
n
d
s
o
l
u
t
i
o
n
s
(
e
.
g
.
,
f
r
e
e
f
y
i
n
g
)
.
T
h
e
r
e
h
a
s
b
e
e
n
a
n
a
t
t
e
m
p
t
t
o
l
i
n
k
a
n
d
e
x
-
a
m
i
n
e
t
h
e
m
a
s
a
w
h
o
l
e
.
D
u
r
i
n
g
t
h
e
c
o
d
i
n
g
p
r
o
c
e
s
s
,
t
h
e
a
u
t
h
o
r
a
i
m
e
d
t
o
c
o
d
e
r
e
l
a
t
i
o
n
s
h
i
p
s
s
y
s
-
t
e
m
a
t
i
c
a
l
l
y
w
i
t
h
t
h
e
h
e
l
p
o
f
q
u
a
l
i
t
a
t
i
v
e
a
n
a
l
y
s
i
s
s
o
f
t
w
a
r
e
(
i
.
e
.
,
N
V
i
v
o
)
.
C
3
:
A
r
e
t
h
e
r
e
m
a
n
y
c
o
n
c
e
p
t
u
a
l
l
i
n
k
a
g
e
s
,
a
n
d
a
r
e
t
h
e
c
a
t
e
g
o
-
r
i
e
s
w
e
l
l
d
e
v
e
l
o
p
e
d
?
D
o
t
h
e
y
h
a
v
e
c
o
n
c
e
p
t
u
a
l
d
e
n
s
i
t
y
?
T
h
e
l
i
n
k
a
g
e
s
a
r
e
f
o
r
m
e
d
b
e
t
w
e
e
n
d
i
l
e
m
m
a
s
a
n
d
t
h
e
i
r
s
o
l
u
t
i
o
n
s
.
T
h
e
r
e
i
s
a
l
s
o
w
o
r
k
d
o
n
e
t
o
l
i
n
k
d
i
l
e
m
m
a
s
w
i
t
h
f
o
u
n
d
e
r
s
’
b
i
a
s
e
s
,
b
u
t
t
h
a
t
i
s
o
m
i
t
t
e
d
f
r
o
m
t
h
i
s
r
e
p
o
r
t
a
s
t
h
e
r
e
s
e
a
r
c
h
e
r
d
e
c
i
d
e
d
g
o
f
o
r
a
d
e
e
p
e
r
a
n
a
l
y
s
i
s
o
f
d
i
l
e
m
m
a
s
i
n
e
x
c
h
a
n
g
e
f
o
r
r
a
n
g
e
,
o
r
e
x
p
l
a
n
a
t
o
r
y
p
o
w
e
r
.
M
o
r
e
o
v
e
r
,
t
h
e
r
e
p
o
r
t
i
n
c
l
u
d
e
s
s
e
v
e
r
a
l
l
i
n
k
a
g
e
s
f
r
o
m
t
h
e
a
n
a
l
y
s
i
s
o
f
d
i
l
e
m
m
a
s
t
o
t
h
e
i
r
o
n
l
i
n
e
s
p
e
c
i
f
i
c
i
t
i
e
s
,
n
a
m
e
l
y
t
y
p
o
l
o
g
y
a
n
d
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
.
T
h
e
s
e
f
o
r
t
i
f
y
t
h
e
s
u
b
s
t
a
n
t
i
v
e
n
a
t
u
r
e
o
f
t
h
e
t
h
e
o
r
e
t
i
c
a
l
o
f
f
e
r
i
n
g
.
F
i
n
a
l
l
y
,
a
p
a
r
a
d
i
g
m
m
o
d
e
l
w
a
s
n
o
t
f
o
l
l
o
w
e
d
;
t
h
u
s
,
i
n
t
h
e
S
t
r
a
u
s
s
i
a
n
s
e
n
s
e
,
a
d
e
n
s
i
t
y
o
f
c
a
t
e
g
o
r
i
e
s
m
i
g
h
t
b
e
l
a
c
k
i
n
g
.
C
4
:
I
s
t
h
e
r
e
m
u
c
h
v
a
r
i
a
t
i
o
n
b
u
i
l
t
i
n
t
o
t
h
e
t
h
e
o
r
y
?
V
a
r
i
a
t
i
o
n
i
n
t
h
e
d
a
t
a
w
a
s
c
o
n
s
i
d
e
r
a
b
l
e
d
u
e
t
o
t
h
e
v
a
r
i
e
t
y
o
f
p
r
o
b
l
e
m
s
r
e
p
o
r
t
e
d
b
y
t
h
e
s
t
a
r
t
u
p
s
.
G
i
v
e
n
t
h
e
r
e
l
a
t
i
v
e
l
y
f
e
w
n
u
m
b
e
r
o
f
d
i
l
e
m
m
a
s
,
t
h
e
a
u
t
h
o
r
f
e
e
l
s
h
e
c
h
o
s
e
t
h
o
s
e
t
h
a
t
e
x
p
l
a
i
n
m
o
s
t
o
f
t
h
e
p
r
o
b
l
e
m
s
i
n
t
e
r
m
s
o
f
r
e
l
e
v
a
n
c
e
f
o
r
t
h
e
p
l
a
t
f
o
r
m
m
o
d
e
l
,
c
r
i
t
i
c
a
l
i
t
y
i
n
t
e
r
m
s
o
f
f
a
i
l
-
u
r
e
o
u
t
c
o
m
e
,
a
n
d
p
a
t
t
e
r
n
s
e
m
e
r
g
i
n
g
f
r
o
m
t
h
e
d
a
t
a
.
T
h
u
s
,
t
h
e
a
u
t
h
o
r
f
e
e
l
s
t
h
e
e
n
d
r
e
s
u
l
t
e
s
t
a
b
-
l
i
s
h
e
s
p
a
r
s
i
m
o
n
y
a
n
d
s
c
o
p
e
.
H
o
w
e
v
e
r
,
d
u
e
t
o
e
x
c
l
u
s
i
o
n
o
f
n
o
n
-
p
l
a
t
f
o
r
m
-
s
p
e
c
i
f
i
c
s
t
a
r
t
u
p
d
i
-
l
e
m
m
a
s
a
n
d
f
o
u
n
d
e
r
s
’
b
i
a
s
e
s
,
t
h
e
t
h
e
o
r
y
h
a
s
l
e
s
s
d
i
m
e
n
s
i
o
n
s
a
n
d
e
x
p
l
a
n
a
t
o
r
y
p
o
w
e
r
t
h
a
n
i
t
w
o
u
l
d
h
a
v
e
i
f
t
h
e
s
e
c
a
t
e
g
o
r
i
e
s
w
e
r
e
i
n
c
l
u
d
e
d
.
T
h
e
a
u
t
h
o
r
s
t
i
l
l
b
e
l
i
e
v
e
s
t
h
e
i
r
e
x
c
l
u
s
i
o
n
w
a
s
a
c
o
r
r
e
c
t
c
h
o
i
c
e
,
g
i
v
e
n
t
h
a
t
1
)
t
h
e
e
n
s
u
i
n
g
c
o
m
p
l
e
x
i
t
y
w
o
u
l
d
h
a
v
e
e
x
t
e
n
d
e
d
b
e
y
o
n
d
t
h
e
s
c
o
p
e
o
f
o
n
e
s
t
u
d
y
;
a
n
d
2
)
f
o
r
t
h
e
o
r
e
t
i
c
a
l
i
n
t
e
g
r
a
t
i
o
n
,
i
t
h
a
d
b
e
e
n
n
e
c
e
s
s
a
r
y
t
o
e
x
a
m
i
n
e
t
o
o
m
a
n
y
e
x
t
a
n
t
t
h
e
o
r
i
e
s
,
l
e
a
v
i
n
g
t
h
e
w
o
r
k
’
s
t
h
e
o
r
e
t
i
c
a
l
i
n
t
e
g
r
a
t
i
o
n
f
r
a
g
m
e
n
t
e
d
.
T
h
e
s
e
c
a
t
e
g
o
r
i
e
s
a
r
e
b
e
t
t
e
r
l
e
f
t
f
o
r
f
u
r
t
h
e
r
s
t
u
d
i
e
s
.
C
5
:
A
r
e
t
h
e
b
r
o
a
d
e
r
c
o
n
d
i
t
i
o
n
s
t
h
a
t
a
f
f
e
c
t
t
h
e
p
h
e
n
o
m
e
n
o
n
u
n
d
e
r
s
t
u
d
y
b
u
i
l
t
i
n
t
o
i
t
s
e
x
p
l
a
n
a
t
i
o
n
?
A
n
a
t
t
e
m
p
t
h
a
s
b
e
e
n
m
a
d
e
t
o
e
n
a
b
l
e
t
h
e
r
e
a
d
e
r
t
o
u
n
d
e
r
s
t
a
n
d
h
o
w
W
e
b
2
.
0
s
t
a
r
t
u
p
s
,
i
n
w
h
i
c
h
t
h
i
s
s
t
u
d
y
i
s
i
n
t
e
r
e
s
t
e
d
,
d
i
f
f
e
r
f
r
o
m
e
a
r
l
i
e
r
d
o
t
c
o
m
c
o
m
p
a
n
i
e
s
.
M
o
r
e
o
v
e
r
,
t
h
e
g
e
n
e
r
a
l
r
o
l
e
o
f
c
o
m
p
e
t
i
t
i
o
n
i
s
s
p
e
c
u
l
a
t
e
d
s
o
t
h
a
t
i
t
i
s
s
e
e
n
a
s
a
p
o
t
e
n
t
i
a
l
r
e
a
s
o
n
f
o
r
p
l
a
t
f
o
r
m
s
t
a
r
t
u
p
s
o
n
t
h
e
I
n
t
e
r
n
e
t
a
p
p
l
y
i
n
g
i
n
d
i
r
e
c
t
m
o
n
e
t
i
z
a
t
i
o
n
(
i
.
e
.
,
o
f
f
e
r
f
r
e
e
p
r
o
d
u
c
t
s
)
.
T
h
e
s
e
c
o
n
c
l
u
s
i
o
n
s
a
r
e
n
o
t
d
e
f
i
n
i
t
e
,
a
l
t
h
o
u
g
h
t
h
e
y
s
h
o
w
a
n
a
t
t
e
m
p
t
t
o
u
n
d
e
r
s
t
a
n
d
t
h
e
b
i
g
g
e
r
p
i
c
t
u
r
e
.
336 336
3
3
7
C
6
:
H
a
s
“
p
r
o
c
e
s
s
”
b
e
e
n
t
a
k
e
n
i
n
t
o
a
c
c
o
u
n
t
?
P
r
o
c
e
s
s
h
a
s
b
e
e
n
i
m
p
l
i
c
i
t
l
y
t
a
k
e
n
i
n
t
o
a
c
c
o
u
n
t
s
o
t
h
a
t
,
g
i
v
e
n
t
h
e
i
r
t
y
p
e
a
n
d
s
p
e
c
i
f
i
c
i
t
i
e
s
(
i
.
e
.
,
u
s
e
r
g
e
n
e
r
a
t
i
o
n
a
n
d
i
n
d
i
r
e
c
t
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
s
)
a
r
e
s
e
e
n
a
s
m
e
c
h
a
n
i
s
m
s
a
f
f
e
c
t
i
n
g
p
l
a
t
-
f
o
r
m
s
i
n
t
h
e
o
n
l
i
n
e
c
o
n
t
e
x
t
,
l
e
a
d
i
n
g
t
o
t
h
e
o
u
t
c
o
m
e
o
f
f
a
i
l
u
r
e
.
C
7
:
D
o
t
h
e
t
h
e
o
r
e
t
i
c
a
l
f
i
n
d
i
n
g
s
s
e
e
m
s
i
g
n
i
f
i
c
a
n
t
a
n
d
t
o
w
h
a
t
e
x
t
e
n
t
?
T
h
e
o
r
e
t
i
c
a
l
f
i
n
d
i
n
g
s
o
f
t
h
i
s
s
t
u
d
y
a
r
e
l
e
s
s
s
i
g
n
i
f
i
c
a
n
t
t
h
a
n
i
t
s
i
n
t
e
n
d
e
d
m
a
n
a
g
e
r
i
a
l
u
s
e
f
u
l
n
e
s
s
,
p
a
r
t
i
a
l
l
y
d
u
e
t
o
i
t
s
s
u
b
s
t
a
n
t
i
v
e
n
a
t
u
r
e
.
H
o
w
e
v
e
r
,
i
t
d
o
e
s
e
x
p
a
n
d
t
h
e
c
o
n
c
e
p
t
o
f
t
h
e
-
a
n
d
-
e
g
g
p
r
o
b
l
e
m
i
n
t
h
e
o
n
l
i
n
e
p
l
a
t
f
o
r
m
c
o
n
t
e
x
t
;
t
h
i
s
m
i
g
h
t
e
n
c
o
u
r
a
g
e
o
t
h
e
r
t
h
e
o
r
i
s
t
s
t
o
r
e
c
o
n
s
i
d
e
r
t
h
e
n
a
t
u
r
e
o
f
t
h
e
p
r
o
b
l
e
m
i
n
t
h
e
i
r
p
a
r
t
i
c
u
l
a
r
s
e
t
t
i
n
g
s
.
I
t
a
l
s
o
o
f
f
e
r
s
a
t
y
p
o
l
o
g
y
a
n
d
c
r
i
t
e
r
i
a
f
o
r
t
y
-
p
o
l
o
g
y
o
f
p
l
a
t
f
o
r
m
s
t
h
a
t
i
s
m
e
a
n
i
n
g
f
u
l
,
c
o
n
s
i
d
e
r
i
n
g
t
h
e
i
r
p
r
o
b
l
e
m
s
i
n
r
e
l
a
t
i
o
n
t
o
a
d
o
p
t
i
o
n
a
n
d
a
c
t
i
v
e
u
s
e
.
T
h
i
s
s
t
u
d
y
i
s
l
i
m
i
t
e
d
i
n
t
h
a
t
i
t
d
o
e
s
n
o
t
p
r
e
s
e
n
t
a
“
g
e
n
e
r
a
l
t
h
e
o
r
y
o
f
p
l
a
t
f
o
r
m
s
”
;
h
o
w
e
v
e
r
,
s
u
c
h
a
n
a
t
t
e
m
p
t
w
a
s
n
e
v
e
r
i
t
s
p
u
r
p
o
s
e
i
n
t
h
e
f
i
r
s
t
p
l
a
c
e
.
N
e
v
e
r
t
h
e
l
e
s
s
,
t
h
e
s
t
u
d
y
d
o
e
s
o
f
f
e
r
g
o
o
d
g
r
o
u
n
d
s
f
o
r
c
o
n
t
i
n
u
i
n
g
t
h
e
r
e
s
e
a
r
c
h
o
f
o
n
l
i
n
e
p
l
a
t
f
o
r
m
s
.
337 337
THE FOLLOWING PUBLICATIONS HAVE BEEN RELEASED SINCE 2013
IN TURKU SCHOOL OF ECONOMICS PUBLICATION SERIES A
A-1:2013 Hanna Pitkänen
Theorizing formal and informal feedback practices in
management accounting through three dimensions
A-2:2013 Samppa Suoniemi
The impact of CRM system development on CRM acceptance
A-3:2013 Kirsi Lainema
Managerial interaction – Discussion practices in management
meetings
A-4:2013 Sueila Pedrozo
Consumption, youth and new media: The debate on social issues
in Brazil
A-5:2013 Jani Merikivi
Still believing in virtual worlds: A decomposed approach
A-6:2013 Sanna-Mari Renfors
Myyjän toiminnan laatu kuluttajapalvelujen
myyntikohtaamisessa – Ostajan näkökulma myyjän
suoritusarviointiin
A-7:2013 Maria Höyssä
Where science meets its use – Exploring the emergence of
practical relevance of scientific knowledge in the regional
context
A-8:2013 Karri Rantasila
Measuring logistics costs – Designing a general model for
assessing macro logistics costs in a global context with empirical
evidence from the manufacturing and trading industries
A-9:2013 Taina Eriksson
Dynamic capability of value net management in technology-
based international SMEs
A-10:2013 Jarkko Heinonen
Kunnan yritysilmapiirin vaikutus yritystoiminnan kehittymiseen
A-11:2013 Pekka Matomäki
On two-sided controls of a linear diffusion
A-12:2013 Valtteri Kaartemo
Network development process of international new ventures in
internet-enabled markets: Service ecosystems approach
A-13:2013 Emmi Martikainen
Essays on the demand for information goods
A-14:2013 Elina Pelto
Spillover effects of foreign entry on local firms and business
networks in Russia – A Case study on Fazer Bakeries in St.
Petersburg
A-15:2013 Anna-Maija Kohijoki
ONKO KAUPPA KAUKANA?
Päivittäistavarakaupan palvelujen saavutettavuus Turun seudulla
– Ikääntyvien kuluttajien näkökulma
A-1:2014 Kirsi-Mari Kallio
”Ketä kiinnostaa tuottaa tutkintoja ja julkaisuja
liukuhihnaperiaatteella…?”
– Suoritusmittauksen vaikutukset tulosohjattujen yliopistojen
tutkimus- ja opetushenkilökunnan työhön
A-2:2014 Marika Parvinen
Taiteen ja liiketoiminnan välinen jännite ja sen vaikutus
organisaation ohjaukseen – Case-tutkimus taiteellisen
organisaation kokonaisohjauksesta
A-3:2014 Terhi Tevameri
Matriisirakenteen omaksuminen sairaalaorganisaatioissa
– Rakenteeseen päätyminen, organisaatiosuunnittelu ja
toimintalogiikan hyväksyminen
A-4:2014 Tomi Solakivi
The connection between supply chain practices and firm
performance – Evidence from multiple surveys and financial
reporting data
A-5:2014 Salla-Tuulia Siivonen
“Holding all the cards”
The associations between management accounting,
strategy and strategic change
A-6:2014 Sirpa Hänti
Markkinointi arvon muodostamisen prosessina ja sen yhteys
yrittäjyyden mahdollisuusprosessiin
– Tapaustutkimus kuuden yrityksen alkutaipaleelta
A-7:2014 Kimmo Laakso
Management of major accidents
– Communication challenges and solutions in the preparedness
and response phases for both authorities and companies
A-8:2014 Piia Haavisto
Discussion forums
– From idea creation to incremental innovations. Focus on heart-
rate monitors
A-9:2014 Sini Jokiniemi
"Once again I gained so much"
– Understanding the value of business-to-business sales
interactions from an individual viewpoint
A-10:2014 Xiaoyu Xu
Understanding online game players’ post-adoption behavior: an
investigation of social network games in
A-11:2014 Helena Rusanen
Resource access and creation in networks for service innovation
A-12:2014 Joni Salminen
Startup dilemmas - Strategic problems of early-stage platforms
on the Internet
All the publications can be ordered from
KY-Dealing Oy
Rehtorinpellonkatu 3
20500 Turku, Finland
Phone +358-2-333 9422
E-mail: [email protected]
doc_187405272.pdf
Platform startups on the Internet are an under-researched type of business, although some have demonstrated exceptionally high growth and global penetration (e.g., Google, Facebook, and Tinder). As growth-oriented new ventures are generally important for economic development, and given that their failure rate is generally high, the study focuses on a meaningful topic.
Joni Salminen
Sarja/Series A-12:2014
STARTUP DILEMMAS - STRATEGIC PROBLEMS
OF EARLY-STAGE PLATFORMS ON THE INTERNET
Turun kauppakorkeakoulu
Turku School of Economics
Custos: Professor Rami Olkkonen
Turku School of Economics
Supervisors: Professor Rami Olkkonen
Turku School of Economics
Pre-examiners: Professor Daniel Beimborn
Frankfurt School of Finance & Management
Professor Riitta Katila
Stanford University
Opponent: Professor Daniel Beimborn
Frankfurt School of Finance & Management
Copyright Joni Salminen & Turku School of Economics
The originality of this thesis has been checked in accordance with the University of Turku
quality assurance system using the Turnitin OriginalityCheck service.
ISBN 978-952-249-370-5 (print) 978-952-249-371-2 (PDF)
ISSN 0357-4652 (print) 1459-4870 (PDF)
Publications of Turku School of Economics, Series A
Suomen yliopistopaino Oy – Juvenes Print, Turku 2014
SUMMARY
Startup dilemmas - Strategic problems of early-stage platforms on the
Internet
Platform startups on the Internet are an under-researched type of business, alt-
hough some have demonstrated exceptionally high growth and global penetra-
tion (e.g., Google, Facebook, and Tinder). As growth-oriented new ventures
are generally important for economic development, and given that their failure
rate is generally high, the study focuses on a meaningful topic.
Platforms are defined as “places of interaction”, in which technology is em-
ployed to connect different user groups. Based on interaction, four platform
types are defined: 1) content platforms, 2) social platforms, 3) exchange plat-
forms, and 4) infrastructure. Their complements and usage motives vary; how-
ever, they inherit similar problems from the platform model.
The theoretical basis of this work derives from the literature on two-sided
markets. In addition, works analyzing platforms in the field of marketing, in-
formation system sciences, and network economics are included. Key con-
structs, many of which originate from earlier research on networks and stand-
ards, include network effects, critical mass, winner-takes-all, tipping, two-sid-
edness, and the chicken-and-egg problem. To complete the theoretical per-
spective, concepts on online specificities are also included; namely, user-
generated content, indirect monetization, and the freemium business model.
The methodological approach is grounded theory (GT) which is an induc-
tive method emphasizing a close connection of the researcher to the analyzed
material. GT includes coding, that is, raising the abstraction level by concep-
tualizing central themes (i.e., categories); constant comparison of novel find-
ings to those previously found; and theoretical sampling, that is, the systematic
attempt to challenge and extend ideas by collecting more material. A heavy
emphasis in this method is placed on the researcher’s ability to conceptualize
and theorize (i.e., theoretical sensitivity).
Most often, the method is applied to qualitative data, as also is the case in
this study. In total, 29 post-mortem stories written by founders of failed startup
ventures are analyzed. These data, publicly available on the Internet, comprise
the study’s principal material. Additional data include secondary startup inter-
views, interviews of six founders conducted by the author, and discussions at
numerous startup events in Finland, Sweden, and the USA over the four year
research process. Participation comprised discussions with founders in the at-
tempt to verify ideas on the emerging theoretical model, and gain valuable
industry insight.
In the analysis, four dilemmas of platform startups, emerged: 1) cold start
dilemma, 2) lonely user dilemma, 3) monetization dilemma, and 4) remora's
curse. The first two are variants of the chicken-and-egg problem; in a cold
start, there is a lack of content due to the lack of content, so users are unwill-
ing to join the platform. In the lonely user dilemma, there might be registered
users, but none are present at a given time or place; thus, there is no reason to
use the platform. In the monetization dilemma, users are given free access and
usage, but at the expense of revenue. When fees are introduced, the users flee.
In the remora's curse, a startup is able to solve the cold start dilemma by inte-
grating into a larger platform, but becomes vulnerable to its strategic behavior.
Essentially, the problems are interlinked. By solving the chicken-and-egg
problem through subsidization, a commonly applied strategy, a startup moves
toward the monetization dilemma and eventually fails for this reason. It might
also solve the problem by platform integration, or becoming a complementor
for a larger platform. It has been previously argued in the platform literature
that, if successful, the remora can perform envelopment, whereby it perma-
nently captures the host’s users. However, in this study it is argued that the
power dynamics do not favor the startup, which remains vulnerable to the
dominant platform’s opportunism. In this study, “selective integration” (i.e.,
content and value envelopment) is suggested as an alternative. In addition, the
merits and downsides of multihoming and the freemium model are discussed.
The study makes several contributions. First, the scope of the chicken-and-
egg problem, and also its solutions, is expanded to fit the realities of online
startups. This enables more useful approaches than most other studies focusing
on larger firms, exchange platforms, and pricing as a solution. Second, it is
shown that the strategic problems of early-stage platforms are connected,
which should be considered in studies. For practitioners, this implies the
recognition of the “dilemma roadmap” as a tool for strategic planning. Third, a
total of 19 different solutions are analyzed, and the requirements for a “perfect
solution” are characterized. Fourth, novel constructs are introduced for use and
further development by scholars. Finally, several avenues for further research
are put forward; for example, integration of founders’ biases into a theory,
expansion of platform theory, and the pursuit of more solutions.
Although “one size fits all” solutions are unlikely, theoretical analysis of
the solutions, even if complicated by reality, is a useful process to understand
the fundamental forces by which they are affected. Eventually, models can be
expanded to cover more aspects, thus enabling better solutions to emerge from
the cooperation of theory and practice.
Keywords: platforms, two-sided markets, startups, chicken-and-egg problem
TIIVISTELMÄ
Startup-dilemmat - Alkuvaiheen alustojen strategiset ongelmat
Internetissä
Tutkin alkuvaiheen Internet-alustojen strategisia ongelmia. Niitä on tutkittu
suhteellisen vähän, vaikka osa alustoista on saavuttanut poikkeuksellisen no-
pean kasvun (esim. Google, Facebook ja Tinder). Koska kasvuhakuiset yritys-
hankkeet ovat yleisesti ottaen tärkeitä taloudellisen kehityksen ja innovaatioi-
den kannalta ja koska niiden kuolleisuus on yleensä korkea, tutkimuksen aihe
on mielekäs. Aihe ei kuitenkaan kosketa pelkästään yhteiskuntaa, vaan myös
yksittäisiä yrityksiä ja yrittäjiä, jotka kamppailevat näiden strategisten ongel-
mien ja globaalin kilpailun parissa.
Alusta määritellään vuorovaikutusympäristöksi, jossa teknologia yhdistää
eri käyttäjäryhmiä. Käyttömotiivin perusteella työ jakaa Internet-pohjaiset
alustat neljään tyyppiin: 1) sisältöalustat, 2) sosiaaliset alustat, 3) vaihdanta-
alustat ja 4) infrastruktuuri. Vaikka niiden välillä on eroja, em. alustat jakavat
samat alustaliiketoimintamallin ongelmat.
Työn teoreettinen pohja on peräisin kaksipuolisten markkinoiden kirjalli-
suudesta. Lisäksi hyödynnetään markkinoinnin, tietojärjestelmätieteiden ja
verkostotaloustieteen kirjallisuutta. Keskeisiä käsitteitä ovat verkostovaikutuk-
set, kriittinen massa, "voittaja vie kaiken" -ilmiö, kaksi-suuntaisuus, ja muna-
kanaongelma. Em. kirjallisuushaarat ymmärretään tässä työssä alustakirjalli-
suutena. Internet-liiketoiminnan erityispiirteet, kuten käyttäjien tuottama si-
sältö ja epäsuora ansaintalogiikka, ovat myös mukana.
Metodina on grounded theory (GT), jonka soveltamiseen kuuluu koodaus,
eli käsitteellisen tason nostaminen aineiston keskeisiä teemoja nimeämällä ja
arvioimalla, jatkuva vertailu uusien ja edellisten löydösten välillä ja teoreetti-
nen otanta, eli luotujen teoreettisten konstruktioiden (teorian) systemaattinen
täydentäminen ja haastaminen lisämateriaalin avulla. Metodi painottaa vah-
vasti tutkijan kykyä käsitteellistää ja teoretisoida laadullista ja määrällistä ai-
neistoa, ts. teoreettista herkkyyttä. Useimmiten GT-menetelmää sovelletaan
laadulliseen aineistoon, ja niin tässäkin tutkimuksessa. Pääasiallisen aineiston
muodostaa 29 epäonnistumiskertomusta; lisäksi aineistoon kuuluu haastatte-
luja ja keskusteluja useissa startup-tapahtumissa Suomessa, Ruotsissa ja Yh-
dysvalloissa nelivuotisen tutkimusprosessin aikana.
Analyysin yhteydessä päätettiin keskittyä alustastartupeille ominaisiin on-
gelmiin, eli startup-dilemmoihin, erityisesti: 1) kylmän alun ongelma, 2) yksi-
näisen käyttäjän dilemma, 3) monetisointidilemma ja 4) remoran kirous. En-
simmäiset kaksi ovat muunnelmia muna-kanaongelmasta, jota on laajalti kä-
sitelty kirjallisuudessa. Kylmän alun ongelmassa sisällönpuute estää sisältöä
luovien käyttäjien alustaan liittymisen, ja näin yhdelläkään käyttäjällä ei ole
motiivia rekisteröityä. Yksinäinen käyttäjä saattaa sen sijaan olla jo rekiste-
röitynyt, mutta koska vastapuoli ei ole läsnä tietyssä ajassa tai paikassa, alus-
tan käytölle ei ole mahdollisuutta.
Monetisointidilemman mukaan käyttäjät houkutellaan alustaan tarjoamalla
ilmainen pääsy ja käyttö, mutta taloudellisen kannattavuuden kustannuksella.
Heti kun maksut otetaan käyttöön, käyttäjät pakenevat alustalta. Remoran ki-
rouksessa muna-kanaongelma on ratkaistu integroimalla suurempaan alustaan,
mutta vastineeksi joudutaan alttiiksi tämän alustajan omistajan strategiselle
käyttäytymiselle. Näiden dilemmojen analysointi, niiden sovittaminen aiem-
paan kirjallisuuteen sekä yritys löytää mahdollisia ratkaisuja ovat tämän työn
keskeistä antia.
Tutkimus tekee useita kontribuutioita. Ensinnäkin muna-kanaongelma laa-
jennetaan Internet-liiketoiminnan yhteyteen. Useimmat muut tutkimukset kes-
kittyvät suuriin yrityksiin, vaihdanta-alustoihin, ja hinnoitteluun muna-kana-
ongelman ratkaisuna. Toiseksi työ osoittaa, että alkuvaiheen alustojen strategi-
set ongelmat ovat sidoksissa toisiinsa. Yrityksille tämä merkitsee "dilemma-
tiekartan" hyödyntämistä strategisen toiminnan työkaluna, kun taas tutkijoille
se korostaa tarvetta lähestyä alustaongelmia kokonaisvaltaisesti. Myös mah-
dollisia ratkaisuja pohditaan laajalti: yhteensä tarkastellaan 19 eri strategian
soveltuvuutta dilemmojen ratkaisuun. Tutkimus esittää "täydellisen ratkaisun"
ominaispiirteet ja useita lupaavia mahdollisuuksia lisätutkimuksiin, mm. yh-
distämällä perustajien "harhat" osaksi teoriaa, laajentamalla alustateoriaa ja
etsimällä uusia ratkaisuja metodologisen pluralismin avulla.
Tutkimuksen mukaan muna-kanaongelmalla on kytkeytynyt luonne Inter-
net-alustojen yhteydessä; toisin sanoen yhden ongelman ratkaisu johtaa pian
toiseen. Tilannetta pahentaa ns. "kiitorata-efekti", jonka mukaan aloittelevalla
yrityksellä on rajallinen aika saavuttaa onnistumisia ennen sen lopettamista.
Olennainen lähtökohta on, että alusta ei kykene aina yhdistämään sopivia
käyttäjiä tai sisältöjä toistensa kanssa ja siten ratkaisemaan muna-kanaongel-
maa. Lisäksi ongelmaa ei voida ratkaista kerralla ja sitten jatkaa eteenpäin,
vaan yksinäisen käyttäjän dilemma seuraa alustaa niin kauan kuin se on ole-
massa. Ellei yritys tunnista ongelmia ajoissa, havaitse niiden välisiä yhteyksiä
ja sovella oikeita ratkaisuja, epäonnistumisen todennäköisyys kasvaa. Siksi
dilemmojen luonteen ymmärtäminen on ensiarvoisen tärkeää alustaliiketoi-
minnassa menestymisen kannalta.
Asiasanat: alustat, kaksisuuntaiset markkinat, startupit, muna-kanaongelma
ACKNOWLEDGEMENTS
I would like to thank the following persons: Professor Rami Olkkonen for
guiding me through the research process. His formidable attitude to teaching,
life, and academia has greatly inspired me. The pre-examiners, Professor
Daniel Beimborn and Professor Riitta Katila. Professor Aino Halinen-Kaila,
for guiding me through the doctoral studies. My colleagues, Mekhail Mustak
and Lauri Pitkänen, for their continuous support and fun times in the depart-
ment. (You guys will follow shortly!)
The founders who have tried and failed, but keep on trying. The fact they
shared their stories with the world enabled this research in the first place.
Special thanks are owed to Boost Turku Entrepreneurship Society and the
inspiring people I have met there. In addition, the founders I have met along
the way deserve respect for their hard work. Here are just a few of them, in no
particular order: Ajay Garg, Timo Herttua, Rasmus Kevin, Renato Kern, Tiina
J aatinen, Igor Burattini, Tatu Laine, J ose Teixeira, Marcos Tong, Ville Tapani,
Toni Perämäki, Linas Ceikus, J uho Vaiste, Ville Sirkiä, Timo Hänninen, Kari
Vuorinen, J anne Loiske, Ismo Karjalainen, Antero J ärvi, Ville Kaituri, J asu
Koponen, Camilla af Hällström, Antti Lundstedt, Hazzan Ajao, Denis
Duvauchelle, J eremy Boom, Guillaume de Dorlodot, Victor Vulovic, and Ben
Adamson. Thank you all. I hope to stay in touch!
The following have helped with administrative issues: Auli Rahkala-Toivo-
nen, Sanna Kuusjärvi, Riikka Harikkala, and J enni Heervä. These issues are
trickier than research problems, so I appreciate your help. I would also like to
thank Dr. Elina J aakkola, Professor Leila Hurmerinta, Dr. Hannu Makkonen,
Dr. Harri Terho, and Professor Nicole Coviello for their comments in research
seminars at the Turku School of Economics.
I would to express my gratitude to the following foundations for their sup-
port: Liikesivistysrahasto, J enny ja Antti Wihurin rahasto, Leonard Gestrinin
muistorahasto, OP Tukisäätiö, TOP-säätiö, Turun kauppaopetus-säätiö, Mar-
cus Wallenbergin säätiö, and Turun Kauppakorkeakouluseura. Without them,
there would be less knowledge in the world.
Finally, thanks to my family: to my mother Helena Salminen, for always
caring for me. She is still the best mother in the world. To my younger brother
Niko Salminen, for driving me to the airport when leaving for conference
trips, and also for other help. And to my father, Timo Salminen, for teaching
us that education matters. I wish all children could have the same lesson.
Turku, 14
th
September 2014
J oni Salminen
TABLE OF CONTENTS
SUMMARY
TIIVISTELMÄ
ACKNOWLEDGMENTS
1 INTRODUCTION ................................................................................................ 17
1.1 Research background ........................................................................ 17
1.2 Key concepts .................................................................................... 19
1.3 Research gap ..................................................................................... 20
1.4 Purpose and research questions ......................................................... 22
1.5 Positioning ........................................................................................ 26
1.6 Structure ........................................................................................... 31
2 METHODOLOGY................................................................................................ 33
2.1 Research strategy .............................................................................. 33
2.1.1 Introduction to research strategy ............................................ 33
2.1.2 What is GT? .......................................................................... 33
2.1.3 Why was GT selected as research method? ............................ 36
2.2 Research process ............................................................................... 38
2.3 Research data .................................................................................... 40
2.3.1 Data collection ....................................................................... 40
2.3.2 Selection criteria .................................................................... 42
2.3.3 Description of the startups ..................................................... 43
2.4 Analytical approach .......................................................................... 46
2.4.1 Coding process in GT ............................................................ 46
2.4.2 Application of GT in this study .............................................. 48
2.4.3 Coding guide ......................................................................... 51
2.5 Literature approach ........................................................................... 53
3 THEORETICAL BACKGROUND ....................................................................... 57
3.1 Concept of platform .......................................................................... 57
3.1.1 Platform theory and platform literature .................................. 57
3.1.2 Defining platforms ................................................................. 57
3.1.3 Markets vs. platforms ............................................................ 60
3.1.4 Mediation vs. coordination .................................................... 62
3.1.5 Direct and indirect effects of interaction ................................ 64
3.1.6 Networks vs. platforms .......................................................... 67
3.1.7 Websites vs. platforms........................................................... 67
3.2 Platform definition of this study ....................................................... 68
3.3 Typology for online platforms .......................................................... 70
3.4 Online platforms and user generation ............................................... 76
3.4.1 Why is UG included in the study? ......................................... 76
3.4.2 User-generated content .......................................................... 76
3.4.3 UG in online platforms .......................................................... 77
3.4.4 Ideal user-generation model .................................................. 78
3.4.5 Functional view to UG .......................................................... 80
3.4.6 Implications to startups .......................................................... 83
3.4.7 Limitations of UG ................................................................. 84
4 STARTUP DILEMMAS........................................................................................ 87
4.1 Introduction to dilemmas .................................................................. 87
4.1.1 What is meant by dilemmas? ................................................. 87
4.1.2 The use of dilemmas in this study .......................................... 87
4.2 Dilemmas in the platform literature .................................................. 89
4.3 Dilemmas emerging from analysis.................................................... 93
4.3.1 Results from the black box analysis ....................................... 93
4.3.2 Narrowing the focus of the study ........................................... 95
4.3.3 Chosen dilemmas and their treatment .................................... 97
4.4 Cold start dilemma ........................................................................... 99
4.4.1 Definition and exhibits .......................................................... 99
4.4.2 The literature ....................................................................... 105
4.4.3 Solution: Subsidies .............................................................. 111
4.4.4 Discussion ........................................................................... 115
4.5 Lonely user dilemma ...................................................................... 119
4.5.1 Definition and exhibits ........................................................ 119
4.5.2 The literature ....................................................................... 124
4.5.3 Solution: Remora ................................................................ 134
4.5.4 Discussion ........................................................................... 138
4.6 Monetization dilemma .................................................................... 142
4.6.1 Definition and exhibits ........................................................ 142
4.6.2 The literature ....................................................................... 148
4.6.3 Solution: Freemium ............................................................. 158
4.6.4 Discussion ........................................................................... 164
4.7 Remora’s curse ............................................................................... 168
4.7.1 Definition and exhibits ........................................................ 168
4.7.2 The literature ....................................................................... 180
4.7.3 Solution: Diversification ...................................................... 186
4.7.4 Discussion ........................................................................... 190
4.8 Summary and discussion on dilemmas ............................................ 193
5 SOLVING THE DILEMMAS ............................................................................. 201
5.1 Introduction .................................................................................... 201
5.2 Solutions ......................................................................................... 201
5.2.1 Exhibits ............................................................................... 201
5.2.2 Advertising .......................................................................... 203
5.2.3 Aggregation ......................................................................... 205
5.2.4 Community .......................................................................... 205
5.2.5 Exclusivity ........................................................................... 206
5.2.6 Facilitation ........................................................................... 207
5.2.7 Funding ............................................................................... 208
5.2.8 Get big fast .......................................................................... 209
5.2.9 Influencers ........................................................................... 209
5.2.10 Legitimacy .......................................................................... 210
5.2.11 Market-making .................................................................... 210
5.2.12 Marketing skills ................................................................... 213
5.2.13 Open source......................................................................... 214
5.2.14 Partnering ............................................................................ 216
5.2.15 Scarcity ............................................................................... 219
5.2.16 Search-engine marketing ..................................................... 220
5.2.17 Sequential approaches ......................................................... 221
5.2.18 Standalone value ................................................................. 227
5.2.19 Performance-based compensation ........................................ 228
5.2.20 Personal selling ................................................................... 228
5.3 Summary and discussion on solutions ............................................. 230
6 CONCLUSIONS................................................................................................. 237
6.1 Theoretical contribution .................................................................. 237
6.1.1 Addressing research gaps and questions ............................... 237
6.1.2 Expansion of the chicken-and-egg problem ......................... 239
6.1.3 Interrelatedness of platform dilemmas ................................. 240
6.1.4 Strengths and weaknesses of common solutions .................. 241
6.1.5 Conceptual expansion .......................................................... 243
6.1.6 Substantive theory: dilemmas of platform startups ............... 245
6.2 Managerial implications .................................................................. 247
6.2.1 Think, plan, and utilize the roadmap .................................... 247
6.2.2 Avoid the free trap............................................................... 248
6.2.3 Beware of theoretical UG and network effects ..................... 248
6.2.4 Beware of the internalization problem ................................. 250
6.2.5 Concluding advice ............................................................... 251
6.3 Marketing implications ................................................................... 252
6.4 Suggestions for further research ..................................................... 253
6.4.1 Comparison to success ........................................................ 253
6.4.2 More dilemmas.................................................................... 254
6.4.3 Strategic decision-making biases ......................................... 254
6.4.4 More solutions from practice ............................................... 256
6.4.5 Power dynamics .................................................................. 257
6.4.6 Synthesis of marketing and platform theories ...................... 258
6.4.7 Literature integration ........................................................... 259
6.4.8 Introducing other contexts ................................................... 260
6.5 Credibility ...................................................................................... 260
6.5.1 Evaluative criteria ............................................................... 260
6.5.2 Evaluation of credibility ...................................................... 261
6.5.3 Success with theory ............................................................. 264
6.5.4 Saturation ............................................................................ 265
6.5.5 Risks relating to data ........................................................... 267
6.5.6 Risks relating to method ...................................................... 277
6.5.7 Risks relating to researcher .................................................. 278
6.5.8 Generalizability ................................................................... 281
6.5.9 Overall assessment of credibility ......................................... 286
REFERENCES ......................................................................................................... 289
APPENDIX 1 CODING GUIDE ............................................................................ 323
APPENDIX 2 IS THE COLD START DILEMMA REALLY A DILEMMA? ....... 332
APPENDIX 3 SUPPORT FOR EARLY AND LATE LAUNCHES ....................... 333
APPENDIX 4 STRAUSSIAN EVALUATION OF CREDIBILITY ....................... 335
LIST OF FIGURES
Figure 1 Black box of startup failure ............................................................ 24
Figure 2 The platform literature ................................................................... 26
Figure 3 Research process............................................................................ 38
Figure 4 Historical positioning of the analyzed startups ............................... 46
Figure 5 Application of grounded theory ..................................................... 49
Figure 6 Difference between a reseller and a platform ................................. 62
Figure 7 Market coordination and platforms ................................................ 63
Figure 8 Interactions in an advertising-based online platform ...................... 65
Figure 9 Ideal user generation model ........................................................... 79
Figure 10 Exploratory outcomes – opening the black box of failure ............. 93
Figure 11 Strategic actions and their consequences ...................................... 98
Figure 12 Remora and envelopment ........................................................... 137
Figure 13 Weak remora.............................................................................. 175
Figure 14 Strong remora ............................................................................ 176
Figure 15 Dilemmas and associated problems ............................................ 195
Figure 16 Zigzag to a critical mass (Evans 2009a) ..................................... 225
Figure 17 Logic of “keep on trying”........................................................... 251
Figure 18 Spheres of applicability .............................................................. 282
Figure 19 A tentative formal theory ........................................................... 283
LIST OF TABLES
Table 1 Industry examples ........................................................................... 17
Table 2 Descriptions of analyzed startups .................................................... 44
Table 3 Examples from coding guide .......................................................... 52
Table 4 The literature keywords .................................................................. 54
Table 5 Definitions of a platform (i.e., two- or multisided market) .............. 58
Table 6 Types of network effects................................................................. 59
Table 7 Online platform types ..................................................................... 73
Table 8 Online platforms, interaction, and goals .......................................... 75
Table 9 Functional comparison of users and the firm................................... 81
Table 10 Analysis of dilemmas .................................................................... 96
Table 11 Exhibits of cold start dilemma .................................................... 101
Table 12 Too many consumers (of content) ............................................... 102
Table 13 Consumers and generators .......................................................... 103
Table 14 Basic solution of subsidization .................................................... 113
Table 15 Startup platform .......................................................................... 119
Table 16 Incumbent platform (with a critical mass) ................................... 120
Table 17 Exhibits of the lonely user dilemma ............................................ 121
Table 18 Exhibits of the monetization dilemma ......................................... 142
Table 19 Monetization dilemma simplified................................................ 143
Table 20 Price sensitive (both) .................................................................. 144
Table 21 Quality sensitive (A) ................................................................... 145
Table 22 Truth and assumptions ................................................................ 147
Table 23 Strategic pricing (adapted from Chakravorti & Roson 2006) ....... 156
Table 24 Remora’s choice ......................................................................... 170
Table 25 Exhibits of remora’s curse .......................................................... 171
Table 26 Risks of delegation ..................................................................... 174
Table 27 Applicability of dilemmas across platform types ......................... 193
Table 28 Exhibits from post-hoc analysis .................................................. 202
Table 29 Evaluating applicability of solutions ........................................... 231
Table 30 Addressing research gaps ............................................................ 237
Table 31 Answers to research questions .................................................... 238
Table 32 ‘Build it and they will come’....................................................... 255
Table 33 Evaluation of credibility.............................................................. 262
Table 34 Reasons for writing post-mortems............................................... 268
Table 35 Examples of self-attribution ........................................................ 270
Table 36 Examples of different interpretations .......................................... 274
GLOSSARY
Chicken-and-egg problem: the tendency of users not to join a platform if
others are not joining.
Complement (complementary good): an offering such as a service, program,
game, or other type of application provided by a first- or third-party in a plat-
form.
Complementor: a provider of a complement to a given platform.
Content platform: a platform for distributing, consuming, and sharing con-
tent.
Critical mass: the number and quality of members/complements required to
convince others to join a platform.
Cross-side network externality: seeindirect network effect.
Demand side: end users of a platform.
Direct network effect: a network effect applying to users of the same kind
that can be grouped as one based on their motives or interests (e.g., fans of the
same topic).
Envelopment: a strategy for one platform to capture users of another plat-
form.
Exchange platform: a platform connecting buyers and sellers.
Exclusivity: a rule requiring complementors to single-home (seesingle-hom-
ing).
Feedback loop: a positive or negative reinforcing effect according to which
members of a platform follow other members’ strategic actions.
First-party complement: a complement provided by the platform owner to its
platform.
Freemium: an Internet business model combining free and premium offer-
ings.
Indirect network effect: a network effect between different kinds of user that
should not be grouped as one based on their motives or interests (e.g., buyers
and sellers).
Installed user base: the number of users who have adopted a given platform.
Inter-operability: compatibility between platforms.
Inter-platform competition: competition between platforms.
Intra-platform competition: competition between same side members of a
platform.
Liquidity: amount of interaction occurring between members of a platform.
Monetization: converting free services into revenue.
Multihoming: diversifying strategy of platform members (i.e., both demand
and supply side) to commit to more than one platform by utilizing or distrib-
uting offerings.
Network effect: the more members/complements in a platform, the more use-
ful it is for its current and future users.
Platform: a place of interaction connecting two or more complementing
groups.
Single-homing: committing to only one platform (i.e., opposite of multi-
homing)
Social platform: a platform for interacting with other members due to social
motivation.
Standalone value: utility provided by the platform without its complements.
Supply side: complementors of a platform (seecomplementor).
Third-party complement: complements independent of the platform owner.
Two-sidedness: interaction of two groups complementing each other (e.g.,
buyers and sellers in exchange).
User-generated content (UGC): digital content provided by end users of a
platform.
Web 2.0: a term referring to interaction-enhancing features of the Internet.
Winner-takes-all: a competitive case in which the dominant player receives a
disproportionally high share of the total gains provided by a platform.
17
1 INTRODUCTION
1.1 Research background
The development of information technology in the late 1990s resulted in the
hype of e-commerce platforms (i.e., two-sided marketplaces connecting buy-
ers and sellers), only to be quickly followed by their demise (Evans 2009a).
However, the dotcom period spread seeds of a new beginning, and new plat-
form companies such as Google, Facebook, and Twitter quickly dominated
technology users’ everyday lives, and created new job types
1
and business op-
portunities. Many of them also command high returns: Eisenmann, Parker, and
Van Alstyne (2011, 1272) found that “60 of the world’s 100 largest corpora-
tions earn at least half of their revenue from platform markets.” The recent
surge of platforms has been observed in many industries, and businesses have
swiftly adopted the platform strategy and terminology (Hagiu 2009). Concepts
such asnetwork effects andcritical mass capture critical ideas relating to plat-
form-driven change in the business landscape.
The impact of platforms can be seen in the speed and scope of their growth.
Several platform startups have achieved fast growth and resulted in successful
exits for their founding teams (see Table 1).
Table 1 Industry examples
Platform Founded Type Exit*
eBay 1995 exchange platform IPO in 1998
Google 1998 content platform IPO in 2004
MySpace 2003 social platform Sold to News Corp. in 2005
LinkedIn 2003 social platform IPO in 2011
Facebook 2004 social platform IPO in 2012
YouTube 2005 content platform Sold to Google in 2006
Twitter 2006 social platform IPO in 2013
*IPO or trade sale
The success and vigor of platforms make them an interesting topic to study.
As shown by examples in Table 1, it is well established that platform startups
1
Consider a “search-engine marketer”, a profession that did not exist a decade ago.
18
can be immensely successful. However, it is also commonly known that many,
if not the majority, fail. In fact, observations on the hard reality of platform
startups are the inspiration for this study. These observations were made in the
local startup community in Turku, Finland, and in startup-focused online me-
dia, including influential technology and startup-related blogs, such as
TechCrunch
2
by Michael Arrington, AVC
3
by Fred Wilson, Startup Lessons
Learned
4
by Eric Ries, and also startup-oriented discussion forums such as
Quora
5
and Hacker News
6
; all of which include deep and interesting discus-
sions on various startup problems.
Post-mortem stories (i.e., failure narratives) of unsuccessful ventures are of
particular interest as they contain descriptions, written by their founders, of
problems faced by startups. Reading more than a dozen stories of startup
failure raised the author’s interest in the topic of failure, which evolved into
this study. Focusing on failure was deemed important for two reasons: 1) the
early stage of a startup firm is commonly termed the “valley of death” during
which many startups fail, and 2) success stories often include a “survivorship
bias”, as documented in the literature (e.g., Brown, Goetzmann, Ibbotson, &
Ross 1992). This reasoning led the author to believe that a better picture of
business challenges on the Web could be achieved by studying failures instead
of successes. If, indeed, the majority of startups fail, does it not make more
sense to study them rather than relying on data from the few successful ones?
According to this rationale, failures will help us to understand the Web as a
(sometimes hostile) business environment, reasons for why Internet companies
fail, and, ultimately, also some potential explanations for the rare successes.
The first focus of the study was on online business models, with the prem-
ise that the lack of a proper business model leads to a high failure rate. After
an initial inquiry, this assumption was rejected as there were, in fact, a re-
markable number of well-functioning business models, created especially after
the dotcom bust (see Rappa 2013 for a list). Therefore, it was concluded that
the lack of business models probably was not a major explanatory factor. After
reading a few post-mortem stories on the Web, the final research plan began to
take shape. It was clear that failure is a somewhat complex phenomenon and
involves variables at many levels, including commonly cited managerial
shortcomings, lack of marketing, and not solving real customer problems (see
e.g., Sharma & Mahajan 1980). Indeed, these are quite well understood rea-
sons for failure.
2
www.techcrunch.com
3
www.avc.com
4
www.startuplessonslearned.com
5
www.quora.com
6
news.ycombinator.com
19
In contrast, startup-specific business problems had 1) been examined less
extensively in previous research, and 2) not been well solved by practitioners,
as proven by the failure stories disseminated in the startup community. The
study, which initially was to identify critical success factors in different busi-
ness models that could explain why some firms succeed in online business
while others fail, therefore transformed into its current form. In fact, the final
topic of platform dilemmas emerged from the collected material (cf. Glaser &
Strauss 1967) as it became obvious that the majority of startups in the stories
were failed attempts to create a platform-based business. This discovery led to
synthesizing the practical problems into something more: startup dilemmas,
which require deep analysis and creativity to be solved.
Finally, the research conducted here includes a forward approach. Not only
were the problems defined, but also their solutions were actively sought by the
author while conducting research. McCarthy, Plantholt, and Riordan (1981)
wrote in their thesis, “Success versus survival?: the dilemma of high technol-
ogy firms” more than 30 years ago:
“The result of [earlier] studies has been to tell the reader:
This is what happened and,
This is how the companies responded.
We wanted a study that would tell us:
This is what happened and why.
This is how the companies responded, and
This is how the companies should have responded.”
A similar ambition has driven this study, in that it aims to provide useful in-
sight for founders and researchers by enhancing their ability to identify key
strategic problems and their interconnections in platform business.
1.2 Key concepts
The key concepts of this study are defined here. See the glossary for further
terminology relating to online business and platforms.
? Startup: an early-stage business organization. ‘Early-stage’ is defined as
no more than five years of age; ‘organization’ implies that the startup is not
necessarily incorporated; ‘business’ separates non-profit and open-source
projects from commercial ventures; and ‘marketing’ and ‘distribution’ refer to
the acquisition and serving of customers. In particular, a platform startup is a
startup attempting to create a platform-based business.
20
? Dilemma: a contradictory decision-making situation in which all alterna-
tives seemingly lead to an undesirable outcome (see Oxford Dictionary 2013).
In this dissertation, a dilemma is understood as a conceptualized strategic
problem, meaning that it has been given a name and definition. Whereas stra-
tegic problems emerge in idiosyncratic situations, dilemmas are generally
formulated and, thus, more abstract.
? Strategic problem: a decision-making problem of a strategic agent; in this
study, the platform startup. Strategy is defined in this dissertation as a chosen
course of action by a strategic agent that is therefore preceded by decision-
making. In general, any behavior that considers costs and benefits and makes a
choice, given the available information and assumptions, can be considered
strategic (Lyles & Howard 1988). Thus, the pros and cons of strategic prob-
lems are required to be weighed (Schwenk 1984). Lyles and Howard (1985,
131) describe these as “not the everyday, routine problems but the problems
and issues that are unique, important, and frequently ambiguous”. In other
words, strategic problems can be deep and complex.
? Platform: a place of interaction (cf. Evans 2003) that connects two or
more types of actor. Platforms are often associated with network effects (Katz
& Shapiro 1985), critical mass (Rohlfs 1974), and other associated constructs
discussed in Chapter 3. Many scholars interchangeably employ the terms
platform andtwo-sided markets (Rochet & Tirole 2003).
1.3 Research gap
Approximately a decade ago, platforms began to receive the attention of
scholars. Entering the vocabulary of economics, two-sided markets (Rochet &
Tirole 2003) have become the focus of increased research interest in the field
of industrial economics. Related terms such as app marketplaces and ecosys-
tem have gained popularity in other fields (see J ansen & Bloemendal 2013).
Overall, the implications of two-sidedness have spread from economics to
other disciplines such as information systems (Casey & Töyli 2012), strategic
management (Economides & Katsamakas 2006), and marketing (Sawhney,
Verona, & Prandelli 2005).
However, much remains to be discovered, in particular regarding platform
development (Piezunka 2011). These issues are more closely associated with
the platform business model than generic business problems that can be per-
ceived to apply across industries and firm sizes. Problems such as a lack of
marketing, running out of funds, changes in the business environment or
macro-economy, or management errors have been considered in the extant
literature (e.g., Miller 1977; Gaskill, Van Auken, & Manning 1993; Lussier
21
1996; Dimitras, Zanakis, & Zopounidis 1996). Similarly, there are multiple
studies dedicated to challenges faced by new ventures or startups (e.g.,
McCarthy et al. 1981; Zacharakis 1999; Honjo 2000; Azoulay & Shane 2001),
and also in the online context (Han & Noh 1999; Cochran, Darrat, & Elkhal
2006). |The strategic problems concerning platforms are less well-known. In
particular, five gaps exist.
First, not much is known concerning platform-specific business problems
beyond the chicken-and-egg dilemma. Despite some extensions to other stra-
tegic issues (see Chapter 4), the chicken-and-egg problem is perceived as the
fundamental issue in platform business (Evans 2002; Rochet & Tirole 2003).
However, as this study shows, there are other important problems faced by
platform startups.
Second, the perspective taken in the platform literature often neglects the
startup condition, mainly regarding the lack of resources or pricing power.
This tradition can be seen to stem from Farrell and Saloner (1985) and Katz
and Shapiro (1985), often cited by platform scholars, who focus on industry
standards and “monoliths”, not startup firms. For example, Farrell and Saloner
(1986) discuss the “penguin effect” relating to the adoption of standards, im-
plying that none of the players are willing to take the first step. Exceptions in
the more recent literature are Caillaud and J ullien (2003), Evans (2009a),
Evans and Schmalensee (2010), and Mas and Radcliffe (2011) who consider
the chicken-and-egg problem particularly from a startup/entrant perspective.
However, when strategies for solving the chicken-and-egg problem focus on
pricing (e.g., Caillaud & J ullien 2003) and advertising (see Chapter 4), they
might not be effective for startups lacking the means to execute either.
Eisenmann et al. (2011) propose envelopment as a strategy for capturing com-
petitors’ users (see Chapter 4.7). However, more strategies are needed.
Third, most studies relating to the chicken-and-egg problem are theoretical.
Mas and Radcliffe (2011) and Raivio and Luukkainen (2011) are exceptions
as they approach the problem through an empirical case study. Curchod and
Neysen's (2009) working paper is methodologically closest to this study as it
applies grounded theory. Although theoretical and analytical works have a lot
of intuitive appeal, empirical studies can help ground their concepts more
firmly in the reality of platforms.
Fourth, without closer examination on associated problems relating to its
antecedents or arising from its potential solutions, the chicken-and-egg prob-
lem is typically treated as being isolated. Such a narrow focus concerns most
other strategic problems in the platform literature. Strategy scholars such as
Lyles and Howard (1988) discuss interrelatedness of strategic problems. Thus,
a more holistic approach that recognizes the relations of strategic problems
and their solutions can be perceived as necessary.
22
Fifth, strategic solutions considered by the economist-dominated platform
literature are narrow and focus mostly on pricing (Shy 2011). The importance
of pricing in the online environment is negligible as de facto pricing of many
online platforms approaches zero in terms of both access and usage fees
(Teece 2010). Rochet and Tirole (2005), for example, make a case proving
that a platform can offer negative pricing to one market side and remain prof-
itable overall as a consequence of what Evans (2003) terms “internalizing the
externalities” of platform coordination. However, if entry pricing is set at zero
and the platform is still unable to attract users (i.e., solve the chicken-and-egg
problem), what can be done? It seems that answering this question requires an
answer not centered on pricing strategies.
There have also been calls by practitioners for the type of research at which
this study aims (The Entrepreneurial Enlightenment 2012):
"Most founders don’t know what they should be focusing on and
consequently dilute their focus or run in the wrong direction.
They are regularly bombarded with advice that seems contra-
dictory, which is often paralyzing."
Consequently, platforms’ strategic problems can be regarded as having both
empirical and theoretical relevance. In particular, startup founders, managers,
and investors are interested in learning more concerning specific challenges of
platform companies, as such insight can offer competitive advantage in their
respective markets.
1.4 Purpose and research questions
This purpose of this study is to address some of the previously mentioned re-
search gaps with appropriate research questions, and, in so doing, improve the
chances of platform startups to identify and solve central strategic problems
pertaining to the platform business model, thereby also increasing their
chances of survival.
Although platform startups are currently flourishing in the online market
space, every platform must bypass its early stage to become a viable and prof-
itable business. Therein lies the problem, as most startups ultimately fail
(Haltiwanger, J armin, & Miranda 2009; Watson & Everett 1999). To the au-
thor’s knowledge, whether platform businesses have a higher or lower failure
rate than other types of business has not been studied; however, startup ven-
tures tend to generally suffer from high failure rates. It can be assumed that
platform startups are no exception in this sense, and thus studying their prob-
lems forms the research purpose of this study.
23
The research problem can be formulated as the following research ques-
tions
7
:
RQ 1: What strategic problems are encountered by early-stage online plat-
forms?
RQ 2: How can the problems be conceptualized as dilemmas?
RQ 3: Are the dilemmas interrelated? If so, how?
RQ 4: How can the platform literature and founders’ experiences help solve
the dilemmas?
The research questions therefore relate to strategic problems of platform
startups, which are conceptualized as dilemmas. The interrelations of these
dilemmas are examined, and the study analyzes their potential solutions. Due
to the aforementioned high failure rate, it is meaningful to conduct such a
study that aims at improving the survival rate of platform startups by provid-
ing knowledge on potential challenges they are likely to face, and also offering
a basis for solution building.
By asking the first research question, the study extends beyond the chicken-
and-egg problem, and shows that there is more depth and complexity in plat-
form dilemmas than is generally considered in the literature. The study chal-
lenges the simplification of “getting both sides on board” as a solution (Evans
2002), and argues that even if this is accomplished, a platform does not neces-
sarily fulfill its economic goals in terms of becoming profitable. The study
investigates solutions beyond pricing and subsidies that are more suitable for
startups, given their constraints of time and resources. It is particularly useful
to see how theoretical approaches match the reality of startup founders, and
therefore the examined solutions stem from both the literature and empirical
material.
The second research question addresses conceptualization, which is a form
of abstraction; that is, moving from the particular to the general. Conceptual-
ization facilitates 1) communication of strategic problems, 2) their further
treatment, and 3) theory generation. Communication of novel concepts takes
place both among practitioners and scholars. When a theoretical concept has
reached a state of general knowledge within a field, communication relating to
the phenomenon becomes more efficient and advanced (Hunt 2002). The
benefits of conceptualization relating to further treatment can be seen, for ex-
ample, in the famous prisoner's dilemma (Axelrod 2006); defining this
7
Note that, due to the inductive nature of the study, the precise questions were formulated after
the analysis. Whatever their initial form, theoretical sampling of grounded theory tends to reshape
research questions (Urquhart, Lehmann, & Myers 2010).
24
problem and labeling it as it is has led many scholars to attempt to create
variations, find solutions, and apply it to different contexts. According to
Glaser (2002), conceptualization precedes theory generation, and is therefore
sine qua non in academic work. Instead of constantly reformulating the same
problem, practitioners are able to identify the situation in other contexts, and
therefore also consider general and particular solutions, proposed by others, in
their own context. In sum, conceptualization of startup dilemmas can assist in
bringing these benefits jointly to scholars and practitioners.
Furthermore, the identification, conceptualization, and analysis of platform-
specific strategic problems is a worthwhile research purpose because the for-
mulation of strategic problems influences the solution process
8
(Ackoff 1969;
Lyles 1981). As pointed out by strategic management scholars, strategic issues
are rarely isolated cases, and merit a wider perspective (Lyles & Howard
1988). By thoroughly understanding the problem and its associations, founders
are able to elicit appropriate solutions (Lyles 1981). By careful conceptualiza-
tion (i.e., naming and defining) of the strategic problems, this study raises the
abstraction level and provides a deep insight on them.
Although this study does not show exact relationships between specific
problems and outcomes, based on the material, it assumes that strategic prob-
lems impact the failure outcome; that is, discontinuance of business (Watson
& Everett 1993). The following figure displays the idea of a black box be-
tween the startup beginning and the failure outcome, of which the strategic
problems are a part.
Figure 1 Black box of startup failure
Solving strategic problems, therefore, is part of the process leading to the
outcomes of success or failure. Inference can also be drawn from Pawson and
8
Ackoff's (1969) elevator problem is a good example: if waiting for an elevator is defined as a
technical problem, the company needs to engineer faster elevators. If, however, it is defined as a
behavioral problem, people can be given an activity while they are waiting.
Startup Failure ?
Strategic problem Strategic problem
Strategic problem
Strategic problem
25
Tilley's (2009) model by stating that, in thecontext of online startups, strategic
problems act as mechanisms to the outcome of failure. Adopting this logic
highlights the importance of strategic problems as a research problem; as they
are assumed to be associated with failure, the ability of a firm to identify and
solve them through correct strategic choices is likely to have a positive impact
on the firm’s survival. In other words, solving all issues leads to a viable plat-
form in terms of both interaction and revenue.
To accomplish its purpose, the study is based on empirical evidence from
early-stage startups, not on incumbents or established industry firms that al-
ready have a stable position in the market, and can afford to solve the chicken-
and-egg dilemma and other strategic problems by mass marketing or other re-
source-intensive approaches. Therefore, the focus of the study is on early-
stage online startups that employ the Internet as their marketing and distribu-
tion channel, and follow the platform business model by enabling interaction
between two or more groups of users. The units of analysis are failed early-
stage Internet startups, and the empirical material comprises 29 post-mortem
reports by founders of failed Internet startups, originally published on the Web
(see Chapter 2). The material is analyzed by employing GT in an attempt to
answer the research questions, and thus fulfill its purpose.
The importance of the research purpose can be shown in many ways. Gen-
erally, it is accepted that startup companies are important for the economy
(Audretsch & Acs 1994): they create a large share of new jobs (Kane 2010),
develop innovations to improve people’s lives (Almeida, Dokko, & Rosenkopf
2003), redeploy resources by creative destruction (Schumpeter 1961), fill gaps
in customer needs, and tackle problems efficiently and relentlessly from new
perspectives (Shepherd & Kuratko 2009). Therefore, a study aimed at improv-
ing the conditions upon which startups will thrive can be regarded as im-
portant for 1) the society in general, 2) entrepreneurs and managers of plat-
form startups, and 3) investors seeking the best investment opportunities
among competing ventures.
At the same time, the research focus excludes particular types of issue out-
side its scope. Because the study focuses on problems of early-stage platform
startups on the Internet, other types of platforms and problems are beyond the
scope of this study. The excluded platforms exist in various forms; for exam-
ple, shopping malls, credit cards, and newspapers (Rysman 2009). Their
problems might be different from those of online platforms. Other excluded
problems include, for instance, general managerial problems (e.g., lack of ex-
perience) and general startup-related problems such as liability of newness
(Stinchcombe 1965). Details on how the researcher narrowed the focus on par-
ticular dilemmas can be found in Subchapter 4.3.2.
26
1.5 Positioning
This study is positioned to the platform literature, and particularly to its strate-
gic management stream (i.e., strategic management of platforms). Figure 2
demonstrates how the platform literature is understood in this study.
Figure 2 The platform literature
As depicted in the figure, the platform literature comprises:
· Economics literature specializing in two-sided markets (i.e., two-
sided platforms; multisided platforms).
· Information systems (IS) literature relating to electronic market-
places and mobile application marketplaces.
· Marketing literature focusing on consumer interaction within plat-
forms.
· Network literature on network effects and platform/network struc-
tures.
It is argued here that all of the previous streams contribute to the strategic
management of platforms, which also includes their design (Bakos &
Katsamakas 2008). A discussion on their contents and key areas of interest
follows.
It transpires that platforms are studied across disciplines. Rochet and Tirole
(2003) form the basis of the economics literature on two-sided markets. This
research tradition can be seen to originate from prior studies on network ef-
fects, standards, and technology adoption (Church & Gandal 1992; Farrell &
Saloner 1985; Katz & Shapiro 1986). The economists’ agenda often relates to
pricing, regulatory issues, and antitrust policies (Rysman 2009). However, it
Platform
literature
Economics
(two-sided
markets)
Marketing
(consumer
participation
platforms)
Information systems
(e-marketplaces,
app marketplaces)
Network literature
(network effects)
27
also relates to network effects that stem from the earlier literature on network
effects (Katz & Shapiro 1985), critical mass (Rohlfs 1974), and ‘tipping’
(Shapiro & Varian 1998).
In particular, the concept of tipping has been applied to inter-platform com-
petition, underlying the idea of Internet markets as winner-takes-all markets
(Noe & Parker 2005). Whether two-sided markets result in quasi-monopoly
situations and whether this is good or bad (Luchetta 2012) are central ques-
tions in this debate. Subsidies can be regarded as “dumping” in a one-sided
market, whereas they can be regarded as a necessity for creating liquidity in a
two-sided market (Evans 2009b). Another notable aspect for economic analy-
sis is that, in two-sided markets, both price levels and their structure are sig-
nificant (Rochet & Tirole 2003). For example, a sub-optimal solution in one
side (e.g., price subvention) can result in an improved overall solution regard-
ing the two-sided structure (Evans 2003). Thus, the economists’ agenda links
with themes such as pricing, inter- and intra-platform competition, coopetition,
and monopoly (Roson 2005; Birke 2008; Shy 2011).
Second, network theorists tend to consider one-sided platforms in which all
users are the same type (Wright 2004). This makes sense as they are interested
in graphing the network (Viégas & Donath 2004) or examining the diffusion
process (Valente 1995), and not necessarily the transactional implications or
quality of interaction occurring within the platform. Structurally, platforms
can be perceived as networks of connected actors, or “nodes” (Banerji & Dutta
2009). Rysman (2009, 127) notes that although, technically, the economics
literature on two-sided markets can be regarded as “a subset of network ef-
fects”, it tends to focus on pricing, whereas studies on network effects “typi-
cally focus on adoption by users and optimal network size.” In any case, net-
work effects in their positive or negative form remain a central concept (Shy
2011).
The contribution of network theory to the platform literature is exemplified
by Westland (2010) who combines network laws with willingness to pay. Net-
work theory does not usually consider strategic dimensions of platform man-
agement or two-sided implications; rather, it is interested in describing various
network structures, and explaining their growth and diffusion (e.g., Westland
2010). The overlap in the interests of network and platform theories can be
seen in adoption or diffusion. Although adoption can be studied from a social
perspective in line with Rogers’ (1995) seminal book, network theory tends to
focus on descriptive models as opposed to theorizing on the reasons for
diffusion/adoption. A different subset of studies from the information systems
(IS) tradition relates to technology acceptance, such as the technology ac-
ceptance model (Venkatesh & Davis 2000). These approaches have not been
considered in the platform literature that perceives network effects as the main
28
driver of adoption (Katz & Shapiro 1986), and, although they might provide
valuable insight on the complexity of adoption, they are also not considered in
this study.
Third, regarding the IS literature, Hyrynsalmi et al. (2012), Salminen and
Teixeira (2013), and J ansen and Bloemendal (2013, 3) address the recent
stream of mobile application marketplaces which they define as “an online
curated marketplace that allows developers to sell and distribute their prod-
ucts to actors within one or more multi-sided software platform ecosystems".
The focus has shifted from e-marketplaces (i.e., late 1990s to early 2000s) to
app markets. This shift has followed the change in business markets as espe-
cially consumers have adopted various app stores, and their significance has
therefore increased (Hyrynsalmi et al. 2012). The focus of earlier IS research
was often on business-to-business (B2B) exchange platforms, and especially
on the concept of liquidity (Evans 2009a). Contrary to earlier e-marketplace
research, modern app markets such as mobile phone applications are typically
business-to-consumer (B2C)-oriented (J ansen & Bloemendal 2013). A survey
on the electronic marketplace
9
literature can be found in Standing, Standing,
and Love (2010).
Fourth, marketers are interested in platforms. In 1998, Sawhney had already
highlighted the importance of moving from portfolio thinking to platform
thinking in his commentary for the Journal of the Academy of Marketing
Science, and argued that “marketers who master platform thinking may find
the 21st century to be a somewhat more inviting prospect.” Generally, the
marketing literature tends to focus on the consumer perspective of platform
interaction and strategies relating to marketing problems such as finding and
influencing particular lead users to propagate messages (Hinz, Skiera, Barrot,
& Becker 2011) or otherwise participate in platform interaction: co-creation
platforms (i.e., firms leveraging consumers in their value-creation activities) or
peer-marketing platforms for customers voicing their opinions. Hennig-
Thurau, Gwinner, Walsh, and Gremler (2004) studied consumer opinion plat-
forms, combining virtual communities and the traditional word-of-mouth liter-
ature. In particular, seeding is regarded as a viral marketing strategy to attract
the most prominent users to join a platform (Hinz et al. 2011). Marketing
studies overlap with network studies that aim to find the most influential
“nodes” (Hill 2006). However, marketing adds the actual persuasion of these
nodes to join the platform as "real people". Another special interest of
marketers is the relationship between a platform owner, advertisers, and
9
As discussed later, marketplaces are aspecial type, but not the only type, of platforms relying on
economic exchange as the form of interaction. For other types of platform, the interaction might take a
different form.
29
consumers in an advertising-enabled platform (e.g., Salminen 2010; Reisinger
2012).
Sawhney et al. (2005) examine the Internet as a platform for value co-crea-
tion with customers. This collaborative innovation can take advantage of the
Internet’s distinct features and be exploited, for example, in new product de-
velopment. Cova and Dalli (2009) discuss how marketing theory is developing
towards working consumers as it focuses on value co-creation and customer
participation. Platforms seem to offer opportunities for marketers to leverage
and monetize customer input, and marketing scholars have been showing in-
terest incrowds (Howe 2006) as resources. While marketing studies tend to be
applied, the idea of consumers as an extension of the firm can be perceived, at
a higher abstraction level, as linking to the Coasean theory of the firm (Coase
1937) and transaction cost analysis (Williamson 1981). The overlap shows
how economists and marketers are often interested in the same phenomenon,
although at a different degree of abstraction. In platform terminology, cus-
tomer participation can be expressed with the concept of user-generated con-
tent (UGC), or actions of a platform’s users, such as participating, writing,
uploading, and sharing (Daugherty, Eastin, & Bright 2008). As most of the
studied startups apply user generation (UG), this dissertation considers impli-
cations of the UG model. However, matching the approach with marketing
research (e.g., customer participation) is left for future studies.
The strategic perspective, taken by this study, is to conceive and evaluate
strategic choices for platform stakeholders (cf. Cusumano 2010). As a
perspective, it is not limited to any discipline but to all related works that per-
ceive actors as strategic in a platform context. While a lot of attention has been
paid to the platform owner’s perspective (Birke 2008), including management
of an installed base of users, standards, and complements (McIntyre &
Subramaniam 2009), research also discusses other stakeholders' strategies,
such as those of software developers (Salminen & Teixeira 2013), that is,
complementors. In addition, platforms’ end users are mostly assumed to react
to pricing (e.g., Rochet & Tirole 2003), the number and quality of comple-
ments, and network effects; in other words, the presence of other users (Evans
2002). Of special interest is users' adoption choice, which is perceived to be
constrained by exclusivity versus diversification (Roson 2005) and resource
constraints (Iacovou, Benbasat, & Dexter 1995).
Multihoming, the practice of participants to diversify their investments
across platforms, is an example of the strategic perspective in the platform lit-
erature (Armstrong 2006). Such behavior can be seen to occur in both the
supply- and demand-side, and its feasibility relates to cost functions; namely,
whether it is wise to commit to several platforms in terms of time, effort, and
financial cost, or whether focusing on one platform is sufficient to satisfy the
30
participant’s goals. Multihoming has been studied, for example, by
Hyrynsalmi et al. (2012) in mobile application markets. Another example of
the strategic perspective is the analysis of open versus closed platforms; that
is, which setting is more suitable for the platform owner/complementor
(Boudreau 2010; Eisenmann, Parker, & Van Alstyne 2009; Gawer &
Henderson 2007; Parker & Van Alstyne 2008). The strategic perspective
therefore focuses on platform stakeholders’ choices in reaching particular
goals and outcomes. An overview of strategic problems in the platform litera-
ture can be found in Chapter 4.
Due to the focus on startup dilemmas, or contradictory decision-making sit-
uations, this study is particularly positioned to the strategic management
stream of the platform literature. However, the study utilizes the related liter-
ature such as economics papers with a strategic focus; for example, determin-
ing the correct pricing, design, entry strategy, and subsidization. Such aspects
can contribute to solving startup dilemmas relating to online platforms. The
strategic perspective does not in itself exclude any actors as they can all be-
have strategically; users consider their own benefits, as do platform owners
and complementors.
Perhaps the distinctive feature here is that the question of which actors to
examine varies across streams of the platform literature. As economists are
interested in markets, they tend to examine buyers and sellers, or demand and
supply sides (Evans 2009a). IS researchers are currently paying increasing in-
terest to app marketplaces, so their two sides comprise app developers and end
users (e.g., Mian, Teixeira, & Koskivaara 2011). The strategic choice of a
marketing manager is how to allocate marketing budget (Fischer, Albers,
Wagner, & Frie 2011), whereas it is the CEO/founder’s problematic role to
organize the whole business. This study takes the latter perspective and fo-
cuses on dilemmas that are inherent to the startup’s business strategy, and not
only to its marketing strategy, pricing, or capability to attract developers. The
perspective is that of the startup/founder: What can it/he do to solve the
startup dilemmas and avoid failure?
In sum, it can be seen that platform studies across disciplines are interre-
lated and contain overlapping interests. For example, marketers are interested
in graph theory to find potentially influential users, and economists and mar-
keters share the interest of users/customers as “resources” or assets of the firm.
The chicken-and-egg problem is common to all; as economists aim to solve it
through pricing while marketers propose sales and persuasion as the solution,
these studies underlie adoption as the fundamental phenomenon. In a similar
vein, a strategic focus can be found across disciplines; it is more dependent on
which strategic problems within the platform are of interest to scholars and
what kind of solutions they examine. The present study focuses on strategic
31
problems grounded on the material, and can be perceived as critical for the
survival of platform startups on the Internet.
1.6 Structure
The study proceeds as follows. Chapter 2 describes the methodology, includ-
ing research strategy, research process, data collection and analysis, and also
the literature approach. The theoretical framework, including conceptual un-
derpinnings and their connection to platform literature, is discussed in Chapter
3. This chapter defines platforms as a concept, explains particularities of
online platforms, and presents critical assumptions; namely, user-generation
effects, which will be referred to in the empirical part.
Startup dilemmas in Chapter 4 contain the empirical part of this study, and
are based on post-mortem stories written by founders of failed startups. The
chapter investigates specific problems of platform startups that are conceptu-
alized into dilemmas and then analyzed with the help of the platform
literature. The treatment of each dilemma is divided into four sections:
definition and exhibits, literature review, potential solution, and overall
discussion. Chapter 5 elaborates further solutions based on the second-round
analysis, arising from both the empirical material and the scholarly literature.
Finally, Chapter 6 presents the contribution to theory and practice, further
research ideas, and evaluates the credibility of this study.
33
2 METHODOLOGY
2.1 Research strategy
2.1.1 Introduction to research strategy
This study aims at the creation of substantive theory (Glaser & Strauss 1965)
relating to strategic problems of platform startups on the Internet. This is ac-
complished by conceptualizing and increasing the abstraction level of the an-
alyzed post-mortem stories. The grounded theory (GT) methodology, outlined
by Glaser and Strauss (1967) is applied as an instrument of data collection and
analysis. The following sections explain the method, why it was chosen, and
how it was applied throughout the research process.
2.1.2 What is GT?
Grounded theory is aset of methods to systematically analyze empirical mate-
rial (Finch 2002). This data can be both quantitative and qualitative (Glaser
2004), although GT is most often associated with qualitative data (Kempster &
Parry 2011). Partington (2000) notes that the foundations of GT include
theoretical sampling, or a process of data collection guided by the emerging
theory andconstant comparison, or simultaneous coding and analysis of data.
Suddaby (2006, 634) confirms this perspective, and adds that “oth concepts
violate longstanding positivist assumptions about how the research process
should work.” This contradiction relates to the method’s history of countering
deductive methods in favor of theory generation from data (see e.g., Locke
1996, for a more detailed discussion). The coding process if more thoroughly
discussed in Chapter 2.4.
By nature, GT is an inductive
10
method, intended to help the researcher
elicit answers to his or her research problem from the empirical material
(Eisenhardt 1989). Contrary to deductive reasoning, in which the
presumptions are stronger and the researcher is narrowing the scope of
10
Note that by ‘inductive method’, an inductivetendency or emphasis is implied. Pure induction
and pure deduction, for that matter, are generally considered impossible; new ideas arise from their
combination, or abduction (Suddaby 2006).
34
inquiry, in inductive logic, the scope of inquiry is broader and central issues
are gradually revealed by scrutiny, which in GT is represented by the coding
process (Strauss & Corbin 1994). Eisenhardt (1989, 541) points out that, in
inductive studies, “researchers constantly compare theory and data-iterating
toward a theory which closely fits the data.” This fit between data and emerg-
ing concepts is perceived as important because it reduces the risk of the latter
being detached from empirical relevance (Eisenhardt 1989).
In contrast, by employing hypothetico-deductive logic, the researcher first
develops hypotheses; that is, assumptions concerning what is likely to happen
or be found in the analyzed data (Laudan 1981). After this, the hypotheses are
tested with a specific method such as experiments or statistical analysis, and
the results are discussed. The hypotheses are created either by observing real-
world phenomena or by analyzing the literature for theoretical gaps (Davis
2009). Grounded theory differs from this logic in at least three aspects.
First, there are no initial hypotheses that prove, disprove, or generate a the-
ory, and the theory is generated with the fewest presumptions possible (Glaser
& Strauss 1967). Second, GT does not rely on identifying a theoretical gap
prior to analysis (Heath & Cowley 2004). The defense of the method comes
from the self-proclaimed novelty of the phenomenon and idiosyncrasy of the
utilized material, as a consequence of theoretical sampling. The self-pro-
claimed novelty implies that there is indeed some reason for inquiry as not all
is yet known (Pandit 1996). This derives either from uniqueness of the data or
from newness of the phenomenon
11
. Although the literature is not employed as
a starting point, the relationship to previous theory needs to be considered, and
this is done subsequent to data collection and analysis (Goulding 2005).
According to Glaser (1978, 51), “[r]eading the theoretical literature should
be avoided when possible until after the discovered framework is stabilized”.
Therefore, in GT, the literature review is conducted after the formulation of
categories, which might seem unconventional for researchers trained only on
hypothesis testing (Kempster & Parry 2011). Finally, whereas research ques-
tions in hypothetico-deductive studies tend to be fixed, GT allows for chang-
ing the initial questions if found irrelevant in the field (Charmaz 1990). Such
flexibility is advantageous for theory generation as it reduces the impact of
preconceived constructs and encourages the discovery of new concepts.
Two points emerge from the previous explanations. First, it was posited that
GT is inductive by nature; second, there are no initial hypotheses. The two
words, emphasized in italics, have important implications. The nature of GT is
not pure induction, but more of abduction (as so-called inductive studies tend
11
For example, one cannot have conducted research on Web 2.0 startups prior to 2005 because the
concept did not exist. Thus, a substantive theory is a contemporary outcome, placed in time.
35
to be). This means deductive reasoning is employed in the course of the re-
search; but there is a strong emphasis on “letting the data speak for itself”
(Glaser 1978), as opposed to forcing it into preconceived hypotheses. How-
ever, over the course of the research process, central themes become more ap-
parent, at which point the researcher is encouraged to compare new findings
with intermediary conclusions; this is a form of deductive reasoning.
As an example, consider the following observation: “A cat is in a tree, be-
cause a dog chased it there.” The researcher can employ this piece of data to
formulate a general hypothesis: “Dogs don’t like cats.” To confirm this hy-
pothesis, however, he needs to conduct theoretical sampling. Thus, he seeks
more empirical descriptions on the relationship between cats and dogs, and
discovers the following description: “Today it was so nice to come home and
see my cat Jim and my baby dog Bozo asleep in the same basket.” Aha! Cats
and dogs can get along, so the original hypothesis is incorrect. This method of
comparing new and previous data through tentative assumptions enables cor-
rection and modification of our hypotheses. For example, we can observe a
specific condition and say that “dogs don’t like cats, unless they are accus-
tomed to them from an early age”, and then seek to validate or refute this hy-
pothesis through theoretical sampling. This simple example shows how hy-
potheses emerge from the analysis of data, not precognition; in other words,
they are grounded (Glaser & Strauss 1967).
Due to the somewhat vexing issue of induction/deduction, the GT method
can be best labeled as data-oriented. The rationale of data orientation is to
connect theory with the real world (Glaser & Strauss 1967). In other words, if
the only way to generate new theory were to examine existing theory, one
would never originate ideas “outside the system”. In contrast, the source of
theory can bein praxis or, in effect, the data, as opposed to its fitting into a
priori theoretical framework. Here it is not claimed that one or the other is
better; in the author’s opinion, the research gap can be found both in the liter-
ature andempiricism. For the latter, the challenge is to ensure the phenomenon
has not been exhausted prior to engaging deeply in research activities
12
while,
for the former, it is to ensure the topicality of the research purpose in real life.
Overall, GT aims to avoid theoretical exercises detached from real problems
(Glaser & Strauss 1967).
Thus, ideas come from informants themselves and are labeled to match
their use of language in vivo (Charmaz 2006). The sense-making of the
informants is translated by the researcher to match the discourse in the
12
This would waste resources as enough is already known on the topic, and the researcher would
be unable to provide new insights. In practice, however, novelty is debatable as many phenomena
occasionally reoccur in the literature without being rejected.
36
academic literature. This matching process is a requisite for a) finding the the-
oretical discourse in which the study can be positioned, and b) formulating the
theoretical contribution by employing established research language and con-
cepts (McGhee, Marland, & Atkinson 2007). If this translation is not per-
formed adequately, the researcher risks remaining isolated from academic dis-
course. The successful employment of GT results in a theory with unique as-
pects, although parts of it might overlap with existing theory (Glaser & Strauss
1967)
13
.
2.1.3 Why was GT selected as research method?
First, the researcher was interested in problems of post-dotcom Internet
startups, a phenomenon not well studied (see Chapter 1). When there are no
exact presumptions and the research topic is quite new, a method aiming to
discover central topics is beneficial (Glaser 1978). According to Finch (2002,
220), grounded theory fits well with “the development of novel knowledge
claims of under-researched phenomena.” As identified in the previous
chapter, there are several gaps relating to platform-creation activities, and
managers actively seek to understand why particular strategies work while
others do not. To determine the answer, theoretical analysis is needed.
Grounded theory (GT) is particularly useful when data are in qualitative
form and the researcher still seeks a systematic methodology (Glaser 2004).
GT gives good grounds for conceptualization and raising central topics and
patterns from the data (Charmaz 1990). These features are compatible with the
objectives of this study; thus, GT provides a good methodological match for
solving the research problem.
Second, the richness of the type of data on which this study is based is sim-
ultaneously both an advantage and a disadvantage. Qualitative data requires a
significant amount of sense-making and structuration (Suddaby 2006); how-
ever, the reduction process offers good grounds for theorization (Miles &
Huberman 1994). These properties support the choice of a method that aims to
generate explanations from data. GT is compatible with this need because it is
presented as a systematic method of analysis (e.g., Glaser & Strauss 1967;
Eisenhardt 1989; Kempster & Parry 2011) and often applied to qualitative data
(Goulding 2005).
Third, GT seemed a good fit for the author’s tendency to conceptualize.
Heath and Cowley (2004) argue that the researcher’s cognitive style should
13
Of course, the researcher needs to identify these overlaps and position his/her contribution
towards the extant literature. However, this process can take placeafter the analysis.
37
play a role in the selection of method. As explained in the previous chapter,
the research began with qualitative material from failed startups. The analysis
started by finding general themes in the material. As knowledge on different
methods increased, the author understood that he was in fact coding (e.g.,
Strauss & Corbin 1994). Shortly thereafter, the author learned about grounded
theory. As the method corresponded to the approach taken up to that point, it
was not a large step to adopt GT’s principles.
Fourth, the research construct, strategic problem, is not a fact that can be
quantified, observed, or measured as a variable in an empirical model. Nor can
it be regarded as a latent variable that might be constructed by utilizing other
variables; at least, not without complex interpretations. Rather, a strategic
problem is a concept, or a conceptual construction of reality. This study as-
sumes that strategic problems exist in the real world and, once defined in the
correct strategic situation, can be perceived as a relatively stable form of real-
ity by all interpreters with adequate understanding on their nature. Thus, stra-
tegic problems are situational patterns that emerge when specific contextual
conditions are met. Such an ontological position implies critical realism, in
which real events remain dormant until triggered by particular conditions,
upon which they become actual and, if observed, empirical (Partington 2000).
To produce such understanding that is required to identify and explain strate-
gic problems – as it is not obvious that everyone, even experienced managers,
will identify thema priori – GT has to offer a set of highly useful principles.
Fifth, GT enables an easy expansion from one topic dimension to another,
not being required to remain within the scope of the initial data (i.e., “all is
data”). In this study, this feature of GT shows in expansion from problems to
solutions; that is, additional interviews focusing on solutions and analysis of
online material relating to them. The approach is much different compared to
research designs where the data are static (i.e., remains as what is collected)
and then is employed to answer a priori research questions. In GT, a priori
research questions can change. In this study, the focus changed from business
models to failure to strategic problems, and was finally annexed by the discov-
ery of solutions. Therefore, GT's approach to see “all as data”, in addition to
its flexibility in terms of theoretical sampling and constant comparison, en-
couraged the researcher to be led by the phenomenon instead of his initial pre-
conceptions.
Relating to strategic issues of platforms, Gawer (2009) highlights the im-
portance of a firm’s capabilities and also industry- and firm-specific circum-
stances; that is, thecontext. An emphasis on context supports the choice of GT
as a methodology, as it is often applied to generate understanding on a re-
search problem in a particular context (Chesler 1987); that is, a substantive
theory (Glaser & Strauss 1965). For example, Orton (1997) reported the use of
38
GT in studying strategic change processes in loosely coupled systems. Kan
and Parry (2004) examined resistance to change in a hospital setting, and
identified paradoxical thinking as an influencer. According to Wagner,
Lukassen, and Mahlendorf (2010, 9), “grounded studies are especially
appropriate for gaining an initial understanding of complex transitions”; ar-
guably, strategic problems associated with startup failure can be categorized as
such.
The extant platform literature has approached the chicken-and-egg problem
(i.e., getting both sides on board) mainly from the pricing perspective, and
focused on analytical modeling (Piezunka 2011). Few inductive studies have
been conducted to understand the roots of the problem, or how it might be
solved (e.g., Birke 2008). This study provides a step in that direction. As will
be shown, mere pricing (i.e., levels or structure) is insufficient as a solution to
the cold start problem; in fact, several studied startups offered their products
for free, and still failed to gain growth. The lack of participation is only par-
tially explained by overly high prices; fundamentally, it is a much more com-
plex phenomenon. This study is geared towards the interpretations of failed
startup founders. In these stories, founders explain why their ventures failed.
The inductive nature of the study will provide a needed empirical grounding
for the treatment of strategic problems.
2.2 Research process
This chapter describes how the study was conducted, and how it evolved over
the course of the analysis. Figure 3 depicts the research process.
Figure 3 Research process
First, post-mortems were collected through online searches and by follow-
ing links from aggregators and curators of post-mortem stories [1]. Note that
Data
collection
Data
analysis
Literature
collection
Theoretical
integration
Post-hoc
coding
Constant comparison & theoretical sampling
1
2
3
4
5
Additional
interviews 6
Finalizing
7
0
39
post-mortems represent first-hand data for the analysis, whereas other readings
and discussions with founders in various startup events correspond to theoreti-
cal sampling of GT [0]. In turn, adjusting the learning from the literature, dis-
cussions, and interviews to preliminary findings corresponds to constant com-
parison of GT [0].
In the first phase of data collection, everything relating to failure of Internet
startups was retrieved. In the second phase, criteria for inclusion and exclusion
were developed and narratives were filtered, which is explained in the fol-
lowing subchapter. In total, 29 failure narratives remained at this point. Then,
to analyze the material [2], several phases of coding were conducted according
to the GT method (see Subchapter 2.4).
During the coding phases, strategic problems emerged as the key theme of
the study. Thereafter, the conceptualization of the dilemmas began. At this
point, various streams of the literature were collected and read [3] to
determine the study’s theoretical framework positioning. Note that at this
stage, several alternatives for placing the findings in the literature existed. The
studied research streams comprised the literature focusing on business models,
business failure, and platforms (two-sided markets). Eventually, based on the
author’s judgment, it was decided that strategic problems of the studied
startups, conceptualized as “dilemmas”, had most in common with the plat-
form literature. The research focus was therefore narrowed down, and a sys-
tematic integration of the platform dilemmas into the extant literature began
[4].
Consequently, a more thorough retrieval and review of the platform litera-
ture began (see Chapter 2.5). The researcher made multiple searches and col-
lected the literature by snowball sampling the found papers. The literature was
read keeping the theoretical constructs (i.e., dilemmas) in mind, and the syn-
thesizing of startup dilemmas and the platform literature began. This process is
described astheoretical integration by Urquhart, Lehmann, and Myers (2010).
Commonalities with the findings and extant theory could be found quite eas-
ily, which reassured the researcher that the correct literature had been chosen.
In other words, the literature and empirical accounts seemed to discuss the
same phenomena, despite utilizing different words.
In general, conducting the literature review and positioning after the analy-
sis is in strict accordance with the principles of GT (Strauss & Corbin 1994;
Glaser 2004). Such a choice is intended to facilitate inductive theory for-
mation: that is, to avoid preconceptions arising from the literature to shape the
conceptualization, understanding, and interpretation of the initial findings to
the degree where their originality is lost.
The author found it purposeful also to consider some potential solutions in
addition to the extensive analysis of the problems. However, solution finding
40
is an extension to the work; its main focus is on the dilemmas. To find addi-
tional solutions to the dilemmas, two steps were taken. First, post-hoc coding
was conducted after the literature analysis [5]; this time focusing on solutions
to the dilemmas. This step included coding of the original material for “what
if” statements in which the founders expressed what they would have done
differently, had they been given the choice. Second, the author decided to uti-
lize the principle of theoretical sampling by conducting additional interviews
[6] with founders.
After these efforts, the report was finalized [7]. The report was written in a
conventional format, outlining research questions, then methodology, the liter-
ature, and results. Suddaby (2006) mentions that this is a common way to re-
port a GT study. Although the exact research problems were formulated ex
post, presenting them in the introduction helps readers understand the study’s
purpose. The study itself began with no preconceived theory, as is the case
with GT studies (Glaser & Holton 2004), and the research gap or research
problem did not initially exist in the way described in Chapter 1. Prior to the
analysis, the researcher was interested in a different purpose than that which
emerged from the material over the course of the analysis. However, Suddaby
(2006) notes that although it can at times seem confusing, reporting GT by the
conventional “deductive study structure” is normal.
2.3 Research data
2.3.1 Data collection
The analyzed material comprises 29 failure reports by founders of failed
startups. The narratives, or “post-mortems” as termed by startups, were written
by founders to reflect the startup’s failure, in particular to identify reasons for
that failure. Thus, post-mortem is defined here as a story analyzing a failed
startup venture. The stories were collected from the Internet by following
links from various blog articles listing and publishing post-mortem analyses,
and conducting searches via Google search engine and two startup-centered
online communities.
Keyword phrases for Web searches included:
· startup failure story
· startup postmortem/post-mortem
· startup failure analysis
· business postmortem/post-mortem
· business failure analysis.
41
The data collection process began by gathering all post-mortem stories the
researcher could find. The search was conducted by finding aggregated blog-
posts listing startup failure stories and then following links to original posts,
similar to “snowball” sampling (see Biernacki & Waldorf 1981), and by per-
forming Web search queries. In particular, ChubbyBrain (2011) contained
links to several post-mortem stories. Following links, post-mortem stories
were captured for further filtering and analysis. Additionally, Google
14
was
utilized to find post-mortems; this is because Google’s search algorithms tend
to be the most accurate of current search engines (Uyar 2009), and its index of
Web pages is commonly judged as current and extensive (e.g., Gulli &
Signorini 2005).
Moreover, searches were conducted on two startup-focused online commu-
nities: Quora and Hacker News. These communities contain a substantial
number of discussions relating to Web startups, and also included discussion
threads on startup failure. Reading these discussions helped the researcher to
become familiarized with the phenomenon and find links to still new post-
mortem stories.
In addition to reading the post-mortems, the author sought additional ways
to deepen his knowledge on the startup industry or, as it is commonly termed,
the “startup scene”. The steps for doing so comprised the following.
First, approximately a dozen interviews by a startup-focused journalist
Andrew Warner
15
were read. As these interviews were freely available on the
website in transcribed form, they were read to find confirmation, contradic-
tion, or complements to the post-mortems’ findings. The interviews were a
good source of secondary data because they included both successful and
failed startups, and therefore provided useful background information on the
industry and the startups’ founders’ decision-making and ways of thinking.
Second, to deepen the knowledge on the sampled startups, the comment
sections of the post-mortem stories were read; the stories were published in
blogs, and therefore could be commented upon. There were some cases in
which other founders participated by questioning parts of the analysis or by
sharing their own stories. In addition, in a couple of cases, customers disa-
greed with the story, and also the content suppliers of one startup were bitter
(i.e., in platform terms, the “other side”). Although fascinating, analyzing the
discourse between founders and other interpreters was not the goal of this
study, so the researcher did not go deeply into the question of “who is right”.
14
www.google.com
15
www.mixergy.com
42
However, familiarization was enriched when founders’ replies brought further
clarification to the cases.
Finally, six additional interviews were conducted with founders. Each of
the interviews lasted for approximately an hour and was theme-based, the
theme being the strategies and tactics the founders had employed, or knew
about, in solving the chicken-and-egg dilemma. The founders were asked
questions such as “How are you solving the problem for side A(/B)?”, “What
is your most successful (/unsuccessful) solution?”, and “How are you planning
to grow in the future?” All founders were knowledgeable of the topic, and
could express advanced ideas relating to it. During the interviews, the author
made notes and mentally compared the emerging points to previous findings.
Later, the notes were integrated into the solutions section of this study.
Although the process of theoretical sampling can be continued for a very
long time (in fact, infinitely), within the frame of this study (i.e., its focus on
dilemmas; time constraints) six interviews, in addition to post-hoc coding,
were regarded as adequate for the discovery of solutions. Due to GT's ac-
ceptance of additional data collection and its comparison to earlier findings,
and also its acceptance of different types of data, the research process can be
continued in the future.
2.3.2 Selection criteria
All post-mortems were filtered for further analysis. The selection criteria com-
prised:
· Internet-based commercial venture, but not necessarily incorporated.
· Post-mortem written by one of the founders.
· Can be defined as a platform, connecting two or more groups.
· Established between 2004 and 2010 (i.e., Web 2.0, after the dotcom
period)
· No more than 60 months old (i.e., early-stage startup).
The “Internet-based” criterion stems from the research purpose, which is to
study online business, not offline-with-online-extension, or hybrids (i.e.,
“click-and-mortars”). A general definition of a platform was applied to iden-
tify appropriate startups; moreover, the process resulted in the emergence of
four online platform types.
Additionally, the depth and length of stories were considered, so that the
accepted stories had at least approximately 1,000 words to ensure some
“thickness” (Neilsen & Rao 1987). On average, a post-mortem story
43
comprised 3,037 words. Post-mortems were preferred to be as candid and
unbiased as possible, although this is a subjective measure; potential biases
will be considered later. The stories were not anonymously written as they
included authors’ names. To maintain somewhat consistent interpretations,
only stories written by founders were included; for example, there were some
that recounted interviews with founders, but these were judged less authentic
than had the founders actually written the stories. Tracking the authors in
social media services ensured authenticity of the stories. Most founders were
found via LinkedIn
16
, and they provided more information on their cases.
According to the previously mentioned principles, non-Internet businesses,
seemingly short and superficial stories, those not personally written by found-
ers, and those written in an editorial style or by a journalist were filtered out.
Filtering was conducted to limit the scope of study to self-reflection that was
inherently honest, authentic, and of some depth. However, for selection, incor-
poration (i.e., being a registered company) was not required as it was per-
ceived that this would rule out very early-stage startups on which the study
focuses.
2.3.3 Description of the startups
Overall, after 12 stories were excluded based on the aforementioned criteria,
29 failure stories remained for analysis. Short descriptions were written to
summarize the startups’ purpose in an easily understandable way. Such de-
scriptions facilitate the examination by third parties unfamiliar with the
startups; crystallization is also helpful for analytical purposes. Descriptions
were retrieved from two startup databases, CrunchBase
17
and ChubbyBrain
18
,
or, when neither of the databases contained data on a startup, Google search
engine was employed to find a description, preferably from the founder’s web-
site or blog. The general descriptions can be found in the following table.
16
www.linkedin.com
17
www.crunchbase.com
18
www.chubbybrain.com
44
Table 2 Descriptions of analyzed startups
Description Type Side A Side B
Backfence was a hyper-local, community-based
news and information service.
Content Local users Local users
Boompa was a social encyclopedia focusing on
motor vehicles.
Content Users Advertisers
Bricabox was a platform for creating a personal
social content site.
Social Users Users
ChubbyBrain captured and structured infor-
mation on innovation economy and startups.
Content Users -
Contrastream was a social music platform. Content Indie musicians Users
Devver aimed to turn desktop development tools
into cloud-based services.
Infra Users Developers
Diffle was a social networking site centered on
simple flash games.
Social Users Users
eCrowds combined Web content management
and social networking for SMEs.
Infra Consumers SMEs
eHarmony for Hiring aimed to match job-seek-
ers with job-providers.
Exchange J ob-seekers J ob-providers
EventVue was a tool for building conference
communities.
Social Conf.
participants
Conf.
participants
Imercive provided an IM marketing solution to
help brands’ consumer engagement.
Infra Brands Users
Kiko offered anyone a calendar to keep and
share online.
Social Users Users
Lookery helped social networks distribute their
data outside their Web sites.
Content Users Social
networks
Meetro was a location-aware IM client and real-
time social networking application.
Social Users Users
Monitor110 helped institutional investors ac-
cess, analyze, and monetize Web information.
Content Investors -
NewsTilt was a service for journalists to build an
online brand by engaging their readers.
Social Readers Journalists
Nouncer enabled real-time distribution of micro-
content to Web applications.
Infra Users Developers
Overto aggregated information from different
auction platforms to deliver better results.
Content Buyers/sellers -
Pixish was a platform for user-generated graphic
design work.
Exchange Designers Design-seek-
ers
PlayCafe was an online network that streamed
user-generated game shows.
Content Users Users
[Q&A startup] aimed at creating a marketplace
for selling and buying answers.
Content Askers Answerers
RiotVine was a social event guide for discover-
ing and sharing events with friends.
Social Visitors Event
organizers
SMSnoodle was an SMS based entertainment
channel for the Singapore region.
Content Content
providers
Users
SubMate enabled discovering new people and
things to do before and after commuting.
Social Commuters Commuters
Transmutable was a platform for doing 3D sim-
ulations on the Web.
Content Users -
Untitled Partners enabled fractional ownership
of art through cooperative purchasing.
Exchange Art-lovers Art-lovers
Verifiable wasa platform for data visualization. Content Users Users
Wesabe was a finance service for tracking per-
sonal spending patterns.
Exchange Consumers N/A
Xmarks offers a social bookmarking and syn-
chronization service.
Content Users Users
45
As can be seen, all platform types are presented. Their definitions are dis-
cussed in Subchapter 3.3. The frequency of platform types is as follows
(N=29):
· 13 content platforms
· 8 social platforms
· 4 exchange platforms
· 4 infrastructure platforms.
Most platforms studied are two-sided, but there are also one-sided plat-
forms, in which the users are not divided into two mutually complementing
groups. The average lifetime of a startup was 26 months.
The oldest startup was 57 months at the time of failure, the youngest 8
months
19
. The sample comprised both B2C and B2B startups, with the major-
ity being consumer-oriented startups. The mode of team size was 2.5 mem-
bers, with the largest team having 30 members and the smallest one member.
Most teams were male-dominant, and only two reported women in their team.
Approximately half of the founders (57%) were first-time founders, the rest
had earlier startup experience. All teams had technology experience, but only
38% reported prior marketing experience. The vast majority was US-based
startups; there was one startup from Poland and one from Singapore. Almost
all startups (86%) also applied either user-generation (UG) or aggregation as
their content creation model
20
, which makes UG (Chapter 3.4) highly
characteristic of this sample. Other characteristics include offering free ac-
cess/use of the platform, indirect monetization, and the freemium business
model. These features become relevant in Chapter 4.
The contemporary focus (see Figure 4) excludes dotcoms, several of which
were found among all retrieved post-mortems. Coincidentally, the selected
startups are a part of the so-called Web 2.0 era (O’Reilly 2005). The Web 2.0
period can be seen to start from around 2005 when the concept was first intro-
duced (O’Reilly 2005). The dotcom period is generally regarded to have oc-
curred in the late 1990s to the early 2000s (Razi, Tarn, & Siddiqui 2004;
Evans 2009a), including a strong hype cycle of unrealistic expectations for
Web platforms and e-commerce (Lieberman 2005), and then a quick demise
after a large share of these businesses failed to perform (Cochran et al. 2006).
19
Calculated from date founded to the date post-mortem was published.
20
A content model explains how the startup provides content for its users.
46
Figure 4 Historical positioning of the analyzed startups
The platforms in the dotcom era were mostly e-marketplaces (Wang,
Zheng, Xu, Li, & Meng 2008); since then, there have been considerable
changes in online startups’ business models (Rappa 2013) . While the first
wave of platforms included “importing” retail and B2B exchanges to the In-
ternet, Web 2.0 platforms offer purely digital services on their own (Aggarwal
& Yu 2012). This fact does not particularly reflect the research purpose (see
Chapter 1), although it does add to the topicality and novelty of the material
analyzed in this study.
With the relative “freshness” of the sample, the goal was to ensure that
problems remain topical. If problems were already solved in the “latest batch”
of Web startups, there would be no research gap in the empirical sense, which
would be a critical problem for GT that aims at usefulness of the resulting the-
ory (Glaser & Strauss 1967). Such a risk would be higher had the study in-
cluded Web 1.0 startups, and ignored the implicit learning occurring after, and
due to, them. In turn, the research gap in the literature stems from incomplete
understanding on the chicken-and-egg problem, its solutions, and derivative
problems (see Chapter 1.1).
2.4 Analytical approach
2.4.1 Coding process in GT
According to Strauss and Corbin (1994), after data collection, the researcher
should start by open coding; for example, reading through narratives, making
notes, and identifying themes and interesting phenomena. This process leads
to the creation of categories, or groupings of concepts, that appear to relate to
the same phenomenon (Glaser & Strauss 1967).
Dot-coms Web 2.0
ca. 1999-2001
Samplestartups (2004-2010)
ca. 2005à
47
GT reaches theoretical refinement through iteration; once themes begin to
emerge, the researcher re-reads the material while modifying conceptual
codes. Essentially, this leads to an index of codes, organized under categories
based on the nature of the phenomena being described in the text (Strauss &
Corbin 1994). A category can contain subcategories if the researcher interprets
the phenomenon as a hierarchy (Glaser & Strauss 1967).
Next, axial coding procedures are employed to compare extant codes to the
subcategories, and a selected part of the material is modified to reflect the core
category (i.e., selective coding). In general, axial coding refers to looking for
relationships between conceptual constructs, and the conditions in which they
take place; for example, they might coexist or appear only under specific cir-
cumstances (Strauss & Corbin 1994). The idea is to develop connections be-
tween the themes found in earlier coding (Strauss & Corbin 1994).
Through constant comparison, a grounded theorist derives the core category
from the material (Glaser 2004). The method is then employed to build con-
sistency through a mental process of comparing new coding to existing cod-
ing, so that it becomes part of a single theoretical framework (Goulding 2005).
A suggested approach to this is memoing (Strauss & Corbin 1994; Glaser
2004), which means noting down ideas in a form of meta-data.
Constant comparison reveals whether the new data provide anemergent fit
or not, thereby guiding theory generation (Glaser 2008). GT is an iterative pro-
cess in which new themes and relationships emerge, and the researcher is re-
quired to re-code the data (Gasson 2003). Finch (2002) describes this as mov-
ing from description to analysis, and from analysis to explanation.
Grounded theory arises from the interaction between researcher and data;
therefore, becoming intimate with the circumstances is helpful. According to
Glaser (1978), understanding the context can increase theoretical sensitivity,
which is a mixture of theoretical (i.e., the literature) and practical expertise,
and can improve the researcher’s judgment. Charmaz (1990) makes it clear
that topical knowledge improves the researcher’s ability to perform GT analy-
sis. The researcher’s inner ability to conceptualize in a meaningful way is
highlighted by both Glaser (1978) and Strauss and Corbin (1994) through the
concept of theoretical sensitivity.
Constant comparison is coupled with theoretical sampling by maximizing
similarities and differences between coded phenomena to find the boundaries
of theoretical constructs, not only what is apparent in a limited amount of
data
21
(Creswell 2008). The end result should be an abstract theory derived
from non-abstract data; that is, there is an increase in the level of abstraction
(Strauss & Corbin 1994). Together, theoretical sampling (i.e., finding
21
The requirement, therefore, is an adequate amount of data for all major variations to appear.
48
additional evidence to back up intermediary conclusions) and constant
comparison direct the researcher to verify whether the emerging theoretical
model holds as new data are collected, and to modify the model if necessary
(Elliott & Lazenbatt 2005).
A critical part of GT, as in much research, is deciding what to include or
exclude (Wagner et al. 2010). This cannot be known a priori, as central
themes are unknown until open coding. Once they emerge, one resorts to se-
lective coding; that is, discovering the central phenomenon (Glaser & Strauss
1967). The purpose of selective coding is to choose the themes and codes cen-
tral to the theory under development (Corbin & Strauss 1990). The idea is that
they are unknowna priori, as opposed to hypothetico-deductive logic, so that
the researcher does not know before looking what is central in the data (Heath
& Cowley 2004). Having strong presumptions, therefore, is a risk to finding
the core phenomenon in the data
22
. Once the core phenomenon (i.e., category)
emerges, alternative ways of looking at the material become less relevant, and
the work pivots around the central phenomenon (Strauss & Corbin 1994).
2.4.2 Application of GT in this study
In this study, the author maintained an open mind when first becoming ac-
quainted with the stories. The material was imported to QSR NVivo 10 (i.e., a
software package for qualitative data analysis), with which it was coded.
Figure 5 illustrates how GT was applied in this study.
22
In this sense, GT formulation is an inductive process; however, empiricists and rationalists both
agree that there is no pure deduction or induction because deduction by the researcher’s human mind
always carries some preconditions (i.e., no tabula rasa), as also does deriving conclusions from
empirical data. For a treatment on this topic see, for example, Perry and Jensen (2001); for abduction,
see the method section in Aarikka-Stenroos (2011).
49
Figure 5 Application of grounded theory
First, open coding was employed for familiarization with the material. At
this point, the process is exploratory and descriptive, and does not rely on hy-
potheses of earlier research, although there might be theoretical sensitivity
arising from the literatures’ knowledge, and professional and personal experi-
ence (Strauss & Corbin 1994). The process of constant comparison in this
study involved 1) analyzing failure narratives to build grounded theory and 2)
discussion with founders of other Internet startups outside the sample to verify
the intermediary conclusions. Therefore, the results of inquiry are consistent
with what founders deem important, but transcend individual accounts in
comparison and level of abstraction (cf. Finch 2002).
After open coding (Strauss & Corbin 1994), there was the choice to focus
on 1) startup failure, 2) strategic startup dilemmas, or 3) startup fallacies. Ini-
tially, startup failure was chosen as the focus because the author thought it was
an interesting topic. However, reading the literature eventually made it clear
that new venture failure has been quite extensively studied (see Chapter 1),
with results similar to those suggested by the open coding. Therefore, the
study risked confirming earlier findings and not generating much new
knowledge. On realizing this, the researcher became familiar with the platform
literature and decided to change focus. Therefore, “Dilemmas” was chosen as
the core category and other phenomena were excluded from this work.
Generally, GT studies tend to choose one core category, due to reasons of
manageability (Holton 2010). Dilemmas emerged as the core category in this
study due to the prevailing role of strategic problems in founders’ sense-mak-
ing. In particular, as strategic problems emerged as the most explanatory
theme, there was a shift from examining failure, as the core category, to stra-
tegic problems that were conceptualized as startup dilemmas. The kind of
Strategic problems
Strategic
problem 1
Strategic
problem 2
Strategic
problem 4
Strategic
problem 3
Axial coding
Core category
Data
Selective coding
Open coding
50
flexibility this change represents can be regarded as an advantage of the GT
method that encourages transformation in the course of analysis (Goulding
2005). It is better to change focus than to pursue a less fruitful topic (Corbin &
Strauss 1990). Moreover, the earlier analysis is rarely wasted but re-emerges
as insight for the new focus (Bauman 2010). For example, the initial focus on
failure factors helped the author associate the strategic problems with failure;
that is, ensure that they are important to the outcome.
In the axial coding phase, the researcher found both confirmation and ex-
ceptions, resulting in amending, extending, or refuting earlier assumptions; as
well as developed connections between the categories. Throughout the re-
search process, there was an iterative process of theory formation and constant
comparison of new ideas to earlier ones. This process of familiarization, meant
to increase theoretical sensitivity, included discussions with startup founders at
various startup events in Finland (Turku; Helsinki), Sweden (Stockholm) and
United States (San Francisco) between 2010 and 2014. The researcher partici-
pated in events where startup founders, investors, and enthusiasts gathered to
present business ideas and demonstrations, and to network. The events in-
cluded, among others, Good Morning Stockholm (2010; 2011); Slush Helsinki
(2011; 2012); Summer of Startups keynotes (Turku, 2011); Startup Day
(Stockholm, 2012); Launch Festival (San Francisco, 2014); and many other
startup events in Turku and San Francisco
23
.
The discussions, especially in the first years of conducting research, repre-
sent theoretical sampling in this study; that is, verification of intermediary
conclusions. As the study evolved into strategic dilemmas, the researcher be-
gan asking questions such as “What is your startup’s biggest problem at the
moment?” Answers categorically corresponded to one of the identified dilem-
mas, most typically relating to user acquisition. This led to belief that adequate
theoretical saturation had been accomplished. Overall, dialogue with founders
was comfortable as it is conventional to pitch (i.e., present business ideas) at
the events in which the author participated; thus, founders were in a ready
state of mind to discuss their startups. In addition, many of the founders were
interested when the author mentioned he was studying the failure of platform
startups, and they could relate to the dilemmas as explained by the author.
Moreover, the author was involved as an active member and board member
of the Boost Turku Entrepreneurship Society based in Turku (Finland) during
the period of data collection and analysis, and could therefore observe several
early-stage online startups in their infancy.
Overall, discussions with startup founders and also reading relevant back-
ground material increased the researcher’s domain-specific knowledge on
23
The author spent three weeks in San Francisco participating in local startup events.
51
online startups, thus facilitating the conceptualization of dilemmas. The par-
ticipation in startup events provided useful access to founders with different
business ideas, and enabled comparison with the emerging categories based on
the startups in the sample.
2.4.3 Coding guide
Strauss and Corbin (1994) set out an exact procedure, aparadigm model, for
GT analysis. In this study, the paradigm model requiring the analysis of con-
ditions, context, action, and consequences (Strauss & Corbin 1994) is regarded
as overly complex and not relevant to the research topic. Partington (2000, 95)
notes that this is a common concern and discourages the literal use of the
Straussian approach:
"The difficulty of applying universal grounded theory prescrip-
tions is borne out by experience with doctoral students working
in the field of organization and management who have attempted
to follow the Strauss and Corbin approach but have abandoned
it because of its bewildering complexity."
Glaser (1992) strongly attacks the Straussian approach for what he terms
“forcing conceptual categories”. This disagreement has been discussed exten-
sively elsewhere (e.g., Heath & Cowley 2004; Locke 1996), and will not be
repeated here. Instead of employing Strauss and Corbin’s (1994) paradigm
model, a coding guide is generated based on subcategories of the core cate-
gory; namely, “Dilemmas” as the core category with each dilemma as a sub-
category. Based on this classification, the material was re-coded. Table 3
shows examples from the coding guide.
52
Table 3 Examples from coding guide
Code Meaning Example
Coldstart
dilemma
Inability to get
content with-
out users.
underestimated the “Cold Start” problem, I read this article by
Bokado Social Design which talks about a big issue you face with a
social site, especially when it relies on user-generated content. The
value you provide to your users centers around the content on the
site, so to build a user-base you need a lot of content created by the
first users to kick-off the community.
Lonely user
dilemma
Inability to get
users without
other users.
If someone wasn’t online when you were online, they were no good
to you. While the real-time chat aspect of the application made for
some really serendipitous meetings, it also made it harder for people
to gauge the activity of their communities, especially if they logged
in at odd hours, people were set as away, etc.
Monetiza-
tion di-
lemma
Inability to
charge money
and get users.
For four years we have offered the synchronization service for no
charge, predicated on the hypothesis that a business model would
emerge to support the free service. With that investment thesis
thwarted, there is no way to pay expenses, primarily salary and
hosting costs. Without the resources to keep the service going, we
must shut it down.
The comprehensive coding guide can be found in Appendix 1. In total, 162
codes were utilized to discover meanings from the data. At the open coding
phase, general reasons for failure were coded. In the axial coding phase, they
were grouped into larger categories such as “Marketing”, “Team”, and “Busi-
ness model”. These described the founders’ reasons for failure. However, two
theoretically interesting categories also emerged at this stage: “Dilemmas” and
“Fallacies”, respectively referring to strategic problems and erratic thinking.
This is in line with GT, whereby coding proceeds from description to abstrac-
tion from time, place, and people (Glaser 2008).
The author was unable to find the approach of coding guide in the GT
methodology literature; the closest is Schmidt's (2004) description of employ-
ing a coding guide in the analysis of semi-structured interviews. Nevertheless,
this approach was useful
24
. Additionally, the online platform typology and
ideal UG model (see Chapter 3) reflect the conditional parameters that axial
coding, and also its paradigm model and conditional matrix, aim to discover
from the data. In other words, the spirit of GT is followed. At the same time,
this feature of opting towards more flexibility positions this work more closely
to classic GT, according to the prevailing interpretations of these two schools
(e.g., Heath & Cowley 2004; Locke 1996). Therefore, this study can be per-
ceived as being closer to the Glaserian school, advocating creativity instead of
rigor of analysis, although it does not explicitly subscribe to either school. In
fact, the commonalities of the two approaches seem to exceed their differences
24
It can be argued that writing a coding guide taps into the same cognitive processes as memoing;
namely, articulating and explicating the nature of the emerging constructs.
53
and, as shown by later editions of Strauss and Corbin’s book first released in
1994, they do not necessarily require the categorical following of their proce-
dure
25
. Therefore, it is not seen that two approaches are mutually exclusive
and, consequently, there is no need for a strict adherence to either at the ex-
pense of the other.
Finally, in the course of the analysis, the author applied game-theoretic il-
lustrations (i.e., strategic game situations) as an analytical tool. Regarding di-
lemmas as “games” facilitated their systematic analysis. These illustrations
can be seen in Chapter 4.
2.5 Literature approach
This chapter describes the process of the literature review. Note that the liter-
ature searches were conducted only after the initial analysis (see Chapter 2.2).
In other words, the analysis guided the selection of this particular theoretical
framework, and therefore positioning towards the literature. The overall work
is positioned to the platform literature, which can be perceived as a multi-dis-
ciplinary field. This enabled the researcher to selectively “borrow” the litera-
ture from other areas, such as entrepreneurship and strategic management; that
is, beyond the contribution of these disciplines to the platform literature. How-
ever, this study is positioned in the platform literature, from which theoretical
constructs are drawn.
The base concepts were as follows:
· Online platforms and ‘internet platforms’
· Platforms
· Two-sided markets and ‘two-sided platforms’
· Double-sided markets
· Dual-sided markets
· Multi-sided markets and ‘multisided markets’
· Multi-sided platforms and ‘multisided platforms’.
The concepts were deduced from the platform literature. Table 4 contains
keywords that were combined with the base concepts to produce search que-
ries.
25
“The analytic process should be relaxed, flexible, and driven by insight gained through
interaction with data rather than being overly structured and based only on procedures” (Corbin &
Strauss, 2008, p. 12).
54
Table 4 The literature keywords
Related dilemma Keywords
Cold start dilemma chicken and egg, chicken-and-egg, chicken-egg
user generation, user-generated content
Lonely user dilemma network effects, critical mass
Monetization
dilemma
monetization, monetization
freemium
Remora’s curse power, embedded
Freemium, a term originated by venture capitalist Fred Wilson in 2006 was
selected as a keyword because it has gained increasing interest from practi-
tioners and academicians (see e.g., Teece 2010, 2011), and many online
startups have adopted it as their monetization model. For the same purpose,
‘user generation’ was included. There were not many studies that referred to
these concepts in the platform context; however, studies conducted on other
contexts were chosen, which proved useful in positioning the dilemmas (see
Chapter 4). The literature searches were conducted with the Nelli search en-
gine (i.e., Turku University’s library system) that connects with the major lit-
erature databases including, for example, Science Direct, EBSCO, and Web of
Science. A special approach was employed to generate search queries, which
involved determining base concepts and combining them with keywords re-
lating to selected dilemmas. For example, if a base concept was ‘online plat-
forms’ and the dilemma-specific keyword was ‘power’, then the search query
would be ‘online platforms +power’. Phrase match of search words (e.g.,
“keyword”) was utilized, which resulted in more relevant hits than broad
match (i.e., keyword). All fields of articles were searched, including author-
specified keywords, title, and abstract.
The results were checked for relevance by reading their abstracts to elimi-
nate irrelevant articles (for a similar approach, see e.g., Wiltbank, Dew, Read,
& Sarasvathy 2006), leaving 302 articles that were saved to folders and read in
the process of analysis. The literature was then expanded based on reading the
articles, a form of snowball sampling. In particular, Publish or Perish software
was employed to retrieve articles
26
. This freeware software for Windows ena-
bles the user to run queries onGoogle Scholar, and shows up to 1,000 results
in one view. Additionally, it enables rapid sorting based on rating (i.e., number
of citations). According to Kousha and Thelwall (2007), Google Scholar is a
26
http://www.harzing.com/pop.htm
55
useful complement in retrieving research material, as it can find scholarly
works not included in academic databases.
Priority in selecting articles for a thorough reading was given to recent re-
search as interest in platforms is relatively new. Moreover, classic articles
were read to discover the origins of concepts and theory; for example, the
standards literature from the 1980s (e.g., Katz & Shapiro 1985; Farrell &
Saloner 1985), and network effects from Rohlfs (1974). The classics, and also
more recent seminal papers such as Rochet & Tirole’s (2003), were deduced
from the recent literature. The aim was to utilize the state-of-the-art platform
literature when positioning the dilemmas. Moreover, dissertations were con-
sidered, as some eminent platform theorists wrote their dissertations on plat-
forms (e.g., Hagiu 2008). Working papers were also included, although they
were retrieved beyond the database search.
According to Roson (2005), working papers form a considerable part of the
early (modern
27
) platform literature. Additionally, it was found that the highest
fit for this study were articles explicitly mentioning ‘online’, ‘internet’, or
‘platforms’ in their titles. Departing further from these concepts meant ab-
stracting from the context of online platforms and moving into more general
studies on the phenomenon. Merely including the theme keywords would have
generated a vast amount of the literature relating to the respective
phenomenon (e.g., ‘power’); however, the scope was kept within the platform
literature. Glaser (2004, 12) explicitly mentions that GT treats the literature
“as another source of data to be integrated into the constant comparative
analysis process.” Arguably, this has led to the selection of a theoretical
framework compatible with the emerged phenomenon.
27
“Early modern” refers to the interest following Rochet and Tirole’s seminal working paper in
2001, later published in 2003.
57
3 THEORETICAL BACKGROUND
3.1 Concept of platform
3.1.1 Platform theory and platform literature
Instead of a unified platform theory, scholars rely on similar constructs and
assumptions to study the particularities of platform business. The literature
focusing on the issues of two-sidedness, the chicken-and-egg problem (see
Chapter 4.4), critical mass, network effects, multihoming, and single-homing
(see Chapter 4.7), and associated constructs comprise what can be termed the
platform literature (Rochet & Tirole 2005; Roson 2005; Birke 2008; Shy
2011). These constructs also form the theoretical foundation of this study.
Platforms have been studied in several contexts. The following list contains
examples of some platforms that have been studied: online infomediaries
(Hagiu & J ullien 2011), mobile application marketplaces (e.g., Salminen &
Teixeira 2013), operating systems (Church & Gandal 1992), videogames
(Maruyama & Ohkita 2011), Yellow Pages (Rysman 2004), credit cards
(Rysman 2007), magazines (Kaiser & Wright 2006), and computer industry
(Gawer & Henderson 2007). More examples can be found, for example, in
work by Parker and Van Alstyne (2005). The different contexts are joined by
similar dynamics, including two-sided economics and network effects, which
are crucial for understanding the platform model. These dynamics are dis-
cussed in the following subchapter.
3.1.2 Defining platforms
A platform, or a two-sided market, can be defined in many ways. Table 5
shows definitions judged as the most important based on the literature review.
58
Table 5 Definitions of a platform (i.e., two- or multisided market)
Author(s) Definition
Evans (2003) “[multi-sided] platforms coordinate the demand of distinct groups of customers
who need each other in some way.”
J ullien (2005) “[Two-sided markets are] situations where one or several competing ‘platforms’
provide services that are used by two types of trading partners to interact and
operate an exchange.”
Rochet and
Tirole (2005)
“markets in which one or several platforms enable interactions between end-
users, and try to get the two (or multiple) sides ‘on board’ by appropriately
charging each side […] while attempting to make, or at least not lose, money
overall.”
Armstrong
(2006)
“Many markets involve two groups of agents who interact via ‘platforms,’ where
one group’s bene?t from joining a platform depends on the size of the other
group that joins the platform.”
Evans (2009b) “[Platforms] serve distinct groups of customers who need each other in some
way, and the core business of the two-sided platform is to provide a common
(real or virtual) meeting place and to facilitate interactions between members of
the two distinct customer groups.”
Gawer (2009) “Industry platforms are building blocks […] that act as a foundation upon which
an array of firms (sometimes called a business ecosystem) can develop comple-
mentary products, technologies or services.”
Rysman (2009) “Broadly speaking, a two-sided market is one in which 1) two sets of agents in-
teract through an intermediary or platform, and 2) the decisions of each set of
agents affects the outcomes of the other set of agents, typically through an exter-
nality.”
Hagiu and
Wright (2011)
“an organization that creates value primarily by enabling direct interactions
between two (or more) distinct types of affiliated customers”
First, based on the definitions, some platforms can be described as infra-
structure
28
rather than a market. Markets are what economists consider places
of exchange; that is, where people and companies trade goods and services
(for more definitions, see Diaz Ruiz 2012). Exchange, in other words, is one
form of interaction taking place in a platform of a particular kind (i.e., a mar-
ketplace), but it is not the only form, as will be shown in this study.
Second, a platform can take the shape of a network; consider, for example,
a telephone network or a social network on the Internet. However, it can also
be depicted as a repository of content, from which users retrieve content that
others have contributed. In sum, there are a large number of platforms with
various traits, although they share the same core (i.e., place of interaction
29
).
Third, as can be seen from Table 5, a characteristic commonly associated
with platforms is the presence of so-called network effects (Katz & Shapiro
1985). In a simple form, due to network effects, the more users a platform has,
28
“[A] base of common components around which a company might build a series of related
products” (Cusumano 2010).
29
Moreover, a place indicates a physical or virtual location.
59
the more valuable it becomes. When the platform is subject to direct network
effects
30
, a user’s benefit from utilizing a product increases with the number of
other users of the same kind (Shapiro & Varian 1998). Some physical net-
works, such as railroads or telephone networks, are classic examples of direct
network effects (e.g., Katz and Shapiro 1985). For example, as the railroad
network grows, more destinations become available to passengers. In a similar
vein, the more there are installed telephone connections, the more people one
is able to call. The required network size; that is, an adequate number of users
for a platform to serve its purpose of providing matches, is termed critical
mass.
In addition to direct network effects (i.e., relating to users of the same kind),
a platform can be subject to indirect network effects (i.e., relating to users of
another kind), which are essential for two-sidedness in the platform defini-
tions; that is, there are two distinct groups which influence each other (Rochet
& Tirole 2003). Moreover, an indirect network effect can be positive or nega-
tive, which is strictly a question of perception. Table 6 is based on Shy (2011)
who distinguishes between positive and negative, and direct and indirect di-
mensions.
Table 6 Types of network effects
Direct Indirect
Positive
Positive direct network effects
(e.g., telephone)
Positive indirect network effects
(e.g., auction)
Negative Negative direct network effects
(e.g., spam)
Negative indirect network
effects (e.g., advertisements)
Note that, due to perception, some network effects can be interpreted as
both positive and negative by different people (Shy 2011). For example, some
users might enjoy advertisements, whereas others find them disturbing
31
. Fi-
nally, indirect network effects can be asymmetric, so that one side of an inter-
action appreciates the presence of the other side more than that side appreci-
ates it. For example, if there are only a few buyers in a marketplace, new
buyers are more important to sellers thanvice versa. However, sellers are im-
portant for buyers, as there would be no marketplace without them.
30
Another term, network externalities, is sometimes employed to refer to the same phenomenon. In
this study, they are regarded as interchangeable. This is in line with common terminology in the
literature, while bandwagon effects is also employed. However, strictly speaking, “any economic
effect is an externality only if not internalized” (Farrell & Klemperer 2007, 2021).
31
The perception of advertising is affected by several factors, such as targeting and quality. Thus,
it is a good example of the relative nature of network effects.
60
More precisely, two-sided platforms “coordinate the demands of distinct
groups of customers who are dependent on each other” (Hagiu 2006). J ullien
(2005, 234) defines two-sided markets as “situations where one or several
competing ‘platforms’ provide services that are used by two types of trading
partners to interact and operate an exchange.” Consistent with these defini-
tions, Evans (2003) associates three properties with two-sided platforms: 1)
the presence of two distinct groups; 2) demand coordination benefits, whereby
one group increases the benefits perceived by the second group; and 3) the
necessity for an intermediary to “internalize the externalities”. According to
Evans (2002), it is crucial in two-sided markets to differentiate between price
structures
32
and price levels
33
.
The demand coordination benefits in this definition can be regarded as
equivalent to the concept of network effects
34
. Following Rochet and Tirole
(2003), it can be regarded as typical for platforms to subsidize one group of
users while making profit from the other group. However, because the two
groups experience cross-group linkages, it is not possible to isolate the profits
from the second group without the presence of the first group.
3.1.3 Markets vs. platforms
Hagiu and Wright (2011) state that the platform literature “has constantly
struggled […] with a lack of agreement on a proper definition”, continuing to
state that some authors have implied that retailers, such as grocers, supermar-
kets and department stores would be platforms. Indeed, the basic tenets
35
of
‘two-sidedness’ and ‘interaction between them’ can be satisfied with any mar-
ket, and therefore the platform literature would be no different from the earlier
way of understanding markets.
Relating to this conflict, Rochet & Tirole (2005, 2) refine their original
2003 definition because, based on it, “pretty much any market would be two-
sided, since buyers and sellers need to be brought together for markets to exist
and gains from trade to be realized.” For example, consider a hardware store
that deals with both suppliers and end customers; end customers go there be-
cause the store provides hardware which, in turn, is provided because the store
32
“How to divide the total price [of a transaction] between buyers and sellers” (Evans 2002, 46).
33
“What total price to charge [from] buyers and sellers” (Evans 2002, 46).
34
The concept of network effects differs from economies of scale in that the latter is regarded as a
feature of a single firm, whereas network effects generate benefits for the whole network of firms,
which are compatible with one another (Birke 2008). However, network effects can, in a sense, be
understood as demand- or supply-side economies of scale.
35
Two-sided platforms “serve two types of agents, such that the participation of at least one group
raises the value of participating for the other group” (Li, Liu, & Bandyopadhyay 2010).
61
is frequented by end customers. This and any type of mediated market ex-
change effectively follows the logic of network effects, and will be an “ordi-
nary” market. This definition is also visible in Evans (2009b, 4), who argues
that the fundamental role of platforms is “to enable parties to realize gains
from trade or other interactions by reducing the transactions’ costs of finding
each other and interacting.” In general, such a role can be regarded as being
close to that of a marketplace mediating supply and demand.
How, then, are platforms different from any other market? The platform lit-
erature provides an answer. In their later paper, Rochet and Tirole (2005, 2)
redefine a two-sided market as “one in which the volume of transactions be-
tween end-users depends on the structure and not only on the overall level of
the fees charged by the platform”. Such a particularity does not exist in a one-
sided market. In other words, this definition implies that the price structure
replaces price level as the key focus of interest. For example, one cannot con-
sider how free users are charged in freemium-based
36
online platforms (i.e.,
one side), but has to include paid users (i.e., two sides) to understand the mar-
ket. As there are expected network effects between the two groups,
influencing how price is distributed between them will either increase or
decrease the number of interactions.
This definition also avoids some of the other shortcomings. First, it refers to
users as opposed to trading partners, which is more appropriate for some non-
exchange platforms. In other words, the scope and type of interaction in plat-
forms exceeds the notion of exchange; it can be exchange, but it can also be
something else while still having indirect economic implications. By strict
definition, when the type of interaction moves from economic exchange to
other forms, the platform is no longer a marketplace. For example, it is not fair
to argue that a social network would be a marketplace, because users most of-
ten interact out of non-economic motives.
Are, therefore, all markets platforms? Essentially, “Yes”. As they require
both buyers and sellers to be present, they are two-sided platforms or two-
sided marketplaces. However, not all platforms are marketplaces. A market-
place is defined by exchange while, for example, a content platform host ac-
tivities relating to the content without engaging in exchange with other users.
However, although all platforms do not require economic exchange, they re-
quire some form of interaction. It might be discussion, sharing, content pro-
duction, and consumption, or more; thus, not all platforms are marketplaces.
36
Freemium, a portmanteau of ‘free’ and ‘premium’ (Wilson 2006), refers to one group of users
paying for a Web service and another utilizing it for free.
62
3.1.4 Mediation vs. coordination
Rochet and Tirole’s (2003) price coordination, however, is not foolproof as
their definition does not solve the problem of mediation; that is, how is inter-
action, such as exchange, organized in a platform. For example, consider a
merchant who subsidies some suppliers to sell products cheaper to end users;
here, price structure is affected and also the volume of products is likely to
change as consumers buy more due to low prices.
A satisfactory solution to this issue is offered by Hagiu and Wright (2011)
who distinguish resellers, or classic intermediation, from platforms, so that the
former deals with each market side separately; for example, a hardware store
first negotiates the inventory with suppliers, and then sells it to customers via
its retail locations. Essentially, the platform owner is an enabler of interaction,
but its active participation is not required for the participants to self-organize
interaction, as participants are engaged in direct communication. The situation
of intermediation versus coordination is illustrated in the following figure.
Figure 6 Difference between a reseller and a platform
In both cases, the presence of the other side is beneficial (i.e., there are net-
work effects); however, the coordination structure is different. In regular in-
termediation, the intermediary first creates dyadic relationships with both par-
ties individually, and only then enables the transaction. In a platform, the in-
termediary provides the platform for “open” interaction between the parties.
This has strong implications; for example, relating to customer power (i.e.,
who holds the customer relationship?) and the quality of interaction. As such,
consider how the platform owner is able to filter out negative externalities
(this topic is revisited in Subchapter 4.5.2.).
Regardless of it being in direct control of the interaction, the platform still
has to attract both participating sides (see Chapter 4.4), and enable their
interaction through some medium; for example, a website, bar, or a shopping
mall. Often, it also has to monitor the quality of interaction to prevent negative
Reseller
Supply-side
Platform
Demand-side
Supply-side
Demand-side
63
externalities
37
. The argument presented in Figure 6 is compatible with
Piezunka (2011) who differentiates between intermediation and coordination
as two distinct platform activities. Rysman (2009) follows a similar logic,
noting that in the intermediary model, which he terms a one-sided market, the
reseller takes ownership of the goods, whereas in a two-sided market the plat-
form owner enables transactions between exchange partners.
This distinctive feature has also been noted by other scholars (e.g., Luchetta
2012). Essentially, mediation by the platform is not necessarily of a transac-
tional nature, and the parties interact directly with one another. For example,
users send messages to one another, and the platform merely enables this di-
rect interaction. According to this perspective, shopping malls are platforms
because they enable direct interaction between shopkeepers and shoppers; su-
permarkets are not, because they are reselling the suppliers’ products. The im-
plication is that a shopping mall enables a direct relationship with the cus-
tomer, whereas supermarkets take control of customer relationships; such im-
plications are discussed in Chapter 4.7.
Moreover, platform coordination has some distinctive features. The evolu-
tion from market coordination to intermediation and, finally, to platforms can
be understood with the help of the following figure.
Figure 7 Market coordination and platforms
37
The exception is when the groups are self-coordinating, which follows the ideal UG model.
Side A Side B
Isolated
market
Intermediation
Platform
64
1. In the beginning, there are many isolated marketplaces where interaction
between market actors takes place (i.e., isolated marketplaces phase).
2. Then, there is a mediator that cuts the transaction costs relating to search,
negotiation, and interaction between market actors (i.e., mediated market-
places phase).
3. Finally, there is aunified platform where, again, parties self-coordinate;
however, here, the platform technology provides them with tools to cut trans-
action costs. The unified platform is independent of time and place, unlike
isolated marketplaces.
In this evolutionary perspective, the match-making model moves from dis-
intermediation to mediation to platform. In the first, matches are random and
ad hoc, whereas in a platform they are controlled by the platform technology.
Between these two, the intermediary cuts the number of connections required
from a market actor to successfully transact with another actor; these are clas-
sic intermediation benefits widely known in the strategic management and
marketing literature
38
(see e.g., Bergern Dutta, & Walker 1992; Bakos 1998).
In brief, the platform model enables scaling of market coordination without
losing the intermediation benefits, as self-organization is highly efficient
through the platform technology.
3.1.5 Direct and indirect effects of interaction
Luchetta (2012) employs the aforementioned distinction to divide platforms
into two-sided transaction and two-sided non-transaction markets. Indeed, the
distinction makes intuitive sense because a transaction does not always include
only the two sides; for example, in media markets where buyers of the media
space (advertisers) transact with both the advertising network and the consum-
ers seeing the ads. However, Luchetta (2012, 11) then claims that
"In media markets, the two sides are not necessary, they repre-
sent a business strategy. Television channels are a good exam-
ple: there are channels whose business model is two-sided, that
is based on free content and advertising revenues, alongside of
pay-per-view channels which earn revenues from subscription
fees."
38
If there arex supply agents andy demand agents in an isolated market, the number of potential
connections in an isolated market isx*y. In the intermediation model, they only need to deal with one.
The intermediation benefit is then (x*y)-1 for the whole market, and x-1 or y-1 for the agents,
respectively. The number of potential connections is the same in a platform as in isolated markets, but
it is assumed there exists a factor k, according to which both the overall efficiency and efficiency of an
agent to find matches is better than in the isolated market model.
65
However, this study disagrees with the argument, simply because of net-
work effects; that is, benefits derived by advertisers from displaying ads to
users. The advertising-based platform would not exist without advertising and
therefore, even assuming no direct interaction between advertisers and users
39
,
users derive indirect benefit from the existence of advertisers in the platform.
Moreover, it is not essential to delimit the type of interaction to transacting. In
an online media, interaction between advertisers and users takes place, for ex-
ample, through impressions (i.e., views), clicks, and email subscriptions.
Eventually, users might purchase; however, before that, the advertiser is inter-
ested in conversion-supporting actions in the sales process, as is generally un-
derstood in online marketing theory (Soonsawad 2013). Two sides are there-
fore necessary from the advertiser’s perspective.
However, if we think thoroughly, they are also necessary from the user’s
perspective. Namely, users benefit indirectly from the advertisers’ presence,
even if they might have direct negative “mind harm” from advertising. By de-
limiting ‘interaction’ to only taking place between users and content providers,
one misses the further layer of interaction between advertisers and content
providers. The logic is depicted in the following figure.
Figure 8 Interactions in an advertising-based online platform
39
This is not categorically the case; consider a user clicking an ad on the website, which is a direct
form of interaction. Further, the goal of advertising is to derive deferred interaction benefits “down the
road”; for example, when a user is next changing a car and recalls the banner ad.
User
2nd degree interaction
(primary motiveof
advertiser)
3rd degree interaction
(business logic)
Advertiser
Platform
owner
PLATFORM
(content)
1st degree interaction
(primary motiveof user)
66
Direct interaction is when advertising is shown to the user
40
. A third degree
of interaction is that which takes place between the advertiser and platform
owner
41
, termed business logic. In sum, advertising-based platforms can, ac-
cording to our reasoning, be termed platforms. However, the remark by
Luchetta (2012) is useful in that it recognizes the difference between a market,
as an entity, and a platform as a goal-driven firm. Other notable key distinc-
tions are that the platform interaction is not always exchange but that its goal
can be non-economic gains and that, in a platform triad, the user simultane-
ously might indirectly enjoy the advertiser’s presence, for enabling free con-
tent, and directly dislike advertising. This clearly fortifies the rule of no free
lunch and tends to be generally accepted by users (cf. Pauwels & Weiss 2008).
Therefore, the separating factor can be seen in the direct connection be-
tween users from different sides. In the hardware example, the suppliers have
already delivered products prior to a visit; it would be a platform if users could
see the inventory in advance and order directly from the suppliers. Ordering
from the retailer, as opposed to ordering directly from the supplier via a web-
site, does not constitute a platform but a reseller
42
. Second, the platform differs
from a standard market with intermediaries, in that a standard market advo-
cates ‘one-to-many’ structures (see e.g., Dwyer, Schurr, & Oh 1987), whereby
the market-maker is transacting with suppliers and end users separately at dif-
ferent touch points.
In contrast, in a platform, the parties interact with each other at one touch
point; that is, not with the platform but still via the platform, which is a dis-
tinction from dyadic transactions. In platforms, the major task for the platform
owner is to build liquidity; referring to the number of interactions, sometimes
this might involve creating a critical mass of participants on both sides (Evans
2009a).
This solves the conceptual problem as in the hardware example, buyers and
sellers would not be in direct interaction with one another and, therefore, the
market would not be a platform. The argument is consistent with Hagiu and
Wright's (2011, 2) definition: "[Multisided platforms] enable direct interac-
tions between the multiple customer types which are affiliated to them." As
such, platforms are not intermediaries in the traditional sense; rather, they are
places of interaction, offering facilitating service or features to their users.
40
However, consider the limitations; for example, banner blindness (Benway & Lane 1999) makes
users effectively ignore advertising, in which case the desired interaction benefit does not materialize.
This works both ways; as noted by Evans (2002), the platform is unable to tax transactions beyond the
platform (i.e., the value capture problem), no matter how large the deals made by users and
advertisers.
41
Note that ‘platform owner’ is employed interchangeably with ‘platform sponsor’ in this study.
42
In this type of interaction, the intermediary is directly interfering in transactions between users in
a way other than offering the medium.
67
3.1.6 Networks vs. platforms
How do platforms differ from networks? Structurally, a platform can be un-
derstood as a set of interlinked nodes (e.g., Westland 2010). It is only when we
segment the network users into various groups when the two-sided or multi-
sided dynamics of platforms become relevant. For example, a one-sided plat-
form is the same as any network in terms of network effects: the more there is
any type of users, the more new users are willing to join. The same does not
apply in two- or multisided platforms in which the extant users need to be of a
complementing (i.e., different) type to encourage joining (i.e., positive indirect
network effects). For example, if there are many sellers in a platform but few
buyers, adding more sellers does not increase other seller’s willingness to join;
in fact, it might be reduced by such an increase (i.e., negative direct network
effect). Therefore, while platforms, like almost any social construct, can be
modeled and understood as a network, there are particular dynamics for which
perceiving platform startups as platform startups and not network startups is
meaningful. The analysis is likely to become more useful as a consequence of
such a decision.
3.1.7 Websites vs. platforms
Another defining question is: what differentiates a platform from a regular
website? Following the earlier argumentation, in a normal website, a visitor
does not interact with other visitors. If this is so, then the websiteis a platform.
In practice, the interaction has a purpose, such as engaging with content-re-
lated activities, social interaction, or exchange. As the reader might remember,
this also clarifies why a shopping mall is a platform but a department store is
not. It also marks why research is needed; the nature of interaction taking
place in platforms is believed to be different to that in an intermediary setting.
For example, consider two firms: ActivityGifts
43
and Gidsy
44
. Both ecom-
merce sites sell experiences
45
. In both cases, the end “product” is sold by the
startup and provided by a supplier; however, who coordinates the exchange is
crucially different. In Gidsy, the buyer contacts the service provider directly,
and the website is merely a platform where anyone (i.e., users) can join and
place experiences for sale. In contrast, ActivityGifts first contracts individual
suppliers and then resells their services on the website, taking care of the
43
www.activitygifts.com
44
www.gidsy.com
45
E.g., a tandem jump in Prague.
68
customer interface (i.e., an intermediary model). This has considerable impli-
cations for the two firms; for example, ActivityGifts’ model is much more dif-
ficult to scale up than Gidsy’s model because it requires adding additional
sales personnel to contract suppliers, and customer service personnel to man-
age the customer relationships and support, whereas Gidsy’s model might re-
quire more trust from buyers, as suppliers are not verified, and leave the plat-
form owner vulnerable to direct transacting between buyers and sellers outside
the platform (i.e., the value capture problem). In sum, ActivityGifts is a re-
seller (i.e., a place of buying) and Gidsy is a platform (i.e., a place for interac-
tion that is, in this, case buying)
46
. A website, whether providing products or
not, must enable direct interaction between actors to be definable as a plat-
form.
3.2 Platform definition of this study
A few notions arise from the previous definitions that influence how platforms
are defined in this work. First, “trading partner” (J ullien 2005) is a narrow
conceptualization of the activity taking place in some platforms, especially at
the consumer side. If the basic unit is interaction
47
as opposed to exchange, it
is possible to examine two-sided interactions between users of the same or
different kind, users and platform, and users and advertisers. Instead of trading
partners, therefore, this study mostly refers to ‘users’ as actors in online plat-
forms. However, user is not a synonym for customer, and therefore the defini-
tiontwo groups of customers is different from our purpose; in practice, either
of the groups might be treated differently based on their ability or willingness
to pay (WTP).
Second, there is a problem with perceiving the growth of indirect network
effects as the growth of the number of actors in the complementing part. This
feature is inherited from the early network effects literature focusing on in-
dustry standards (Katz & Shapiro 1985), whereby the number of participants is
restricted, and from network theories such as Metcalfe’s law
48
, which focus on
the growth of an infrastructure such as the Ethernet or telephone networks.
The fallibility of size equals critical mass will be discussed in Subchapter
4.5.2.
46
However, both have the same revenue model: they make sales and retain a commission prior to
forwarding payments to suppliers.
47
Any type of activity necessitating two or more people to take place. Note that ‘interaction’, in a
similar sense, is also employed by Rochet and Tirole (2005).
48
The value of a network increases proportionally ton
2
, whenn is the number of individual inter-
connected nodes.
69
Based on the previous discussion, platforms in this study are defined as
follows:
A platform is a place of interaction among one or many groups
of users whose interaction benefits, known as network effects, the
platform owner aims to increase and monetize.
The ‘place’ attribute is similar to Evans' (2009b) definition of platforms as
physical or virtual meeting places for two distinct groups. The place can be a
marketplace, in which case the interaction is in a form of exchange between
buyers and sellers, or it can be a content platform in which users create content
such as discussions or video. Furthermore, it can be a social network where
users engage in social interaction. What makes Internet platforms interesting,
from both practical and theoretical perspectives, is their immense potential for
scaling (as in: increase in size). For example, Facebook grew from zero to one
billion users in eight years (Shaughnessy 2012). Clearly, this potential opens
new types of business opportunity. Startups are among the first to experiment
with these opportunities in platform markets. Their role is either to create new
platforms or join existing ones as a complement; both strategies are considered
in this study.
Monetization in this definition is similar to “internalizing externalities”
(Evans 2002; Rochet & Tirole 2003), in that the platform extracts rents for the
coordination benefits it provides for its members. As noted, delimiting the
‘two groups’ required for platforms to buyers and sellers (e.g., Li et al. 2010)
is not appropriate in the online context because other divisions are equally rel-
evant, such as, for example, free users and paid users, users and advertisers,
and contributors and consumers of content. Buyers and sellers are associated
with exchange platforms in this study, and other groups under their proprietary
platforms. Therefore, the scope of platforms extends beyond markets, and the
definition by Hagiu and Wright (2011) is the most influential for the definition
of a platform in this study. Note, however, that this does not render the two-
sidedmarket literature obsolete; most of it applies even when the interaction in
the market is other than economic exchange.
Critical to this notion, in the context of online platforms, is that match-
making is performed through programmatic means; that is by algorithms
matching complementing parties (e.g., buyers and sellers; men and women)
based on the criteria and other information they have willingly given, or in-
formation that is retrieved from their behavior or other context to facilitate
match-making. These actions of the platform owner aim to increase the num-
ber of matches and subsequent interaction; outcomes that are associated with
the viability of the platform, both in terms of liquidity and monetary gains.
70
3.3 Typology for online platforms
This section presents the types of platform examined in this study, and relates
them to other platform types studied in the literature. Classification of plat-
form types can be employed to understand their strategic problems. In partic-
ular, different complements and motives of participation are associated with
different platforms. Because complements are associated with the strength of
network effects (Parker & Van Alstyne 2010), and motives with the use of a
platform (Boudreau & Lakhani 2012), recognizing that these vary by the type
of platform might lead to important discoveries for their strategic manage-
ment.
Platform classifications relevant to this study can be divided into a hierar-
chy of three domains:
General platforms ? Technology-enabled platforms ? Online platforms
These domains are now explored in more detail. First, Evans (2003) identi-
fies three types of platform: 1) market makers, 2) audience makers, and 3) de-
mand coordinators. Market makers create an environment for economic ex-
change in which parties are involved in transactions with each other. Market
makers reduce transaction costs relating to searching for and negotiating with
trading partners (Evans 2003). Second, audience makers create matches be-
tween advertisers and end users to enable advertisers to send messages to their
desired target audience that comprises users of the platform. The platform
owner needs to determine how much advertising is allowed in the platform,
especially when it interrupts the consumers’ usage (Rysman 2009). Third, de-
mand coordinators enable members to interact by providing services in the
background, such as operating systems and payment cards (Evans 2003).
Gawer (2009) divides platforms into 1) internal platforms, addressing one
firm’s offerings (e.g., Sony Walkman); 2) supply-chain platforms, addressing
chain-wide execution (e.g., the Renault-Nissan alliance); and 3) industry plat-
forms where the focus is on industry-level coordination (e.g., Microsoft Win-
dows). A similar typology is presented by Piezunka (2011) who distinguishes
that streams of the literature tend to focus on 1) product platforms; 2) industry
platforms; and 3) two-sided markets. The difference is marked by the role of
the platform sponsor: in the first, it offers products; in the second, it coordi-
nates complements and end-users who might directly interact outside the plat-
form; and in the third, it coordinates two sides interacting in a platform.
One of the most cited general platform classifications is that by
Schmalensee and Evans (2007). It includes 1) exchanges, 2) advertiser-sup-
ported platforms, 3) transaction platforms; and 4) software platforms.
71
Exchanges include coordinating dyadic interactions between buyers and
sellers; any stock exchange would qualify. On the Web, for example, auction
websites such as eBay are included. Examples of advertiser-supported
platforms include TV or radio stations that show content for free and monetize
by selling advertising space. Online equivalents are, for example, search
engines such as Google (see Salminen 2010). Transaction platforms, such as
credit card providers, mediate transactions between merchants and consumers;
PayPal is a good example on the Internet. Software platforms are tied to
specific hardware, and sometimes referred to with the concept of ecosystem;
for example, mobile application platforms such as Nokia’s Ovi Store. Huang,
Ceccagnoli, Forman, and Wu (2009) define ecosystems as communities of
innovation networks in which industry leaders coordinate collective efforts of
developers and other partners towards shared goals.
Second, technology-enabled platforms mentioned by Saha, Mantena, and
Tilson (2012) include computer and mobile operating systems, online advertis-
ing networks, job boards, real estate brokers, electronic marketplaces, and
payment systems, both mobile and online, such as Paypal. Technology plays a
role in facilitating connections between supply and demand sides, and also
computes optimal routes and allocations between parties, a property that is
useful when mapping potential connections. Internet platforms are built on top
of Web technologies that are typically based on open standards, thus excluding
competitive strategies based on patents and standards, and enabling social
connections and scaling effects associated with technology (for illustration,
see Horowitz & Kamvar 2010).
Third, online platforms are a subset of technological platforms; they can
also be termed Internet platforms (Sawhney et al. 2005) due to the fact that the
Internet is the medium through which participants interact. Consequently,
Web technologies play a major role in how match-making is executed by the
platform owner, and also in how the platform scales; for example, very rapidly
extending across national boundaries. Some authors also employ the term
electronic intermediary (Bakos 1998). However, intermediation refers to a
value chain deviating from the market-making function; for example, enabling
parties to independently find one another (Evans 2003). Gazé and Vaubourg
(2011) distinguish online auctions, traveling intermediaries, online media,
massive multiplayer online role-playing games, and e-payment platforms.
Saha et al. (2012) mention the following online platforms: 1) electronic la-
bor markets, 2) ecommerce sites, 3) online advertising platforms, 4) online
auctions, and 5) group-buying platforms. Caillaud and J ullien (2003) point out
that online platforms are able to monitor individual transactions, and therefore
charge transaction fees tied to the number of interactions, not only access fees.
Gazé and Vaubourg (2011) discuss another trait they perceive as typical for
72
online marketplaces: side-switching, which is changing roles from buyer to
seller as it is easy, reversible, and has no financial cost. However, it is unclear
if this feature only applies to online markets as, also in brick-and-mortar cases,
one can act as a seller and buyer within the same market space (consider e.g., a
flea market).
Conceptual classifications can be regarded as arbitrary
49
because typologies
can employ alternative criteria while classifying the same phenomenon; nei-
ther being wrong nor correct (Kotha & Vadlamani 1995). For example,
Schmalensee and Evans (2007) employ criteria such as the form of product
(i.e., software platform), business model (i.e., advertiser-supported platform),
and the operating level (i.e., transaction platform, which can be regarded as a
form of infrastructure). Thus, we can posit that it is difficult to define plat-
forms in a mutually exclusive way as they might be embedded, so that, for
example, a software platform provides advertiser-supported products.
There are some reasons why the extant classifications are insufficient in the
context of this study. First, although we can detach the embedded platform
from its parent and examine it in isolation, depending on whether we are inter-
ested in the infrastructure level, business model, or type of interactions taking
place within it, based on particularities in these dimensions, it can be argued
that online platforms merit their own classification. For example, software
platforms (Schmalensee & Evans 2007) would combine hardware and soft-
ware, which is not relevant in pure online business
50
. This liaison arises from
the hardware-software paradigm introduced by Katz & Shapiro (1985), and
relates to the complementarity of the two; hardware being more valuable ac-
companied by useful software, whereas, clearly, software cannot be run with-
out hardware. The result is a different kind of chicken-and-egg problem than
the one analyzed in this study, and is often implied when explaining computer
industry dynamics
51
(e.g., Boudreau & Hagiu 2009).
Second, many classifications employ a revenue model as the defining fac-
tor. However, the revenue model only explicates how the platform generates
revenue (e.g., by advertising); this information is not very relevant for solving
the chicken-and-egg problem. Much more central, in this respect, is examining
why users join and participate in platform interaction, and even pay for access
49
In the sense that, although consistent mutually exclusive items, classifications can be equally
valid but different.
50
However, as previously noted, similar dynamics to other two-sided markets can be assumed;
thus, the platform literature is highly relevant. Furthermore, some strategies applied in the mobile
application markets by key players are associated with the fact they are also hardware vendors.
Therefore, even if the application marketplaces can be modeled in isolation as their own two-sided
markets, in some cases, theories might want to consider the hardware liaison.
51
Clearly, the platform perspective would explain the difference between rivals through the
concept of network effects: one is able to leverage them while the other is not.
73
and usage. For this purpose, the classification presented next considers poten-
tial motives of users of online platforms.
In an attempt to provide a specific classification for online platforms, com-
bining both the digital environment and the two-sided structure, the following
classification is proposed. Overall, it is based on the grounded theory analysis
and thus on the nature of the startups examined in this study.
Table 7 Online platform types
Type Sides Focus of interaction Industry example
Exchange platform Buyers and sellers Exchange motive / transac-
tions
eBay
Content platform Creators and con-
sumers of content
Content creation and con-
sumption
Social platform (one-sided) Social motive Facebook
Infrastructure Providers and devel-
opers
Enable other products and
services
The Internet
This classification matches well with the nature of startups sampled in this
study (see Table 2). Note that the infrastructure platform is a special case
which involves no interaction between parties. Infrastructure is compatible
with Gawer and Henderson’s (2007) definition
52
of a platform as a structure to
build on top of, but not with Hagiu and Wright’s (2011). In this study, interac-
tion between parties involved in the platform is regarded as more important
than hardware-level interaction, and thus the infrastructure model will not be
considered in later parts of this study.
The exchange platform connecting buyers and sellers is the most widely
documented case in the platform literature, and most authors refer to it when
discussing platforms. Based on the literature review, especially in the field of
economics, the focus is on marketplaces (i.e., exchange platforms). Some au-
thors see a platform in the sense of infrastructure (i.e., enabling to be built
upon), which is not the most fruitful approach when considering how the plat-
form owner can solve business problems as a strategic agent.
There can be some overlap between the types. For example, Kim and Tse
(2011) analyze knowledge-sharing platforms, which would classify either as
52
“We define a product as a ‘platform’ when it is one component or subsystem of an evolving
technological system, when it is strongly functionally interdependent with most of the other
components of this system, and when end user demand is for the overall system, so that there is no
demand for components when they are isolated from the overall system” (Gawer & Henderson 2007,
54).
74
content or exchange platforms depending on the type of interaction. If re-
spondents are provided payments, not “payments in-kind” as in Mungamuru
and Weis (2008), the interaction is exchange, and therefore the “laws of ex-
change” should apply. If, however, interaction is voluntary and driven by in-
trinsic motivation such as the status of knowing a lot, it is a content platform
and visitors are interested in receiving content benefits. Furthermore, if there
is a relatively stable community and users engage in lengthy discussions and
roles, the platform can be classified as a social platform (Mital & Sarkar
2011).
The overlap problem can be solved by splitting the user motives into pri-
mary and secondary motives. For example, a person searching the Web to find
information has a primary ‘search intent’ (cf. Schlosser, White, & Lloyd
2006), even though he/she might end up sharing the results of the search in a
social platform. Therefore, even when they relate to content, primary motiva-
tions to visit a social platform can be of a social kind. For example, sharing
content is arguably more about sharing (i.e., ‘social intent’) and less about
content. As the purpose of this platform is to provide content, most people
visit it because of that. However, some people might visit it for the primary
purpose of sharing, or some other motive for social interaction. Thus, in our
typology, exchanging content is a spillover effect, not the primary motive for
participating in the platform.
This discussion is not merely semantic. In some cases, the secondary mo-
tive can become an even more powerful predictor of conversion than the ac-
tual motive of visiting the platform. For example, Oestreicher-Singer and
Zalmanson (2009, 39) observe that “in the context of music content, commu-
nity activity is more strongly associated with the likelihood of subscription
than is the music consumption itself.” Furthermore, the interplay between
content and social features can be a critical part in finding solutions to thecold
start dilemma through spillover effects (see Chapter 4.4) and remora’s curse
(Chapter 4.7)
53
. If the motive for participation is known, the platform owner
can provide appropriate incentives, including implicit incentives (e.g., social
satisfaction) and explicit incentives (e.g., monetary compensation), or a com-
bination of both (Ren, Park, & van der Schaar 2011). A place of exchange can
be regarded as being subject to different rules and “economic laws” (i.e., type
of reciprocity associated with interaction
54
) than a content platform.
Therefore, motives for interaction differ in content platforms and social
platforms. Different motivations are also applied in Luchetta's (2012) typology
53
Throughenvelopment (Eisenmann et al., 2011) or embedded platforms, in which a platform of a
different type is built on top of the host to gain access to its user base.
54
For example, Porter (2004) argues that “[r]elationships in networked-based communities are
often of short duration and driven by utilitarian needs.”
75
of platforms. By adopting his style of presentation, online platforms can be
specified as follows.
Table 8 Online platforms, interaction, and goals
Platform Interaction User A Goal of A User B Goal of B
Content Consumption Consumer To consume
content
Contributor To contribute
content
Social Communication Individual A To connect
with B
Individual B To connect
with A
Exchange Transaction Buyer To buy Seller To sell
It follows that the purpose for interaction is symmetric for social platform
users, and the resulting problems are coordinating problems (e.g., who is
online) that the platform will efficiently solve (e.g., by storing messages). The
same applies to an exchange platform; the needs are symmetric, although in-
versely, so that they complement each other. However, participants in a plat-
form relying on user-generated content face some asymmetry; there are users
interested in consuming content and others who produce it. This suggests
some dynamics that will be considered in Chapter 4.4; namely, the goals of the
two sides cannot always be solved by simple match-making.
In sum, exchange platforms address buyers and sellers whose primary in-
teraction is a transaction (i.e., trade). Content platforms comprise users con-
tributing content and users consuming it
55
, and content-related activities (e.g.,
consumption, reading, writing, and watching) are the type of interactions for
which the platform exists. Social interaction (i.e., discussions, chats, messag-
ing, and communication) defines social platforms and their users. A social
platform is one-sided when the user is only interested in others sharing his/her
traits or interests, whereas, in a dual-sided social platform, the user seeks a
complementing party.
55
Whether they are seen as one heterogeneous group or two distinct groups (i.e. one- or two-sided
platform) is another arbitrary choice enabled by flexibility of the two-sided framework.
76
3.4 Online platforms and user generation
3.4.1 Why is UG included in the study?
In this chapter, the author outlines an ideal user generation model to capture
the potential benefits of user generation (UG) targeted by startup founders.
Wishful UG effects are characteristic to the startups in the sample, and influ-
ence the emergence of strategic problems and their solutions. User generation
in the theoretical treatment enhances the outcome more than omitting this
critical aspect, which is why UG is considered a part of the startups’ business
logic.
The lack of desired behavior from users is, based on analyzing the post-
mortem stories, associated with the failure outcome, and therefore needs to be
considered in this substantive theory of strategic problems. The online plat-
form typology and the ideal UG model are Internet-based specificities, and
considering them deepens the perspective on platform strategies in the online
business context.
3.4.2 User-generated content
In the literature, UG effects are often approached through the concept of user-
generated content (UGC), also commonly utilized by startup founders. A
widely accepted definition (e.g., Hermida & Thurman 2008; Ochoa & Duval
2008; Banks & Deuze 2009) can be found in Wunsch-Vincent and Vickery
(2007), defining UGC as a) content made available through the Internet, b)
involving creativity by end-users of a Web service, and c) non-commercial
motives of creation.
Earlier papers relating two-sided markets/platforms with UGC include
Albuquerque, Pavlidis, Chatow, Chen, and J amal (2012), who compare firm-
initiated promotion with user-initiated promotional activities, and Evans
(2009b), Ren et al. (2011), and Calvano and J ullien (2012) who consider UGC
as a means to connect consumers and advertisers. Yoo (2010) also mentions
UGC in his treatise on network providers and two-sided markets, but only to
differentiate it from peer-to-peer networks. Kim and Tse (2011, 42) give an
example of UG effects in a platform context:
"[A] knowledge-sharing platform may overcome an initially
small membership by starting with some prebuilt knowledge da-
tabase to attract new questioner members. New questions posted
by these questioners will attract some answerers, whose answers
will in turn attract more questioners. This cyclic process can
77
help a knowledge-sharing platform to overcome the chicken-
and-egg problem that is commonly found in a general two-sided
market."
However, none of these papers consider the theoretical implications of UG
in online platforms, or fully extend its meaning beyond content-related activi-
ties. First, considering content only as content, as opposed to a complement,
does not capture its implications for online startups. Following Chapter 3.1,
apart from installed user base, complements are associated with network ef-
fects; thus, the more content in a content platform, the more valuable it is to its
users, all else being equal (cf. Varian 2003). At a general level, the same ap-
plies to interaction: its volume positively correlates with the usefulness of a
platform (see Chapter 4.5).
Second, the functions of UG are not limited to user acquisition as is typi-
cally considered (e.g., Kim & Tse 2011). In contrast, the users are seen to
adopt more roles that influence the viability, growth, and success of a plat-
form. These functions are discussed in later sections. Third, even if Web 2.0
marked a “revolutionary” disruption of the earlier “version” of the Web (see
O’Reilly 2005), the notion of UG has not developed much since. For example,
Beuscart and Mellet (2009) note that Web 2.0 is associated with UGC sites,
blogs, social news sites, and social network sites. UG, therefore, is commonly
perceived as “video and photo sharing” or similar activities, the meaning of
which to the platform is not regarded as critical. However, the importance of
UG seems obvious when observing the scale and diffusion of online platforms
in the real world. For example, consider Facebook, originating from a small
community and, a few years later, having over one billion registered users
(Shaughnessy 2012)
56
. Such results were obtained without advertising, very
minimal user support and staff-per-user ratio (Kirkpatrick 2010), instead rely-
ing on UG effects.
3.4.3 UG in online platforms
When, therefore, considering UG not from a static perspective but as a com-
plement, a new definition is needed. As network effects are highly important
for two-sided markets, a definition for UG relating to platforms is proposed:
User generation is provision of content or other direct or indi-
rect benefit through actions of a user of an online platform to
56
Edwards (2013) notes, correctly, that a portion of this includes fake profiles and double
registrations; nevertheless, the growth is phenomenal.
78
others users characterized by the platform owner’s attempt to
monetize it.
This definition is compatible with the two-sided framework in the following
ways. First, it states that users provide benefit for other users, thus implying
network effects (e.g., Katz & Shapiro 1985). Second, the benefit can be either
direct or indirect (Shy 2011), based on the type of interaction (e.g., communi-
cation vs. trade) and the mechanism of benefit provision; for example, peer-
marketing can increase the installed user base, which indirectly benefits other
users. It can be seen that members of the same sub-group or dyad of a platform
benefit from direct interaction, whereas participants included in another group,
such as advertisers, the platform owner, and other potential stakeholders,
might derive indirect benefit from the existence of the user base (cf. Clements
2004). Third, it includes the aspect of the platform owner aiming to monetize
the UG outcomes, which is compatible with Rochet and Tirole’s (2005) and
Evans’ (2003) perceptions on a commercial platform and the “internalizing of
externalities”.
After establishing this definition, the totality of UG effects, which are no
longer limited to content, is explored. In this conceptual inquiry, suitable theo-
retical frameworks are applied, reaching beyond the immediate platform liter-
ature. This is necessary to base our theoretical argument on the appropriate
literature. As noted, the platform literature is lacking in this aspect, as it does
not consider UG effects with the same gravity that, as the analysis revealed, is
shown by platform startups. Based on the grounded theory (GT) analysis,
founders implicitly assume these UG benefits in their platform strategy.
3.4.4 Ideal user-generation model
User generation was an essential concept for the studied online platforms, and
its emergence can be employed to explain inadequate strategic responses by
founders. To understand the logic applied by founders, we generate a compre-
hensible model going beyond what is understood in prior platform research as
thepotential benefits of UG (i.e., UG effects).
‘Ideal’ UG refers to properties associated with UG in the online context.
Namely, it is the optimal model at which startups aim, but which most will not
reach, as proven by the sample. Central to this idea is that UG aims to replace
functions typically assumed by the firm, and delegates them to the user base.
While the full treatise of this notion goes beyond the scope of this study, the
79
user base is, in the ideal sense, utilized to extend the “boundaries of the firm”
57
(Coase 1937). Namely, startups applying this implicit and theoretical model
aim at 1) minimizing human intervention and labor cost through designing a
technological solution (i.e., the platform); 2) facilitating the creation of
socially desirable high-quality content, or activating a virtuous cycle of
network effects; 3) scaling their technology and business beyond the limits of
the startup’s internal resources with the help of platform users; and 4), above
all, delegating critical business functions to the user base.
The following figure depicts the ideal model of UG effects.
Figure 9 Ideal user generation model
The ideal UG model is associated with network effects, so that the content
actually forms a complement in the platform, making it more valuable to all
users (Arroyo-Barrigüete, Ernst, López-Sánchez, & Orero-Giménez 2010). In
other words, the first set of users’ actions is expected to initiate the virtuous
cycle. More precisely, in the ideal UG model, it is implied that the user inter-
nalizes some of the critical tasks of the startup that, in turn, by externalizing
them, will reach better operational efficiency. The user base is, in theory, able
to produce effects several magnitudes larger than the startup’s resources would
enable; thus, it can be regarded as the startup’s resource. This can be proven
by assuming that the cost to the platform owner to create content, marketing,
customer service, and other functional activities necessary to launch and
maintain the platform increases with, for example, labor costs and marketing
investments, and thus has a realistic limit (i.e., budget), whereas similar
57
It goes beyond the study’s scope as the study cannot prove that firms/startups systematically
delegate their critical functions to users, which is why the model is ideal or theoretical. However,
startups that apply state-of-the-art methods to survive seem to indicate this possibility. This is
definitely a topic for further research.
UGC
content
e-WOM
new users
P2P support
support
Social indicators
conversions/ revenue
USER
GENERATION
80
activities without transaction cost due to platform coordination, self-
organization, or labor cost are effectively cost-free for the platform owner.
Therefore, harnessing UG and network effects leads to a virtuous cycle in
which the willingness to adopt increases as a function of earlier accepted in-
vitations, which then increases the willingness, for example, to send more in-
vitations (Trusov, Bucklin, & Pauwels 2008). This is compatible with network
effects as defined by Farrell and Klemperer (2007, 2007): “there are network
effects if one agent’s adoption of a good (a) benefits other adopters of the
good (i.e., a ‘total effect’) and (b) increases others’ incentives to adopt it (i.e.,
a ‘marginal effect’).” Through UGC, the startup aims to increase the ‘total
effect’, while user-generated user acquisition is planned to increase the ‘mar-
ginal effect’. Given that the diffusion/propagation has an upper boundary on
the limits of a market (Salminen & Hytönen 2012), the outcome can become a
winner-takes-all situation (Noe & Parker 2005) in which the platform replaces
its rivals as a side-effect of expansion (cf. Facebook replacing MySpace). This
effect is noted by Arroyo-Barrigüete et al. (2010, 643), who describe it as a
“re-alimentation schema that makes strong products ever stronger (virtuous
circle) and weak products ever weaker (vicious circle).” This exponential in-
flationary trait can be explained by the "small world" characteristics of the
Internet that hosts online platforms (Schnettler 2009), and the desire to propa-
gate messages, such as invitations and content. Zhang and Zhu (2011) find that
social effects lead to an increase in contribution when the installed base of
Wikipedia increases, therefore supporting UG’s virtuous cycle.
The ideal UG model is grounded in the data; namely, in the assumptions of
the founders. They implicitly and explicitly devise their businesses to support
the idea of UG as this is perceived to be the interaction taking place in the
platform. Indeed, the concepts of interaction and UG are closely associated
and refer to the same phenomenon, which is the activity taking place in the
platform. Note that the ideal UG model differs greatly from what is
understood in the literature by “user-generated content”. Namely, the literature
examines UG almost proprietarily as a function of content creation, whereas
this study’s definition, derived from the two-sided platform literature, extends
far beyond mere content.
3.4.5 Functional view to UG
To provide a more granular perspective to the idea, general functions of a firm
are related to UG. The following list parallels an organizational structure (i.e.,
the firm) in which functions mostly have a clear purpose (Mathur 1979). The
analogy becomes even more distinct when presenting a functional comparison.
81
Table 9 Functional comparison of users and the firm
User function Firm function
Content creation Content production
Moderation Quality control
User acquisition Marketing
Support Customer service
Feedback Market research
The paralleled functions are general functions in which many firms, in-
cluding platforms, need to engage. A platform needs to provide content or li-
quidity (Evans 2009a); it needs to moderate the quality of UGC or other com-
plements for spam (Moh & Murmann 2010), or low-quality complementors
(Boudreau & Hagiu 2009); it needs to provide support/customer service for its
users/customers (Rochet & Tirole 2003); it should take user requests into con-
sideration when modifying platform design (Stanoevska-Slabeva 2002) and,
thus, spontaneous feedback by the community or platform users is a form of
market research; and, most importantly, it needs to acquire users/customers,
that is, conduct marketing (Eisenmann, Parker, & Van Alstyne 2006). These
are general functions that need to be organized in one way or another; accord-
ing to the UG model, users are given most if not all of these tasks.
As explicated, users obtain control over content generation that enables not
only cost-free production, from the startup’s perspective, but a quality that
matches the audience’s tastes
58
. To ensure quality, users monitor each other
and report negative behavior. By automating the system to respond to user re-
ports, the startup avoids any labor relating to quality control
59
. Moreover, in a
platform demonstrating community traits, users face social disproval from
other users as a consequence of misconduct, which they are therefore likely to
avoid (Sheridan 2011). In exchange, users help each other to learn the plat-
form, and also mediate commonly agreed rules of behavior. Support can arise
from earlier platform-specific investments by other users who behave in an
altruistic way (Boudreau & Lakhani 2012).
Arroyo-Barrigüete et al. (2010, 644) refer to the “learning network effect”,
which “derives from the fact that an increase in network size will increase the
number of users with specific knowledge of the related technology.” Essen-
tially, expert users provide a form of “after-sales service” to new users,
58
Kim and Tse (2011, 41) describe this effect: “[a]s a result of knowledge sharing between
members, knowledge-sharing Web sites have an accumulated knowledge database of answered
questions that attracts people who have questions.”
59
In its simplest form, there are report functions. It is more complex when the user becomes a
moderator; that is, an active agent who scouts the platform for low-quality interaction.
82
thereby increasing the platform diffusion. According to this argument, the
presence of peer support can facilitate bothex ante adoption andex post inter-
action.
User-generated invitations, a form of peer marketing (see e.g., Smith,
Menon, & Sivakumar 2005) are, in fact, also beneficial for other users, given
they match their preferences; if not, they will be interpreted as unsolicited
messages (i.e., spam). This is because they reduce the recipients search cost
for interesting content. Therefore, sharing links of content among peers is an
efficient dissemination mechanism and, in theory, resolves the need for any
other marketing
60
, which will be revisited in Chapter 4. It is widely acknowl-
edged that peer-to-peer propagation plays a critical role in the diffusion of
most online platforms that are currently dominant, and that this effect relates
to UG (Albuquerque et al. 2012, 406):
"[C]ontent creators, besides populating the platform with mate-
rials, serve as marketing agents by advertising their own content
or generating referrals and links to uploaded content in other
websites. Given the interconnectedness and viral community
structure of the Internet, the relation between marketing activi-
ties by the firm and the decisions of content creators is likely to
play an essential role in the development of most user-generated
content platforms."
Moreover, consider the customer acquisition function that, as per the ideal
UG model, relies at least partly on search engines providing a marketing chan-
nel by automatically indexing content, and thereby providing free organic traf-
fic (i.e., visitors) to the website, a process here termedsearch-engine external-
ity
61
. Host platforms can provide a stream of users; that is, act as marketing
channels based on actions of the startup and in respect to the platform type
(e.g., content platforms ? Google traffic; social platforms ? Facebook traf-
fic). This strategy is similar to envelopment (Eisenmann et al., 2011), and its
merits and risks are discussed in Chapter 4.7.
60
Interaction that provides direct network value is described by Oestreicher-Singer and Zalmanson
(2009, 14): “small acts of structured contribution that can be perceived as adding value to the user’s
own content consumption but that can also add value to the community […] for example, tagging
content with keywords to ease its discovery, or rating content in order to promote its popularity and
reputation.”
61
This is an externality as it is not a reason for interaction between content consumers and content
creators. For the startup, it represents an externality that can be internalized.
83
3.4.6 Implications to startups
What does the ideal model imply for a platform startup? First, the immediate
costs of providing the platform are radically reduced. Second, effective emer-
gence of UG implies self-organization, similar to projects such as Wikipedia
(see Stvilia, Twidale, Smith, & Gasser 2008) and Linux (Benkler 2002). Third,
transaction costs relating to coordination of the business are externalized,
meaning that the users will begin to coordinatefor the firm (Hagiu 2006).
In particular, UG can reduce the cost of scaling; namely, increasing the
connections and activity in a platform. Imagine the platform owner’s cost
structure comprising user acquisition, denoted a, platform maintenance, de-
notedm, and user support (i.e., customer service), denoteds. To grow the user
base, the platform owner incurs the cost of a multiplied by each acquired user.
If the user base grows by a factor of k, the support cost also grows by this
factor. However, if the user acquisition and support functions are performed
by the current user base at their own expense (i.e., time), only the fixed
62
cost
m remains for the platform owner. By applying UG, the platform owner is able
to maintain itsneutrality, thus avoiding costly marketing and support activities
while monetizing the increased usage of the platform driven by UG’s expo-
nential dynamics and network effects. For example, a content platform exists
to produce and disseminate content to other users. In a regular website, the
owner creates the content; in a platform, it originates from the users. The
startup as a platform owner is not a ‘side’ of the market but still benefits from
the content, as it helps to attract more users on which the platform owner can
capitalize (i.e., monetize).
As previously described, self-moderating effects reduce a startup’s work-
load when scaling to millions of content units. By reducing the transaction
costs of its members, a platform is able to attract users (Hagiu 2006). This fact
has been established in the platform literature. However, less attention have
been paid to other costs. In particular, introducing UG reveals its importance
for online platforms, but also makes it easier to understand the founders’ logic
in pursuing UG in their platform strategies. Hagiu (2006, 2) claims that “any
MSP [multisided platform] performs one or both among two fundamental
functions: reducing search costs and reducing shared transaction costs among
its multiple sides.” By combining parties of interaction, the platform is able to
lower the cost of finding matches, negotiating, and validating their quality. In
a UG platform, these activities originate from the user base. Search costs for
users are lowered because they are able to find new interesting content or
62
In this example, platform maintenance cost is assumed to be fixed; for example, startups
applying cloud-based hosting can convert their server costs from fixed to variable.
84
connections, and the benefit is sustainable because the platform keeps auto-up-
dating as a consequence of other users’ actions. Further, the platform does not
need to invest in advertising because users promote the service to other users.
Finally, startups can integrate mechanisms into their platform design to
nurture UG activities, such as 1) community building, 2) viral mechanisms, 3)
search-engine externalities, and 4) frictionless sharing (Darwell 2013). If these
efforts are either built-in to the product or come at minimal cost by inviting
friends and sending emails to gather the critical mass, then it is assumed that
marketing is free and the customer acquisition cost equals zero. These kinds of
economy makes it possible to replace paid match-making services. Moreover,
indirect monetization might be a requisite to 1) maintain social norms, not
economic norms (Fehr, Kirchler, Weichbold, & Gächter 1998), and 2) encour-
age reciprocity; as users get a free platform, they might feel the need to con-
tribute or “pay” through UG. Susarla, Oh, and Tan (2012) studied diffusion in
a UGC platform (i.e., YouTube) and concluded that diffusion is influenced by
social contagion rather than user heterogeneity. Clearly, user-to-user dynamics
play a role in the adoption/diffusion of a platform, and therefore should be
considered in relation to the chicken-and-egg problem.
3.4.7 Limitations of UG
In sum, within the online platform, users are expected to engage in multiple
roles such as content creation, moderating (i.e., quality control), customer ac-
quisition (i.e., inviting other users), and providing support (i.e., customer ser-
vice). As a result, the required input from the platform owner is, in theory,
greatly reduced through coordination features programmed into the platform.
Indeed, in an ideal situation, the platform becomes self-sustaining and self-
propagating, while the platform owner is still able to internalize some of the
benefits from user interaction, typically indirectly (e.g., by selling data or ad-
vertising space), whether the interaction is content creation and consumption,
exchange, or social interaction.
Clearly, this is where the problems begin, as the ideal UG model often re-
mains just that – ideal. The rest of this study, especially the section discussing
dilemmas, demonstrates some of the central challenges in applying UG as a
critical part of a platform startup’s business model. The ideal model rarely, or
almost never, materializes
63
, and in practice the platform owner’s intervention
63
The examples provided earlier suffer from survivorship bias, which means that they are not
adequate for generalizing empirical meaning. However, nothing prevents taking them as part of a
theoretical model, given that the limitations are properly understood.
85
is needed in most parts to guarantee progress and frictionless operation. How-
ever, the few examples, in addition to assumptions of transaction-costless
functioning, of the ideal platform are compelling thoughts. Therefore, alt-
hough it might seem counterintuitive with hindsight, ignorance concerning
marketing
64
as exhibited by several founders in the sample, and reported after-
wards
65
, has logic; had users accepted those critical functions, the startup
might have succeeded. The ideas that the users replace marketing and that
search costs are categorically low are debunked in Chapter 4.5 and Subchapter
4.6.2 respectively; “no need for marketing” is challenged in Chapter 5; and the
negative effects of removing quality control from the platform owner are
discussed in Subchapter 4.5.2. Therefore, despite its theoretical merits, the
ideal UG becomes dangerous when applied literally.
Overall, the ideal model leaves many aspects uncovered and relies on unre-
alistic assumptions of users’ willingness and ability to manage critical func-
tions of the firm. In its pure form, the winner-takes-all outcome can also be
unrealistic, as it fails to consider multihoming behavior and other aspects of
adoption apart from network effects, such as differentiation through features
or marketing. Nevertheless, it helps us to understand how platform startups ap-
proach strategic thinking. Understanding UG is important as it is associated
with startups’ strategic decision-making. For example, consider Facebook,
Twitter, and Google that have accumulated users not by paying but by peer
effects. These success stories formreference points that orientate founders
towards choosing similar elements in their own startup
66
. This effect is close
to an anchoring bias (Bunn 1975) according to which decisions are made
based on prominence and not averages or suitability of the case context.
Finally, some UGC-based companies (e.g., Twitter) are struggling with
monetization, which indicates that even a working UG model might not be
sufficient to guarantee economic sustainability. As such, popularity does not
indicate profitability (see Chapter 4.6), and solving the chicken-and-egg prob-
lem through UG might leave a firm vulnerable to other strategic dilemmas.
With this premise in mind, the study will now move to discuss the strategic
problems of platform startups in more detail.
64
“The next issue to tackle was marketing. How do we make them aware of it? We decided to use
blogs. What better way to expand than to piggyback on an existing network? […] It rarely works.
Everyone wants to do it, but it isn’t easy to get bloggers to write about something.” (May 2007).
65
“We fell into the ‘build it and they will come’ school of thought (although even when they came,
we still weren’t in good shape).” (The Chubby Team 2010).
66
“When it came to certain website design elements, we didn’t know what customers wanted […]
and so instead, we thought ‘let’s take elements from sites we like and tweak them’ and we’ll get the
same magical effects on our site that they’ve gotten. Wrong. Features don’t work in a vacuum. They
work because you take time to understand your customer and then build features to accommodate
them.” (The Chubby Team 2010).
87
4 STARTUP DILEMMAS
4.1 Introduction to dilemmas
4.1.1 What is meant by dilemmas?
A ‘dilemma’ is a situation of conflict, in which a decision maker usually faces
two mutually exclusive choices that both lead to a seemingly undesirable out-
come. In the Oxford Dictionary (2013), dilemma is defined as “a situation in
which a difficult choice has to be made between two or more alternatives, es-
pecially ones that are equally undesirable.” Although, in everyday life, indi-
viduals often face contradictory decision-making situations, researchers in ac-
ademia tend to model decisions through preferences and weights; thus, out-
comes that are perceived more costly are avoided while those with higher ex-
pected gains will be sought (e.g., Layard, Layard, & Glaister 1994). In
psychology, one speaks of cognitive dissonance, a state of contradictory emo-
tions relating to situations, persons, or outcomes (Festinger 1962).
In economics, scholars examine various tradeoff situations (e.g., Ball et al.
1988; Cohen & Klepper 1992). In particular, economists apply game theory to
examine actors’ strategic choices under a set of assumptions; for example, the
prisoner’s dilemma (see Axelrod 2006) is a famous game-theoretic problem
that is structurally close to the cold start dilemma, presented in Chapter 4.4, in
that participants are driven to a dissatisfactory solution. In the strategic man-
agement literature, strategic or wicked problems (Mason & Mitroff 1981) are
characterized by associations with other problems, recursive feedback, envi-
ronmental uncertainty, ambiguity in definition, conflicting tradeoff in their
solutions, and societal constraints upon theoretically effective solutions (Lyles
& Howard 1988). Proper definition of a problem and assessment of the strate-
gic situation are regarded as important for finding a solution (Klein 2012), alt-
hough individuals are perceived to be constrained by their cognitive capabili-
ties andbounded rationality (Simon 1956).
4.1.2 The use of dilemmas in this study
Here, the concept of dilemma will be applied to examine various challenges to
platform startups. More precisely, four startup dilemmas are conceptualized in
88
this study that derive from the material through the inductive grounded theory
approach, and formed through discussions with founders (i.e., contextualiza-
tion), and the support from the theoretical framework (see Chapter 2). There-
fore, the origin of these dilemmas is inductive while their conceptualization
and treatment follow a deductive process, based on the aforementioned con-
textualization and support from the literature.
A typical dilemma for online companies is the monetization of their offer-
ings. The assumption by founders can be termed expectation of free; more
precisely, they believe consumers prefer free Web services to paid ones be-
cause, it is argued, consumers are very reluctant to pay for digital goods. Note,
this is an assumption that drives founders’ thinking, regardless of whether it is
true or not. If this assumption is true, startups monetizing their products di-
rectly risk commercial flight as soon as fees are introduced. Due to hyper-
competition (D’aveni 1994) andlow switching cost
67
, startups have few means
with which to generate lock-ins. However, they need both users and revenue.
Therefore, they offer the product for free to solve the dilemma of adoption,
and everything is presumably fine.
Except that, consequently, they fall into afree-rider trap; a paradoxical sit-
uation in which the success of a product, measured by installed user base,
leads to growing economic losses
68
. In other words, the demand for a Web
service is high while financial returns are low due to free provision (Lee &
Brandyberry 2003
69
). Formally, we can argue for the logic of Shy (2011), in
that startups form a belief of willingness to pay; if all startups assume that
willingness to pay is zero (i.e., demand is zero with any price), this is a case of
equilibrium and the belief becomes self-fulfilling (Evans 2002).
In this simple illustration, three lessons can be found. First, that dilemmas
are associated with assumptions in real human situations. Second, that they are
often related, so that one problem and its solution lead to a new problem; thus,
there is a need for new thinking. Third, that solutions are most often tradeoffs
between “two evils” or the selection of only one desirable option, such as
when selecting between users or revenue. Finally, one can observe that
dilemmas offer a fruitful ground for various conceptualizations; for example,
thefree-rider trap. This dimension makes them appealing to various groups,
67
Switching cost is defined here according to Shy (2011, 120): “When firms capture market share
before they encounter competition, the network effects that are associated with their installed bases
generate switching costs, which are the costs of switching from one brand to another incompatible
brand”. Refer to an alternative definition in Subchapter 4.6.1.
68
This is easy to prove by assuming that each user costs to acquire and serve while producing no
revenue in return. In such a scenario, exponential growth leads to exponential losses.
69
”There are ample examples where commonly employed metrics (unique visitors, page views,
sales, etc.) suggest success whilecompanies struggleto obtain profitability” (Lee & Brandyberry
2003, 10).
89
both researchers and managers. In fact, similar assumptions are applied in
Chapter 4.6, in which the ‘monetization dilemma’ is discussed in depth.
4.2 Dilemmas in the platform literature
Several dilemmas have been identified in the platform literature. The most
important of them (Rochet & Tirole 2003), the chicken-and-egg problem, is
part of this dissertation, separated into cold start and lonely user dilemmas.
This issue is discussed thoroughly in Chapters 4.4 and 4.5. This section will
provide a literature overview on other strategic problems in the platform liter-
ature.
In general, Eisenmann et al. (2006) mention three challenges faced by a
platform owner: 1) the pricing problem, or setting prices so that overall profit
is optimized and takes two-sided dynamics into account; 2) the winner-takes-
all problem, which is topical for platforms not dominating due to a tendency of
markets to tip, which is the tendency of one system to dominate its rivals in
popularity after gaining an initial edge (Katz and Shapiro 1994); and 3) the
envelopment problem, involving rival platforms integrating the platform as
part of their offering and thus capturing users. These issues are discussed in
the following chapters with regard to dilemmas that emerged from the mate-
rial. Envelopment can be perceived as a solution for startups to fight dominant
platforms with regard to the cold start dilemma, whereas the price-setting
problem relates to the monetization dilemma (see Chapters 4.4 and 4.6).
Cennamo and Santalo (2013) discuss two particular problems: coring ver-
sus tipping and positioning dilemma. Coring implies exclusive contribution
(e.g., apps exclusivity) by complementors to a specific platform
70
. If the com-
plementor gives exclusive rights to the platform owner, he/she loses the op-
portunity to multihome; therefore, there is conflict of interest between the plat-
form owner who prefers exclusive complements and the complementors who
prefer multihoming to maximize profits. The more there are a) exclusive com-
plements and b) complements overall, adding to intra-platform competition,
the less feasible it is for new entrants to join; thus, the coring dynamics are
against tipping dynamics (Cennamo & Santalo 2013).
In a similar vein, Lee (2013) discusses exclusivity as a strategic problem;
when possible, forcing exclusivity is beneficial to the platform owner.
However, this is done at the expense of competitiveness. If gains from an
70
Gawer and Cusumano (2008) define the terms as follows: “‘Coring’ is using a set of techniques
to create a platform by making a technology ‘core’ to a particular technological system and market.
‘Tipping’ is the set of activities that helps a company ‘tip’ a market toward its platform rather than
some other potential one.” Cennamo and Santalo (2013) employ these terms in an applied sense.
90
exclusive platform fall short of combined gains from other smaller platforms,
rational complementors will switch. Therefore, any case with several equally
or near-equally strong rival platforms that requires exclusivity from third-party
complementors might be ineffective. Modern app marketplaces, for example,
tend not to require exclusivity (Hyrynsalmi et al. 2012). Multihoming is also
typical for video game platforms, in which game makers publish their titles on
many platforms simultaneously (Idu, van de Zande, & J ansen 2011). However,
there are exclusive first-party titles that do not prevent third-party publishers
from multihoming (Clements & Ohashi 2005). However, as the platform
engages in direct competition with its complementors by offering first-party
supply, there is a conflict of interests. If first-party titles comprise the majority
of sales within a platform, third party vendors have less incentive to join than
if there were no exclusive first party titles. This, as argued, forms a dilemma
of first-party exclusivity versus non-exclusivity (Lee 2013).
Another strategic problem relating to complementors is how collaborative
versus competitive it should be in terms of complementors (Economides &
Katsamakas 2006). By definition, a platform can be either neutral or competi-
tive (Hsiao 2003). If the platform owner competes with application providers,
future providers have less incentive to join as, when given a choice, they are
likely to avoid predatory platform owners. However, assimilation through ac-
quisition might be regarded as preferable from an economic perspective, as
proved by several purchases by dominant online platforms (e.g., Facebook
acquiring Instagram; Google acquiring J aiku). Huang et al. (2009, 3) refer to
this problem as “the fine line that platform sponsors must walk between max-
imizing profits and leaving sufficient residual profit opportunities to encour-
age complementary innovation.” They mention that absorbing complements
can increase a platform owner’s profit in the short term while discouraging
other complementors from making platform-specific investments. The strate-
gic problem of the complementor is avoiding to be absorbed or made obsolete
by integration into the core platform (Huang et al. 2009).
A related problem discussed by Hagiu and Wright (2011) is disintermedia-
tion; when the platform has performed its duty and matched two member
groups (e.g., buyers and sellers), the two can, in some cases, continue their
interaction without utilizing the platform, thereby eliminating the possibility of
lifetime revenue. Thus, the following strategic dilemma can be formulated: if
the platform owner enables transparency and uncontrolled communication
among its members, it receives more interaction because it is easier for mem-
bers to interact; this is beneficial for growth but leads to loss of lifetime gains.
However, if the platform enforces non-transparency and strict control on
communication between members, it can retain interaction within the platform
at the cost of interaction levels (Hagiu & Wright 2011).
91
Relating to quality, Wu and Lin (2012) discuss the problem of governing
diversity. There are multiple problems associated with diversity, which is de-
sired by the demand side but problematic for the supply side. One problem is
that the more competition focuses on a particular niche, the less overall benefit
the platform owner receives. Competition is likely to drive down prices and
platform owner’s profits, insofar as they depend on the pricing of its comple-
ments (e.g., through revenue sharing), while intensive focus on particular cat-
egories of complements foregoes long tail effects; that is, larger sales volume
based on diverse tastes and needs of end users. The second problem is the is-
sue of quality. The platform’s reputation is affected by its complements’ spill-
over effects, so that reputable and popular complements elevate a platform’s
image, whereas low-quality complements reduce its appeal to end users. These
problems are not dilemmas, as they lack contradiction. However, they require
proper strategic response in controlling quality without repelling complement-
ors, and encouraging variety in terms of niches and categories to fulfill differ-
ent end-user needs and thus reap long tail benefits. Wu and Lin (2012) propose
discriminatory support based on quality to enhance heterogeneity that, in their
model, leads to higher overall profits for the platform owner.
The positioning dilemma assumes two rival platforms, a generalist and a
specialist, focusing on mass markets and niches respectively (Cennamo &
Santalo 2013). To differentiate from the competition and create a distinct po-
sitioning in the minds of users, the platform must decide between the two
approaches. If it chooses the generalist, it will lose distinction, and a potential
niche market. However, if it chooses the specialist approach, it risks losing
users who are interested in both generalist and specialist content. Because of
winner-takes-all dynamics, users are inclined to adopt the generalist platform,
with widest selection of content (Cennamo & Santalo 2013).
Reisinger (2004) mentions a strategic problem relating to subsidization, a
common strategy in two-sided markets. The competitive dynamics can lead to
a prisoner’s dilemma situation in which competing platforms set negative
prices, thus eroding their profits (Reisinger 2004). However, it is unclear
whether this is simply a manifestation of competition in general, in that com-
petition tends to lower prices and profits, or a unique problem for platforms.
Nevertheless, its effect can be seen in the analyzed startups that typically ap-
plied an indirect monetization model without, however, a working plan to ex-
tract sufficient revenue from either side (see Chapter 4.6). Relating to indirect
monetization, and particularly to the audience maker model (Evans 2003), a
special strategic dilemma takes place when the demand side perceives adver-
tising negatively but is employed as a monetization model. Logically, the
more advertising the end users see, the more revenue the platform owner
earns, although at the cost of end users’ dissatisfaction (Anderson &
92
Gabszewicz 2006). Therefore, the dilemma involves setting the level of
advertising so that it fulfills both economic goals and, if not serving, at least
not repelling users from the platform.
Church and Gandal (2004) identify four types of demand-side issues relat-
ing to platforms: 1) coordination problems, 2) tipping/standardization, 3)
multiple equilibria, and 4) lock-in. They explain coordination problems from
the customer’s perspective, so that a customer choosing the wrong platform or
standard risks “being stranded” as the expected network effects do not actual-
ize (Church & Gandal 2004). There is a coordination problem because the
customers cannot communicate their willingness to joinex ante, and therefore
each is hesitant to join. As can be seen, this is indeed the chicken-and-egg
problem (e.g., Evans 2009a). Customers cannot redeploy their platform-spe-
cific investment towards adoption of another platform. It is unlikely, however,
that the platform owner would be able to exercise power because users simply
have no incentive to stay, despite any sunk cost. Thus, thehold-up problem is
unlikely to arise (cf. Klein 1998). The choice of the platform is a demand-side
strategic problem; it helps understand why users are cautious when adopting
platforms. In tipping, after a particular threshold, one platform becomes domi-
nant and all users convert to being its customers (Shapiro & Varian 1998).
Tipping becomes a problem if inferior technology is chosen, in which case
the opportunity cost is the loss of superior technology in achieving platform
users’ goals (Church & Gandal 2004). The Qwerty keyboard layout is an often
employed example that, according to some, is not the optimal layout in terms
of writing speed but is practically impossible to replace due to its wide adop-
tion; that is, network effects (Parker & Van Alstyne 2010). Once a standard
has been widely diffused, it is hard to abolish; however, before that, its dis-
semination is difficult due to the chicken-and-egg problem. In the literature,
this is referred to as the standardization problem (Besen & Farrell 1994;
Weitzel, Beimborn, & König 2006). Multiple equilibria is the opposite of tip-
ping, so that customers are unable to commit to any competing platforms due
to fear of choosing the wrong one (Church & Gandal 2004). In this case, all
competing platforms lose as adoption is delayed to the last possible moment.
Finally, lock-in can become an issue for users adopting the winning design
(Church & Gandal 2004). Multiple types of power play can arise; for example,
the aforementioned hold-up problem whereby the platform owner can raise
prices as long as the switching cost remains higher or there are no de facto re-
placements, or the quality of the platform’s operations or technology might
suffer due to lack of competition. These effects are similar to monopoly, and
are naturally associated with locked-in customers (Farrell & Klemperer 2007).
As can be seen, the platform literature has discussed a variety of strategic
problems relating to strategic choices of platform owners, complementors, or
93
demand-side users. The dilemmas presented here were identified through a
literature inquiry and represent the current state of research. However, any
number of new dilemmas can be created based on alternative situations. This
study focuses on four specific strategic dilemmas that are presented in the
following chapter. The four dilemmas emerged from the GT analysis, and are
chosen because they represent the issues identified by the studied platform
startups’ founders. Moreover, they seem to respond to the platform literature,
in which the chicken-and-egg problem typical for startups is central.
4.3 Dilemmas emerging from analysis
4.3.1 Results from the black box analysis
The following figure reveals a model of the “black box” of failure (Chapter
1.2) based on grounded theory (GT) analysis. After deciding to focus on
dilemmas (i.e., emergence of the core category), selective coding was con-
ducted to find support for dilemmas, with new dilemmas also being found.
Figure 10 Exploratory outcomes – opening the black box of failure
Startup
Biases Dilemmas
Failure
· Pioneer’s dilemma
· Cold start dilemma
· Lonely user dilemma
· Monetization dilemma
· Remora’s curse
· Pivot dilemma
· Peter Pan’s dilemma
· Juggernaut dilemma
· Illusion of scale
· Illusion of free
· Technology bias
· Build it and they will come
· Dog food blindness
· Sunk code fallacy
· Reference point bias
”Bl ack Box”
94
The purpose of Figure 10 is to show the larger framework in which the cho-
sen dilemmas are rooted. Briefly, the dilemmas are defined as follows:
· Pioneer’s dilemma: if the startup launches too early, it will pay the
pioneer’s cost and is likely to fail due to insufficient resources; if it
launches too late, it is unable to capture users from incumbents.
· Cold start dilemma: without content, users are unwilling to join and
generate content.
· Lonely user dilemma: without other users available at a given time,
users are unable to use the platform.
· Monetization dilemma: if access and usage of a platform is provided
for a fee, users are unwilling to join; if access and usage is free, the
platform is economically non-viable.
· Remora’s curse: if users or content is sourced from a host platform,
the cold start problem can be solved; however, at the loss of power
relating to customer relationships, monetization, and so on.
· Pivot dilemma: if the startup accommodates its user’s wishes in prod-
uct development, it loses focus; if it does not, it loses the user.
· Peter Pan’s dilemma: if the startup accepts external funding, it loses
decisive authority and becomes vulnerable to hasty decisions; if it
does not, it loses against competitors with funding.
· Juggernaut dilemma: due to lack of legitimacy, the startup is unable
to convert enterprise clients which would grant it legitimacy.
Following earlier research outlining failure as a combination of reasons
(Lussier 1996), it can be stated that the failure of the sampled startups com-
prises 1) general business problems (e.g., management issues; lack of market-
ing), 2) startup-related problems, arising from the fact of being a startup (e.g.,
“liability of newness”), and 3) platform-specific problems.
“Illusions”, which were mentioned by some founders, are perceived as fal-
lacies and observed also potentially to exist in other cases. It was also found
that founders typically associated biases (i.e., their own thinking errors) as rea-
sons for why they could not properly address the dilemmas, or even identify
them in time. They are perceived to relate to dilemmas because they affect the
assumptions of strategic decision-making. For example, assuming that all us-
ers prefer freeness over quality will more likely lead to a monetization di-
lemma than a contrary premise.
The identified fallacies are defined as follows:
95
· Illusion of scale: the tendency of startup founders to assume online
businesses require less effort to succeed than offline businesses.
· Illusion of free: the non-validated assumption that users are unwilling
to pay for online products.
· Technology bias: the tendency of startup founders to assume that all
startup problems can be solved by technological means.
· Build it and they will come: the tendency of startup founders to as-
sume that the product will market itself.
· Dog food blindness: the refusal of accepting fault in one’s product.
· Sunk code fallacy: the tendency of startup founders to refuse to make
drastic business changes (i.e., pivots) due to the time and effort spent
making the current version of the product.
· Reference point bias: the tendency of startup founders to assume that
successful implementation of a particular strategy or tactic in another
context would automatically work in their context (e.g., “because it
works for x, it will work for us”).
Due to limitations on the scope of this study, fallacies were left for further
research. It was considered that including them would 1) take away the focus
of dilemmas, and 2) expand the required theoretical basis to become too exten-
sive for one study. In other words, to maintain depth of the analysis, it was not
perceived possible to thoroughly discuss dilemmas and biases, and so the latter
are only briefly discussed as preliminary observations.
Further clarification in the next section will explain why a subset of prob-
lems was chosen for detailed treatment. Consistent with GT principles (Glaser
2004), all dilemmas and illusions were captured by the author. Particular
names, including “remora”, “cold start”, “dog food”, “build it and they will
come”, and “sunk code” were taken from founders’ post-mortems and industry
terminology.
4.3.2 Narrowing the focus of the study
As can be seen, the GT analysis identified many phenomena that remain out-
side this report. All research can be regarded as a tradeoff leading to the neces-
sity of restraining the research focus (Eisenhardt & Graebner 2007), and
focusing on dilemmas was simply the author’s choice. The author preferred a
deeper focus on dilemmas, albeit this decision omitting the treatment of biases
that, according to the analysis, are equally important when considering the
failure outcome.
96
The analysis showed that platformstartups struggle with many other prob-
lems relating to their startup nature (e.g., Wasserman 2013). However, given
its positioning, this study focuses on platform-specific problems. Most other
startup problems are well documented in the literature. For example, liability
of newness (Bruderl & Schussler 1990; Freeman, Carroll, & Hannan 1983;
Singh, Tucker, & House 1986; Stinchcombe 1965) is associated with the prob-
lem of legitimacy (i.e., “J uggernaut dilemma”). The entrepreneurship literature
has analyzed problems of adaptation and related turnaround strategies (e.g.,
Boyle & Desai 1991; Hofer 1980; Melin 1985).
In a similar vein, the strategic management literature has identified glitches
between venture capitalists and founders. Also, growth pains such as thecash
flow problem
71
(Mears 1966; Wilcox 1971) are associated with Peter Pan’s
dilemma. Katila, Rosenberger, and Eisenhardt (1998) studied the “shark’s di-
lemma”; that is, how a startup can collaborate with a larger organization while
retaining its competitive advantage. Pioneer’s advantages and disadvantages,
and also those of early movers, have been extensively covered in the literature
(e.g., Agarwal & Gort 2001; Golder & Tellis 1993; Kerin, Varadarajan, &
Peterson 1992; Lieberman & Montgomery 1988; Robinson, Fornell, &
Sullivan 1992).
The following table classifies the strategic dilemmas based on their applica-
bility.
Table 10 Analysis of dilemmas
Dilemma Specific to plat-
form startups
Specific to
startups
Specific to online
business
Applies to any
business
Cold start x
Lonely user x
Monetization x x
Remora’s x
Pivot x
Peter Pan’s x
Pioneer’s x
J uggernaut x
The cold start dilemma can be regarded as a specific problem for platforms,
regardless of whether they are online or offline (see Chapter 4.4). The lonely
user dilemma relates not only to activating users, but also to time
72
; thus, it is a
problem of real-time social services. Monetization is a general problem of
71
The cost of customer acquisition needs to be covered instantly, while customer lifetime revenue
is received in the future. Therefore, the faster the company grows, the more it accumulates loss.
72
Finding available matches at any given time complicates coordination, and thus aggravates the
chicken-and-egg problem.
97
online offerings, and also applies to platforms, although not necessarily to all
startups beyond Internet markets. Remora’s curse applies when the platform
startup employs the remora strategy to obtain users or content from another
platform. The pivot dilemma
73
applies to all businesses, but is not a specific
problem of platform startups. Peter Pan’s dilemma is a problem for startups
that need to decide whether to remain small and be consumed by competition,
or grow big and be consumed by expenses. The pioneer’s dilemma relates to
launching an unfinished product and failing to gain adoption, or waiting for it
to be perfected and losing competitive advantage. Similarly, the juggernaut
dilemma is proprietary to startups: they cannot get customers due to a lack of
legitimacy, and due to lack of customers, they cannot get legitimacy (cf.
Stinchcombe 1965).
Detailed treatises on all dilemmas, although enticing, would have severely
fragmented the study as they clearly connect with multiple streams of the liter-
ature. In other words, breadth was sacrificed for depth. This decision was rein-
forced by the fact that it proved difficult to find a common denominator that
would have enabled building a unified theoretical framework, as now has been
achieved by relying on the platform literature. Finally, due to the relative re-
cency of the platform/two-sided markets literature, it was concluded that there
is more room for contribution than other identified streams, especially on the
strategic management of platforms.
4.3.3 Chosen dilemmas and their treatment
Following the aforementioned rationale, the focus of this study is on strategic
problems proprietary to platform startups on the Internet, particularly on the
following dilemmas:
· Cold start dilemma
· Lonely user dilemma
· Monetization dilemma
· Remora’s curse.
These dilemmas will be discussed in detail in the following sections, while
other dilemmas are omitted. The presentation of dilemmas follows the struc-
ture of: 1) definition and exhibits, 2) the literature positioning of the dilemma,
and 3) solutions derived from theory. Solution is intuitively defined as a
73
If a startup heeds customer feedback when developing a product, it loses its raison d’être, or its
original vision; if it ignores customer feedback, it loses the customers.
98
solution to a problem that, in this case, satisfactorily solves one or both parts
of the dilemma.
It is argued that by solving the cold start dilemma through subsidization
(e.g., offering free access and usage), the startup will face the monetization
dilemma, whereby it is unable to capture economic value from the interaction
taking place in the platform. Whereas, when solving the lonely user dilemma
by applying the remora model (i.e., ‘envelopment’ in the platform literature),
the startup faces what is termed ‘remora’s curse’ (i.e., dependence of the host
platform). The latter condition bears similarity to the classic hold-up problem,
which is explained in Subchapter 4.7.2. Cold start and lonely user dilemmas
are understood as different realizations of the chicken-and-egg problem pre-
sented in the dissertation’s introductory chapter.
The following figure illustrates the idea.
Figure 11 Strategic actions and their consequences
Therefore, the following strategic decisions apply:
1. When facing the cold start dilemma, the startup solves it by subsidiza-
tion or remora.
2. When facing the lonely user dilemma, the startup solves it with therem-
ora model or subsidization.
3. When facing the monetization dilemma, the startup solves it with the
freemium model.
4. When facing remora’s curse, the startup solves it by diversifying.
The selection of solutions arises from the literature and analysis of the em-
pirical material. Subsidization is commonly considered a solution to the cold
Cold start
dilemma
Lonely user
dilemma
Monetization
dilemma
Remora’s curse
Subsidization Remora
Freemium Diversifying
Types of chicken-
and-egg problem
Derivative
problems
99
start problem in the platform literature (e.g., Rochet & Tirole 2005), while the
remora model is conceptually similar to the envelopment strategy presented by
Eisenmann et al. (2011). Freemium, however, is a special form of subsidiza-
tion that is commonly applied by Web startups (Wilson 2006; Niculescu &
Wu 2013). Notice that subsidization and the remora model can both be applied
in relation to the cold start and lonely user dilemmas. In the following sub-
chapters, this particular order has been chosen for the purpose of presentation,
that is, not to repeat their treatment.
Diversifying is synonymous to multihoming, which is a central concept in
platform theory (Armstrong 2006). Also, because there is generally a high de-
gree of interoperability between Web platforms, for example, through applica-
tion programming interfaces, or APIs (see Rochet & Tirole 2003), both envel-
opment and multihoming are common in online markets (Mital & Sarkar
2011). Therefore, the considered solutions depict both platform theory and
practice. However, they have not been integrated into one framework in the
extant literature.
4.4 Cold start dilemma
4.4.1 Definition and exhibits
The cold start dilemma is a specific problem for content platform startups re-
lying on UG. The dilemma can be defined as follows: when there is a lack of
existing content, no users are motivated to create new content, and so there
remains a lack of content. As a result, the ideal UG model fails, the platform
will fail, and the startup will fail. These assumptions will be examined next.
First, it is assumed that existing content has a relationship with new content;
that is, the reason why other content is created. Note that the content also in-
volves externalities that are not other content, such as sharing and ‘liking’.
These spillover effects will be discussed in Subchapter 4.4.2.
Second, the cold start dilemma might differ through the assumption of
multi-sidedness, depending on whether or not user homogeneity is assumed:
1) in a one-sided content platform, users provide content that is beneficial for
the same type of users, and 2) in a two-sided content platform, users provide
content that is beneficial for other types of user. The interaction between user
groups defines the type of the platform, which is logically derived from the
fact that users’ interests vary: for example, buyers seek sellers, not other buy-
ers in an auction platform; males (typically) seek females in a dating site;
however, people interested in mobile phones are looking for other people of a
similar type in a mobile phone discussion forum. This is relevant due to
100
motivational factors; namely, creating a community, which will be revisited
when discussing solutions. If the interests of the users are common, the plat-
form can be defined as a community, and therefore tactics to acquire users
from this particular niche should varyvis-à-vis a mass audience.
The question of motives and incentives is paramount for getting the desired
response (see Table 11 [2]). Based on the ideal UG model (see Chapter 3.4),
the goal is that the content and actions of first-arrived users lead to the re-
cruitment of second-generation users either directly (e.g., invitations) or indi-
rectly (e.g., content indexed by search engines), as opposed to the startup ac-
quiring new users, which requires marketing investments and, potentially,
skills (Table 13, [6]). Building such assets can be prevented by thebuild it and
they will come fallacy (Table 13, [2]), defined as a tendency of technology-
oriented founders to avoid marketing.
There are two types of participation behavior: contribution and consump-
tion
74
. Contribution is feasible if expected benefits are larger than the cost of
contribution. Consumption has a lower cost but also a low switching cost due
to a generally high number of alternative sources of content (i.e., substitutes)
on the Internet. Further, the benefit of consumption arises from the
informative or entertainment properties of the content; the supply side benefit
comes from search engine externalities, which is compatible with online
search behavior (Hsieh-Yee 2001), and also the social spillover effect, such as
sharing or commenting on the content. The difference in participating
behavior enables analysis of the setting as a two-sided platform, whereas
including single-user motivation would result in a one-sided platform.
The analyzed startups report the cold start problem as follows.
74
Note the similarity to types of monetization behavior: joining without paying (i.e., free users)
and paying for joining (i.e., customers).
101
Table 11 Exhibits of cold start dilemma
Example
[1] "(We) underestimated the “cold start” problem […] especially when it relies on user-generated
content. The value you provide to your users centers around the content on the site, so to build a
user-base you need a lot of content created by the first users to kick-off the community." (Dickens
2010).
[2] "We fell into the “build it and they will” come school of thought (although even when they came,
we still weren’t in good shape). Users didn’t review because there was no enlightened self-
interest for them to do so. Nobody wanted to edit our data for the same reason." (The Chubby
Team 2010).
[5] "[A] place where people can find and provide information about private companies/startups, and
also review them. Awesome idea, right?! WRONG! Where the heck are we going to get all this
data about startups?" (The Chubby Team 2010).
[6] "[H]aving to consistently find new content was probably the biggest hit to my motivation for the
site. As much as I loved indie music it was draining to constantly find new albums to post up."
(McGrady 2008).
[7] "[The startup] was a product designed to connect journalists with readers. As such, we had two
sets of customers, which means we need to do customer development twice. I spent a great deal
of time designing the ultimate solution for journalists, and almost no time on what readers
wanted. As such, I didn’t really know what to make, or what to say to the journalists about what
they should write." (Biggar 2010).
[8] "Our proposition was made even more complicated because we were trying to create a market-
place. When a magazine opens for submissions, you’re submitting to that magazine. But [the
startup] was one step removed – anyone could make an assignment. So even if you trusted [us],
you didn’t necessarily trust the person who posted the assignment." (Powazek 2008).
[9] A close friend reading my early business presentation told me he liked it a lot but was worried
about one line. He said it sounded like I’m building infrastructure and that is going to create
problems with most investors. At the time I was focused on building a website and the comment
didn’t register. Now, with more perspective, I can certainly appreciate the advice. (Hammer
2008).
[10] In community-generated media, trust is everything. When you ask for submissions, contributors
go through an instant internal calculation: “Do I trust these people with my work?” When your
site is brand new, you’ve got no record to rely on. And with more shady “user-generated
content” schemes popping up every day, people have their defenses up. (Powazek 2008).
As noted, and demonstrated by exhibits, a cold start is a specific problem
for Internet startups trying to leverage UGC [1]. Examples include discussion
forums, blogs, and various crowdsourcing services, in which the startup offers
a platform for discussion or other forms of social interaction; for example,
dissemination of information, pictures, videos, or ratings [5]
75
. Such startups
depend on relevant and updated content (e.g., articles, comments, pictures,
reviews, and ratings) to acquire visitors, convert them into repeat users, and
encourage them to produce more content
76
.
75
For example, YouTube is a content platform, so is Flickr. Whether they are one-sided or two-
sided platforms is arbitrary.
76
In contrast, in-house content-generation does not require users to actively produce content;
albeit, startups following this model might enable UGC, hoping for UG benefits.
102
Consequently, if there is little or no UGC in the platform, new participants
have no or only small incentive to join; if no participants join, no new content
is created, and so forth. As no visitors and new content are created, there is no
reason for the platform to exist and the startup will fail. In other words, the
cold start dilemma is a variation of the well-known chicken-and-egg problem,
and quite a typical reason for platform startups to fail.
The cold start dilemma can be demonstrated with a simple game. The fol-
lowing considers a one-sided platform in which users are all the same type,
and gain benefit from each other’s participation.
Table 12 Too many consumers (of content)
C
2
Contribute Not
C
1
Contribute 0, 0 -1, 1
Not 1, -1 0, 0
In the game, contributing can be unpleasant for consumers of content; thus,
they incur a cost (i.e., -1). This cost is a function of time, effort, and uncer-
tainty concerning the usefulness of contributing (see Table 13, [10]). Moreo-
ver, if a consumer contributes content, another consumer will not return the
favor but simply consumes the content, thus gaining a payoff of +1. If both
parties contribute, the effort and benefit cancel each other out. Both receive a
payoff of zero. However, this is not stable equilibrium as each party has an
incentive to improve his/her position by not contributing (i.e., moving from 0
to +1). Because not contributing yields the same benefit as contributing if the
other side makes the same choice, players might be indifferent to contribution.
The safest strategy, which minimizes potential cost, known as minimax (see
Camerer 2003), is not to contribute, as contributing risks a negative payoff.
Therefore, users will not contribute when they expect others not to return
the favor. However, more importantly, they might not contribute especially
when they expect others to contribute.
Now consider changing the game into a two-sided platform, where there are
two different groups of users, both of which derive additional benefit from
complementing interactions.
103
Table 13 Consumers and generators
G
Contribute Consume
C Contribute 2, -1 0, -1
Consume 3, 1 0, 0
Generators enjoy contributing (i.e., creating content), although it is more
valuable to them when there are consumers of content. In contrast, they are
indifferent to consuming content. Consumers of content do not prefer contrib-
uting, so it is costly to them: however, they derive benefit from the content
produced by generators. This is clearly demonstrated in Table 13, case [7],
where the startup focused on one side while neglecting the other; as a conse-
quence, it was unable to provide network effects for either.
Generators receive intrinsic benefit from contributing and are indifferent to
other players’ contributions, but not consumption. In fact, consuming content
is a type of invited free-riding: the creator of the content wishes it to be con-
sumed, and is indifferent to whether others create content or not; although,
they might appreciate “side payments” such as praise and criticism. Further, in
contrast to the previous example, the payoff for the consumer no longer relates
to his/her own choice of contributing and, given he/she has information on the
generators’ payoffs, the consumer will always prefer not to contribute. A sta-
ble equilibrium
77
is when generators contribute and consumers do not.
Furthermore, this explains why merely joining a platform is insufficient in
the absence of active usage. The users who join will quickly churn if the plat-
form is “cold”. This observation is critical in terms of determining which ac-
tion to follow: joining or participating
78
. Therefore, it becomes important for
the startup to find contributors and offer them a convenient platform. Contrib-
utors and consumers can have complementary needs, which is why both par-
ties are needed. Moreover, the startup needs both to consider the critical mass
of any users (see Chapter 4.5) and also the correct proportion of participants.
This is because, in a two-sided setting, the types are interdependent, and rather
than preferring the existence of a similar kind of participant, users prefer a dif-
ferent kind to join.
A two-sided platform, in particular, functions on reciprocal utility: the util-
ity of groupA for groupB is in proportion to the utility of groupB to groupA
77
Neither party would gain a better payoff by switching.
78
As noted by one of the founders (Roseman 2010): “good Google foo [website traffic] won’t save
you. You need that traffic to translate into a community. A visit is not an interesting statistic,
especially in a business that requires the community to produce content.”
104
(for details, see Subchapter 4.4.2). The chicken-and-egg dilemma is therefore
associated with the quality, amount, and type of activity of respective groups
to their counterparts. Advertising is a special variation as there is no content
without advertisers
79
; without content, no visitors; and without visitors, no
benefit for advertisers. Therefore, although often considered negative network
effects vis-à-vis end users, advertisers can, in fact, generate indirect utility by
funding content creation. However, this does not apply under UG; thus, ad-
vertisers are not considered in the dilemma.
Further, a two-sided platform requires strategies for managing both coun-
terparts, whereas a one-sided platform considers the needs of only one set of
customers or users (Table 13, [6]). The implication is that the startup needs to
consider both sides in its strategies. Finally, although the startup can mediate
interaction between two parties, there needs to be a degree of mutual trust for
the interaction to take place, as exemplified in Table 13, [8]. This trust might
not automatically transfer from the platform to all of its users.
Often, the monetization model for content platforms is indirect due to high
competition in most content markets, and also the considerations presented
below; thus, the content is provided for free and monetization is achieved by
showing advertisements (see monetization dilemma, Chapter 4.6). Note that
UG introduces some restrictions to monetization; firms cannot readily charge
for amateur content due to uncertain quality, expected unwillingness to pay
(refer to Subchapter 4.6.1), and resistance from the users creating the content
who might feel that their rights are violated if their content is monetized with-
out revenue sharing. The typical monetization model is therefore indirect:
content platforms delegate content creation to users instead of utilizing the
firm’s own resources, design a process through which the content is re-used to
attract new users such as search engines and social sharing functions, and then
monetize the content through indirect revenue models, typically advertising.
Due to the problem between UGC and direct monetization, that is, the
startup cannot directly monetize the free content provided by users without
consumers’ retaliatory effects (i.e., churn), and by providers of the content
(i.e., resistance to charge for their input), the strategies for monetization re-
main limited. In fact, even after solving the cold start dilemma, a startup needs
to solve the monetization dilemma to become what can be termed a viable
business. Moreover, platforms dealing with real-time interaction need to con-
sider the lonely user dilemma.
79
As the firm is unable to provide content without indirect monetization; note there is an inverse
proportion to advertising, so that users typically respond negatively to its increment.
105
Finally, Appendix 2 includes a brief meta-discussion on the definition of
cold start; namely, if it can be considered a dilemma based on the definition
given here.
4.4.2 The literature
The chicken-and-egg problem of acquiring content, participants, or liquidity
80
is widely recognized in the platform literature, and also the preceding litera-
ture on e-marketplaces (see e.g., Caillaud & J ullien 2003; Parker & Van
Alstyne 2005; Sun & Tse 2007; Kim & Tse 2011; Raivio & Luukkainen
2011).
In fact, the chicken-and-egg dilemma is an inherent consequence of the
two-sided nature of a platform due to the requirement of “getting both sides on
board” (Evans 2002), or an “essential feature of two-sided market analysis”
(Luchetta 2012, 11). Therefore, the phenomenon is not novel; however, rela-
tively little empirical research is focused on its solutions. Furthermore, it is
often neglected by economic models that assume simultaneous entry (Hagiu &
Spulber 2012), and consider pricing the key strategy for encouraging entry
(Piezunka 2011).
The cold start dilemma can be described as a coordination problem, in
which either all or no users adopt a platform (Farrell & Klemperer 2007). This
is similar to herd behavior (see Banerjee 1992) or circular logic, as a user’s
action depends on that of collective action, and can lead to tipping in an in-
dustry-wide setting (see Katz & Shapiro 1994). As a theoretical extreme, it
displays how network effects can dominate adoption if all other determinants
of adoption are ignored. However, it does not apply well to real contexts.
Farrell and Klemperer (2007) give an example of a photography market, com-
prising photographers and film developers who both favor each other, in
which they insert more realistic assumptions, so that groups are not making
all-or-nothing choices; rather, some users might adopt and others not in re-
sponse to individual preferences and heterogeneity among users, and negative
network effects termedintra-group congestion.
This network term, congestion, translates into negative network effects in
the platform context. Also, Shy (2011) notes that network effects are not al-
ways positive, and thus do not always increase willingness to adopt. For ex-
ample, consider a network in which users are evaluated as being, for instance,
“shady” or unreliable. Such a network tends to attract a similar kind of
80
The termliquidity is often employed in the context of e-marketplaces (e.g., Ordanini, Micelli, &
Di Maria 2004), and refers to transaction volumes.
106
participant while high-quality users will refrain from adoption
81
. With regard
only to the platform, the perceived quality of participants is likely to play a
role of adoption. The startup therefore needs to pay attention to not merely
attracting any users but such users who increase the propensity of desired
users (i.e., the target market) to join. This granularity is often neglected when
discussing “traffic driving” and “community building” as strategies for
attracting users. Negative network effects are discussed more thoroughly in
Subchapter 4.5.2.
Although the chicken-and-egg problem lacks a substantial amount of em-
pirical work, it has been focused on by some studies; for example, Mas and
Radcliffe (2011) studied it in the context of a mobile payment platform in a
developing country, Funk (2006) contrasted J apanese and Western efforts in
building a mobile Internet, and Raivio and Luukkainen (2011) documented
challenges relating to Open Telco, a project inviting mobile carriers to collab-
orate. Mas and Radcliffe (2011) identify three factors that prevent payment
platforms from scaling up: 1) lack of effective network effects, 2) the “sub-
scale trap” (i.e., cold start dilemma), and 3) lack of trust from both parties.
They argue (ibid., 305) that
"At first, all these elements work against a deployment. The ben-
efit to a customer of joining the system is minimal when few oth-
ers are connected (network effects) and the merchant network is
not sufficiently dense […] to meet their […] needs. Meanwhile,
merchants remain reluctant to tie-up scarce working capital […]
because they do not yet see enough demand from customers […]
And customers lack trust in the system, because they know few
people who can vouch for the service."
The literature refers to a critical mass (e.g., Rohlfs 1974; Evans &
Schmalensee 2010), which can be defined as the point where the perceived
cost of participation is lower than the benefits of participation, and the benefits
of generating content are superior to only consuming it (for a more detailed
approach to critical mass, see Subchapter 4.5.2). Two points are important
relating to the timeliness of a critical mass: 1) user expectations, and 2) com-
petitive dynamics. User expectations are important because if users are pre-
sented with a “cold platform” they might quickly notice it to be of no use and
81
A real example might beblack hat search engine optimization communities, which basically aim
to divert search engine algorithms by applying unethical practices such as link farms. The existence of
such communities is strictly a negative externality for honest search-engine optimizers, as the “rotten
apples” ruin the reputation of the industry and force search engines to tighten their rules concerning
optimization.
107
never return
82
. The cold start period is formulated by Evans (2009a, 102) as
the “ignition phase”, in which
"[C]ustomers are trying the platform and assessing its value;
these early adopters will stop coming back, and stop recom-
mending it to their friends, if the platform does not grow quickly
enough."
This approach would seem to be compatible with the notion of a ‘one-shot
game’, which users either join instantly or not at all, and which, in turn, can
become problematic when associated with the “perfect product” fallacy
83
. Es-
sentially, while the beginning of a platform startup is critical, it is questionable
whether, for fear of not generating a critical mass, its launch should be delayed
or not.
Second, in a competitive setting where rivals compete for the same users,
whichever reaches a critical mass first can become the dominant platform. In
the case of strong network effects, it is expected that reaching a critical mass
will lead to tipping (Shapiro & Varian 1994), in which users of competing
platforms switch to the focal platform, and thus the market is what is termed
winner-takes-all (Sun & Tse 2007). In contrast, the critical mass can be re-
garded as isolated from competition, when assuming multihoming and less-
than-strong network effects, so that internal consistency
84
of the user base is
sufficient, regardless of coexisting platforms.
Moreover, trust creates a relationship with the legitimacy of new ventures
(Stinchcombe 1965) and the vast body of the literature in which it is discussed
(for a review of online context, see Grabner-Kräuter & Kaluscha 2003). Trust
is, however, implied in the two-sided markets literature through expectations.
An example of negative expectations is given by Galbreth, March, Scudder,
and Shor (2003, 316):
"If buyers and sellers are skeptical of the prospects of an e-mar-
ketplace, these expectations might be lowered [which] could
cause participation levels to move toward the empty e-market-
place equilibrium as opposed to the internal one. […] The gen-
eral lack of confidence in Web-based initiatives in recent years
could explain why many e-marketplaces failed to grow to the
projected participation levels ? when potential participants ex-
pect the e-marketplace to fail, it will."
82
A strategy for overcoming this is to avoid a large-scale launch (i.e., marketing launch) prior to
reaching a critical mass (see Chapter 4.8).
83
A tendency to delay launch until the product is ready. See Appendix 3 for comparison between
late and early launch.
84
Internal consistency refers to the possibility of a user finding a match with a relative ease.
108
In fact, expectations are in the nature of the dilemma; if expected adoption
is low, then it remains low and the expectation becomes a self-fulfilling
prophecy (Shy 2011). In contrast, if the platform gains more powerful advo-
cates, expectations are higher and the prophecy takes a positive direction, even
hype. The case of hype was present in the earlier stage of online platforms,
namely the dotcom era. Evans (2009a) argues that the failure of dotcom plat-
forms was due to three reasons: 1) existing bilateral relationships and other
offline arrangements “got the job done”, so there was no need for online
platforms; 2) participants perceived that e-marketplaces aimed to depreciate
their value. Thus, they mainly focused on price competition between
participants, which dilutes brand and services; and 3) lack of liquidity. Due to
the fact that sellers were skeptical about joining, buyers lost interest, and there
was no reason for the platforms to exist
85
.
Rohlfs (1974) and Funk (2006) refer to the cold start dilemma as astartup
problem
86
and associate it with direct or indirect network effects. Funk (2006)
studied how J apanese firms were able to overcome this problem when intro-
ducing a mobile internet, whereas Western companies failed to do so, and dis-
covered that this was due to 1) generating entertainment content by smart
partnering, and 2) offering the content via a micro-payment system in which
fees were collected and redistributed through a standard system. Evans (2009a,
102) associates this with a critical mass by stating that “[t]he challenge that
catalyst entrepreneurs face is how to achieve the critical mass necessary for
ignition […] over some reasonable space of time.” This thought will be revis-
ited in Chapter 4.8.
Schilling (2009, 195) notes that “if the user must invest considerable effort
in learning to use a computer platform, he/she will probably choose to invest
this effort in learning the format they believe will be most widely used.” This
approach, again, highlights the importance of network effects in creating plat-
form pull. Furthermore, it links the cold start problem to adoption barriers.
Adoption of technology, or ‘technology acceptance’, has been widely studied
within information systems science (for a literature survey, see e.g.,
Mäntymäki 2011) and especially by economists who consider benefits versus
costs; ultimately, leading to preference concerning the platforms as not all of
them can be adopted. Their adoption costs can include psychological re-
sistance, time, and effort, which might be crucial for some users, even to the
extent that time is scarcer than money. Some startups highlighting free models
as the primary motive for adoption might apply a sub-optimal pricing strategy,
which will be discussed in Chapter 4.6.
85
This analogy to dotcoms, ten or so years later, will be revisited in Chapter 4.6.
86
Rohlfs (1974, 18): “how to attain such a user set, starting from a small or null initial user set.”
109
Both the number of participants per se and their proportion need to be con-
sidered. This is especially important for match-making platforms such as da-
ting services. An example relating to content platforms, in which the reasons
for participation relate to information goals, might be a knowledge-sharing
platform, here a sub-type of a content platform, in which there needs to be the
correct proportion of questions and answers (Kim & Tse 2011), and providers
of both to guarantee a sustaining process of UG. Similarly, an exchange plat-
form needs to make sure there are sufficient buyers to interest sellers andvice
versa (Teece 2010). Therefore, even regarding content, and not only social
connections, the startup faces coordination issues of a match-making type.
Rationality of adoption does not only reflect the current situation (i.e.,
weighing costs and benefits as they are), but also the future prospects of the
platform; namely risk and uncertainty. Katz and Shapiro (1986, 824) formulate
this position as follows:
"n the presence of network externalities, a consumer in the
market today also cares about the future success of competing
products.[…] total benefits derived from it will depend, in part,
on the number of consumers who adopt compatible products in
the future."
This is also asserted by Köster (1999) who asserts that adoption depends on
both historical interactions (i.e., existing base of content or users) and ex-
pected interaction (i.e., expected bases). Although both Katz and Shapiro’s
(1986) and Köster’s (1999) definitions referred to other complementary dura-
ble goods, digital content is compatible with the implications; essentially, life-
time projections and long-term survival of the chosen platform are likely to
influence the contribution decision. Equally, the demand-side will find this
important, as finding a reliable source of content reduces search costs (e.g.,
time spent employing search engines).
Another point to note from Katz and Shapiro’s (1986) quotation is technol-
ogy-specificity, which is rooted in the notion of asset specificity (Riordan &
Williamson 1985). Specificities might have weight in the adoption decision;
for example, as many startups eventually fail (Haltiwanger et al 2009), users
might experience greater than normal doubts concerning contributing to or
joining them, given they are aware of failure rates. In the case of demise, a
user loses the time and effort invested in learning a new system (i.e., the
learning curve is a platform-specific investment), incurs search costs for find-
ing a replacement, and might experience loss of private data in some cases
87
.
87
Although the author must note that all failed startups observed during the research period
provided a decent export function prior to closing down, this seems to be the industry standard for a
‘graceful exit’.
110
Finally, the user might not only take a passive stance in evaluating the risk of
platform demise, but might take action to influence it; thus showing support,
although based on the notion of improving his/her own gain. Due to awareness
of inter-platform competition, the propensity of user recommendations, also
termed peer marketing, viral marketing, or simply word-of-mouth, can actu-
ally increase, thus enabling the startup to reach the ideal user-generated cus-
tomer acquisition. As put by Katz and Shapiro (1986, 831), “given the network
externalities, each consumer wants all other consumers to purchase his
favored technology.” Therefore, in theory, network effects can encourage a
user to promote the platform to his/her peers; the more who join, the more
useful it also becomes for him/her.
As established, the market does not exist if the chicken-and-egg problem
remains unsolved, as there is no reason for either party to interact (Evans
2002). Moreover, the adoption by one group triggers a cascading growth as the
benefits of adding a new member come both from that new member and the
influence this member has on attracting more members. Evans (2002, 76)
notes that “f we assumed the base of sellers were important to attracting
buyers, (and vice versa), the indirect benefits would be even greater because a
buyer joining the system would induce additional sellers to join (and so on),
which would generate additional indirect benefits on both the seller and buyer
sides.” Hence, we are close to the definition of viral growth (e.g., Salminen &
Hytönen 2012), which is essentially exponential growth due to one member
inviting more than one new member, and so on
88
.
Note that if we do not assume two-sidedness (i.e., distinct user groups) and
network effects (i.e., interdependent interaction), the cold start problem would
not be compatible with the chicken-and-egg problem in the literature. This is
because, if the content platform is a one-sided market, the platform owner
(i.e., startup) might simply provide the content, as is done by media
broadcaster and news websites, among other content portals, and there would
be no cold start problem. This would render the study’s treatment a trivial
exercise. To counter this, we can distinguish 1) two distinct user groups (i.e.,
consumers and contributors of content) and also 2) interdependence (i.e.,
network effect) between them, which is mediated by UG, so that the
consumers derive benefit from the content produced by contributors.
Whether contributors derive benefit from the presence of consumers is ar-
guable; it is possible that the nature of this benefit is intrinsic motivation, de-
sire to create, or some form of altruism. The motives to contribute to platform
88
The simplest definition for viral growth isx*y >1, wherex describes the number of users invited
(by base user) andy the number of accepting users who join (Salminen & Hytönen 2012). When all
users successfully invite more than one new member, the growth is viral.
111
development have been studied, in particular, relating to open-source plat-
forms: Schilling (2009, 202) asserts that “in the software industry, individual
programmers may work on an open-source software program because it re-
sults in solutions to their own problems, provides an opportunity to interact
with peers and improves their reputation as experienced programmers.” Sim-
ilar, but varied, motives can be found in the UG context, although more re-
search is needed. There is an emerging body of the literature relating to the
motives of the crowd which comes close to this purpose; for example, Dow et
al. (2011), Kittur et al. (2013), Pitkänen and Salminen (2012), Zhao and Zhu
(2012), Zheng, Li, and Hou (2011). Drawing from this literature would open
opportunities in utilizing crowds to solve the cold start dilemma.
4.4.3 Solution: Subsidies
Subsidization is the most commonly considered solution in the platform liter-
ature. By definition, a side is subsidized “when the price it faces is lower than
the price it would face in an independent market” (Bakos & Katsamakas 2008,
173). In the lonely user dilemma, to increase possibilities for interaction, the
startup might subsidize one of the user sets for joining. With regard to the cold
start dilemma, the startup might subsidize contributors of content in exchange
for their efforts, or developers for creating extensions or content creation tools.
Because developers or other contributors might require payment in exchange
for their contribution, subsidies can also include negative prices (Parker &
Van Alstyne 2005).
For example, Evans (2002) argues that “providing low prices or transfers to
one side of the market helps the platform solve the chicken-and-egg problem
by encouraging the benefited group’s participation – which in turn, due to
network effects, encourages the non-benefited group’s participation.” In a
similar vein, Spulber (2010, 7) argues that “incentives induce strategic partici-
pation which resolves the cross-market coordination problem.” Free offerings
are not a new invention. Rohlfs (1974, 33) had already proposed that “[t]he
most direct approach is to give the service free to a selected group of people
for a limited time.” Rohlfs (ibid., 33) further argues that the initial user base
must be sufficiently large to achieve a critical mass, and notes that “half
measures are worse than useless”, because demand will be zero without a
critical mass of users. This feature, stemming from network theory, has been
adopted by the platform literature, so that “[a]n important characteristic of
multisided markets is that the demand on each side vanishes if there is no de-
mand on the others, regardless of what the price is” (Evans, Hagiu, &
Schmalensee 2006).
112
Although Piezunka (2011) specifies that the platform owner might not give
negative prices because of a potential moral hazard problem (Gawer &
Henderson 2007), negative prices have been observed, for example, by Parker
and Van Alstyne (2005), and cited by Mas and Radcliffe (2011) as a strategy
utilized by Paypal. Mas and Radcliffe (2011) argue that negative prices (i.e.,
paying for users to join) helped PayPal achieve a critical mass faster than
would have been enabled only by zero prices. Other startups have applied rec-
ommendation fees to incentivize their users to promote the service to their
peers (Libai, Biyalogorsky, & Gerstner 2003). However, this can be defined as
marketing rather than subsidization, which aims to get users to adopt the ser-
vice by not charging them for its use
89
(Lyons, Messinger, Niu, & Stroulia
2012). Free or negative pricing is possible, although with the precisely identi-
fied threat of spamming.
Moreover, the startup might begin by subsidizing one side and, when it is
secured, then moving the subsidization to the other side. The goal is to reach a
state in which no subsidy is needed as the platform has become self-
sustaining. Caillaud and J ullien (2003) refer to this strategy as divide-and-
conquer, by which the market is divided into two markets (i.e., two-sided
markets) with one being captured by subsidization, after which the other side
will follow, enticed by network effects. However, subsidization might become
a permanent strategic choice to maximize participation in a two-sided market
(Rochet & Tirole 2003). In the context of online startups, subsidization refers
to free models in which the users, or a segment of them, are not charged.
Complete free offerings, or non-paid access and usage of the platform, are
referred to as freefying, whereas offering a free version and a paid version
between which the users can move (i.e., convert or downgrade) is termed
freemium (see Chapter 1.5).
Consider basic subsidization. In a sequential form, the startup subsidizes
one party who is more unwilling to join; after securing participation, the other
side can be acquired, for example, through online marketing. Subsidies are, for
example, offering free trial (i.e., direct monetization) and free premium mem-
bership (i.e., mixed monetization). The problems of freefying are discussed
elsewhere (see Chapter 4.6). However, essentially, freefying does not elimi-
nate the cost of adoption although it eliminates the economic part of it; neither
is it an effective competitive strategy, as it is easy to copy and can advocate
unwillingness to pay. Freefying also creates the monetization dilemma when
the startup is unable to capture economic value from its user base.
89
A more advanced form is a dual-sided referral incentive; for example, Dropbox offers additional
storage space for both the referred and referring user.
113
Table 14 Basic solution of subsidization
side A
Not willing Willing
side B Not willing Subsidize DN*
Willing DN* DN*
*Do nothing
The startup only needs to subsidize when either party, or both, are unwilling
to join. For example, if A is willing to join and B is not, the strategy is to sub-
sidize B and do nothing with A. If both are willing, the startup does not need
to give subsidies, which would indicate very strong demand for the platform.
If both are unwilling, the startup needs to subsidize both, which would indicate
either low demand or high competition.
However, complications arise: what should be done when both parties re-
fuse to join, and the subsidized price is already set at zero? Such is the case
with free models
90
. In this case, as proven with our sample, a potential fallback
is failure, or otherwise applying alternative solutions to solve the dilemma. A
third option might be to set a negative price, and pay users to register, alt-
hough this strategy would be poor for many reasons. For example, the quality
of entrants might be low, the costs of adding users grow linearly, and the
problem of active use impacted by users not joining due to intrinsic motiva-
tion
91
.
Moreover, subsidization can quickly become too costly if the market is
large, as it typically is in online consumer markets
92
. Belleflamme and
Toulemonde (2004) show that even in B2B markets with a limited number of
participants, a platform subsidy can make the platform unprofitable to a severe
degree, which is due to its linear property: each member needs to be subsi-
dized to the same extent. Therefore, the solution would be to subsidize until a
critical mass is reached (e.g., by covering this as a form of marketing expense)
and then allow UG to take over user acquisition, as per the ideal UG model.
Another solution would be discriminant subsidies in which the amount of sub-
sidy varies based on the expected utility of the user, (i.e., to other users in a
single-market platform and to the other side of the market in a two-sided
90
Free models refer to free access and usage (i.e., freefying) and the freemium model.
91
Clearly, the cost of adding users is negative, even without including potential subsidies, when
having considered user acquisition costs, such as advertising and other marketing costs. Therefore,
negative price attached with indirect monetization (i.e., free users) and marketing costs can quite
easily be detrimental from a financial perspective.
92
Consider that the total cost of subsidies, c, would be dependent on a fixed cost x (per user) that
would grow exponentially by factor a alongside exponential growth of user basey (subject to a), so
that c = (xy)
a
. Hence, exponential growth would lead to an exponential cost of subsidization.
114
platform); in other words, paying some users more to join. However, this
would raise questions of fairness and might be counter-productive.
The question of subsidies can also be turned around by considering the pro-
spective members’ perspective; perhaps opinion leaders could be recruited by
offering them other types of incentive, for example, of a social nature. Eventu-
ally, the tactics might vary but the principle remains: members who produce
the most content are more valuable in content platform, and members who are
socially connected and willing to propagate other users to join are valuable to
the social platform
93
. Moreover, the user is not necessarily acting in accord-
ance with economic rationality principles when creating content or joining a
platform; even an exchange platform can involve social motives for participa-
tion. Alas, the type of utility they seek might be more fragmented and hetero-
geneous than generally understood by startups or platform theory. Tapping
into social motives might, in these cases, provide gains that exceed the effect
of financial incentives or cost savings.
Furthermore, moving from free to paid products can become problematic.
As noted by Brunn, J ensen, and Skovgaard (2002), penetration pricing is a
common tactic in one-sided markets, although it can have an effect on long-
term profitability. Introducing fees (i.e., ‘bait and switch’), if the form of sub-
sidization is freefying, can be difficult as parties rebel against going from “free
to fee” (Teece 2010). The early platform literature established that the plat-
form sponsor (i.e., platform owner) actively promotes the platform, not only
by subsidizing the cost of its adoption. For example, Katz and Shapiro (1986,
822) define a platform sponsor as one who “is willing to make investments to
promote it”. Further, they argue that when two rival technologies exist, if one
is promoted and the other is not, the promoted one can rank higher in adop-
tion, regardless of whether it is superior
94
in some objective comparison (e.g.,
features). Therefore, subsidization alone cannot be regarded as the optimal
solution for the cold start dilemma as factors other than price of usage influ-
ence adoption. Further, the cost of subsidization might be lost due to rivals
competitively reducing their rates.
Moreover, there is the question concerning which side to subsidize? First,
Belleflame and Toulemonde (2004, 6) argue that “in several categories of two-
sided markets, most agents of one side of the market arrive before most agents
of the other side.” Hagiu (2006, 721) points out that “in the software and vide-
ogame markets, most application developers join platforms (operating systems
and game consoles) before most users do.” In social platforms, the order of
93
The conclusion, therefore, matches that drawn by Li and Penard (2013), in that quality can
replace quantity in the early stage of a platform.
94
This was the case in the early 2000s when Sega first launched its Dreamcast console, but players
hesitated to adopt it due to Sony’s clever promotion of the soon-to-be-launched Playstation 2.
115
entry does not seem to be relevant. However, if this division is assumed, con-
tent creators in content platforms need to arrive first as content must logically
exceed its benefits. In general, the side that derives less benefit from partici-
pation should be subsidized (Curchod & Neysen 2009) as the risk of non-
adoption is assumed greater. However, Parker and Van Alstyne (2005, 1503)
assert that either both sides can be the target in the context of free distribution,
or the side that “contributes more to demand for its complement is the market
to provide with a free good” when network effects are high. This notion does
not consider the risk of adoption but rather the benefit for profit maximization.
Second, some users are more valuable than others, in terms of their network
utility. The two-sided literature terms these marquee customers (Rochet and
Tirole 2003), while in the social networking literature, prominent users are
termed prestige nodes (Evans 2009a); and in marketing such terms as
influencers (Gillin 2009), early adopters (Rogers 1995), or opinion leaders
(Flynn, Goldsmith, & Eastman 1996) are employed; that is, users with whom
many people want to connect. These users influence others to join the
platform.
Therefore, influencers generate significant direct or indirect network
effects; thus, it pays to subsidize their entry (Niculescu & Wu 2013).
Therefore, platforms need to identify and recruit influencers early on both
sides. Finally, Parker and Van Alstyne (2005) assert that the structure of subsi-
dies depends on the industry. They give some examples (ibid., 1496), for
example video streaming services and advertising, in which subsidizing
consumers is an industry norm; and operating systems and videogames, in
which the developers are subsidized while consumers are paying. Therefore,
there might be no universal solution to which party receives subsidies, as
industry conventions need to be taken into account.
4.4.4 Discussion
Subsidization can solve the cold startup problem under some circumstances,
but only as a local solution; there is thefree beer effect based on which giving
a free platform is not viable business unless it is successfully monetized. Price
is not the only matter influencing adoption although it will necessarily limit
the scope of price-related strategies; therefore, even negative pricing might not
be sufficient to solve the cold start dilemma. Subsidization might also lead to a
free-rider problem if the product is structured to enable both free and paid us-
age (e.g., freemium). This takes place by assuming satisficing behavior (Si-
mon 1956) and positive willingness to pay. The user is then able to receive
benefit from the platform without economic cost, while the startup incurs a
116
freefying loss, given that there are users who would have been willing to pay if
no free option was presented. Startups opting for freefying thus perceive the
risk of adoption as greater than the risk of deferred monetization, potentially
by an indirect monetization model
95
.
Consider that the startup is aware of the cold start problem, which, at least
intuitively, it is in most cases. To kick-start the content platform, it produces
some initial content in the hope of initiating the user discovery process that
will lead to users finding, reacting to, and sharing content forward, thereby
recruiting new users. However, if this strategy fails, and the crowds fail to
materialize, the startup will find that it needs more content or that a type of
pivot is required.
However, if its in-house resources do not scale to match content production
or, perhaps, it lacks specific expertise for content production, it might consider
UG, representing a “magic bullet”, the solution to the content problem. How-
ever, this study has already established that UG is not a solution to the cold
start problem (Table 11), but rather the consequence of a need or interest.
Therefore, given that the root cause lies elsewhere, the cold start problem is
not resolved, and the startup returns to the beginning.
It is therefore crucial that a startup recognizes the limits of in-house content
generation, unless its deliberate purpose is to develop a content-production
organization, which is a strategy sometimes termed ‘inbound marketing’
96
. In
a business model reliant on UG, however, in-house content provision cannot
easily extend beyond kick-off due to its costliness; in fact, in the long run, it
would dissolve the benefits of UG if the content community was not be self-
sustained. For leveraging the potential of Internet users in terms of content,
UG is simply superior to any alternatives
97
, which are aggregation and what
Mark Zuckerberg terms frictionless sharing (Darwell 2013), a concept refer-
ring to automated sharing of activities conducted online. At the time of writ-
ing, this approach is in its infancy and facing major resistance regarding pri-
vacy issues.
The route to a solution can, in fact, arise from the fact that not only existing
content, or users, make up for the decision to contribute, or join, but that
expectations of future interactions can attract users to perform first interactions
with the platform. This abstracts the requirement of network effects from
quantity or quality to a signaling problem, essentially an issue of marketing
communications. If the startup is able to communicate the vision of the
95
This is commonly known as “searching for a business model”.
96
The cold start dilemma does not concern a firm that employs in-house content generation,
because it will start deterministically producing content from the beginning. Its challenges relate more
strongly to traffic generation and marketing, both being closely associated in online marketing.
97
Consider replacing all user-generated content in Facebook with editorial content.
117
platform in a credible fashion to, for example, early adopters or “influencers”
and employ this communication to commit them to performing first
interactions with the platform, in this case creating content, the cold start
dilemma is in theory solved. However, user acquisition that relies on content,
as in the ideal UG model, is a major problem, which is why credible signaling
without first-interaction commitment would be fruitless. Thus, a positive
expectation is preceded by awareness, generated by some form of marketing:
Sequence: marketing actions ? awareness ? positive (or negative)
expectation ? adoption (or non-adoption)
In some sense, it would be sensible for the startup to find a mixed solution
in terms of one that is pure. For example, hiring an in-house community man-
ager to coordinate content generation with partners and ensure initial attempts
to achieve a critical mass are successful. Further, the process of content dis-
semination needs to be well considered, so that 1) the startup has ready access
to social media platforms where it can disseminate the content, referred to as
an ‘integrated marketing communications strategy’ (see Mangold & Faulds
2009), and 2) that the website is optimized for participation and sharing to be
as frictionless as possible. However, at the same time, founders need to
acknowledge that technically frictionless does not imply socially frictionless
sharing. Ries (2011) argues that users might assume a social risk related to
sharing their behavior, and therefore the expected propensity to share does not
necessarily materialize.
In theory, the cold start dilemma can be solved by integration into an exist-
ing platform that supplies the much needed users who will generate content.
However, there can be 1) strong intra-platform competition and 2) misalign-
ment of goals between the startup and platform owner, which result in rem-
ora’s curse (Chapter 4.7). Equally, in theory, freefying will remove the obsta-
cle of purchasing as it sets the price at zero. However, it ignores the fact that
the cost of adoption not only includes a financial cost but also the time and
effort of learning a new product (i.e., changing behavior). In addition, when
the solution is effective and new users join, the startup faces the problem of
monetizing the user base. Once free, it is hard to revert to offering paid prod-
ucts without a considerable churn in the user base. As argued throughout the
study, users are not synonymous with customers. This conundrum is discussed
in Chapter 4.6.
A horizontal strategy would be to increase the scope of topics (i.e., econo-
mies of scope) to drive adoption. A vertical strategy would be to find oppor-
tunity niches; that is, unserved content areas, unserved social niches, or buyers
and sellers not being adequately served. Finally, differentiation in terms of
118
features (in simple terms: better execution) can perform an opportunity to grab
customers despite network effects, which is often put forward as the reason for
both Google’s and Facebook’s success, whereby they both provided better
solutions in comparison to alternatives with a critical mass, including Yahoo
and MySpace, both of which are slower and more complex. Thus, in a struc-
tural sense, there are likely to be aspects that are impossible to account for in
the design of a platform such as execution and features, which can have a
stronger influence than network effects
98
. These can either work in favor of or
against the incumbent, depending on whether it has a better product or not.
More research is needed, however, on the critical success factors of competing
platforms’ execution strategies.
It has been discovered that successful adoption addresses overcoming
change resistance relating to routines (Oreg 2003), and that time and effort are
comparable to financial cost (i.e., money) as factors determining the outcome
(Webster 1969). Although the removal of financial cost through freefying can,
in an economic sense, reduce the overall cost of adoption, it cannot be re-
moved as time and effort lead to the necessary existence of a learning curve
(Yelle 1979) that, as stated, is a factor of adoption. This is the reason why the
cold start problem, interpreted as a problem of adoption, cannot be solved by
freefying alone. Even if the subsequent monetization problem were solved, the
adoption problem would potentially return to haunt the startup. Finally, the
potential discrepancy in perception is crucial as it explains why startups per-
ceive freefying as an answer to cold start problems, which is a logical conclu-
sion assuming that the adoption cost only includes financial cost, whereas us-
ers would, in reality, require assistance to decrease the learning curve. Clearly,
from an economic perspective, the learning curve can be increased if expected
benefits are high. Unfortunately, not all startups create products for such a
need that a user is willing to spend considerable time and effort on learning
their systems.
The introduction of negative network externalities that increase with harm-
ful activity is a potential risk. In a two-sided market with indirect
monetization, Internet startups often resort to advertising as their monetization
model. However, advertising typically represents a negative indirect network
effect for end users, which is a manifestation of the “cat and mouse” game
between advertisers and consumers because the latter desire to escape the
former. In contrast, in a one-sided setting, spam and fake profiles result in
direct negative externalities for both parties.
98
As expounded by one of the founders (Dickens 2010): “We were offering information on great
albums and community voting. But other sites like Last.fm and Hype Machine were offering the actual
music. That was a competitive advantage that’s hard to beat, and we lacked a significant user base to
convince enough people.”
119
4.5 Lonely user dilemma
4.5.1 Definition and exhibits
Generally, for users to join a social platform, they expect to find other individ-
uals using it. If none can be found, there is little or no incentive to join the
platform. The logic is equivalent to the cold start dilemma. In contrast, once
the first generation of users have signed up, new users are enticed to join
through the connections of the first group, and so on; startup founders refer to
theviral effect or simply exponential growth. The logic is based on the notion
that the total benefit generated by a social platform can be measured through
the number of connections between users (cf. Metcalfe’s law; see Briscoe,
Odlyzko, & Tilly 2006), and the frequency and quality of activity within these
connections (i.e., the network effects).
The principle of users’ mutual expectations can be demonstrated with a
simple game.
Table 15 Startup platform
S
2
Join Not
S
1
Join S
2
, S
2
-1, 0
Not 0, -1 0, 0
S
2
is the potential number of interactions between members of the platform,
and marks the network effect. Albeit being a bad proxy for network value (see
Subchapter 4.5.2), it is easy to quantify and represents an upper limit for
interactions (Aggarwal & Yu 2012). In other words, parties potentially draw
symmetric benefit from each other’s presence, and payoffs are equal.
J oining has a cost, if not financial then time and effort, which is why ex-
pected non-participation of another party leads to both not joining. Both par-
ties would be advantaged by joining but as it is risky for each of them to do so,
the outcome might be both not joining. This is referred to as the coordination
problem in game theory (Van Huyck, Battalio, & Beil 1990), and describes
well the lack of legitimacy to which the new platform is subject.
In contrast, consider an incumbent platform. This example demonstrates the
importance of a critical mass.
120
Table 16 Incumbent platform (with a critical mass)
S
2
Join Not
S
1
Join S
2
, S
2
S
2
-1, -1
Not -1, S
2
-1 -1, -1
In this case, the incumbent platform already provides a critical mass of us-
ers (or content) for interaction, which is why the dominant strategy for each
party is to join. Even when other users do not join, the entrant receives benefit
from the existing base of users (S
2
-1). Because both parties have the incentive
to join, it is also the Pareto-dominant equilibrium. Therefore, it is much more
difficult for a new platform to attract entrants than for an existing platform
with a critical mass. Consequently, even in the presence of multihoming (see
Chapter 4.8) and low switching cost, the startup can fail to rally users.
However, the basic chicken-and-egg problem becomes more complicated
when introducing dynamic factors, such as time and place. Consistent with the
definition of the cold start dilemma, the lonely user dilemma can be defined as
follows:
In a social platform, when there are no existing users, no new
user will have a motivation to join. Additionally, when there are
no active users at a given time or place, no other users will use it
at that time or place.
If a user has no contacts in a social service, the perceived benefit of the ser-
vice equals zero for that particular user at that particular time or place, re-
gardless of the number of registered users or “static” critical mass, such as
content, that is always available. In practice, these platforms can include social
platforms such as chat services requiring simultaneous presence of parties, and
location-based services whereby the interacting parties need to be available at
the same time and also in the same place.
In the cold start dilemma, the focus is on recruiting new users (e.g., to gen-
erate content) and keeping them active in UG activities. In the lonely user di-
lemma, the focus is on acquiring users for social interactions taking place be-
tween individuals and groups, and keeping this interaction active (i.e., the
problem of active use) while considering the effect of time. As such, at any
given time, not only on average, the platform must have a critical mass to pro-
vide matches and thus be useful
99
.
99
Consider the Facebook platform: if in some given time frame, all of a user’s friends were offline
and had not updated their statuses, eventually the user would permanently stop using the service,
regardless of how many registered users there are.
121
Therefore, the requirement of a critical mass is much more extensive than in
the case of static content
100
. In other words, the demand-side benefit in social
platforms is derived from social interaction (i.e., social exchange) instead or
more or less static content, with the source being topicality, information, en-
tertainment, or other properties of the content. For example, unlike communi-
cation between friends, reviews and videos are not social interaction in a fun-
damental sense
101
. In a content platform, users enter the website for the sake of
the content (e.g., news, reviews, articles, and videos), whereas the lonely user
dilemma is typically associated with social network sites in which availability
of others is conditioned by time and/or physical location. Table 17 exhibits the
dilemma.
Table 17 Exhibits of the lonely user dilemma
Exhibit
[1] "I think you need a critical mass in any community and we didn’t quite achieve that critical
mass. I mean, who wants to go into a forum when there’s really nobody to talk to?" (Warner
2009).
[2] "Lastly, the “real-time problem”. This one is similar to the location problem in that if someone
wasn’t online when you were online, they were no good to you. While the real-time chat aspect
of the application made for some really serendipitous meetings, it also made it harder for people
to gauge the activity of their communities, especially if they logged in at odd hours, people were
set as away." (Bragiel 2008).
[3] "We launched our product and got all of our friends in Chicago on it. We then had the largest
papers in the area do nice detailed write-ups on us. Things were going great. We had hundreds
of active users and you could feel the buzz around it. […] The problem, we would soon find out,
was that having hundreds of active users in Chicago didn’t mean that you would have even two
active users in Milwaukee, less than a hundred miles away, not to mention any in New York or
San Francisco. The software and concept simply didn’t scale beyond its physical borders."
(Bragiel 2008).
[4] "The weakness of the hub strategy was the market players never arrived at the same time. Sellers
would flock but there would be no buyers, or buyers would flock and there would be no sellers."
(Anonymous founder).
[5] "The real tests come at moments like we had about a week after our initial launch. Lots of
people dropped by, told us they loved the site, and didn’t come back. So, there we were, left with
one big question that lead to endless others: Why aren’t they coming back? Is something too
confusing? Is our idea a bad one? Do we just wait and see if they come back later? Do we need
to build another tool?" (Karjaluoto 2009).
100
Here it is assumed that social interaction expires much more rapidly. However, content also
expires. If no new content is added, after a while users will stop using it, although the remaining
content would continue to provide benefit; this is not the case for social interaction. However, in some
cases thetopicality of content approaches the temporality required by social exchange. Consider, for
example, a news portal in which all content has to be fresh.
101
Note that this does not exclude spillover effects between content and social interaction. In fact,
these are generally requisite for UG effects to occur.
122
We can deduce that the number of users required to initiate the self-replica-
tion UG process is often referred to as a critical mass, both in the literature
(see the following chapter) and by practitioners [1].
Thecoordination problem [4] is distinguished from the real-time problem
[2] based on the notion of time. Coordination fails as a result of an overall lack
of participants in the other side. The real-time problem might mean that there
is a potential critical mass in the other side, but that they are momentarily in-
active
102
. As noted previously, the real-time aspect is emphasized in the lonely
user dilemma due to immediacy of social interaction. More precisely, coordi-
nation can relate to the participants’ different needs, which requires under-
standing both sides well and managing their expectations. The timing of con-
verting users can be critical here; thus, if the technology is premature, per-
suading users to join can cause a major disappointment. Basing the platform
design on the premise of self-organization might not take place in reality.
Furthermore, in social environments, match is not simply a question of the
number of members in group A or B, but also their quality (i.e., compatibility).
Match might require a special type of user property relating to, for example,
demographics or offline relations. Not all counterparties willing to interact
will regularly provide a match
103
.
The real-time problem, if defined as ‘getting users on board’, suggests that
solving the cold start problem is insufficient to solve the lonely user dilemma;
that is, registering to a platform does not automatically lead to active use,
without which, the platform will gradually die regardless of adding new users.
This is termed churn in marketing and is parallel to pouring water into a
bucket with a hole. Thus, while loyalty is low, increasing customer base will
only increase cost, relating to lifetime value, as customers constantly abandon
the service. In other words, users of a real-time service need to be simultane-
ously present or coordination will fail. This is crucially different from static
content, whereby coordination is much less affected by timeliness.
The transferability problem [3] implies that a predominant user base in
context A (e.g., location) cannot automatically be generalized as a critical
mass in context B (i.e., another location), even when it fits the notion of criti-
cal mass in its primary context. In particular, the problem relates to hyper-
local platforms such as location-based services. Although our exhibit
102
How is the real-time problem different from the lonely user dilemma? The former is a
manifestation of the latter, in which time is the match-making criterion. However, the lonely user
dilemma can be manifested in relation to other match-making criteria, such as physical location and
preferences. In both cases, the user is “lonely” without an adequate match.
103
However, this type of differentiation also exists in exchange platforms. Consider the following
criteria for match; for example, item being transacted, condition, reputability, and location of the other
party. In general, however, users are more selective in engaging in social interaction with “strangers”
than transacting with them.
123
addresses location, the transferability problem itself can be generalized into
any context in which one group is so distinct from another that direct network
effects will not emerge across the two groups.
Therefore, the startup needs to consider its match-making role
104
and
emphasize user-acquisition based on the development of dynamics between
the groups. For example, if there is a shortage of either female or male mem-
bers in a dating service, more users of the required gender need to be recruited.
When there is disconnection between user bases, for example, niche division
or geographical distance (i.e., local social networks), there is a shortage of
synergy between user segments; that is, no positive network effects arise even
when the groups are connected. Hence, each segment needs to be built indi-
vidually due to proprietary network externalities to that community, although
the transferability problem [3] will not be overcome unless propagated by
members of the community. The benefits for a startup involved in building
multiple communities are therefore limited to learning gains, which can
facilitate replication of critical success factors, and also potential reputation
and brand spillover effects when users in another community become aware of
the platform’s existence, and perhaps start acquiring its community.
Furthermore, users’ homogeneity, defined as similarity of interests, demog-
raphy, or other feature that increases similarity, might influence the perceived
utility of the network by an individual user. These features can include, for
example, location, online status, and similar preferences. Diversification is
needed if the service is match-making between opposite groups with users
looking for counterparts (i.e., buyers for sellers; men for women); thus, homo-
geneity tends to be counterproductive in two-sided markets
105
. However, in
one-sided platforms, users derive benefit from similar users joining the ser-
vice, which implies direct network effects. They might also appreciate com-
plements by other firms, such as plugins, games, or additional content by third
parties within a platform ecosystem, which reflect indirect network effects.
Conversely, when there are users in one side (A) competing for members on
the other side (B), each additional user in A in fact reduces the incentive for
similar users to join. Strictly speaking, assuming that match-making exhibits
rivalry in that connections between members in A and B exclude other
104
Essentially, a marketplace platform is a mediator between two parties, often supply- and
demand-side, so that it offers auxiliary benefits in addition to matching (i.e., coordinating), such as
payment options and vouching.
105
In a sense, this is a trivial observation. The definition of two-sidedness entails the idea that the
groups are distinct. Therefore, similarity merely parallels this state with some actual criteria for
distinctness.
124
connections, this increases competition
106
. Therefore, a high number of
members in A represents negative (i.e., direct) network effects for a prospec-
tive member of A as they are competing for the same resources (i.e., members
of B). The final decision to join is affected by the difference between
perceived negative network effects (i.e., the level of competition) in
comparison to perceived positive network effects (i.e., the number/attractive-
ness of group B). Implications such as these will be further discussed in the
following literature subchapter.
4.5.2 The literature
The problem of active use is expressed in Albuquerque et al. (2012, 407):
“there are usually two (or more) stages in the decision to participate in a user-
generated platform. Users must first opt to visit the site, and once in the site,
they must decide to generate and/or consume the available content.” In other
words, acquiring a user is not alone sufficient to guarantee interaction benefits
to the other side. A similar conclusion is drawn by Xia, Huang, Duan, and
Whinston (2007) who distinguish between the user decision to adopt/join
AND to continue use . Therefore, it is acknowledged that the adoption choice
is not the only requisite for a permanent solution to the chicken-and-egg
problem.
The lonely user problem, therefore, is not only limited to encouraging reg-
istrations or other forms of subscription/joining, from which a startup might
infer that optimizing registration pages (i.e., landing page optimization) is a
top priority, but also to activity taking place after the user has enrolled. As
presented by Boudreau and Hagiu (2009, 171), “Facebook must then activate
the ‘social graph’. Beyond simply establishing linkages among members, it
must keep these linkages active, fresh and compelling”. This is a problem as,
after joining, the success in fact depends on the community’s or platform’s
activity, which is the source of new users. In fact, if the replication rate or viral
coefficient drops below one, the exponential growth stops (Salminen &
Hytönen 2012) and, considering churn, the user base can begin to decrease.
Not only this, but it is also possible that users estimate the degree of activ-
ity, as it is part of their utility function, prior to joining, and employ this esti-
mation, based on the activity they see when visiting the platform for the first
time, as an adoption factor. This would mean that a low anticipated frequency
106
This might or might not be the case, depending on the strategy of users. For example, in a dating
site, a user might stop creating connections to potential dates after finding “the one”. However, it is
also possible that he/she might continue to create further connections.
125
of activity decreases the expected utility. It has been previously argued in this
study that expected benefits are proportional to expected costs; as such, more
time and effort are spent on adopting platforms that are perceived to be genu-
inely useful. This might seem to be a trivial statement but, in fact, it supports
the idea of showing social activity to prospective users in an attempt to per-
suade them to join, which calls for a different strategy to the walled garden, or
bowling pin, strategy often applied to explain Facebook’s success (e.g.,
Spulber 2010)
107
.
Second, the real-time problem can be approached through the concept of
synchronicity. Porter (2004) defines it as the “degree to which a medium ena-
bles real-time interaction”. She distinguishes between 1) synchronous and 2)
asynchronous interaction; the former requires that parties are present simulta-
neously, and therefore corresponds to the definition of the real-time problem.
Asynchronous platforms, such as online forums, enable users to browse and
create messages at their convenience. Porter (ibid.) mentions that a particular
interaction design does not necessarily lead to interactivity among users, as
users might not behave as expected. Hence, the real-time problem can arise if
users are not utilizing real-time features to their advantage. The requisite of
timeliness makes it much harder for the startup to generate matches (see Sub-
chapter 4.5.1) than in a registration-based system.
According to Caillaud and J ullien (2003), users might utilize several service
providers in a case where one platform does not provide a match. Hagiu
(2006) employs the example of Match.com to elaborate the need for registra-
tion and data collection necessary to create matches through some form of
permanence. Overall, when introducing social interaction, as opposed to con-
tent, and time, as opposed to constantly available content, the chicken-and-egg
problem assumes a more complex form.
There are also limitations of network effects both as a concept and as an
automatic solution to chicken-and-egg problems in two-sided markets pre-
sumed by some startup founders. A definition, in the context of social net-
works, is given by Mital and Sarkar (2011, 380): “the probability of a new
user subscribing to an application is proportional to the number of the appli-
cation’s existing users. Thus social networking sites exhibit network effects”.
Network theory has created many constellations of network value. One of the
best known is Metcalfe’s law, stating that the value of a network is propor-
tional to n
2
, where is n is the number of nodes connected in the network
107
Essentially, the question relates to two alternative choices: 1) hide the interaction or show it and
2) let the content be indexed by search engines or not. What has worked for Facebook might, in fact,
be countered by the totally different open garden strategy. As such, this is a vivid example of context
influencing the implication.
126
(Gilder 1993). Later, the law was criticized for the fact that not all connections
are in active use (Briscoe et al. 2006).
For example, Odlyzko and Tilly (2005) refer to the concept of gravity,
which means that local connections are more valuable than those that are more
distant, basically negating the assumption of uniform value. As Samanta
(2009, 3) notes, simply multiplying the market sides is not a realistic measure
for network effects: “in a large network, such as the internet or a credit card,
with billions of potential transactions between buyers and sellers, most are not
used at all. Therefore, it would be wrong to assume that the volume of trans-
actions per buyer will grow linearly with an increase in number of sellers”.
The aim here is not to go too deeply into the discussion of the nature of “net-
work laws”, but to consider how they apply to online platforms.
In fact, the network value debate parallels that in the platform literature. As
such, calls for other criteria in addition to network size have been made by
several authors (e.g., Farrell & Klemperer 2007; Suarez 2005; Birke 2008). In
particular, economists have discovered the limitations of employing the size of
a user base as the predominant proxy of network effects. In his literature sur-
vey, Birke (2008, 24) argues that
"A […] departure from the assumption that total network size
matters can be found in some of the newer empirical papers on
network effects which argue that social networks are mainly
local and that local geographical network size is therefore the
relevant network measure."
Consider that network effects are dependent on contextual factors such as
an industry or market, even a company, as a user base of one company, can
interact more than that of another company and therefore be more valuable.
Suarez (2005, 719) examines network effects at the industry level and argues
that
"[A]n industry that features very strong ties could simply annul
classical network effects […] an industry with moderately strong
ties may allow for both strong ties and classical network effects
to be significant. Finally, an industry in which weak ties pre-
dominate would de facto revert to the classical case, a monolith
that cannot be broken into parts on the basis of tie strength."
Ties are implied by Suarez to increase change resistance (see Coch &
French 1948) or switching costs (Farrell & Klemperer 2007). It is simply
possible that existing industry relationships, or ‘inertia’, exceed the expected
benefits of network effects. In online consumer platforms, barriers for switch-
ing might also involve habitual or behavioral elements relating to learning
costs (Farrell & Klemperer 2007) and loyalty; the latter of which being possi-
bly based on seemingly irrational logic such as brand preference (see
127
Thompson & Sinha 2008). For example, Maicas, Polo, and Sese (2009) refer
to personal network effects which underscore relativity. According to Maicas
et al. (ibid.), personal networks influence switching behavior. Intuitively, it
can be concluded that personal networks are most likely to affect adoption and
usage of a social platform.
In particular, the limitation of network effects in social platforms is noted
by Boudreau and Hagiu (2009, 171): “members care only about their relevant
network rather than the aggregate network [so] growth is about expanding a
mosaic of social networks rather than scale per se.” This argument will be
discussed below. Second, consider Evans’ (2002) requirements for network
effects: 1) one agent’s adoption of a good, product, or service benefits other
adopters, and 2) “his adoption increases others’ incentive to adopt.” Evans
(ibid.) refers to these two effects as the total effect and marginal effect. How-
ever, both assumptions can be contested in online contexts. First, consider in-
clusion of negative network effects. An example is mentioned by Boudreau
and Hagiu (2009, 171): “Facebook has the challenge of minimizing negative
interactions on its platform, ranging from irrelevant interactions, those that
are inappropriate to the context, all the way to ‘fraudsters’ and illicit activ-
ity.”
Clearly, all platform types considered in this study are subject to negative
externalities emerging from low-quality participation: in content platforms, it
is a risk of spam and low-quality content, in social platforms the aforemen-
tioned negative interactions, and in exchange platforms, the risk of frauds and
scams. Consequently, a startup is obliged to monitor the quality of its user
base. In the ideal model, this is assumed to be the task of the self-organizing
user base. However, as argued, the ideal in real cases rarely occurs. In fact, the
quality of a platform’s user base can be defined as a determining factor of in-
teraction.
For these reasons, individuals might prefer to keep their personal core net-
work small (Mital & Sarkar 2011). This, again, is not taken into account when
modeling only positive network effects and network size. In fact, the reverse
can take place: privacy can explain why some social platforms are more suc-
cessful than others, and why open inclusion is not always the optimal choice
(cf. Boudreau 2010). If network effects are tied not only to quantity, but in-
volve qualitative aspects, the idea of benefits being proportional to thenumber
of users is not fully compatible with the notion of the positive nature of net-
work effects. For example, consider spam (i.e., unwanted email messages):
when it increases due to more participants joining email markets, the basic
premise of network effects would imply that user benefits increase, but clearly
this is not the case. Even by restricting the application of the premise to
128
positive network effects, the implications might not be as fruitful as originally
thought.
For example, when a large number of medium-quality content is produced,
the marginal effect on overall benefit is much smaller than if the same content
was of high quality, or, if medium-quality users join a network, other prospec-
tive users are less interested than if high-quality users join
108
. Therefore, alt-
hough strictly speaking “true”, and the network benefit is added proportion-
ally, the proportion is mediated by quality. Obviously, quality is hard to de-
fine
109
, especially as it can differ according to the preferences of individual
users. In sum, applying the notion that network effects increase in proportion
to quantity, of users or content, can be regarded as unnecessarily delimiting
110
.
Although the economic literature tends to focus on positive network effects
(see e.g., Birke 2008), users in online markets are easily affected by negative
effects such as the aforementioned spam (Hinde 2003) and harassment in so-
cial platforms, cluttered or obtrusive advertising on content platforms (Rumbo
2002), and the risk of fraudsters in an exchange platform. Therefore, the issue
of negative network effects is crucial in the online context, and should be un-
derstood as a potential cost for adoption, or even a barrier.
Moreover, the relationship between installed user base and propensity to
adopt is not straightforward. However, heterogeneity of a given user base can
impose strong effects on a user’s willingness to adopt. For example, consider a
20-year-old user who is looking for a date in a large network, A, but each and
every one he/she finds is unsuitable. Then he/she switches to platformB with a
user base of less than half of network A’s, and finds a date immediately.
Clearly, there are noreal network effects for her in network A, even if it has a
larger user base than network B. Of course, matching involves a possibility of
chance, but perhaps network A was targeting elderly users of age 60–70, and
this did not match with the searcher’s intent. Regardless of the reason, it is
insufficient to assume that network effects are uniform across all users, and
that they originate only from the size of the user base.
Similarly, we can consider a one-sided platform where convention argues
that adding a new user benefits other users of the same type. However, by ap-
plying the previous logic from a two-sided market, the marginal increase of
network utility by adding user x in sidek of the platform for user y in network
sidef (i.e., the opposite side) will only be positive if the additional user is a
108
Note that the medium condition is deliberately framed, because low would indicate a negative
network effect, whereas medium is a small positive network effect.
109
See Reeves and Bednar (1994) and Zeithaml (1988) for some attempts.
110
According to the author’s understanding, however, the misconception is not due to the investors
of these laws but to the fact they have been later redeployed to contexts missing fit. For example,
Metcalfe’s law originally described Ethernet connections, not social networks.
129
suitable match (e.g., of the correct kind, quality, type, or age). In other cases,
the increment is either neutral
111
or negative. For example, Farrell and
Klemperer (2007, 1974) consider that
"sers of a communications network or speakers of a lan-
guage gain directly when others adopt it, because they have
more opportunities for (beneficial) interactions with peers."
This condition, albeit intuitively making sense, only applies if individuals
are interested in communicating with speakers of other languages. If one has,
for example, learned a language only to communicate with a spouse, which
might be the case in international marriages, again, adding new members will
not increase the network utility in praxis
112
. Additionally, note the concept of
‘peers’ in the definition. It might be that the new speaker of the language is a
baby on the other side of the world; clearly, proximity in the network influ-
ences how realistic it is that network effects become useful.
Even in electronic networks, such as the Internet through which all users
have a theoretical ability to connect with everyone else, private social net-
works are often concentrated, for example, by region, age, or preferences
(Thelwall 2008), as if to mirror offline proximity conditions. We rarely see
random connections in social life, but they havepurpose that leads to unique,
and perhaps unpredictable, network topographies also in a relatively friction-
less network, such as the Internet. In other words, we cannot automatically
assume that the existence of network size suffices for adoptionper se, and that
if it suffices for user x, it will also suffice for user y. Therefore, the ‘chicken’
might not purely be in the size or structure of the network, but in the underly-
ing differentiating factors
113
of targeted users such as, for example, age, loca-
tion, or preferences.
Banerji and Dutta (2009, 605) come to the same conclusion by stating that
“positive externalities arise from the specific patterns of interaction between
groups of users” as opposed to those that are general. However, given indirect
network effects, the issue is slightly more complicated than this. While direct
utility derived from a distant user might be zero, or diminishingly low, the
111
In fact, from the three types of network effect (i.e. positive, neutral, and negative), neutral,
which is when the user is indifferent to the other side, can be subtracted. As an aggregate condition
this cannot apply, otherwise all users would be indifferent and there would be no interaction, and,
thus, no platform. However, neutrality intuitively applies at an individual level. For example, merely
adding some random member to Facebook is unlikely to increase a user’s interest in the platform.
112
Note that on average, someone from the language community is likely to be interested in
communicating with the new person; therefore, in aggregate, the benefit (slightly) increases.
However, this does not affect the skeptic’s or spouse’s benefit from the network, as it is not the
network that made him/her adopt the language.
113
In fact, even in Farrell and Klemperer's (2007) definition, this is considered beneficial. However,
this elaboration is included to highlight the nature of “beneficial”, to avoid the misinterpretation of
size of user base as being synonymous with success in achieving network effects.
130
indirect effect of having a lot of “zeros” adds up to a user base, which provides
useful network externalities even when a user is interested in interacting with a
selected few. This is elaborated by Arroyo-Barrigüete et al. (2010, 646) who
employ the example of Microsoft Messenger in stating that “at a global level,
[adding new irrelevant users] would have an influence due to the fact that, if
there are a large number of users, the system will be improved over succeed-
ing versions (indirect network effects).” Hence, the pool of other users, alt-
hough individually meaningless to the user, is beneficial as a group.
As in the previous example, locality (in a geographic sense) can be associ-
ated with social factors and therefore to the strength of network effects. Suarez
(2005), referring to Rogers’ (1995) innovation studies, asserts that a local
community might refuse to adopt a technology with a larger overall user base.
This seems to confirm the idea of the transferability problem from one context
to another: a local optimum does not generalize to a global optimum. In other
words, as observed by some startups in the sample, major dominance in one
market might not endure when expanding to other markets. Again, the size of
user base does not matter per se, and a smaller community with high
consistency
114
can endure external pressures to adopt technological innova-
tions.
Moreover, Suarez (2005, 712) claims that thestrength-of-ties perspective is
commonly employed in social network theory to imply superiority or inferior-
ity of connections, so that “relationships among the different actors in a net-
work can be broadly classified into some basic types: strong versus weak ties
and direct versus indirect ties.” If adopting this perspective, weak ties would
seem to indicate less significance to platform adoption
115
. This principle was
also observed by Rohlfs (1974): “If an individual’s demand is contingent on a
few principal contacts’ being users, there may exist many small self- sufficient
user sets.” Indeed, users behave differently than telephone networks, in which
the total utility provided can be parallel to the number of connections it is able
to create between randomly connecting users. However, even there the usage
of those connections is not uniform, or totally random; for example, some
connections are more actively utilized, whereas others are more rarely uti-
lized
116
. Therefore, there is purposefulness behind the usage of pre-existing
connections; similarly, there is determinism in the way users generate
114
Defined as strong network effects, and no multihoming.
115
However, Birke (2008) gives an example concerning why the availability of remote nodes is
important. Consider, for example, the emergency number; although rarely utilized, access to it is
important to users of the telecommunications network.
116
“Texas and Maine may have less to communicate”, as noted by Briscoe et al. (2006).
131
connections
117
, such as friendships, in social platforms. For example,
friendship requests from unknown individuals tend to be rejected on Facebook
(Pempek, Yermolayeva, & Calvert 2009). Alternatively, the reverse can occur,
and the user actively employs social platforms to find people previously
unfamiliar to him or her. It is this ‘hidden intent’ that complicates modeling
network effects.
Consequently, it is important to note that network effects areone criterion
for adoption, and therefore cannot fully explain it. The superiority of a product
can explain cases such as Google overcoming Yahoo, and Facebook over-
throwing MySpace, even in the presence of network effects. This principle is
mentioned by Farrell and Klemperer (2007, 2012) who elaborate the Qwerty
versus Dvorak case by asserting that “f the penalty is small, switching […]
could be privately inefficient for already-trained QWERTY typists even with-
out network effects. And evidently few users find it worth switching given all
the considerations including any network effects.” In a similar vein, Suarez,
(2005, 711) notes that an “obvious explanation” for deviations from expected
network effects are other factors, such as price or technological characteristics.
In other words, network effects are not the only source of benefits derived
from adopting a platform; the platform might have stand-alone value
(Kristiansen, 1998), or its adoption is the consequence of word-of-mouth.
Thus, adoption does not result from network effects but from social effects,
such as the bandwagon effect (Henshel & J ohnston 1987), or from the plat-
form’s stand-alone value. This suggests startups should not overly rely on
network effects as the ultimate goal but also focus on other areas, such as
technology, differentiation, and marketing. All of these are potential variables
explaining network effects, as opposed to network effects magically appearing
from user interaction. This separation principle is at times forgotten also in the
literature, which considers network effects to fully explain adoption while
wondering why, in the presence of strong network effects
118
, one competitor
overcomes another.
The confusion between differentiation and network effects might arise from
the specifications of analytical models; for example, differentiation is not al-
ways considered. In such models, network effects portray a situation in which
two equally differentiated and equally marketed platforms are competing,
which might not always be the case in the real world, by pricing and size of
117
Rochet and Tirole (2005, 5) stress this point through the concept of usage: “The cardholder and
the merchant derive convenience benefits when the former uses a card rather than cash; a caller and
a callee benefit from their communication, not per se from having a phone; and so forth.”
118
This is a special problem of applying the formal models to empirical contexts. In a formal sense,
strong network effects are defined so that “[n]etwork effects are strong if they outweigh each
adopter’s preferences for A versus B, so that each prefers to do whatever others do” (Farrell &
Klemperer, 2007, 2018).
132
user base. Logically, such models tend to give results that favor the im-
portance of network effects.
It is possible to take alternative approaches to defining network effects; for
example, consider a proposed definition:
"In general, the higher the strength of network effects in a two-
sided platform (defined as propensity to find a match), the
smaller the initial user base required to grow."
This definition overlaps with viral marketing theory, in which the growth
idea is approached, for example, through the concept of aviral coefficient (see
Salminen & Hytönen 2012). In such a model, the propensity to send and ac-
cept invitations defined whether the network experiences exponential growth.
Coincidentally, perceived network effects would affect both propensities, so
that users are more likely to send invitations, as added users increase their
benefit, and accept them because of the benefits provided by the existing net-
work. Again, however, we stumble upon the ‘minimum requirement’ concept
(i.e., critical mass), as the expected benefits of the network are considerably
lower if the user base is insufficient to convince invitees to join. Defining the
network effects as the propensity of a user to find a match relating to his/her
intent avoids the ‘quantity versus quality’ problem. In this case, both can con-
tribute positively or negatively to the emergence of network effects.
The idea of propensity (i.e., probability) is somewhat compatible with
Roson (2005) who argues that, from a demand perspective, two sources for
network externalities can be identified: 1) single interaction externality, in
which “matching quality improves when more alternatives become available”
and 2) multiple interaction externality, in which every user gets a benefit from
every interaction by other pairs. However, Roson (ibid.) also assumes that
quality is improved by the number, not type, of participants.
Finally, we consider the notion of critical mass, as it can be regarded as
requisite for the presence of network effects. A two-part definition is offered
by Suarez (2005, 718). First, “A critical mass occurs at the point at which
enough adopters have chosen a particular technology that the technology’s
further rate of adoption becomes self-sustaining.” This indicates self-propaga-
tion that was found to be central in the ideal UG model. Suarez (ibid.) adopts
the second part from Katz and Shapiro (1992): “the system with a lower in-
stalled base enjoys a significant advantage for instance, newer and superior
technological capabilities”. Therefore, incumbents are seen to possess excess
inertia (Farrell & Klemperer 2007) regardless of their quality. There are recent
cases in the online market, however, that neglect this ‘inertia’. Most currently
dominant platforms employed as examples in this study have been
133
early-movers but not first-movers, first-mover disadvantage can arise, for
example, as a result of technological inferiority
119
. Evans and Schmalensee
(2010, 21) show that “even without fixed costs or economies of scale, platform
businesses typically need to attain a critical mass when they are launched in
order even to survive”, which would indicate that a ‘go big or go home’
strategy is suitable for these markets, and that heavy investments in early
marketing to acquire a user base would be required. A critical mass can
therefore be defined as the condition between functioning (i.e., realized) and
non-functioning (i.e., theoretical) network effects, so that:
No critical mass ? no network effects
Critical mass ? positive/negative network effects
This is based on the assumption that the network must have some type of
minimum participation, not necessarily relating to quantity, before it can pro-
duce benefits or matches for users. This is also the position of Shapiro and
Varian (1998, 184) who argue that: “Network externalities make it virtually
impossible for a small network to thrive. The challenge is to overcome the
collective switching costs that such a network requires to grow.” In the net-
work economics literature, successful achievement of network effects, in a
competitive setting with no multihoming, is often termedtipping (e.g., Katz &
Shapiro 1994), meaning that once a rival technology reaches a particular
degree of adoption, all industry participants migrate to support that technol-
ogy. Such is the case concerning standards (e.g., Farrell & Saloner 1985).
When the winning design emerges and industry participants become aware
of its predominance, they will start supporting it and abandon other standards
(Katz & Shapiro 1994). Therefore, multihoming can exist, in this type of set-
ting, only until the dominant technology has been decided, after which all
players will single-home. However, if there is interoperability between tech-
nologies, which is not the case for mutually exclusive standards, the behavior
might be different and the market might eventually become oligopolistic; that
is, comprise many standards or platforms (e.g., Hagiu and Wright 2011). More
precisely, this can be seen to be the case for many online platforms that all
have internal consistency in terms of critical mass (i.e., users who are com-
patible with each other to the extent that matches can easily be created), but
none has an absolute dominance of the market
120
.
119
Technical problems are commonly acknowledged as one of the reasons for Friendster losing to
other social networks.
120
For example, there are several competing online dating platforms, which is possible due to users
multihoming or otherwise preferring one platform to another.
134
However, at the same time, this is contrary to Shapiro and Varian’s (1998)
argument, which supports the notion that industries with strong network ef-
fects gravitate to the leading platform. The difference is in the notion of ‘in-
dustry’ versus ‘market’, and standards and compatibility. If we change the unit
of analysis from an industry, in which it makes inarguable sense to employ
compatible technology through standardization, to different market verticals,
we can better understand the outcome in the case of many two-sided markets,
including those online.
Nevertheless, a critical mass cannot be seen to equal stable market domi-
nance, although it might imply niche dominance, assuming lower competition.
Niches, therefore, are local targets as opposed to mass markets. In the study’s
sample, a startup faced special problems of hyper-local communities (see the
previous chapter). In particular, the issue is that dominating one local niche
does not automatically lead to advantages when moving beyond the niche; that
is, network effects do not generalize. This becomes a problem when a partic-
ular niche is insufficient for viability or the goals of the startup.
Thus, thenon-transferability of network effects can become an issue in the
differentiation strategy, if segmentation is too narrow, or if markets are natu-
rally isolated (e.g., cities in some circumstances) or exhibit social dissimilari-
ties. Hence, the existence of the transferability problem, as described in the
definition, has been noted in the literature.
4.5.3 Solution: Remora
As noted in Subchapter 4.5.2, the practice of a platform drawing benefit from
the user base of another platform bears similarity to envelopment in the plat-
form literature. It is described as follows (Eisenmann et al. 2011, 1271):
"Envelopment entails entry by one platform provider into an-
other’s market by bundling its own platform’s functionality with
that of the target’s so as to leverage shared user relationships
and common components. Dominant firms that otherwise are
sheltered from entry by standalone rivals due to strong network
effects and high switching costs can be vulnerable to an adjacent
platform provider’s envelopment attack."
Envelopment, in this sense, is an aggressive strategy that aims to replace the
target platform by making its pre-emptive assets, namely high switching cost
and strong network effects, ineffective. However, as previously noted, the
remora strategy aims at becoming a complement instead of a substitute. Due to
power imbalance in a remora setting, the remora is not aiming to replace but
accommodate the host (see Figure 12).
135
However, Eisenmann et al. (2011) name the following industry examples of
envelopment: PayPal ? eBay; Google Docs/Chrome/Android ? Google
search. Therefore, envelopment can take place as a market diversification
strategy in which the platform draws from its own user base in a different ver-
tical. Clearly, Google is a strong example of such an expansion, as it currently
provides more than a dozen services that more or less relate to its core func-
tionality (i.e., search). In a similar vein, Ries (2011) notes that “many […] vi-
ral products didn’t really build their own working ecology: they colonized
someone else’s. That was true for PayPal cannibalizing eBay, YouTube and
MySpace, and could still be true of Slide, Zynga, or RockYou – we’ll see.” The
“colonization”, as argued, is beneficial for both parties in the case of comple-
ments.
Even in eBay’s case, PayPal provided a useful auxiliary service that was
missing from eBay’s own selection. eBay tried to introduce its own version,
termed BillPoint, which is an example of substitution by the host, but failed
due to better marketing strategies employed by the remora. According to Mas
and Radcliffe (2011, 311):
"PayPal faced a constant fight with their ecosystem host, eBay,
once eBay realised that some of the value from their customers
was going to PayPal. As the owner of the platform, eBay sought
to derive significant advantage from integrating its own payment
engine into its marketplace website. […] Ultimately, PayPal
knew that their continued success was going to be dependent on
eBay’s not shutting them out of their auction website entirely, so
they sought increasingly to diversify from eBay auctions […],
and eventually sold out to eBay."
This case demonstrates well the hazards of a remora. First, competition
from the host platform that, after discovering the remora is grabbing a dispro-
portionate amount of its user value, reacts by introducing a substitute, then, the
danger of being denied access and, finally, being absorbed by the host, in a
form of acquisition. It also demonstrated diversification as a means to counter
the threat of the host, which will be discussed in the following chapter. Nota-
ble industry cases include Twitter’s acquisition of Tweetdeck (Parr 2011b) and
Facebook’s acquisition of Instagram (Constine & Kutler 2012). In Facebook’s
case, its strategic goal was to make the acquired company a complement for its
own platform, thus enabling better photo sharing. In Twitter’s case, the ra-
tionale had to do with substitution. In other cases, the remora is not so lucky,
with the host eventually absorbing its product ideas as features in the subse-
quent release.
Eisenmann et al. (2009, 225) elaborate the problem of substitution-through-
absorption:
136
"Dependency also came with a danger […] Many software exec-
utives wondered if they could trust ambitious Microsoft employ-
ees with sensitive information. Executives […] had seen the in-
novative features of prior software show up as features in later
versions of Microsoft’s products [and] wondered if their em-
ployees’ conversations with Microsoft’s technical staff would
contribute to seeding a future competitor."
Whereas, in a standards and technology setting, a firm can protect its intel-
lectual property rights (Church & Gandal 2004), abstract ideas are not patenta-
ble in most countries. Features absorbed by Microsoft include, according to
Parker and Van Alstyne (2008), disk defragmentation, encryption, media
streaming, which is also employed as an example by Eisenmann et al. (2011),
and Internet browsing. Absorbing has also been documented in the Web envi-
ronment (Parker & Van Alstyne 2008, 2): “Whether through internal develop-
ment or acquisition, coercive or not, platforms such as Apple, Facebook,
Google, Intel, Microsoft, and SAP have routinely absorbed valuable features
developed by ecosystem partners.” Parker and Van Alstyne (2008) go on to
state that absorbing developers’ innovations can reduce their incentives to
continue developing for the platform, given they remain uncompensated
121
,
and can lead to developers exiting the platform. A case of developer flight in
the videogames market is described by Eisenmann et al. (2009, 151):
"If an incumbent has been too aggressive in extracting value,
demand- and supply-side users may rally around entrants […]
When it dominated the console market, Nintendo dealt with
third-party game developers in a hard-fisted manner. Conse-
quently, developers were pleased to support Sony when it
launched the PlayStation console in 1996."
Consequently, it is a useful tactic to attract remoras for the host not only be-
cause they increase network benefits for the demand-side, but because they
can provide ideas on how to improve the core platform. At the same time,
over-exploitation of their ideas provides a negative incentive to continue col-
laboration. A similar effect can take place in the demand-side where, for ex-
ample, privacy issues might become a concern
122
.
Exclusion, or denial of access, is similar to a counter-envelopment strategy
depicted by Raivio and Luukkainen (2011, 79): “The envelopment threat re-
fers to a case where we have several platform providers and common
121
In contrast, some startups welcome acquisition as an exit strategy, and therefore cases of
absorbing-through-acquisition would increase the incentives of such startups to join.
122
This was noticed by Facebook as a competing project termedDiaspora gained much success in a
crowdfunding platform; Diaspora aimed to attack Facebook’s privacy-related weaknesses by
promising a more secure platform, although it later failed.
137
customers. In this situation one provider may try to exclude other platforms
from the market.” It might not necessarily be that the platforms are competing
in the same market, but if the host perceives the remora taking advantage of it
without reciprocity, it can resort to denial of access. For example, such was the
case of Craigslist (i.e., general exchange platform) cutting access of AirBnB
(i.e., specialized exchange platform), a peer-accommodation service
123
. In
particular, the value of content can be seen in a platform strategy termed
walled garden (e.g., Berners-Lee 2010), in which the platform owner restricts
the visibility of information to search engines
124
that are known profiteers of
capturing economic value from third-party content.
According to Eisenmann et al. (2009), specific results from the hold-up
problem might include 1) limiting quality of cross-platform transactions, 2)
raising prices, and 3) charging for interoperability rights. In particular, the
platform host might deliberately “limit the quality of cross-platform transac-
tions to maintain differentiation” (Eisenmann et al. 2009, 138). Charging for
interoperability rights as a strategy for the host to improve its margins can in-
clude charging startups for API units
125
. In turn, a startup is initially a weak
platform that hinders its ability to utilize the same strategy of monetizing API
usage. The following figure illustrates how the standard remora strategy dif-
fers from envelopment.
Figure 12 Remora and envelopment
Consider that, as soon as the number of complements exceeds a particular
threshold (e.g., they fulfill categories needed for inter-platform competition),
the group itself is enough for the platform owner, and individual developers
become expendable. This logic can be employed to explain power asymmetry
123
In this case, cutting access to posts would negateaggregation, whereas preventing auto-posting
would remove automatic access to users.
124
For example, most Facebook content cannot be accessed by Google; therefore, Google started its
own social networking platform, Google+.
125
For example, such a practice is operated by Google (i.e., Google AdWords).
Envelopment
Remora
138
between the developer community, which is fragmented and competing, and
the platform owner that is concentrated and a monopoly within the platform.
When dissatisfied, the host can more or less replace individual complements
as new entrants are interested in taking their place. However, if the entire pop-
ulation turns against the host for some reason, the momentum will be reversed
and the host will quickly lose popularity. In the case of envelopment, the rela-
tionship is more hostile, and the counterparty is interested in replacing some or
all of the host’s functions.
This chapter has shown how remora’s curse relates to the literature. We
have employed theoretical constructs such as hold-up to understand the prob-
lem, and cited industry examples of both the remora strategy and the host’s
opportunistic behavior. However, we have also argued that the host’s oppor-
tunism is curbed by 1) multihoming behavior and 2) the risk of adverse selec-
tion to restrain its use of power.
4.5.4 Discussion
When the number of users is zero and the benefit derived from the product by
the user depends on the number of other users (i.e., direct network effects
apply), the benefit provided by the product to additional users also equals zero.
Therefore, no rational users join. This notion is applicable to both one- and
two-sided markets
126
. Within two-sided markets, under indirect network ef-
fects, when users of one (i.e., supply) side have no match from the other (i.e.,
demand) side, there is no incentive to join, regardless of the number of other
users in the same side, andvice versa.
As the number of relevant contacts increases, so does the perceived benefit
of joining the network, even up to the point where the user feels social pres-
sures to join. Both buyers and sellers need to be present in an exchange plat-
form, or marketplace, otherwise the service provides no benefit for the user. In
contrast, when users fail to actively utilize the platform, it becomes a cold
platform, which explains why MySpace lost to Facebook, regardless of the
critical, although static, mass acquired. It also implies that a critical mass
should not be measured by the number of connections (i.e., Metcalfe’s law),
but as the number of interactions across time. The number of interactions indi-
cates active use, which is as substantial a problem for a startup as the cold start
problem.
126
Their difference is that, in one-sided markets, users are of the same type (e.g., friends in a social
network), whereas a dual-sided market addresses two different types of user (e.g., sellers and buyers)
who are influenced by one another.
139
The literature survey provided useful approaches to understanding the na-
ture of the lonely user dilemma. Central to this is that the first models focused
on how the network appears to an outsider, not how the connections are em-
ployed. It was later discovered that the structure and usage of a social network
are closely tied to one another; the connections emerge between people who
share interests or are also associated in offline interactions, which forms the
potential of the network usage. Consequently, realized, not theoretical, inter-
action is required for the UG benefits at which a startup aims. Inarguably,
without active use, the potential of the platform does not sustain, which is de-
picted as liquidity in the early platform literature and as theproblem of active
use here; possible even when the cold start problem is solved.
The problem of active use is, in fact, quite important as it relates to sus-
taining the perception of utility. While strong network effects can quickly
grow a user base, with low loyalty the platform faces considerable churn,
which leads to a reverse effect of exponential growth, a sudden decay. For ex-
ample, if lead users or top contributors of content exit the platform, their fol-
lowers might easily follow, after which the followers of those followers exit,
and so on. Diffusion and churn can therefore be perceived as being subject to
herd behavior. This obviously implies that the startup is advantaged by keep-
ing the community active, after it has been established.
The real-time problem adds a temporal component to the lonely user di-
lemma. Put simply, at any moment in time, there needs to be a critical mass of
users active in the platform. As the potential moments of time extend well be-
yond storing the interactions (e.g., messages saved in an inbox), potential con-
nections are much scarcer than in a non-real-time social network. Essentially,
the startup has a major coordination problem, as people do not necessarily
connect to the platform at the same time as other users. At best, this pattern is
hard to control
127
. The real-time problem concerning simultaneous entry of
sides is consistent with the literature. If the requirement of simultaneous pres-
ence is strong, there is no solution other than efficient coordination of user
flows; for example, based on the hour of day, day-part advertising, and focus-
ing on the side lacking members. There is likely to be a threshold up to which
the lack of counterparts is tolerated, with the user retrying at a later time.
Additionally, even the usefulness of content can have a temporal nature, be-
cause new and different content is required by new and different users. Re-
gardless of this “timeliness of utility”, the lifetime of content can be regarded
as long, even infinite. While it remains in the Web server, it exists, as opposed
to users who log out, and can be indexed and shown by search engines,
127
An exception would be when the platform is tied to a specific event (e.g., a football game); then,
all participants would self-coordinate to be simultaneously present.
140
thereby providing visitors with consistent value
128
. Obviously, some pieces of
content (e.g., “classics”) hold their utility longer than others.
Moreover, it was found that the conceptualization of network effects,
namely their contrast to the size of the user base, might be problematic. Un-
derstanding the motive to join seems to be associated with deviations of eco-
nomic rationality, even if optimizing social utility replaced profit-seeking.
Therefore, altruistic behaviors and group dynamics can come into play. The
critical mass phenomenon, the author believes, cannot be explained only by
network effects and the growth of utility. Other phenomena such as the band-
wagon effect and herd behavior are most likely involved. Whether they are
perceived as rational or not, is a different matter.
If network size is not the correct metric to indicate network effects, and lo-
cality has been perceived as more important, what does this imply? In a sense,
that 1) expectation of close proximity (e.g., city; local community) is useful
for the startup, but 2) under UG, the network structure spawns from the user
base. These are complementary remarks in the sense that, given the infor-
mation based on which one can assume greater consistency among the
network unit, the startup should target that unit as opposed to employing mass
marketing, even when mass markets are the goal
129
Relating to other reasons for adoption, consider a user joining the dominant
platform instead of a startup platform. In some cases, the motive for non-
adoption might not lie within the stronger network effects of the dominant
platform but the in the fact that it offers a better solution. This is an important
notion, which arises from the marketing assumption that people join platforms
to satisfy a need, as opposed to the platform assumption that they join because
there are network benefits. Here, theories can confuse the direction of causal-
ity; namely, that network effects would be the sole motive for avoiding
switching while, in reality, loyalty is a more complex phenomenon
130
. As
noted in Subchapter 4.5.1, loyalty is a critical factor in sustaining the level of
active use, which is also a requisite for the realization of network effects.
The platform literature refers to marquee users, whereas this type would be
described in the marketing literature as opinion leaders, individuals character-
ized both by a large number of connections and their associatedsocial capital.
However, the essential conclusion is that marquee users are not only able to
increase the network effects of a given platform but regular “J ohn and J ane
128
An illustration of the lifetime value of content can be seen in J acob Nielsen’s (1998) article, in
which he describes how visitors to his popular blog articles have increased over time.
129
However, for example, mass advertising can bring legitimacy to some platforms, which is
especially important in the B2B context. An interesting case study on the topic can be found in Mas
and Radcliffe (2011).
130
Other criteria such as usability, trust, and features of the platform are likely to affect
adoption/loyalty.
141
Does” can increase the utility of the network as they provide a better match for
other “regular” users, which contrasts somewhat to the traditional marketing
convention of targeting ‘lead users’
131
. Network effects are a coordination
game, not a pure game of numbers.
In sum, through a literature survey, it has been shown that there are specific
cases in which quantity does not apply as a proxy to network effects. These
include, at least, the following:
· The presence of negative network effects (e.g., spam; socially
undesirable connections).
· Whenever the user base is non-homogeneous; that is, in most cases of
content and social connections in which both access and quality of
interaction are important.
· When there areother motives for adoption, so that network effects do
not explain platform adoption, switching, multihoming, or refusal to
adopt to the full extent.
Thus, the effectiveness of network effects in accumulating a user base de-
pends on circumstances such as industry, quality, interoperability, and multi-
homing behavior. Users act purposefully and are influenced by social motives;
modeling their connections without taking motives into account portrays an
inaccurate image. The author has also proposed an alternative approach that,
rather than merely the number, involves the probability of matching as a proxy
for network effects. This approach is conceptually more accurate as it gener-
alizes across a large variety of perceptional factors that influence network ef-
fects such as, for example, demographics, preferences, intents, and location.
In brief, it can be concluded that not only the number of connections is of
consequence, so is their interdependence. This can be termed quality or rele-
vance, depending on the conceptual perspective. However, it is fundamentally
linked to the fact that people value some social connections more highly than
others, and that some connections are more frequently utilized than others.
Understanding these statements from earlier theories can provide useful
guidelines for Internet startups.
131
However, it is not suggested that targeting lead users would not be a worthwhile investment of
marketing efforts; simply that, in the context of network effects, their increment to the network utility
is proportional to the increase of adding probability of matches, not the size of their network.
142
4.6 Monetization dilemma
All companies at some point must start generating revenue to
remain viable. (Lincoln Murphy)
4.6.1 Definition and exhibits
The monetization dilemma occurs when a startup needs to decide whether to
offer its platform for free at the loss of business viability, or charge for the
access and/or usage at the loss of users’ willingness to join. In other words,
willingness to join (WTJ ) and willingness to pay (WTP) are in conflict.
The following table presents exhibits of the dilemma.
Table 18 Exhibits of the monetization dilemma
Exhibit
[1] "This post attempts to summarize the [startup’s] story: how we got to be the most heavily used
browser synchronization service in the world and yet still find ourselves pulling the plug."
(Agulnick 2010).
[2] "For four years we have offered the synchronization service for no charge, predicated on the
hypothesis that a business model would emerge to support the free service. With that investment
thesis thwarted, there is no way to pay expenses, primarily salary and hosting costs. Without the
resources to keep the service going, we must shut it down." (Agulnick 2010).
[3] "The thesis of our business model […] was that there was a need for video producers and
content owners to make money from their videos, and that they could do that by charging their
audience. We found both sides of that equation didn’t really work. […] Video producers are
afraid of charging for content, because they don’t think people will pay. And they’re largely
right. Consumers still don’t like paying for stuff, period." (Diaz 2010).
[4] "[E]ven if enough people wanted the product, the business model around it is something which
we haven’t been able to figure out. We have the product’s version 2.0 sitting ready […] but we
do not see a clear exit yet, so are hesitant to launch it. Being blogged about major tech blogs
[…] we already got that love. If we stayed out in the market more – we’d probably get more
‘love’. But ‘love’ can only keep the servers humming for so long
." (Anonymous founder).[5] "The experience has made me ask myself almost every time I see a cool web app – ‘OK, but how
will it make money?’, and if it can’t, then it would not be more than a short-lived dream for its
founders and backers." (Anonymous founder).
[6] "I felt like getting into the monetization stage was going to be long and difficult. And it was one
of those businesses where I liked the idea, but I didn’t think about monetization before I started,
because it was kind of a sexy idea, for me at least. And, I got some traction. I ended up with a
few thousand subscribers in a few weeks with the help of some larger companies that were
helping me out at the time. And, I kind of realized that to make my first dollar was going to be a
long time away […]." (Warner 2012).
[7] "Despite having over 200 beta testers at launch, it proved difficult to convert them into
customers. My prices started at $10/month, and though in my eyes this was a bargain, my
product didn?t demonstrate enough value to enough of my market quickly enough to justify the
operational costs of the business and my personal expenses." (Newberry 2010).
143
As the monetization dilemma relates to generating revenue, it takes place
independently of the size of user base ([1], [2], [4], and [5]), regardless of how
substantial this is; therefore, it concerns even popular platforms, such as
132
. Consequently, popularity among users does not automatically lead
into financial success, unless successful monetization occurs. Other key tenets
of this dilemma are that the willingness to pay (WTP) of users in online
platforms is low [3], monetization requires time and effort [6], and even low
prices may not be "low enough" to attain WTP [7].
The dilemma is based on two critical conditions, which here are termed the
payment and revenue hypotheses:
a. Payment hypothesis: If a startup offers a paid product, it acquires zero
or very few customers; the potential risk here beingillusion of free.
b. Revenue hypothesis: If the startup offers a free product, it earns zero or
very little revenue (i.e., problem of free).
Seemingly, the startup cannot win. The dilemma therefore addresses diffi-
culties of both direct monetization (i.e., impossible to gain users) and indirect
monetization (i.e., impossible to create business); the former being impossible
under the premise that users refuse to pay when charged for access or usage of
a platform, and the latter under the premise that the startup is unable to
execute a successful model of indirect monetization even after gaining users,
which leads to an unviable business in the long run
133
.
The following table displays a choice matrix of the dilemma.
Table 19 Monetization dilemma simplified
Paid product Free product
Result No users No revenue
Major underlying assumptions, in economic terms, include high substituta-
bility between products (i.e., competition), low switching cost
134
between
them, so that users can easily switch providers and therefore cannot be locked
in, for example, by ‘bait and switch’, and strict homogenous price sensitivity
132
Twitter’s monetization troubles are well known in the industry (see e.g., Wired 2008).
133
Although, in the short term, this problem can be removed through venture funding; the goal
being to “capture market, monetize later”. This strategy, however, depends on successful
implementation of the indirect monetization model, not assumed in the dilemma’s premises.
134
“A consumer faces a switching cost between sellers when an investment specific to his current
seller must be duplicated for a new seller” (Farrell & Klemperer 2007, 1977).
144
whereby all customers always choose the lowest price, so that users move
from paid to free. Further, it is assumed that differentiation has no impact on
WTP. If even one of these assumptions were incorrect, one of the hypotheses
would fail, and the dilemma would dissolve. Consider the assumptions: first,
substitutability refers to the possibility of replacing the startup’s product with
competing products, which might or might not be true, totally depending on
the product. If the startup has created something that no competitor can
replicate, its product cannot be easily substituted and it would then be fair to
counter the first hypothesis and assume that users would be willing to pay for
the product, assuming that the differentiating property is something they want;
mere differentiation is insufficient.
From this abstraction, the nature of differentiation can be deduced. Whether
or not there is differentiation is irrelevant unless its nature leads to positive
WTP. The payment hypothesis argues that WTP is zero, thus any condition
that negates this argument invalidates the dilemma. The premise of competi-
tion assumes an association between competition, differentiation, and WTP,
which are practical issues that the startup needs to consider in determining the
validity of the hypothesis in its case
135
.
Price sensitivity is a key assumption, and a simple game demonstrates its
meaning. Consider, for example, a game in which players can choose between
free and paid versions of a product. The free version is of inferior but suffi-
cient quality, whereas the paid version is of better quality although costly.
Both players are price sensitive; that is, price is more important to them than
quality.
Table 20 Price sensitive (both)
B
Free Paid
A Free 2, 2 3, 1
Paid 1, 3 1, 1
The paid version can be more desirable if price is not included. However,
because it is, the dreaded cost makes price sensitive users prefer the free over
the paid version. Both users want to avoid a situation in which they are paying
and the other one is free riding. The free rider’s payoff is 3, because the
135
This estimation is required ex ante as the startup needs to decide on the monetization model.
However, the decision is not necessarily irreversible, although moving from free to paid might be
more difficult thanvice versa.
145
paying party helps to keep the platform free. The dominant strategy for each
player is to move from paid to free (1 ? 2 and 1 ? 3). Hence, neither of them
will pay.
Now consider the reverse, when the other player (A) is not price sensitive,
but prefers the premium features of the paid version.
Table 21 Quality sensitive (A)
B
Free Paid
A Free 1, 2 1, 1
Paid 2, 3 2, 1
Because A is not sensitive to price but quality, and the paid version offers
better quality, A will always prefer the paid version, regardless of B’s choice.
This is similar to generators always preferring to produce content, except in a
standalone sense. Whereas generators derive benefit from the existence of
consumers (i.e., positive indirect network effects), paid users are indifferent to
free users. Free users, however, enjoy benefit from paid users, becausein the
long term they help to keep the platform free. Therefore, the free user’s payoff
is 3 when there are paid users. From A’s perspective, however, there is no free
rider problem as he/she is not price sensitive. He/she will not switch from paid
to free, and B will not switch from free to paid; thus, there is a stable Nash
equilibrium
136
. Consider, however, that the free users now have an incentive to
keep paid users on board. The startup can leverage this as a part of their UG
strategy. Moreover, it becomes amarketing problem for the startup to find us-
ers who are more quality- than price-sensitive, and a product-development
problem to provide such quality that satisfies them.
Second, low switching cost implies that any lock-in mechanisms are weak
and, if fees are introduced, users can easily switch between alternative plat-
forms. This invalidates ‘bait and switch’ type of solutions; for example, first
offering free product and then charging for it. Given price sensitivity and low
switching cost, the user would abandon the platform when fees are introduced.
In the case of online platforms, switching costs are typically reduced by high
interoperability (e.g., API access), advanced import/export functions offered
by platforms that exclude proprietary storage of data, and the relatively small
136
Neither player can improve, given the choice of the other player (Nash 1950).
146
learning curve of new platforms as Web services might follow the same con-
ventions (see Cappel & Huang 2007).
Third, price sensitivity implies that users choose the platform based on
price and will therefore prefer free platforms to paid ones, even if this means
sacrificing some quality or features, a type of behavior termedsatisficing (see
Simon 1956). Similarly, if a platform is initially free, to solve the cold start
problem, but later turns to paid, to solve the monetization problem, that is, ap-
plying the bait and switch strategy, users will exit the platform. Given the
number of substitutes, they will always have a fallback. Therefore, if the as-
sumption of price sensitivity is correct, the payment hypothesis is true. Price
sensitivity sets it so that users have low WTP. Note that these premises are
neutral with regard to multihoming (Armstrong 2006); users might multihome
or not, but the introduction of fees would prevent adoption.
Fourth, it must be noted that if a startup is successful in indirect monetiza-
tion, it will not fall into the problem of free and the dilemma will dissolve.
That is, indirect monetization is only a solution when it is successful; a priori,
this can be difficult to determine, which is why the author hypothesizes that
startup founders might be more likely to underestimate their ability to directly
monetize and overestimate their ability to indirectly monetize.
Moreover, the notion of competitiveness in the definition is important, but
only indirectly. As noted, competition only matters if the lack of it is not due
to a lack of demand; that is, in markets where there is no demand, there is no
market although, at times, startups plan to create new markets. Second, the
offsetting factor to competition if differentiation, by which firms are able to
overcome competition. However, differentiation only matters if the differenti-
ating factor is perceived as more beneficial by the user. Some startups are able
to articulate differentiation although in praxis the difference can be trivial to
end users.
Finally, the hypotheses are subject to false confirmation (see e.g.,
Chesbrough 2004), meaning that their truthfulness might be incorrect. Con-
sider a hypothesis that is rejected even if it is true
137
or, conversely, a hypothe-
sis that is accepted when it is false. This relationship between validation and
truth is depicted in the following table.
137
Depending on the ontological position, truth can be defined as an objective property of the nature
(e.g., a law of physics) or an interpretation of the ideal condition (i.e., “social law”).
147
Table 22 Truth and assumptions
Real i t y
Bel i ef s
True False
True True False positive
False False negative True
To elaborate, in some cases users might exhibit behavior that is non-price-
sensitive: behavior that exists both offline and online. For example, every day,
some users pay Spotify for access to music, Netflix for access to movies, and
Dropbox for virtual storage space
138
. In cases where there is a motive to pay
but the startups abandon the alternative based on the payment hypothesis, it is
subject to a special case of confirmation error (i.e., false positive; see Table
22) that the author terms illusion of free (see Chapter 6.4). Under this condi-
tion, the startup is choosing freefying by default as it makes the assumption
that, without facts and contrary to the truth value, users would not be willing
to pay for the use of product, content, or access to a platform
139
. Therefore,
there is a risk of false positive: Users are willing to pay for the product, con-
trary to the founder’s belief. This is quite a significant risk as a startup faces a
choice horizon that might include creating products for which a proportion of
customers would pay, which is an opportunity neglected in the case of the
false positive, and the choice horizon becomes constrained to free offerings.
Therefore, the false negative might not only harm monetization but lead to
neglecting the discovery of other types of product and market space, otherwise
termed missed opportunities.
As previously stated, the dilemma will dissolve if the startup is unable to
successfully implement the indirect monetization model, as Google achieved
by aggregating Web content. There are, however, empirical foundations (i.e.,
reported by failed founders) to believe that indirect monetization is not as
straightforward as many founders assume, which will be discussed in the fol-
lowing subchapter.
138
It must be noted, however, that none of these services incorporate UG as the content production
mechanism. Nevertheless, this is irrelevant with regard to the premise of WTP.
139
In brief, WTP is assumed to correlate with willingness to join. Were the reverse shown, there
would be grounds to discard the dilemma.
148
4.6.2 The literature
In addition to the scholarly literature, there has been substantial discussion
among startup practitioners on the sustainability of free models (see e.g.,
Murphy 2011; Chen 2009; Kopelman 2009). Therefore, both academicians
and practitioners acknowledge the problem. This subchapter will focus on the
conditions of monetization in two-sided markets or, as put by Evans (2003),
“internalizing the externalities”.
As noted, the problem of monetizing online offerings has been identified in
the literature. For example, Beuscart and Mellet (2009, 166) note that viable
businesses must be “built around free access to content and principal
services; economic strategies mostly depend upon the sites’ ability to monetize
their growing audiences”. Also in the platform literature, Rochet and Tirole
(2005, 6) state that “[m]anagers devote considerable time and resources to
figure out which side should bear the pricing burden, and commonly end up
making little money on one side […] and recouping their costs on the other
side.” In a similar vein, Teece (2010, 172) argues that “[w]ithout a well-
developed business model, innovators will fail to either deliver or to capture
value from their innovations. This is particularly true of Internet companies,
where the creation of revenue streams is often most perplexing because of
customer expectations that basic services should be free.”
The technological nature of the Internet, as a communication network, is
compatible with creating platform-type of businesses. Clearly, technology
provides an apt solution for coordination problems that are difficult to handle
via manual coordination and negotiation (see e.g., Argyres 1999), which in-
creases the feasibility of entry. Second, various open-source platforms com-
pete with commercial platforms (Mian et al. 2011). While open source as a
phenomenon generally increases social welfare (e.g., Lerner & Tirole 2004),
open-source platforms are perceived as substitutes to commercial platforms,
and therefore create pressure for free offerings. It is often assumed that the
marginal cost of distributing an information good is close to zero, or negligible
(e.g., Shapiro & Varian 1998; Niculescu & Wu 2013).
However, supporting users in a platform is not equivalent to this assump-
tion. Free users are associated with marketing cost (i.e., user acquisition), ser-
vicing costs (e.g., bandwidth), and support cost (i.e., customer service)
140
.
Therefore, although low, the marginal cost of free users is not zero (Murphy
2011). How close to zero it is depends on the cost structure of the individual
startup; in any case, subsidizing free users comes with a cost. Furthermore, the
140
Although it can be assumed that the ‘free’ property in a good makes an average user less
demanding compared to paying customers.
149
cost is directly proportional to adding users, which means that if there is an
exponential growth of users, the costs also grow exponentially. Hence, a suc-
cessful startup, in terms of user growth, typically requires external funding as
it needs to cover expenses from growing the user base without being able to
monetize it directly. This has been the case for most of the famous Internet
platforms: Facebook, Google, Twitter, and even PayPal (Mas & Radcliffe
2011). Consequently, basing the free pricing strategy on the information-
goods argument does not appear to be valid.
Nevertheless, there are properties that can validate it, which relate to the
nature of two-sided markets. Pricing in a two-sided market can deviate from
that in a one-sided market, in that equilibrium pricing might not be equal to
marginal cost when examining the sides in isolation (Evans 2003). In contrast,
it is a part of a larger equilibrium where both sides are concerned. This makes
subsidization, and making a loss in one side, a possible strategy
141
. In their
model, Parker and Van Alstyne (2005, 1494) show that “even in the absence
of competition, a firm can rationally invest in a product it intends to give away
into perpetuity.” It is assumed that the losses will be recovered either by taxing
the other side of the market, or raising prices after the platform reaches market
dominance. Theoretically, premium sales should cover, and exceed, the costs
of serving free users, so that the firm is profitable (Parker & Van Alstyne
2005). As the user base benefits from a positive feedback-loop, free users at-
tract paid users who, in turn, attract more free users, and so on. This charac-
teristic of the network effects makes it difficult to estimateex ante which price
level is correct; that is, how much subsidization the platform can provide and
still become profitable in the long run.
Despite conflicting perspectives, the argument for low cost of digital ser-
vices is strong (Parker & Van Alstyne 2005, 1503): “This strategy also takes
advantage of information’s near zero marginal cost property as it allows a
firm to subsidize an arbitrarily large market at a modest fixed cost.” Based on
this study, setting up the platform cannot be termed “modest fixed cost”.
However, in a similar vein, Fletcher (2007, 221) notes that “the prices charged
on one side of the market need not reflect the costs incurred to serve that side
of the market.” Finally, Evans (2002, 68) argues that “there is not necessary
relationship between price and marginal cost on either side of the market. In
fact, the price on one side of the market could be well above marginal cost
while the price on the other side of the market could be below marginal cost.”
However, this implies that the rule of marginal cost being zero for information
goods does not apply for pricing the premium product, as there is a
141
However, “[l]osing money initially to buy penetration can also be an important phenomenon in
one-sided networks” (Evans 2002, 57).
150
disconnection between the two sides. In effect, the startup needs to consider
both the acquisition and serving cost of the premium side and also that of the
free side. This is a central conclusion from the disconnection of sides (cf.
Evans, 2002
142
).
Therefore, employing the marginal cost theory of information goods is not
appropriate for two-sided online platforms, the price structure for which is
dual-sided, comprising 1) fixed development costs, 2) acquisition cost of free
users, and 3) service costs (e.g., bandwidth; customer support) of free users, 4)
acquisition cost of paid users, who convert instantly, 5) conversion cost of
paid users; (i.e., cost of converting activities, minus cost of acquiring free us-
ers), and 6) service cost of paid users. This price structure needs to be consid-
ered in pricing strategy, which, as demonstrated by the study’s sample, is un-
fortunately not always the case. Furthermore, in the case of subscription-based
charging, as is the case for many so-called SaaS startups (Xin & Levina 2008),
the pricing strategy needs to consider the expected lifetime revenue of the av-
erage customer, resulting from the lifetime and service level chosen
143
, time
required to convert free users to paid customers, the percentage of this conver-
sion, and the average lifetime of a paid customer.
In the practitioner’s side, Anderson (2009) claims that “practically every-
thing Web technology touches” will end up as free for consumers, as the mar-
ginal costs are approaching zero and prices are approaching the marginal cost,
and as “there’s never been a more competitive market than the Internet”. Alt-
hough not a part of academic debate, this perspective has proved popular
among startup practitioners and managers, even reaching some kind of para-
digmatic status
144
. However, a counter-argument to this claim can also be
found, and it comes from Bekkelund (2011, 16) through a simple but powerful
connection to Hal Varian’s earlier contribution:
"Based on the arguments in Varian (1995) it is likely that the
prediction of free products on the Internet in Anderson (2009)
only yields for purely competitive markets, i.e. when there are
“several” producers of an identical commodity. On the other
hand, it is not necessarily true for markets with monopolistic
competition, i.e. when there are several somewhat different
products, some of which are close substitutes."
142
“Indeed, a key feature of these [two-sided] markets is that, because the product jointly benefits
two parties, there is no basis for separating benefits or costs.” (Evans 2002, 9).
143
Online startups often apply so-called ‘tiered pricing’, by which the customer chooses a service
level. For example, level A gives Y features, and a paid level B gives Y+n features. As customers
frequently switch between the service levels, the average lifetime value of a customer is a mixture of
A and B; generally lower than price level B times the customer’s lifetime.
144
The future will show if this is a temporary phenomenon, or a true change in business logic.
151
Indeed, to assume free is to assume commoditization. Although claimed by
Anderson (2009), and followed by many startups, there is no definitive proof
of the sustainability of the model, even more importantly of its necessity, if
WTP and its association to differentiation strategies can be shown in the real
world. In fact, it is observable that consumers are in fact paying for online
content
145
. It is therefore not categorically true, at this point in time at least,
that all industries converge to free.
The connection made by Bekkelund (2011) is therefore crucial as it implies
that differentiation is a potential strategy against the monetization dilemma. In
fact, Anderson’s (2009) assumption of strict price elasticity abandons the use
of marketing as a differentiating factor in both 1) communicating products and
2) acquiring information on the needs of customers to create differentiated
products. Such a fallacy occurs when marketing is not perceived to contribute
to price elasticity. As economists posit, advertising has the potential to create
“artificial product differentiation” and change tastes (i.e., preferences), and, as
a consequence, advertised products face less elastic demand, associated in the-
ory with higher prices (Beuscart & Mellet 2009).
However, a static set of assumptions such as these can be countered by ar-
guing that regardless of differentiation through marketing or product diversifi-
cation, rivals are able to compete away these benefits rather quickly. Arroyo-
Barrigüete et al. (2010, 644) argue that “[winner-takes-all] does not mean,
however, that competition is scarce; it is quite the opposite, in fact: the com-
petition can be very intense until a company succeeds in establishing its tech-
nology as the dominant one.” Indeed, the inclusion of competitive dynamics is
required; some aspects relating to this are further elaborated in Chapter 4.7.
Moreover, if offering a free product, the user base cannot be substituted in-
definitely, so the problem culminates in decisions such as “which side to
charge?” and “how much?” If the startup is able to create network value be-
tween two sides in the market, it should also internalize that value (Rochet &
Tirole 2003). Therefore, depending on the platform firm’s perspective, its
pricing strategy has to, or can, consider network effects; that is, the indirect
benefits derived by parties from interacting with one another.
Inherently, online markets seem to provide a fruitful ground for free offer-
ing, while making monetization difficult. According to Luchetta (2012), these
can include 1) low technical and financial entry barriers, and 2) strong
network externalities. While the latter premise is subject to specificities, such
as markets and user behavior (e.g., multihoming), entry to online platform
markets is inarguably easier than, for example, offering a shopping mall
145
Refer to the annual reports of Zynga and Spotify for anecdotal but valid evidence on discarding a
categorical rule.
152
platform. While this might increasehalf-hearted efforts and thereby lead to a
naturally high mortality rate
146
, it can also erode serious competitors’ price
levels. In particular, whether or not there are network effects does not, strictly
speaking, matter.
Consider a founder who believes there are network effects to be achieved in
a given market, and therefore sets pricesgratis to obtain a critical mass before
rivals. Whether users multihome or not, or how strongly network effects actu-
ally affect adoption, thus plays no role in the outcome; the startup will make
free offerings. In a sense, network effects can be red herrings, and so the con-
cept is a double-edged sword for a founder who fails to understand its impli-
cations.
It remains true, however, that theoretical support for critical mass is strong
and it is widely regarded as the reason for subsidizing users: “one way to do
this to obtain a critical mass of users on one side of the market by giving them
the service for free or even paying them to take it” (Evans 2002, 50). At the
same time, the fundamental notion is to refer to two-sided markets; however,
the strategy only makes sense if the startup is able to recoup loss leader in-
vestments on the paying side. Rochet and Tirole (2005, 3) imply that the pur-
pose of the two-sided market, in a commercial sense, is to charge total prices
that cannot be negotiated away by participants: “The platforms’ fine design of
the structure of variable and fixed charges is relevant only if the two sides do
not negotiate away the corresponding usage and membership externalities.”
Indeed, if the parties were able to negotiate the externalities, the market would
be feasible. If introducing fees to either side would lead to a collapse, this
would represent a “natural cause of death” because failure is a proper response
to a lack of demand
147
.
We can therefore differentiate ‘true’ or genuine demand from superficial
demand, which is present only when offered free. This is similar to the case of
giving away free beer. Such a business would obviously be successful on the
first day, but if the next day or a few days later it introduces fees and none of
the customers return, there might not be demand at the set price
148
. Applied in
the context of online startups, if the price is zero and still there is no
146
Given that Internet startups might require little sunk investments apart from learning, there is
also a lowexit barrier. In fact, exit is a positive strategy if the learning accumulated can be redeployed
in another startup that is expected to succeed better.
147
This remark relates to long-term scenarios; in the short term, firms might apply penetration
pricing and other loss-leading tactics. However, in the long run, demand will determine the venture’s
viability.
148
In other words, reducing price to zero can introduce pseudo-demand, which can be eliminated by
any price rise.
153
attendance, this might imply there is no demand at any price
149
. However, a
much more plausible explanation, based on the author’s analysis, is that the
startup simply lacks awareness to generate such feedback that would enable it
to develop the platform in the correct direction, eventually gaining legitimacy.
In any event, it is imperative that adoption is preceded by both price
considerations and awareness (i.e., marketing).
The literature lends support to the idea that a lack of demand results in dis-
appearing two-sided markets: “an important characteristic of two-sided mar-
kets is that the demand on each side vanishes if there is no demand on the
other-regardless of what the price is” (Evans 2002, 50). At the same time, su-
perficial demand is not considered as it is demand for the product, albeit in the
“free beer” sense, and would vanish if fees were introduced. Inarguably, if
both sides of the platform are unwilling to pay, it cannot be viable unless it is
non-profit, public, or venture-funded.
In a similar vein, Tajirian (2005, 1) recognizes the efficiency of platforms
in coordinating exchange, and therefore their newness, but still posits the plat-
form should be able to extract rents from its services:
"[A platform] is more efficient in facilitating the exchange coor-
dination than a bilateral relationship between buyers and
sellers. Nevertheless, for the existence of an economically viable
market, the marketplace must to be able to derive economic
profits from facilitating the coordination by appropriately
charging each side of an exchange."
It might also be the case that there is demand, which the startup is unable to
capture. This would be the case when the platform is too open and relies on
users’ goodwill to compensate it for its services. Given rationality, if users
gain advantage (e.g., financial, time, or effort) in bypassing the platform at the
transaction stage, while utilizing its services at the match-making stage, they
will take this opportunity. Therefore, platforms that are unable to capture rents
become easy free-riding targets. As noted by Roson (2005), the interaction
between users might not always be perfectly observed, or it might only be a
part of the interaction in the platform, while continuing elsewhere
150
. For
example, match-making services such as dating sites only benefit while people
149
If there is lack of demand at any price, the lack of demand can be said to be genuine and there
might be no need in the market for that product. In startup language, such a condition can be termed
“vitamin syndrome”, whereas highly demanded products would be “pain killers”.
150
Consider the author’s personal experience of utilizing a freelancer platform to run an auction in
the platform, and then contracting the developer outside the platform. In this case, the platform was
not bypassed, as they charged a listing fee, but had their fee been commission on the transaction, the
author would have been tempted to free-ride.
154
are searching for partners; after a match is made, users no longer need to uti-
lize the service
151
.
Paradoxically, the more efficient the platform is in providing matches, the
less it earns if it is unable to capture rents. Moreover, even when it captures
rents based on transactions, a relationship between users might develop that
will bypass the platform in their future interaction of a similar kind (Rochet &
Tirole 2005, 13): “Buyers and suppliers may find each other and trade once
on a B2B exchange, and then bypass the exchange altogether for future
trade.” Note that this observation is similar to the example given earlier of
‘ActivityGifts vs. Gidsy’. In brief, private knowledge can become an issue.
However, offering some slack on this condition can be appropriate if other UG
effects on average compensate for the lost taxable interaction. Such would be
the case when users still propagate the platform, give feedback, or assist other
users; compensating for opportunistic behavior. In any case, it can be seen that
the choices relating to a platform’s openness (Eisenmann et al. 2009) can
influence its ability to appropriate coordination services.
Moreover, there can be psychological aspects defending the free strategy;
namely, introducing fees, any fees, might result in a disproportionate negative
effect on users’ willingness to adopt (WTA). Shampanier, Mazar, and Ariely
(2007), for example, argue that the benefits of a free product are more highly
appreciated than paid products, regardless of their difference in quality. This
lends support to the idea of satisficing (Simon 1956), so that users fallback on
free products if they provide at least a somewhat satisfactory solution to their
problem (i.e., “get the job done”
152
), despite their inferior performance. This
would indicate considerable friction in adopting paid platforms if substitutes
are available, even if inferior. However, Pauwels and Weiss (2008) study a
successful transitionfrom free to fee, and conclude that a charge can be made
for content, even when the theoretical reference price is zero.
However, even if not appreciated by users, fees might be indirectly im-
portant. As such, pricing can influence the quality of the user base that, in turn,
influences the benefits users derive from the platforms. Thus, prices can have
an indirect effect on increasing network effects, and pricing is an important
connection as an adverse selection control mechanism (Akerlof 1970;
Cennamo & Santalo 2013; Dushnitsky & Klueter 2011). Therefore, pricing
can be regarded as a governance mechanism that filters out low-quality par-
ticipants from the platform. Removing low-quality interaction that influences
151
In some cases, this poses a moral hazard for the platform. For example, Hagiu and J ullien (2011)
discuss a price-comparison site’s incentives to divert searchers toward more profitable products.
152
Consider a mobile app for note-taking versus pen and paper. Clearly, they are not competitors in
the application marketplace. However, from a user’s perspective, they can be substitutes. In this sense,
substitutes and indirect competition can be similar (see Chen, Esteban, & Shum 2008).
155
perceived network effects is critical, and can increase user satisfaction,
loyalty, and finally, the basis for rent capturing. Prices can also signify quality
to users; consider the abundance of free services, and the problem of
determining quality in the absence of prices
153
.
In particular, low-quality users might generate negative cross-group exter-
nalities that would harm the other group of users. This can occur regardless of
which side is being charged. For example, too much low-quality advertising
on Facebook might turn users to different social platforms, whereas low-qual-
ity visitors resulting from Facebook advertising might turn advertisers to other
platforms such as Google’s AdWords
154
. Therefore, the lack of pricing (i.e.,
two-way free access) can lead to a situation of adverse selection, which is det-
rimental to the survival of the platform. There are two important issues here:
first, in a freemium model, is there a spillover from free users to paid users so
that the former might, under some circumstances, create negative externalities
for the latter (e.g., congestion; use of support resources)? If this is the case, the
freemium model risks adverse selection, as paid users become annoyed by the
presence of free users.
Second, price is not the only mechanism with which to prevent adverse se-
lection problems. For example, authentication through identity can reduce
spam caused by anonymity. This was discovered by the popular technology
blog TechCrunch; after introducing obligatory authentication through reveal-
ing identity when logging-in to Facebook, abusive comments decreased
(Burns & Blesener 2013). The theory being that, in the presence of social
penalties, people avoid abusive behavior when their identity is revealed,
whereas anonymity enables such behavior. Some support for this notion can
be found in Lea, Spears, and de Groot (2001). However, the latter is limited to
the inherently low quality of maliciousness, whereas quality problems can
arise regardless of malicious intent (see the earlier discussion in Chapter 4.5).
Sides can also be considered as separate markets so that, for example, there
are both developer and consumer markets. As such, the competitive dynamics
can result in interesting findings. For example, Chakravorti and Roson (2006)
consider a situation in which there are two competing platforms, A andB, that
compete over two market sides, x and y. They show that a price decrease by
platformA in one side of the market (x), apart from competition, will lead to a
153
Note that with the lack of prices, the startup omits revenue from its measures of success. If
growth of revenue is replaced by growth of user base, the startup replaces customers with users as a
proxy for success, and therefore commits to a fallacy of false popularity (see Chapter 3.4 for users and
customers). Briefly, as users do not reveal a product’s economic viability, or genuine demand, their
use as a decision criterion can give incorrect information on where to allocate resources.
154
This example is not only hypothetical. In the online advertising industry, it is commonly
acknowledged that Facebook ads are of lower quality than search advertisements on Google. As a
result, Google’s advertising revenue was ten times larger than Facebook’s in 2011.
156
price increase in the opposite side
. This is the basic notion of subsidization(Rochet & Tirole, 2003). However, in a competitive setting, platform B will
respond by reverse action; that is, by increasing the price of x to capture that
market, and lowering it for y to regain losses. This is shown in Table 23,
which depicts the platforms’ strategic choices.
Table 23 Strategic pricing (adapted from Chakravorti & Roson 2006)
Platform A Platform B
Side x Lower prices Raise prices
Side y Raise prices Lower prices
Consequently, lowering the price for side x will gain relative advantage
over the other platform in that segment, and vice versa, and acquiring market
share there will attract more users to sidey, which can be taxed based on the
network value provided by side x. Note that this assumes sidex is willing to
pay; it might be that introducing fees offset the perceived network value.
The major contribution of the literature is that the two sides of the market
are interconnected, and that the startup might produce a loss on one side. This
principle is paraphrased by Parker & Van Alstyne (2005, 1498):
"If the increment to profit on one complementary good exceeds
the lost profit on the other good, then a discount or even subsidy
becomes profit maximizing. Free-goods markets can therefore
exist whenever the profit-maximizing price of zero or less gener-
ates cross-market network externality benefits greater than in-
tramarket losses."
However, in terms of monetization, this is not much help. Farrell and
Klemperer (2007, 2020) provide a more useful approach: “It is efficient to
subsidize a marginal adopter for whom the cost of service exceeds his private
willingness to pay, but exceeds it by less than the increase in other adopters’
value.” This effectively implies, transferred to the monetization context, that
subsidization can be a profitable strategy while the price paid by premium us-
ers exceeds the overall subsidization cost for free users, including acquisition,
serving, and support. Equally, the same applies for as long as advertising reve-
nue, when monetizing through adverts, exceeds the subsidies.
Finally, it is relevant to draw an analogy to dotcom failures as some argu-
ments presented here exhibit similar features to those made at the time. Essen-
tially, in the dotcom era (ca. 1999-2001), two-sided markets were known as
electronic marketplaces. Most of the literature addresses the failure of B2B
marketplaces with, arguably, a similar conclusion to that regarding the B2C
157
context, given that the platforms have similar dynamics. Some of the failure
factors included, for example, 1) lack of quality indication, so that buyers
could not distinguish between reputable and non-reputable sellers, 2) exces-
sive competition on price among platform supply-side participants, 3) brand
dilution, and 4) existing industry relationship (Evans 2009a). Describing the
hype at the time, Evans (ibid., 115) states:
"Various researchers forecasted that B2Bs would come to ac-
count for a large fraction of commerce. Goldman Sachs pre-
dicted in 2000 that B2B e-commerce transactions would equal
$4.5 trillion worldwide by 2005. […] Entrepreneurs and venture
capitalists poured into this new industry. Between 1995 and
2001 there were more than 1,500 B2B sites.[…] Most of them
collapsed in the early 2000s as investors realized that they did
not have a viable business model and as the expected buyers and
sellers failed to turn up.” (Present author’s emphasis)
Therefore, unrealistic business expectations of the new economy are posited
as a reason for the demise of the dotcom era. The argument, based on an anal-
ogy between dotcom e-marketplaces, which arguably displayed similar char-
acteristics such as multi-sidedness, and modern online platforms is that free
offerings should be monetized, otherwise modern startups will share a fate
similar to dotcoms.
However, if the company’s goal is not sustainability but acquisition, the
strategic choice of freefying can be better understood. It can be seen that the
platform sacrifices profitability, even in the long term, as an attempt to raise
interest from investors and larger companies, and then to monetize the, albeit
hyped, interest. If this is the goal then profit is secondary to liquidity, as noted
by Brunn, J ensen, and Skovgaard (2002).
The analogy to dotcoms is somewhat alarming as the key tenets for failure,
based on both the sample and the theoretical survey, still very much surround
online markets. It has been argued that the nature of information goods, with a
low marginal distribution cost, or the nature of two-sided markets are insuffi-
cient for free offeringsper se, and require a realistic plan to monetize. It seems
that the literature shows mixed approaches, and cannot debunk or positively
confirm the premises made in the definition of the dilemma. Therefore, it is
fair to argue that the dilemma endures quite a multi-faceted treatment. Next, a
potential solution, the freemium model, is considered.
158
4.6.3 Solution: Freemium
When one side is subsidized (refer to Subchapter 4.4.3), the startup is forced to
find a party that is willing to pay for the demand-side’s use of the product,
which is termed indirect monetization. When the monetization model is free-
mium, the startup has one user base that is split into free and paid users, and
the model is direct monetization, with price discrimination. Free users are of-
fered the basic service while paid customers receive extra features, quota, or
support (Pujol, 2010). Freemium is a widespread model in the context of Web
startups, as noted by, for example, Niculescu and Wu (2010) and Teece
(2010). Therefore, it is suitable to consider freemium a potential solution to
the monetization dilemma
155
.
According to Wilson (2006), freemium is defined as follows:
"Give your service away for free, possibly ad supported but
maybe not, acquire a lot of customers very efficiently through
word of mouth, referral networks, organic search marketing,
etc., then offer premium priced value added services or an en-
hanced version of your service to your customer base."
The definition can differ based on the type of platform. For example,
Riggins (2003, 70) considers content platforms: “What these information pro-
viders are essentially doing is degrading their information product to create a
free version of the good that satisfies low-type consumers, but holding back
enough content so that high-type consumers are not entirely satisfied and,
therefore, are willing to pay for the fee-based site.” Riggins divides users into
low- and high types based on their WTP; however, he does not consider that
users can move from one side to the other (i.e., downgrade or upgrade). Free-
mium can also be regarded as second-degree price discrimination or version-
ing (Varian 1983), whereby users are given a choice between low-quality and
high-quality products
156
. Note that the equivalents are free and paid product in
the freemium setting, and that the quality does not refer to “bad” quality but,
for example, that the other product has less features. The low quality still has
to be sufficiently substantial to invite adoption, as discussed previously.
Beuscart and Mellet (2009) suggests that although Internet platforms
categorically give free access, they are able to monetize through four means:
1) advertising (e.g., Facebook), 2) freemium (e.g., Evernote), 3) transaction
fees (e.g., eBay), and 4) donations (e.g., Wikipedia). Donations apply to
155
In general, any solution that either directly or indirectly increases WTP (i.e., converting from
free to paid users) is effective.
156
In contrast, first-degree price discrimination occurs when the startup is able to identify WTP, and
therefore targets users with precise products. In second-degree discrimination, users are presented
with both options, and they can self-select (Riggins 2003).
159
non-profit projects rather than commercial ventures, and are not considered
here
157
. Advertising is a form of indirect monetization; thus, it effectively
circumvents WTP. As WTP is made irrelevant, the problem can be solved by
finding advertisers that are willing to pay for access to users. However,
advertising is not generally considered a good option, unless the platform
generates a substantial mass of traffic; therefore, it is a winner’s choice in a
winner-takes-all market.
Riggins (2003, 81) notes that “sponsored sites have struggled to find a
profitable business model based on advertising revenues.” The performance of
advertising is also criticized by Beuscart and Mellet (2009, 165), who describe
advertising revenues of Web 2.0 companies as “weak and disappointing, espe-
cially related to their audience”
158
. Advertisers often seek economies of scale
that an early-stage platform is unable to provide and, due to transaction costs,
it does not make sense for them to contract a large number of weak plat-
forms.
159
A negative position is also taken by Clemons (2009) who concludes
that advertising will eventually fail as the primary business model because it is
distractive, consumers do not trust it, and its informative capability is being
replaced by recommendation platforms.
Additionally, transaction fees are a form of direct monetization and there-
fore beyond the assumptions of the monetization dilemma. If transaction fees
were applied successfully, there would be no problem with monetization,
given sufficient liquidity; however, contrary to the hypotheses, this is not the
case in the monetization dilemma
160
. In contrast, freemium aims to overcome
WTP by a form of second-degree price discrimination. This differs from ‘bait
and switch’ in that a startup does not normally change the price levels after
users join, but expects usage to grow naturally and, therefore, free users to
convert to paid users (i.e., customers)
161
. As a solution, freemium relies on its
ability to change negative WTP into positive WTP.
As noted, freemium is not the same as subsidization, in which, according to
the two-sided markets theory, it is assumed that two sides interact. In
157
There are initiatives (e.g., a startup termedFlattr) that propose voluntary micro-payments (i.e., a
form of donations) in exchange for the consumption of content. However, they have not reached the
mainstream at this point in time.
158
Google is a notable exception; its revenue, mostly from advertising, amounted to $50Bn in 2012
(Google Inc. Announces Fourth Quarter and Fiscal Year 2012 Results 2013).
159
However, any type of platform can access the advertising market though online advertising
networks (see Salminen 2010), which largely reduce transaction costs for both parties in finding,
negotiating, and monitoring performance of their counterparty. In exchange, advertising platforms
take a commission based on some revenue sharing principles; typically, the publisher, that is the
connecting platform, retains the majority of click-based revenues.
160
Moreover, transaction as a term only applies to exchange platforms.
161
In contrast, it might increase features over time; however, the free version typically remains an
option.
160
freemium, there is no such assumption; instead, users form a priori one
homogenous group, and then self-select into ‘free’ and ‘premium’ groups.
Even if this results in two groups of users (i.e., free and paid users), these
groups derive no immediate benefit from each other’s presence
162
. Although
the goal of growing the user base can be shared by a two-sided market in
general and a startup applying a freemium model, the difference is that free
subsidization in a two-sided market aims to provide positive network effects
for another side of the market, whereas freemium attracts free users who can
later be converted to paid customers. However, in a one-sided market,
freemium can be regarded as a type of subsidy. As such, it is similar to
differentiation
163
and versioning
164
.
There is support for the idea of converting free users to paying customers.
Traditionally, marketers have employed sampling to penetrate a market, so
that giving free samples converts users into loyal customers (e.g., Milgrom &
Roberts 1986). Peitz and Waelbroeck (2006, 907) argue that, as a result of
sampling, consumers are willing to pay more “because the match between
product characteristics and buyers’ tastes is improved”. This is especially true
for Web services that are similar experience goods, so that consumers need to
try before they buy because quality is difficult to determine prior to testing
(Shapiro, 1983). As noted by Niculescu and Wu (2013, 2), “y trying (sam-
pling) the product or part of it before committing to any purchase, consumers
could learn more about the quality and other attributes (such as performance,
functionality, interface, and features) of the software, capabilities of related
modules, compatibility issues, hardware requirements, etc.” However, in con-
trast to physical goods, with which sampling is limited due to replication and
distribution costs (Niculescu & Wu 2013), freemium benefits from infor-
mation goods properties (see Subchapter 4.6.2), and therefore scales much bet-
ter.
Moreover, there can be positive spillover effects relating to free users. For
example, Oestreicher-Singer and Zalmanson (2009) found that social features
built alongside content interaction increases the propensity to convert to a free
user, so that the more active users of social features were also more likely to
convert. This suggests a relationship between content interaction and social
interaction; that is, spillover effects between them. Potentially, particular types
of platform might benefit from building structures that support other types of
interaction; for example, exchange platforms compatible with sharing and
content platforms that enable exchange. In a similar vein, Albuquerque et al.
162
However, free users can derive long-term benefits, as shown in Chapter 4.4.1.
163
Creating features that make the product special in the eyes of users.
164
Creating several versions of the product.
161
(2012, 408) found that free users’ content creation activities led to increased
profits of the user-generated platform they studied, whereby “free marketing
activities and referrals bring in about 50% of the sales of the platform, and we
suggest that [the company] should provide additional incentives to content
creators to increase their referral behavior.” Although, in general, this study
advises against overly optimistic expectations concerning the role of peer
marketing, they seem to prevail in some circumstances.
Platform theorists have employed the concept of differentiation to explain
the coexistence of several platforms in a given market. For example,
Tanriverdi and Lee (2008, 382) note that “heterogeneity in customer prefer-
ences allows differentiation, limits market tipping, and leads to the coexistence
of multiple [OS] platforms.” It is seen that users might require different feature
sets. This enables platforms to attract different types of user within the same
markets, which relates to the assumption that while network effects hold
within a group, they might not generalize (see Subchapter 4.5.1). Cennamo
and Santalo (2013) refer to distinctive positioning in inter-platform competi-
tion, which means that the differentiating features and target markets depend
on the choices of other platforms. Over-differentiation can result in a niche
trap, whereby the mass-market provider is ultimately also able to capture the
niche users (cf. tipping; Farrell & Klemperer 2007). Applied to freemium, the
startup can create features that are distinct from other platforms while main-
taining the same minimum requirements set by the competition. Such an ap-
proach aims to simultaneously maintain differentiation while appealing to us-
ers’ standard expectations.
However, there are some limitations to freemium. First, Oestreicher-Singer
and Zalmanson (2009, 2) note that “conversion rates vary and are often very
low, and firms continue to seek effective strategies for converting consumers
‘from free to fee’.” Indeed, many authors note the ‘expectancy of free’ by
customers (see Subchapter 4.6.2), thus complicating reversion to paid offer-
ings. There are also varying accounts on how easy it is to convert free users to
paid customers; Murphy (2011) notes that industry averages are approximately
three to five percent of conversion rates. This study will not consider the tacti-
cal methods of conversion optimization; however, based on the author’s expe-
rience, it can be noted that a substantial amount of time and effort is spent by
startups on optimizing their conversion rates. In particular, Pauwels and Weiss
(2008) highlight, among other factors, the importance of 1) the correct price
point, given the competition, 2) accumulating a substantial base of free users
before attempting conversion-enhancing actions, and 3) properly executed
marketing communications.
Second, based on an extensive sample of users of the freemium-based web-
site Last.fm that is classified as a content platform, Oestreicher-Singer and
162
Zalmanson (2009, 19) found that that it took 652 days on average for a free
user to convert to a paid user. They concluded that “the typical subscription
decision […] requires deep familiarity with the website and its features.”
Therefore, even when the conversion might take place from free to paid, the
process can be long and require consistent persuasion (i.e., marketing) by the
startup. Then, the startup needs to consider supporting costs for the free users
during their free period (Subchapter 4.6.2), to find which user types are more
likely to convert, and to create potential tactics to facilitate earlier rather than
later conversion.
Third, Bakos and Katsamakas (2008) examine the optimal platform design
structure and conclude that a platform would be advantaged by focusing on
one side, and then charging that side. By applying this logic to freemium, this
would mean that the startup should serve paid users better than free users. Alt-
hough this is achieved with premium features that offer higher quality (i.e.,
more features), the free version needs to be able to solve the cold start prob-
lem. Under freemium, platform design issues therefore concentrate on choos-
ing the appropriate structure for product variations and ‘tiering’ (Semenzin,
Meulendijks, Seele, Wagner, & Brinkkemper 2012). Such design choices re-
quire a tradeoff, whereby the platform needs to determine what is sufficient to
include in the free version to attract free users while keeping paid users satis-
fied.
This is known as thecannibalization problem (e.g., Riggins 2003), which
has been studied extensively in the extant literature. Generally, a firm facing
the problem must balance its allocations so that customers do not have an in-
centive to fall back to their second-best choice (see Subchapter 4.6.1). How-
ever, not offering a free version might leave the cold start problem unresolved
(Subchapter 4.4.3). As offering a free version to some users is a strong form of
subsidization and creates the cannibalization problem, the startup needs to bal-
ance this with sufficient investments to paid users, or risk “spillovers from the
intermediary’s investments in the other side of the network” (Bakos &
Katsamakas 2008, 192). The spillovers would be, for example, overly gener-
ous features or usage quotas, depending on the type of premium constraints
enforced, which would reduce the incentive for paid users to stay on the paid
user side or free users converting to paid customers, when this would other-
wise be required by the stricter conditions.
Pujol (2010, 2) refers to the cannibalization problem as the reflective
competition dilemma: “Feature differentiation can be challenging as it
requires tradeoffs between growing the free user base and generating reve-
nues, [i.e.,] the reflective competition dilemma.” Riggins (2003, 70) describes
the problem as follows: “For two-tier sites […], the challenge is to provide
enough free content to keep users coming back in order to increase banner ad
163
revenues, but, at the same time, limit the free content such that high-end users
will still be willing to pay to access the premium services and information.”
Moreover, Riggins (2003) assumes that advertising is a sort of control mecha-
nism, so that it can be utilized by a platform to increase inconvenience up to
the point at which users are willing to convert. In a pure freemium strategy,
this option does not exist. The problem is also perceived by practitioners
(Chen 2009): “the key is to create the right mix of features to segment out the
people who are willing to pay, but without alienating the users who make up
your free audience.” For the sake of clarity, the problem can be conceptualized
in this study as the feature definition problem in the specific context of the
freemium business model
165
. This conceptualization captures the freemium
business model as opposed to, for example, advertising, and refers to measures
that need to be considered by the startup when designing its offering.
Bekkelund (2011, 16) argues the use of “observable characteristics, such as
memberships in particular social or demographic groups; or unobservable
characteristics, such as the quality of the choice the consumer purchases”.
These are needed because the WTP is not known to the startup in second-de-
gree price discrimination (Riggins 2003). Bekkelund (2011) notes that free-
mium enables startups to experiment with pricing plans to discover users’ true
WTP, which is compatible with Ries’ (2010) proposal. Consequently, by
identifying common characteristics of users who are willing to pay, the startup
might be able to move to first-degree price discrimination, in which it would
directly offer discriminatory pricing plans based on WTP (Laffont, Rey, &
Tirole 1998).
In other words, a potential move for startups is to find a proper niche to tar-
get. In saturated markets, it is generally more challenging to discover needs
that other platforms have not satisfied (cf. Parrish, Cassill, & Oxenham 2006).
To speculate whether, and to what extent, online markets have become satu-
rated or not goes beyond the scope of this study; however, generally, given the
low cost of experimenting with different platform models, many mass markets
tend to be difficult for challengers. Nevertheless, by employing different fac-
tors as differentiation criteria (e.g., geographic location; language; highly spe-
cific interests), startups might be able to create niche platforms that are able to
reach a critical mass for both self-propagation and active use.
Teece (2010, 178) mentions that it is also possible to apply a hybrid
strategy and puts forward Flickr as an example: “Flickr’s multiple revenue
stream business model involves collecting subscription fees, charging
165
The problem relates to two conflicting incentives: 1) when giving free content, paid users might
be willing to downgrade, and 2), when only offering paid content, users who would convert after a
trial period will not adopt the platform.
164
advertisers for contextual advertising, and receiving sponsorship and revenue-
sharing fees from partnerships with retail chains and complementary photo
service companies.” Several methods can therefore be applied alongside
freemium. As also mentioned by Riggins (2003), the platform can utilize
negative network effects associated with advertising to drive up the number of
users’ willingness to convert. However, such a strategy is risky because users
might also switch instead of converting. Thus, utilizingadvertising as a threat
might not be an effective tactic.
In sum, freemium is a way to split users into free and paid users. It is simi-
lar to, although distinct from, bait and switch. The startup does not expect us-
ers to pay for the initial adoption but, as their usage grows, they are offered
extra paid service. As such, the functionality of the method as a solution is
directly linked to the proportion of free and paid users. If the paid user base is
sufficiently significant to sustain free users and satisfy the startup’s financial
goals, the solution is successful. However, more research is needed to under-
stand the antecedents to conditions in which users can be made to convert.
This relates to emerging studies on conversion optimization (see e.g.,
J ankowski 2013; Paden 2011; Soonsawad 2013). For example, Pauwels and
Weiss (2008) document that their case company employed e-mail marketing
and price promotions to upsell content subscriptions to their base of free users.
4.6.4 Discussion
By offering access and usage of its platform for free (i.e., freefying), a startup
is able to attract users, but is unable to monetize (i.e., attract paid customers).
If monetized, users opt for free substitutes. Therefore, should a startup aim for
free users or charge for its product? How valuable are “users” actually? How
to capitalize on popularity? These are questions that arise from this dilemma.
In essence, the literature shows two camps. The first argues that free models
are fundamentally different from the old rules of doing business (i.e., a new
economy), whereas the other argues for “business laws” such as revenue, via-
bility, and sustainability. The strong analogy to dotcoms cannot be ignored in
this debate as there is a risk of history repeating itself.
By applying the freemium model to a product startup (e.g., one selling a
uniform product to all customers), it can be transformed into a two-sided
setting (see Chapter 6.1). This is because the startup’s user base can now be
divided into two groups, which is a condition for platforms, of free users (i.e.,
users) and paid users (i.e., customers). The question of the analysis then be-
comes: are there network effects between these groups? At first sight, the an-
swer is “No”, as the product qualities do not change for paid users regardless
165
of the number or quality of free users, or vice versa. However, it can be argued
that the free users gain long-term benefits from paid users as this guarantees
their free usage. Ultimately, if the startup is unable to attract a sufficient
number of paid users, it will close down, and both free and paid users will
lose. If free users are aware of this, they might become motivated to promote
the platform to their peers in the hope of converting some to paid users.
However, although free users gain benefit from the existence of paid users,
it is unclear whether paid users benefit from the existence of free users. If not,
the network effect is asymmetric: one side benefits more than the other. In
fact, this is not necessarily uncommon in online business. Consider advertisers
and free users (i.e., consumers of content); advertisers, in theory
166
, benefit
from the presence of users, and would not engage with a platform without
them. However, the reverse does not apply, and users would in most cases
happily frequent the platform without advertisers
167
. This condition, as previ-
ously stated, is a negative indirect network effect, and it is not clear whether it
exists between free and paid users (e.g., through congestion). While it is often
assumed that the marginal cost of distributing a digital product is “close to
zero”, if the startup invests its resources in acquiring free users, the overall
user acquisition cost is transferred to prices that paid users will ultimately
pay
168
. In other words, there is a potential free rider problem.
This is, in fact, well understood in the platform literature as it is perceived
fait accompli that one side is being subsidized by charging the other. To the
author’s knowledge, this only becomes an issue in two cases: first, when as-
suming ideal user generation (UG); in other words, when the startup expects a
positive non-economic contribution from free users, such as peer marketing.
According to the study’s analysis, there are startups that make this assumption
at least implicitly. The second case, in which free riding is problematic, is
when paid users are tempted to join free riders, and leave the cost to a de-
creasing group of paid users, or when they avoid conversion in the first place.
166
It is important to note here that the relationship between the number of users and advertisers’
increased utility (i.e., the network effect relationship) is not necessarily linear. From an advertiser’s
perspective, much attention is paid to whether users 1) are from the proper target group, whether
defined by demographics of interests, and 2) that they are actively processing advertisements or, more
preferably, clicking them. Furthermore, while clicks are relatively easy to track, banner blindness
(Benway & Lane 1999) complicates the processing of online advertising, and therefore can be a major
obstacle in the relationship of user base versus its worth to well-informed advertisers.
167
We can, therefore, argue for an implicit contract between the platform and its free users: users
accept advertisers in exchange for free content.
168
Consider, for example, that the startup pays 100 money units to acquire 100 users, of which 5
convert (i.e., shift from free to paid user). Assuming that acquisition costs are evenly distributed (i.e.,
acquiring each user costs 1 money unit), and that acquisition costs are transferred to prices, which they
will be in the long run according to economic rationality, the 100 money units will thus be added to
the total cost of the product that will be paid by the 5 converted users; therefore, each will pay 19
more money units than without free users: (100-5)/5=19.
166
This type of strategic behavior can occur, for example, when users reduce their
usage on purpose to avoid quotas; that is, where premium features are linked
to consumption
169
. In sum, freemium can have negative spillover effects on
conversion that restrict its use as a solution to the monetization dilemma.
Clearly, freemium and advertising can be combined so that free users, and
not paid users, are shown advertisements. As advertising can represent a nega-
tive indirect network effect for free users, some would convert to paid users to
avoid advertisements. In theory, this would provide the startup with two bene-
fits: 1) revenue from advertisers, and 2) a higher conversion rate from ‘free to
paid’ than otherwise would have been the case. The problem with advertising,
by applying the platform literature, is simply that the startup needs to be a
critical mass of users, impressions, or clicks to attract advertisers; otherwise,
there are no network effects. This critical mass can be considerably larger than
some startups assume, as those in the advertising side of the market are only
willing to reduce their transaction costs by dealing with the largest platforms.
Moreover, advertising can result in switching instead of conversion if users are
ad-sensitive.
A more advanced tactic is not to employ indirect monetization at all but, in-
stead, leverage the actions of free users in a frictional non-user-generated
manner, to increase the conversion rate from free to paid users. This is identi-
cal to creating positive direct network effects. For example, the startup can
gather the usage of free users as “templates” and offer them to paid users as
complements. Therefore, the network benefits will only be available to pre-
mium side users, which would be an incentive for free users to convert to paid
customers
170
. Because the generation of such content is ‘frictionless’, it is free
from the free rider dilemma depicted earlier.
If freefying (i.e., offering a free access and usage) is applied as an ex ante
solution for the cold start dilemma in an attempt to increase the adoption rate,
turning from freemium to paid access is an ex post solution to the
monetization dilemma. For example, some content portals have begun
introducing pay walls for access to content (Spulber 2010). It is still too early
to judge the results of these attempts; nevertheless, the publishing industry is
facing strong pressure to monetize free content. This highlights the dynamic
interaction, or vicious cycle, between dilemmas; that is, if fees are introduced,
solving the cold start dilemma might become more difficult, whereas if
169
Note that the definition of the freemium model applied here is based on versioning and tiered
pricing (i.e., where free is the base level) not, for example, on time limitations. Although some authors
include this as a variation of freemium, trial software stems from desktop software and does not offer
a permanent free version as offered by the pure freemium model.
170
In other words, the free side would be neutral to network effects but the standalone benefit would
be the same, regardless of how many users are in the paid side, which is an interesting juxtaposition
whereby the product simultaneously exhibits and does not exhibit network effects.
167
freefying is applied, monetization becomes more difficult. As in most cases,
the startup is seemingly forced to make a tradeoff between the strategies.
However, opting for a paid monetization model might serve as astress test
for quick failure. If no users are willing to pay for a product, even after be-
coming aware of it, perhaps there is no true demand. Offering for free is
equivalent to offering free beer; if it succeeds and the idea is later to charge for
beer, there is no guarantee that the business will be sustainable once fees are
introduced. Of course, this does not apply to indirect business models, alt-
hough these might face the issues discussed previously in this chapter. In ad-
dition, freefying makes it impossible for the startup to compensate its content
creators. Although this is not a major concern for a process of UG with mo-
tives deviating from economic rationality, it can become an issue in special
cases. For example, one startup in the sample aimed to market journalists’
writings, essentially relying on them to write for free, thereby commoditizing
the labor they were paid to perform in traditional industry.
Competition with free can become hazardous in an industry setting with
low distribution and entry costs, such as many online markets, because en-
trants are 1) many, causing increased pressure for subsidization across verti-
cals, and 2) able to subsidize one side on the expectation of an emerging se-
cond sideex post (cf. finding a business model), and because this is facilitated
by their cost structure. Eventually, this strategy can lead to arace to the bot-
tom, whereby all platforms subsidize some sides by lowering prices, resulting
in all markets becoming subsidized. In the medium term, the startup might be
able to cover the loss of subsidization through venture capital, in the hope of
either finding a sidey to tax, if all of its users are initially x, or converting us-
ers from the subsidized side to the paid side, if they apply the freemium
model. The situation will be resolved when entrants are unable to recoup their
losses and exit both markets, which will enable the remaining platforms to
adjust their prices upward and become profitable.
The conditions for a race to the bottom include theex ante undefined side y,
if sidex is subsidized, andvice versa, and also the allowance period of expec-
tations (e.g., hype) that enables new entrants to consume previously profitable
sides (i.e., markets). As the required initial investments are generally low in
the Internet, venture capitalists might favor disruption by sponsoring ventures
that offer free substitutes to previously paid product categories to gain quick
traction, and then sell these ventures based on the expected value of their user
base.
The monetization dilemma is connected to the cold start dilemma; a startup
is unable to subsidize users or pay for content if it employs a free business
model. Hence, external funding is needed. However, startups without external
funding might simply neglect paid actions, such as marketing, that would
168
facilitate solving the cold start or lonely user dilemma and thus, are unable to
test whether the product would gain a critical mass
171
. This can lead to failure
without even getting a chance. Thus, the solution of free models can lead to
far-reaching consequences with regard to other startup dilemmas.
Regardless of the increased interest in monetization problems, the conclu-
sions seem to be mixed. One side of the argument seems todefend the use and
viability of free models in platform markets, while the other side contests it.
The two-sided nature of the market, and network effects associated with it, are
critical to this discussion. Pricing decisions (i.e., level and structure) in two-
sided markets are distinct from those in one-sided market considerations.
Therefore, subsidizing free users can be defended as a strategy, as rents can, in
theory, be extracted from the other side with higher WTP. However, the mar-
ginal distribution cost (e.g., nearly zero in information goods) is too loose a
definition, as it misses this subsidization cost, and also, typically, the cost of
supporting free users.
Essentially, a startup can suffer from hidden information if the platform is
too open and enables parties to negotiate and transact on their own, which is a
particular risk for exchange platforms that often need to retain control over
transactions. For other platform types, interaction beyond the platform (i.e.,
bypassing the platform) is not a major issue unless users are transacting (e.g.,
dating sites). Social platforms and content platforms, which often apply an
indirect monetization model, need to consider the negative indirect network
effect imposed by advertising, although theory shows that the effect can be in
both directions, and also its particular challenges in providing sufficient re-
turns for long-term business viability. In particular, advertising is appropriate
for platforms with strong dominance, so that they can create a sufficient num-
ber of impressions and clicks to attract advertisers. Although the problem can
be alleviated by delegating negotiation and coordination to an advertising plat-
form, the risk of returns remaining low exists for this monetization model.
4.7 Remora’s curse
4.7.1 Definition and exhibits
The remora is a type of fish that attaches itself to a larger fish
like a shark or even a boat. It rides along with its host and feeds
on whatever comes by. The remora can also detach from its host,
swim on its own, and survive. (Don Dodge)
171
The classic marketing maxim: if customers don’t know about the product, how can they buy it?
169
As established in Chapter 4.4, startups can face greater than expected difficul-
ties in achieving a sufficient degree of user-generated content (UGC); “suffi-
cient” being enough to launch a self-sustaining process of content replication,
or a critical mass. To overcome this hurdle, some startups opt for theremora
strategy, which is to join an existing platform to gain access to its predominant
user base or content, and in this way solve the cold start dilemma. In practice,
this might mean developing applications on top of existing platforms, such as
Facebook, Google, or Twitter, and leveraging their application programming
interfaces (API)
172
and user bases; essentially, gaining access to network ef-
fects without generating a critical mass. The solution might appear solid in
theory, and there are several cases in which it has worked well (e.g., acquisi-
tion
173
or direct monetization
174
); however, our sample of failed startups also
showed its limitations (for exhibits, see Table 25). The purpose of this chapter
is to analyze these limitations.
The dilemma of a remora’s curse takes place when a platform entrant needs
to decide whether to integrate a critical functionality relating to distribution,
marketing, or monetization to a predominant platform at the cost of losing
power in those areas, or to develop an independent solution at the cost of los-
ing access to the platform host’s pre-existing user base, content, distribution,
monetization system, or any other asset to which the integration would grant
access.
Consequently, remora’s curse addresses the choice of either developing a
product on top of existing platform (i.e., become a ‘remora’) or not (i.e., start
an independent platform); the former gives access to a pre-existing user base
or content while the latter requires that the user base or content be created sep-
arately without the “kick-off” provided by the host platform. In both choices,
the startup pays a tradeoff cost, as depicted in Table 24.
172
API, in the case of Web applications, enables access between applications; in platforms terms,
interoperability.
173
Instagram, for example, was built to be compatible with Facebook, and was acquired by
Facebook for $1Bn in 2012 (Constine & Kutler 2012).
174
Zynga, for example, charges the user for virtual goods sold on the Facebook platform, and
generated revenue of $1.2Bn in 2012 (Zynga Inc. 2013).
170
Table 24 Remora’s choice
Join Not join
Tradeoff Lose power over tech-
nology, marketing, and
monetization
Lose access to the pre-existing user
base or content (i.e., cold star di-
lemma)
Therefore, by joining an existing platform
175
as a supply-side participant, a
startup gains access to an existing user base or content but increases its de-
pendency on the platform owner, thereby in effect trading off 1) technology
power, 2) marketing power, and 3) monetization power to a) the distribution
function and b) the marketing function, which are delegated to the host plat-
form
176
. Technology power implies that the host influences the startup’s
technology choices, and the startup incurs initial integration costs and, when-
ever the host platform’s specification changes, continuous adaptation costs.
This can be regarded as a form of asset specificity, as discussed in the litera-
ture subchapter. Losing control over marketing and monetization refers, re-
spectively, to the inability to differentiate via marketing, as the platform poses
marketing restrictions, and the inability to choose a monetization model as this
is imposed by the host. In effect, the startup will also forego customer relation-
ships because it is the platform owner that retains customer information
177
.
The host has more information on the users but it restricts sharing it due to 1)
privacy concerns and 2) the competitive value of information. Capturing value
is another conflict: while in the platform, a remora can never reach an outcome
by which its revenue supersedes that of the host. This is theoretically impossi-
ble when the host imposes a revenue sharing scheme and blocks all other
means of monetization
178
.
Exhibits of remora’s curse are presented in the following table.
175
Such as Facebook, Google, and Twitter, or iOS, Android, and Windows mobile.
176
Distribution is delegated as the startup’s platform is accessed through the parent platform’s
interface. For the same reason, marketing is expected to be self-organizing as users will find the
startup’s product inside the host platform.
177
This was a major concern for record labels considering distribution with iTunes, as Apple would
hold customer information (i.e., customer relationship); eventually, Steve J obs was able to convince
them of the mutual benefit (Isaacson 2011).
178
The revenue of a remora grows proportionally to the host’s share; so that a/b àk(a/b), in which
k =growth factor, a =host’s share, b =remora’s share; given that a >b anda +b =1.
171
Table 25 Exhibits of remora’s curse
Example
[1] "We exposed ourselves to a huge single point of failure called Facebook. I’ve ranted for years
about how bad an idea it is for startups to be mobile-carrier dependent. In retrospect, there is
no difference between Verizon Wireless and Facebook in this context. To succeed in that kind of
environment requires any number of resources." (Rafer 2009).
[2] "The killing blow was when Facebook changed its app platform to make things less spammy,
and thus less viral. We were toast." (Parr 2011a).
[3] "Predictably and reasonably, Facebook acted in their own interest rather than ours. Their
Summer 2008 redesign supported Facebook’s goals elegantly but hurt our publishers and us in
ways that became clear just weeks after we’d raised another ~$2M." (Rafer 2009).
[4] "We were doing some time-consuming processing on gathered data so there wasn’t a big time
buffer we could use. With each downtime (the website worked but with no actual data it didn’t
make much sense to use it anyway) we had to wait until a backlog was cleared. Chances were
good that by this time we had another issue to deal with – a bug in the code, on of auction plat-
form’s changing the structure of their data, a simple hardware malfunction, or running out of
disk space." (Brodzinski 2009).
[5] "[Facebook] wasn’t a perfect fit for the Nouncer services, but it still fit in with the overall strat-
egy and philosophy. It also looked like an easy thing to do with a big marketing potential. The
result was JabAbout, a Facebook application using the social graph to propagate short mes-
sages by following the friends-of-friends paths. JabAbout failed to build a user base and was
eventually shutdown." (Hammer 2008).
[6] "Mint’s dependence on Yodlee apparently suppressed their acquisition interest among compa-
nies that knew Yodlee well (such as Microsoft, Yahoo, and Google); since we had developed our
own technology for aggregation, we didn't have that particular problem, and in fact had some
acquisition interest simply for the aggregator we’d built." (Hedlund 2009).
The exhibits demonstrate remora's curse from several angles. First, atech-
nology lock-in [1] indicates a situation in which continuous investments from
the startup are required to keep its product up-to-date according to the tech-
nological specifications of the platform owner. This might limit the available
technologies to some extent while increasing dependence on the host’s tech-
nological choices. If the choices are not optimal for the startup’s product, this
will reduce its competitiveness. Further, changing functionality [2] requires
the startup to react and organize its product development according to that of
the platform owner, and it pays adaptation costs [4].
The bigger issue, however, is the lack of control with regard to the user
base. At any time, the platform owner can restrict or deny the startup’s access
to users, justified as a change of service terms [3] or platform design [2].
Losing access to users may also occur due to a technical breakdown. In
consequence, the solution to the chicken-and-egg problem dissolves [5].
Moreover, the platform sets rules for marketing over which a startup has
little control. For example, the platform might give additional visibility to
particular products and not others, thereby distributing competitive advantage.
A startup can have little control over its visibility in the platform as it cannot
172
influence the rules
179
, and in general advertising is not allowed
180
. Overall,
these limitations may reduce investors' willingness to invest in a startup [6].
As noted, adaptation costs arise when the startup is dependent on the plat-
form as a source of data. First, it has to build the product so as to be compati-
ble with the platform. Second, it has to account for changes that might easily
break the flow of data and, therefore, its own product. Third, the platform
owner can restrict access to data, rendering the product useless. Coordination
problems of this kind, therefore, relate to the functionality of the product, and
apply especially to startups following the aggregator content model
181
whereby, in theory, the startup’s product integrates into several host platforms
to fetch data. This solves the cold start problem well as the fetched content
will enable demand-side benefits; for example, the more websites indexed, the
better the search engine, all else being equal.
By aggregating data from several websites, the startup might gain an in
praxis a solution to the cold start problem. However, at the same time, it be-
comes dependent on these data sources; any change in which necessitates an
adaptive response or the startup’s platform loses its ability to function
182
. The
more aggregated platforms (i.e., data sources), the higher the risk for coordi-
nation problems; however, the less the dependency on individual sources, as
they become expendable in a large selection, and the more changes by plat-
form owners, the higher the risk of coordination problems. Further, the startup
is forced to constantly monitor the health of the third party data source. Note
that aggregation is a special case when joining a platform; its purpose is not to
acquire users directly (as in: host platform ? startup’s platform), but to pro-
vide benefit for, often, existing users by offering them content from other
sources, or to utilize the content indirectly through social interaction spillovers
or search-engine indexing, which can lead to website traffic.
179
However, if the platform is fair and the rules are transparent, the startup is able to increase its
position by adapting to them and outperforming competitors. Equally in this case, it is not affecting
the marketing variables set by the platform owner, but only adapting to them.
180
To compensate for the lack of marketing tools provided by platform owners, some developers
have created tools for peer marketing. In them, applications exchange users on a ‘give one, receive
one’ basis; the revenue comes from selling a small portion of slots to advertisers.
181
Assume that all websites in the world suddenly deny Google’s access to their content; the search
engine would instantly become worthless. As Google provides indirect network benefits (i.e., a large
number of searchers) this is unlikely to happen. Further, Google is inherently hedging its risk by
diversifying the aggregation to billions of sites; therefore, its dependence on an individual host
approaches zero.
182
A real-time service loses matching ability; a static platform becomes outdated.
173
Consider two “degrees” of integration:
· Full integration: building the product inside the host platform (i.e.,
turning to a full complement).
· Selective integration: accessing the host platform’s functions and
user base but retaining, for example, distribution and marketing
183
.
These types of service are sometimes termed ‘mashups’.
Due to its definition, remora’s curse applies to both degrees of integration.
The severity of dependence might be less in full integration as user base and
marketing freedom is retained. However, if the access provided by the host
platform is critical for the functionality of the remora platform, as is assumed
in the definition, the dependency is also critical.
Moreover, it is assumed that most users find products, including those of-
fered by the startup, within the platform. That is, the remora retains beyond-
platform marketing capabilities, although they are mostly irrelevant when dis-
tribution is delegated to the host. For example, currently, leaderboards and
rankings are controlled by the host in most online platforms However, if this
assumption was denied and the startup was able to successfully market so that
users connect to the platform to find the product, the marketing dependence
would be broken. This is not, however, a solution to the dilemma as the plat-
form owner retains control of technology, distribution
184
, and monetization. If
revenue sharing works in favor of a startupin praxis, this does not remove the
fact that, in theory, the host can change the terms; although, while there is
competition for complements, a choice such as that would most likely result in
inter-platform competition.
While any of the above functions are considered critical, removing them
partially from the host’s control does not solve the dilemma. However, partial
integration can be sufficient in solving the cold start dilemma; more precisely,
the startup might be able to draw users from its host to an extent whereby it
obtains a critical mass. Even if the host then exercises its power, this is not
detrimental to the startup as it has already gained a critical mass and is now
self-sustained.
The risks associated with delegation are presented in the following table.
183
In selective integration, delegated functions can be arbitrary based on functionalities offered by
the host and the startup’s strategy.
184
Even in selective or partial integration, whereby distribution would not be delegated, the problem
will persist while the host controls any of the critical functions.
174
Table 26 Risks of delegation
Delegated function Risk
Technology Technology lock-in
Marketing Favoritism
Monetization Unequal revenue sharing / no revenue
sharing
Distribution Breakdowns, changing terms
In a typical setting, the remora’s expected benefit of joining relates to dis-
tribution. In aggregation, the product is distributed outside the platform, there-
fore with distribution and marketing costs, whereas the platform brings, in
theory, delegation benefits. However, this matching, from the perspective of
any startup other than the category leader, is not automatic, and herein lays the
fallacy of believing that marketing investments are not required. In other
words, intra-platform competition exists even in the presence of network ef-
fects, and due to the host’s incentives to promote the strong remora at the ex-
pense of weak remoras, participating in a platform as opposed to being inde-
pendent can in fact become detrimental; that is, the required cost for differen-
tiation exceeds coordination benefits provided by the platform, which is easily
perceived when understanding that fair treatment is not a profit-maximizing
strategy of the host. Rather, it benefits from favoritism; particular killer apps
bring much more revenue, and are much more difficult to replace, than the
long tail of complements
185
.
Consider, for example, a simple game with two players: remora and the
host. Two versions will be presented: first, a version in which the remora is
weak, meaning that the host does not believe it will sell. In the second, the
remora is strong in the sense that the host believes in it and will give it addi-
tional marketing support (i.e., exposure). This is a sequential game with three
turns: first, the remora decides whether to join or not; second, the host will
either sell its product or not
186
; and third, the remora will decide to stay or
leave.
The players make investments which they might lose, and gain benefits
which they might keep. Sales are recurring (i.e., third round) and parties
185
Consider App Store with millions of applications. The existence of this many complements is
beneficial to the platform owner and also the end user, given that his search cost of having so much
choice is not paramount, which is another reason for the platform owner to apply favoritism.
However, the majority of developers are disadvantaged as their offering cannot be easily discovered
(Salminen & Teixeira 2013).
186
This simplification equals the remora’s expected benefits described earlier; that is, acquiring
users or content.
175
engage in revenue sharing. Network effects are assumed, as the following
figure explains.
Figure 13 Weak remora
In the first stage, the payoff is expected benefits. As the remora will avoid
marketing investments, such as advertising and hiring a marketing manager, it
has a positive payoff. The product as a stand-alone would have some intrinsic
value, but less than when combined with the host platform’s assets (i.e., ex-
pected network effects). If the host makes sales, each party’s payoff increases
in proportion to that of the other party (i.e., revenue sharing
187
).
Not selling a weak remora’s product gives a higher payoff to the host as it
can keep the incremental network value without extra effort; comparatively, it
incurs an opportunity cost of not selling the strong complement, which is why
the payoff for not selling is higher than for selling. However, the host gets a
positive payoff for the remora joining as the remora provides an increment to
its complement base
188
(i.e., marginal network effect).
If the remora defects, it will lose its platform-specific investment. It will
also need to redeploy its product and compensate for loss of marketing dele-
gation, which is similar to the hold-up problem. However, it is assumed that
the remora can recover some learning effects by redeploying the product either
to independence or to another platform. Its departure will cause the host to
lose the incremental value. If it stays, it incurs no additional cost, but can also
resort to multihoming, which is not considered in the game.
At this point, keeping the remora will not produce additional gains for the
host as it does not expect the remora to sell but to provide perceptible value.
However, losing the remora would mean the loss of its incremental value.
187
For example, Apple shares revenue with its App Store developers using a70/30 ratio, in favor of
developers (Gans 2012).
188
According to the indirect network effects assumption, the complement base, as a whole, provides
a sales argument for the demand-side users.
Remora
Remora
Host
Join
Not
Makesales
Not
Leave
Stay
(1, 0)
(2, 2)
(-2, -1)
(0, 1)
176
Figure 14 Strong remora
By joining, the strong remora gets the same expected delegation benefits as
the weak one. At this point, it will only provide the incremental network util-
ity. Not joining will also produce similar effects as in the case of a weak rem-
ora.
In this case, the host can make high sales and is incentivized to sell. If
changing, the host would lose both the incremental network effect and the
sales effect. The strong remora would lose its platform-specific investment
and sales effect. Thus, both parties have an incentive to continue
collaboration
189
. Here the demand-side user base (i.e., indirect network effects)
becomes important for the strong remora; while, due to lack of exposure,
theoretical network effects are important for the weak remora. Strong remoras
actually realize high payoffs from participation.
Essentially, expected network effects are crucial with regard to the failure
of a weak remora. While it provides an actual marginal increase as a network
effect for the demand-side that the host can monetize, the weak remora gains
nothing in return. For the weak remora, if it is a possible strategy, becoming
strong before joining the host platform would provide a potentially better way
of investing its resources than joining as a weak player. From the host’s per-
spective, because payoffs are similar in the first step, it would need to distin-
guish between strong and weak remoras (i.e., “cherries and lemons”).
For the host, intra-platform competition is often desirable; startups
represent supply-side complements that increase demand-side utility
190
. For
startups, the reverse can apply: the greater the competition, the more difficult
it is to acquire users or customers, and the remora’s marketing delegation
advantage dissolves. The more the startup commits relationship-specific
189
Alternatively, the strong remora might consider multihoming to several platforms, which is not
considered in this game.
190
The logic is such that greater selection increases customer benefit, a standard assumption of
indirect network effects.
Remora
Remora
Host
Join
Not
Makesales
Not
Leave
Stay
(1, 0)
(3, 3)
(-4, -3)
(3, 3)
177
investments to the platform, the higher the degree of lock-in. In addition, moti-
vation to join a platform might arise from the expectation that user acquisition
is less costly within than outside the platform. However, when intra-platform
competition is high, this is less likely to be the case because other startups and
established firms compete over the same users. The competition can, in fact,
lead to an outcome whereby user acquisition is equally, or more, costly than
outside the platform
191
. As a result, the perceived marketing benefits relating
to customer acquisition can dissolve, as demonstrated in Table 25 [5].
This implies that even if there is a potential market, and network effects ap-
ply so that the increase in end users is due theoretically to the startup, the
startup is forced to compete within the platform. Therefore, these types of
network effect are here referred to as ‘theoretical network effects’, which are
theoretical (i.e., potential) as they do not realize under high intra-platform
competitionunless the startup is a category leader. In other words, the network
effects are not shared equally; some participants enjoy them, while others,
perhaps the majority, depending on the competition, do not. Therefore, net-
work effects that do not take place in the real-world setting are worthless to
the startup, and it gains no advantage in joining a platform with strong plat-
form effects compared to the situation of starting a platform without a critical
mass
192
.
Furthermore, the remora strategy is distinct from utilizing a platform as a
traditional marketing channel because of the integration of one or many criti-
cal functions into the host platform. For example, a user cannot access the
startup’s platform in a specific platform without first joining the host platform
(i.e., full integration), or the user might not access it beyond the host platform
if API access is not available (i.e., partial integration). If the host platform is a
monopoly, then joining it might give the remora access to some monopoly
benefits
193
. In contrast, when there is effective inter-platform competition
whereby users are distributed between several competing platforms, it makes
sense for the startup to follow this pattern by diversifying. While a subset of
users will treat platforms as mutually exclusive and choose one among them,
another subset will adopt several competing platforms simultaneously,
regardless of interoperability. The platform literature respectively refers this to
191
Such a situation is exacerbated when the platform owner reduces diffusion subsidies within the
platform, thereby increasing friction between the startup and potential users; for example, when
Facebook reduces visibility of application invites or organic post visibility in user streams.
192
Theoretically, the start-up gains a diminishingly small advantage compared to a pure cold start;
although, the more competitive the market becomes, the more the start-up becomes a “long tail”
provider. In brief, such a market exhibits winner-takes-all dynamics; however, not due to network
effects but to favoritism and user preferences.
193
Such as user adoption, so that in the absence of alternatives, the host platform keeps growing the
number of users.
178
singlehoming and multihoming (see Subchapter 4.7.3). By multihoming, in the
supply side, the startup can gain access to both multihoming and single-hom-
ing users, given that the host platform does not require exclusivity. In contrast,
choosing one host platform (i.e., single-homing) excludes users who single-
home to a different platform than that chosen by the startup.
As the author has argued, under some circumstances, the expected benefits
of the remora model do not materialize. If there is a reason for the startup to
believe soex ante, the dilemma dissolves as there is no rational reason to join.
However, under no conditions will the potential power of the host be negated,
regardless of whether it is enforced or not. The sole relaxation of the di-
lemma’s validity from this side would be when integration only touches non-
critical functions, but this is not in accordance with the definition presented
here. The host choosing not to exercise its power is not a relaxation because,
although it leads to a favorable position for the remorain praxis, the benefits
are not stable as there is uncertainty concerning the host changing its strategy.
Another case is take it all, when the host is lazy in exercising its power in
any of the critical dimensions. In such cases, the remora can in effect trans-
form into a leech, gaining users while retaining all benefits. However, again
note that the dilemma in effect persists as, at any time, the host can change the
rules of the game. Twitter is a well-known industry example of a “lazy host”
that grants free access, does not enforce revenue sharing, and is built as a very
open communication platform with low lock-ins to the website. For example,
Facebook has implemented strong lock-ins because users have to log in to its
interface for each interaction, and are shown advertising; Twitter can be ac-
cessed from anywhere without realizing additional revenue
194
. Nevertheless,
Twitter has also been known to break the rules of sound business logic in other
areas, mainly monetization. In general, platforms expect reciprocity; even if
not charging their complements, they expect them to provide indirect network
effects that can be monetized according to their monetization model.
However, the loss of user base must be discussed; more precisely, the defi-
nition is ‘users’ not ‘customers’. In other words, we return to the issue of ‘user
versus customer’. It is therefore possible to argue against the premises of the
problem by stating that users are only desirable if they can be converted to
revenue, because there is an implicit assumption that the startup wants users.
In fact, this becomes an issue when the platform owner is possessive about the
opportunities to monetize; for example, as is the case with Apple, but currently
not Facebook, restricting available monetization methods. As such, assuming
194
However, the behavior of platform complements is not “opportunistic”, as they have no other
choice. In opportunistic behavior, the startup chooses a strategy, among other strategies, which
maximizes its profit at the platform host’s cost. By this logic, Twitter is a “non-profit” platform, and
should be excluded from commercial analyses.
179
indirect monetization is not possible, free users would not be worth the
startup’s efforts as the platform can internalize all complement benefits relat-
ing to monetization; that is, there is no revenue sharing.
In sum, remora’s curse addresses situations in which platform participants
are at the mercy of the platform owner, which often aims to control revenue
sharing within the platform; for example, Apple’s App Store dictates the reve-
nue sharing terms for developers. However, the reverse might occur if the plat-
form is open; that is, it enables free access to its data and does not control
monetization. This case is clearly demonstrated by Twitter: for a long time,
third party service providers tapped into tweets generated by Twitter without
contributing anything in return; applying the animal analogy, the remoras had
become leeches. These included services to monitor tweets, set alerts, and
manage tweet streams. Counter-examples such as these do not remove the ex-
istence of the dilemma because Twitter can willingly exercise its power, which
it has begun to do (Nickinson 2013).
The main risks of falling victim to the platform owner’s strategic behavior
can be attributed to platform design (i.e., rules, terms, and specifications) and
unpredictable changes, in which central functionality is altered with no influ-
ence from the participants. Although joining an existing platform, or becoming
a ‘remora’, might appear to be alow hanging fruit, or an easy solution to the
cold start problem, the startup should be cautious about the potential hazards.
As established in the introduction to this chapter, startups, as is the case with
most organizations, are obliged to trade off strategic alternatives. Therefore,
joining a platform is not a trivial matter as it can lead to strong lock-in effects
and might be irreversible, especially considering the startup constraint of lim-
ited runway (i.e., depletion of time and resources).
Nevertheless, benefits of joining an existing platform probably exist in
some form. It can be assumed that the advantages are strongest when the
product category or industry is unfamiliar to potential customers, and therefore
requires strong persuasion, market education, and heavy investments in pro-
motional activities. However, each platform has its own competitive dynamics
that might not always be fair and which can, in fact, lead to complete dissolu-
tion of the expected benefits. Although it might appear to be a good strategy
for solving cold start problems, joining a platform does not automatically se-
cure more customers due to intra-platform competition and the above-men-
tioned dynamics (Table 25). The platform owner’s aim is, in most cases, to
encourage competition among participants within its platform. An exception to
this goal exists when protecting category leaders (i.e., killer apps) due to their
higher benefit to the platform. In such cases, new entrants will have difficulty
because incumbents are protected by the platform owner, for example, through
180
dominant ranks in application listings, thereby increasing the risk of awinner
take all outcome.
4.7.2 The literature
This subchapter will examine the remora strategy from the perspective of the
platform literature, which refers to remora-type structures as complements
(Farrell & Klemperer 2007; Rochet & Tirole 2003); as such, becoming a rem-
ora is to become a complement. The problems derive from not having owner-
ship of the platform, while the benefits originate from the existence of the host
platform (i.e., coordination effects) and its end users (i.e., network effects).
The large number of participants in the other side of the market increases at-
tractiveness to join, whereas the platform’s specialized coordination abilities
increase matching to a point at which transaction costs of finding, negotiating,
and monitoring the other side of the interaction can considerably decrease.
The platform owner’s tendency to exercise power, and also the risks relat-
ing to the remora’s position have been extensively discussed in both the plat-
form and economic literatures; transaction cost-related concepts are especially
applicable. In general, despite the fact that Internet startups did not even exist
at the time of its invention, the classic hold-up problem (see Klein 1998) ad-
dresses this type of issue at a general level, although not necessarily from the
same perspective
195
.
More specifically, Hagiu and Yoffie (2009) identify three hold-up risks: 1)
the host raising prices after becoming successful, 2) vertical integration into
the remora’s business, and 3) losing the ability to differentiate. Hagiu and
Yoffie (ibid.) give respective examples: 1) after reaching dominance with
Windows, Microsoft raised OEM licensing prices; 2) Google has been bun-
dling applications into its core offerings; and 3) Toys ‘R’ Us was unable to
differentiate against small players in Amazon’s marketplace. In general, these
risks are compatible with the concerns voiced by the founders (see Subchapter
4.7.1).
The platform owner becoming the startup’s direct competitor is another
risk. Due to asymmetric information in favor of the platform owner, it is able
to monitor each product and decide whether or not to provide a substitute.
Such examples have been documented in the industry (Honan 2012). How-
ever, as mentioned earlier, there are also documented success cases of
195
The hold-up problem requires 1) asset-specificity, 2) incomplete contracts, and 3) incentive to
“hold” (Klein 1998). In Web platforms, these arise if the complement cannot reuse its platform-
specific investments. In general, no contractual agreements protect the complement, and the host can
treat individual complements as expendable when they are large in number.
181
employing the remora strategy to rapidly acquire new customers (Campbell
2012), although, even in these cases, the power imbalance and, therefore, the
dilemma is present (Kelly 2009).
Remora is a strategy that aims to internalize externalities of a larger net-
work (i.e., envelopment). In platform markets, a remora relates to implications
of compatibility. As opposed to competing technologies, especially rival
standards, online platforms invite compatibility through their application pro-
gramming interfaces, or APIs (Evans et al. 2006). This behavior might be
different to that in other industries where “t is unlikely that the sponsor(s) of
a network with a large installed base will grant compatibility. Doing so en-
hances intra-network competition and […] provides very little benefit to the
system sponsor. Compatibility eliminates the installed base advantage of the
incumbent, reducing its market power and profits” (Church & Gandal 2004,
21). The reverse is argued here, based on different assumptions. Church and
Gandal (2004) imply that compatibility enables substitution and envelopment.
Here, this is not primarily considered by remora’s strategy because the host
can either prevent access or absorb remoras, and therefore counter remoras
that aim at becoming substitutes
196
.
Church and Gandal’s (ibid.) concern relates mainly to standards and tech-
nology. Once a standard or technology is open, the host cannot cancel the de-
cision as the technology has become public knowledge. Therefore, accepting
remoras can be perceived as reversible, while opening technology can be irre-
versible. The exception is when a startup performs aggregation as its content
model, and can envelop through content; as such, it can envelop the target if
the user is not motivated to visit the host website because information is given
by the remora
197
. However, even in the case of standards, inviting competition
can actually lead to a better outcome from the technology-holder’s perspec-
tive. As noted by Shy (2011, 131), “Sony did not use [open] strategy and as a
result it had to abandon its Betamax video technology in 1988 because it re-
fused to license it to competitors, thereby paving the way to VHS standards.”
In sum, it has been established that the host has an incentive to offer its plat-
form, and remoras have incentives to join it.
196
Furthermore, the risk of envelopment only applies to complements that are platforms. Most
complements (i.e., apps) in social platforms, for example, are stand-alone products that together
increase the utility of the platform. Envelopment would take place if popular apps were to move away
from the social platform, taking users with them. Such a coordinated move seems unlikely and, even
in this case, the user would most likely multihome given that, even without the complements, the
social platform offers intrinsic benefit while sufficient friends, who are not associated with apps,
remain.
197
Google is an example of a remora; it scans and indexes host sites, and then displays information
in search results. It utilizes content from other websites to monetize.
182
The purpose of the remora strategy is to gain benefit from an existing in-
stalled base, and therefore it is assumed that users’ switching cost is low,
thereby moving from the host platform to the remora platform
198
. Eisenmann
et al. (2011, 136) assert that “f users switch between rival providers of a
shared platform, they do not forfeit platform-specific investments in comple-
ments or in learning the platform’s rules.” The interface remains similar and
users remain in a “trusted” environment. For example, building an application
on top of Facebook does not require users to migrate from Facebook. They can
find the app through Facebook and utilize the familiar interface to access it
(i.e., low learning curve), and therefore the cost of adoption can be less.
Additionally, the remora can gain brand spillover effects (Olson 2008) from
the host platform’s enhanced adoption. The host’s strategy, in contrast, is to
prevent over-excessive brand spillovers that might compromise its platform
through abuse or dilution by the remora (ibid.). The platform owner also aims
to benefit from direct monetization while avoiding thecommodity trap; that is,
offering infrastructure without control of customer relationships. Those
joining the platform want to benefit from the owner’s reputation. This conflict
is shown in practice by, for example, Facebook’s rules prohibiting the use of
its supposed endorsement, and the willingness of competitive organizations to
associate with Facebook by utilizing its logo or other signaling devices (see
Facebook 2013).
Essentially, by committing to a platform, a remora makes relationship-spe-
cific investments, and will therefore be vulnerable to related problems: sunk
costs (i.e., technology development that cannot be redeployed), power abuse
(e.g., host changing the terms), the hold-up problem (i.e., difficulty of switch-
ing in the case of abuse), and even the free-rider problem whereby the plat-
form owner employs remoras to increase its popularity among end customers
while retaining all associated economic gains
199
. Cennamo and Santalo (2013)
note that “[h]igher sunk costs that are relationship specific imply […] a
higher exposure to hold-up problems.” Applied to startups, a learning curve
can emerge for platform-specific skills if the host platform’s technology dif-
fers from the startup team’s skill set.
Compatibility with the current team’s skills is an influential factor as a high
degree of compatibility requires little adaptation with regard to product devel-
opment. However, developing for a single platform (i.e., single-homing) might
become highly asset-specific (see ‘multihoming’ in Subchapter 4.7.3). When
198
In fact, there is no switching as the startup will become a complement not a substitute. We can
refer to this as the ‘conversion cost’, essentially implying the same propensity for a user to join the
startup’s platform.
199
For example, free apps increase the attractiveness of Apple’s App Store, although an Apple
device is needed to access them. Developers do not receive revenue from hardware sales.
183
joining the platform requires skills that can be redeployed in the case of exit,
asset-specificity through skills will not become an issue. In fact, many current
online platforms utilize open Web standards and programming technologies
(Zeldman & Marcotte 2009); therefore, learning them, although being a sunk
cost, does not lead to asset specificity.
Further problems, from the remora’s perspective, include substitution by
acquisition or rivalry (i.e., absorption through substitution). The former hurts
non-acquired competing startups while the latter is harmful for all firms in the
vertical entered by the platform. The platform owner can utilize its marketing
power to secure better positions within the platform for its own features or
those of the acquisition target, when acquired. It is in a far superior position
regarding download trends and other types of information than remoras, from
which this type of information can remain hidden. In the presence of
asymmetric information (i.e., host advantage) and delegation of marketing and
distribution, a natural condition of moral hazard arises (e.g., Pauly 1968). In
other words, by utilizing its power to exercisefavoritism for developer A, the
host will neglect the delegated tasks from developer B. In reality, this is a
common practice
200
, although platform owners tend to build it as amarketing
mechanism, so that the most popular applications receive the most prominent
positions in leader boards and category views
201
.
The issues of power and dependency have been widely discussed in the lit-
erature beyond two-sided markets theory. For example, Yli-Renko and
J anakiraman (2008, 134) argue that “resource interdependencies with other
organizations are viewed as constraints and restrictions; that is, being de-
pendent on an exchange partner means that the partner has increased bar-
gaining power. Therefore, to survive and succeed, firms should take action to
minimize threats to organizational autonomy and attempt to control the re-
sources needed by other organizations to make others more dependent on
themselves.” In the platform context, whether to depend on the platform’s re-
sources or become the platform on which others are dependent is precisely the
question; both include risks, hence the dilemma. Exchanging power for dele-
gation, becoming dependent on sunk costs, and the opportunity cost of “going
solo” are hazards of the remora strategy. In contrast, opting for an independent
launch in platform markets is problematic when the platform is incompatible
with an incumbent platform.
200
“Sorting applications on the basis of popularity, the platform sponsor can choose to own the
highest rank order items, as Microsoft has chosen to do for its operating system and game platforms”
(Eisenmann et al. 2009, 147).
201
The classic conundrum for the non-favored application, therefore, is: How to get visibility
without downloads, and how to get downloads without visibility? Hence, the need for a marketing
function re-emerges.
184
This is noted by Farrell and Klemperer (2007, 2045): “Switching costs and
network effects can work in tandem to discourage incompatible entry: switch-
ing costs discourage large-scale entry […] while network effects discourage
gradual, small-scale entry.” A large-scale entry can be ineffective because the
installed base is unwilling to become a new platform, whereas a small-scale
entry would initially provide the platform for a small network of users, alt-
hough their adoption is prevented by the critical mass in the incumbent plat-
form. Therefore, the startup ends up in the familiar double-bind: the cold start
dilemma, which also represents the tendency of going “back to square one”, as
discussed in Chapter 4.8.
Although strategic thinking influences the behavior of the platform owner,
it is not entirely sovereign in its use of power. Instead, it needs to consider 1)
inter-platform competition, and 2) quality. Consider the negative effects that a
large-scale exit by high-quality supply-side actors would have on the demand-
side as a result of power abuse. This might also lead to high-quality users, in
terms of their high willingness to pay (WTP), exiting the platform; the re-
mainder would be low-quality complements (e.g., apps) and low-quality users
(e.g., free users). This type of escalating chain of events, led by the exit of ac-
tors from one side, has been highlighted by Akerlof (1970) who argued that a
lemon’s market can arise if high-quality actors from one market-side abandon
a market, followed by high-quality actors in the corresponding side, leaving
only low-quality actors in each side.
The degree to which hosts utilize power varies to a great extent. Open plat-
forms, such as Linux in the operating system market, allow the greatest free-
dom, although often the least business support, whereas more closed platforms
(e.g., App Store) can include users with higher willingness to pay (Developer
Economics 2012). If the startup monetizes directly, the feasibility of joining a
platform relates to its user base’s WTP. If network effects are a factor in WTP,
opening the platform might increase aggregate WTP because interoperability
enables access to a larger user base (Eisenmann et al. 2009).
It is relevant to note that the remora’s achieved network will not compete
against that of the platform owner while the remora’s users continue to emerge
from the network; this is because its user base will always remain a subset of
the platform’s user base. In contrast, envelopment aims to take the users away
from the remora. This is thecomplement effect, which makes it feasible for the
platform owner to attract new remoras; in other words, new complements in-
vite new subsets
202
, and the entire network size expands.
202
Although the host network will grow and feed remoras, the reason it will grow is because of
remoras. Effectively, this is a solution for the cold start dilemma, as suggested in Subchapter 4.5.3.
185
The characteristic of social networks to initiate sub-networks (see e.g.,
Ganley & Lampe 2009) in fact gives support to the remora strategy. This is
because they become a powerful entry barrier; for example, consider an en-
trant who would like to create a social network for a particular niche, such as
dog lovers. The entrant will soon discover that the dominant platform most
likely includes a sub-group sharing this interest. If not, then the entrant can
begin such a sub-group and become a complement, which explains why the
diffusion effects are so strong in the social platform field. In addition to net-
work effects, complements, including user-generated groups and applications
created by developers, increase the platform’s benefits for existing and new
users. At the same time, sub-groups introduce an entry barrier for new social
platforms; therefore, the remora strategy becomes feasible, given that the plat-
form allows monetization. For example, in Facebook this is possible inde-
pendently, but Apple controls it in App Store. Although application developers
might enjoy monetization gains, users who create new social sub-groups are
not typically included in revenue sharing (cf. Facebook); this is not in conflict
with their motivation, which is more intrinsic than profit-driven motives.
Some authors argue that the Internet is characterized by winner-takes-all ef-
fects, which are more or less stable in the presence of strong network exter-
nalities. For example, Herings and Schinkel (2001, 25) state:
"The fabulous dynamics of the information and communication
technology sector makes monopoly positions to be temporary. As
soon as the speed of technological innovation diminishes, […] it
is nearly impossible to enter into a sector with strong network
externalities and one monopolist."
The first argument is easily acceptable as seen by quick transformations in a
few key Internet markets; for example, MySpace replacing Friendster, then
Facebook replacing MySpace as the most popular social network, and simi-
larly, Yahoo being replaced by Google in a relatively short period of time
(Gawer & Cusumano 2008). However, a major relaxation to the second argu-
ment of impossible entry, it can be argued, is when the market is suitable for
network externalities in general, not only in the case of the dominant com-
pany; that is, multihoming takes place (Mital & Sarkar 2011). Thus, if we con-
sider multihoming behavior, which is customers willing to join several com-
peting platforms, the argument of impossible entry disappears. This is because
network externalities are not mutually exclusive and can be utilized by many
companies in the market, given that they are able to provide benefits that in-
terest users. If we were to apply a third assumption, namely interoperability,
the initial argument would become even weaker. In an environment of strong
interoperability between platforms (e.g., through APIs), there is less incentive
to remain a proprietary user, or provider, of a single platform, given that the
186
users are active in taking advantage of this feature, and that supply-side par-
ticipants can monetize within the platform. In fact, this can be seen in the an-
ecdotal evidence of people subscribing to several social networks and porting
contacts between them, and also developers creating products for several com-
peting platforms.
4.7.3 Solution: Diversification
A common strategy to reduce dependence on a single platform is diversifica-
tion to several host platforms, similar to multihoming in the platform literature
(Rochet & Tirole 2003; Armstrong 2006). In this strategy, the startup utilizes
several host platforms instead of only one. Salminen and Teixeira (2013) sug-
gest that developers should multihome to avoid being trapped in a single ap-
plication marketplace. Hagiu and Yoffie (2009) provide the example that firms
can advertise on both Google and Yahoo platforms; that is, employ them as
marketing channels to drive search traffic. This idea is now developed further
through a concept calledselective integration that, along with associated strat-
egies, is defined below:
· Selective integration: a strategy of choosing which parts or functions
of a platform are integrated with a host platform.
· Content envelopment: using aggregation from one or several host
platforms to gather a critical mass of content, after which UG negate
any dependence.
· Value envelopment: changing the monetization model when passing
users from the host to the remora.
First, diversification takes place when the remora sources content or users
through aggregation from several hosts. In aggregation, the remora feeds from
several sources; as such, the host might either not necessarily be aware of the
remora’s existence
203
or welcome it
204
. Aggregation reduces dependence from
a single source. A simple rule for dependence of the remora on the host can be
given to illustrate diversification effects through aggregation:
203
As in the case of the auction platform startup in the sample that aggregated results from several
sites. However, lack of awareness might not necessarily help, as it effectively prevents cooperation. If
hosts are cooperative, hiding from them achieves smaller payoffs.
204
As in the case of Google, whereby all websites want to be included regardless of the fact that
Google monetizes their content, the benefits of getting free traffic overcome this nuisance.
187
D =1/x, in whichD =Dependence, x =number of hosts
As x approaches infinity, D approaches zero, assuming equal performance
acrossx. In reality, we observe this effect, for example, through search engines
that aggregate the content of billions of websites; their dependence on one site
is diminishingly small, whereas a developer’s dependence on Facebook or
Twitter, given that the number of available hosts is much smaller, is naturally
bigger. A good example is Google: because it indexes billions of websites, its
dependence on a single site, no matter how big the size of the site, remains
very low; therefore, Google as a remora has the power advantage.
Second, as part of diversification, a startup might opt for selective integra-
tion, a type of mixed strategy that would take place when the startup partially
leverages one or many host platforms; for example, as a source of content or
users, while maintaining, for instance, monetization alternatives in its own
platform. This is the case when the monetization model changes in transition
from host to remora, so that:
Host ? indirect monetization
Remora ? direct monetization, without revenue sharing
The transformation of a monetization model can be referred to as value
envelopment. An example is when the users of a free platform become paid
users of a remora (e.g., Zynga as the remora and Facebook as the host). The
platform owner applies an indirect monetization model (e.g., advertising),
while the remora applies a direct monetization model (e.g., selling virtual
goods), without the host being part of revenue sharing. In another setting, the
host monetizes by distributing its complements (e.g., free apps), and then ap-
plies a more or less generous scheme of revenue sharing.
The reverse can also occur, whereby monetization is delegated to the host.
This is the case when the remora joins an online advertising network, such as
Google AdSense. The agent will then resell the advertising inventory
205
, and
the startup is able to capitalize on the aggregated content. A special case is
termedarbitrage, in which the platform simultaneously buys cheap clicks (i.e.,
visitors) from the network and sells more expensive clicks in return, profiting
from the price difference (Gunawardana, Meek, & Biggs 2008).
Third, content envelopment takes place when aggregation occurs for a lim-
ited time: that is, sufficient to obtain a critical mass, after which UG effects
begin to take place. In this option, the startup employs technology to aggregate
205
In practice, AdSense employs an algorithm based on, for example, semantic matching of content
and keywords, and advertisers’ placement preferences (see Salminen 2010).
188
content to solve the cold start problem, but then relies on user generation
(UG). Aggregation is a means of employing technology to retrieve content.
The goal of employing aggregation
206
as a content model is to solve the cold
start dilemma because the host provides the content. Additionally, when ag-
gregation is employed in relation to diversification, it has the potential to solve
the remora's curse by reducing dependence on a single host. After this, specific
problems relate to monetization and active use, and also, in some cases, the
access towalled garden systems, which are non-accessible by Web crawlers.
Bi-directionality and selectivity under the selective integration require
closer examination. First, both sourcing and spreading content back to content
or social platforms is possible. For example, a startup called AirBnB famously
applied this tactic by spreading its product listings to another much larger plat-
form. Through such efforts, the startup can reach potential users in relevant
verticals; however, the host might perceive programmatic solutions in a nega-
tive light, and block access, as was the case with AirBnB and its target. Man-
ual efforts, however, scale relatively poorly.
Second, technical solutions can be created to facilitate users’ interaction
with the content; a typical example being social media buttons that enable
sharing with various social networks. Because social platforms are dependent
on fresh and interesting content, its provision generates organic traffic for the
platform without the need of integrating its product to the host. Mital and
Sarkar (2011) mention two examples of mutual benefit among platforms:
Facebook and YouTube, which enable the sharing of videos on the social plat-
forms and, simultaneously, increase views as more people click to view the
videos. This is a form of symbiosis between content and social features.
Moreover, when platforms are differentiated, they can “share” users even
when outsiders consider them competitors in the same market. Mital and
Sarkar (ibid.) put forward the example of Facebook and LinkedIn; both are
social platforms, but the creation of connections is complementary as the for-
mer is specialized in private (i.e., strong) ties whereas the latter is for profes-
sional (i.e., weak) ties. Consequently, these features explain why platforms do
not expect exclusivity.
Third, the startup might aim to create embedded platforms (i.e., a platform
within a platform); for example, a dating application in both Facebook and
Google Plus. In this strategy, the payoff results from a spillover effect; some
fraction of the overall user base of the host platform is expected to convert to
206
A technological means to elicit information from various sources, such as public websites and
databases.
189
users of the embedded platform. In addition, brand effects can carry over when
a presence is established in a host platform
207
.
In one of the post-mortems, selective integration (i.e., sourcing data but not
customers) was employed to explain how a competitor was able to solve tech-
nical problems more rapidly and, therefore, produce more benefit to custom-
ers
208
. This suggests that it is possible to join a platform to source information
while retaining control over customer relationships. A potential application of
this solution is to keep the platform’s core technology proprietary while
reaching into online marketing channels; for example, by applying search
engine optimization (Berman & Katona 2012), social media marketing, and
inbound marketing
209
. Although successful application of these tactics might
not be easy to achieve, dominant online platforms are currently easy to access;
for example, search engines (i.e., content platforms) index all websites
210
, and
social media sites enable creation of open communities and fan pages.
From the host’s perspective, remoras are complements. Should they become
substitutes, the host’s attitude might quickly change to being overtly hostile.
As a consequence, within-platform integration sets specific boundaries to plat-
form design; the more active the platform owner is in defending its interests,
the less maneuvering space remoras generally have
211
. Diversification is facili-
tated by inter-platform competition and interoperability through APIs. Host
platforms are open to invite remoras because complements add demand-side
utility (i.e., indirect network effects) that the platform can tax. It is seemingly
a win-win situation, although exclusivity to one platform reduces space for a
remora’s strategic maneuvering, and might be demanded by hosts under some
circumstances (cf. Armstrong & Wright 2007). Aggarwal and Yu (2012) note
that interoperability can be utilized to replicate the host platform’s network
effects due to the fact that the remora accesses the same user base as the host.
They give the example of Google Social Circles that suggests to a new user
207
Consider the popular game Mafia Wars which started independently but become known, and
after integration to Facebook multiplied its user base.
208
“That one mistake (not using or replacing Yodlee [platform] before Mint had a chance to launch
on Yodlee) was probably enough to kill Wesabe alone. […] Everything I’ve mentioned […] are great,
rational reasons to pursue what we pursued. But none of them matter if […] a shorter-term
alternative is available.” (Hedlund 2010).
209
SEO aims to increase one’s search engine ranking, whereas inbound marketing aims to reach a
strong presence in communities relating to the startup’s industry (Halligan & Shah 2009).
210
For example, Google scans websites that provide their content, and in exchange receive free
traffic, also known as organic as opposed to paid traffic. In this model, Google is a remora that
employs aggregation to tap into content platforms to retrieve sites in which its user base is interested.
Any platform that has content which it allows Google to index is also a remora that receives the
organic traffic.
211
Consider Microsoft that declined to support Intel’s hardware solution due to its proprietary
software which could have risked Microsoft’s hegemony over the platform (Evans et al. 2006).
190
the option of replicating his or her existing social network structure, retrieved
by accessing Facebook.
There are also some downsides to diversification. Most importantly, the
remora faces additional costs of accommodating different platforms. As noted
by Hagiu and Yoffie (2009), downsides include “the extra engineering, mar-
keting, and support required to play with several MSPs [multisided plat-
forms].” The integration costs depend on the scope and depth of integration. In
marketing integration, the costs are relatively low, whereas building separate
functionality for different platforms is costly. This limits the effectiveness of
the solution, especially for startups subject to resource constraints. In general,
the startup’s ability to maintain a portfolio of different technologies in various
platforms involves resource constraints relating to team skills, time-to-market,
and finances. Finally, as a technical solution to social problems (e.g., content
sharing and user-driven user acquisition) aggregation has severe limitations
because these problems cannot be solved programmatically. In other words,
aggregation as a means to collect and treat data can be highly beneficial, but it
might not necessarily be sufficient to kick-off user interaction with the content
that is necessary to harvest UG effects.
In sum, aggregation from many sources is more effective in reducing the
host’s power than multihoming when there are a limited number of choices
(i.e., the market is oligopolistic). If, for example, there is an oligopoly in social
platforms, the dependence is larger due to fewer substitutes; however, assum-
ing that these platforms compete, they have a strong incentive to provide at-
tractive terms to complements, and thus dependence on fewer hosts can lead to
better bargaining power for remoras. As such, intense inter-platform competi-
tion is likely to curb hosts’ opportunistic behavior.
4.7.4 Discussion
In platform markets, remora’s curse can be regarded as a tradeoff between cre-
ating a platform and joining an existing one
212
, whereby the opportunity cost
of creating a platform is foregoing the customer base of an existing platform,
and the opportunity cost of joining a platform is the loss of power relating to
technology, monetization, and marketing choices, which are dictated by the
platform owner at its convenience. For example, the platform owner might
decide to favor some startups over others in terms of visibility within the plat-
form. It might also study them and decide to offer a substitute, sometimes
212
A platform startup will effectively build “a platform within a platform”, although it inherits the
rules of its parent platform, and therefore remora’s curse applies.
191
termed ‘absorption strategy’. The remora tradeoff is therefore a strategic
choice; instead of creating a platform, a startup decides to join one, thereby, in
theory, eliminating the cold start problem because the platform functions as a
channel for distributing the service and acquiring new users. More precisely,
the strategic choice of building on an existing platform grants benefits such as
saved development time, cost, and direct access to an existing user base and
content. Some of these benefits can become a source of competitive ad-
vantage. In other words, the choice to engage or not is also influenced by a
startup’s competitive strategy.
If the platform is “fair”, it does not take advantage of its power advantage.
However, the state of fairness is not necessarily stable, and eventually the plat-
form might behave strategically; for example, it would prevent a startup from
gaining such excessive rents that the platform itself would become dependent
on the startup. In other words, although it can be achieved, success in a plat-
form market is capped for remoras. Finally, although not explicitly mentioned
by informants, platform-reliant startups might be involved in a race to the
bottom if the platform they have chosen is replaced by another platform
213
.
Although platform dependence is embodied in remora’s curse, the benefits
might counterbalance this risk, at least in the medium term, which seems to be
suggested by some anecdotal evidence, including success cases and what if
statements
214
. By not leveraging a pre-existing platform, a startup faces other
dilemmas, mainly monetization and cold start dilemmas, in addition to its
variation of the lonely user dilemma. The benefits of the platform relate to
customer acquisition and monetization
215
that, for both, the platform offers a
potential channel with a critical mass. In other environments, monetization can
be more challenging and customer acquisition requires seemingly more ef-
forts
216
. Further, the platform can provide competitive advantages over rivals
if it is exclusive instead of inclusive; when it is not, a remora faces intra-plat-
form competition.
Therefore, if the startup applies a platform business model, it can become a
competing platform itself by choosing the independent option, or it might turn
into aplatform within a platform by choosing to join an existing platform. In
the latter case, however, its power is restricted in the same manner as if it
213
For example, within a couple of years, Facebook replaced MySpace as the most popular social
network, growing from zero to 600 million users in a few years (Hartung 2011).
214
“So how would I do things differently today? […] I would wait until [the] location is all clean
and dandy with the carriers and build on top of that.” (Bragiel 2008).
215
The host platform’s coordination can increase WTP by making it attractive and easy; often
termed ‘attributes’ of Apple’s App Store, which collects credit card numbers upon registration and
enables one-click purchases.
216
Although within the platform, the competition is typically comparable to that outside the
platform ecosystem; that is, the startup cannot escape the need for differentiation.
192
applied a non-platform business model. In other words, the strategic choice of
integrating the product on top of a platform leads to a degree of platform de-
pendence that, in turn, increases competition against other firms within the
platform
217
, and also puts the startup at risk of the platform’s strategic deci-
sions, which naturally deviate from the startup’s goals in some aspects. This
can result in changing, for example, the rules (i.e., terms of service), technol-
ogy, design, or behavior of the platform, increasing API cost or restricting ac-
cess, introducing a directly competitive feature (i.e., substitute), thereby mak-
ing the startup’s product redundant, or acquiring a competitor and then favor-
ing it within the platform, for example, by providing it with more visibility. As
there are some vivid examples of this strategic behavior, it has been widely
discussed in the startup community. Further, building a proprietary technology
to integrate into the platform imposes the opportunity cost of developing an
independent product, although in some cases a startup might resort to diversi-
fication to solve the dilemma
218
.
However, the sample demonstrates that the multihoming process is not au-
tomatic, even if transference logic is assumed. There is a need for more re-
search to identify the conditions for successful transference of business logic
across contexts. This study, however, demonstrates that such a problem exists
and, consequently, a working business model in one context (e.g., location)
might not generalize, or at least requires “starting from zero”, as the new mar-
ket is not part of the same network.
Especially, technologically oriented founders, who are sometimes charac-
terized by a lack of interest in marketing (see Roberts 1990), might perceive it
as feasible to delegate marketing and distribution functions to the platform
owner. However, they might fail to acknowledge that, within-platform, they
arealso subject to competition, and so the need for differentiation by, for ex-
ample, marketing, will not dissolve. In addition, marketing is needed even
when externalizing user acquisition for the platform owner, due to the plat-
form owner’s incentives to favor high-performing remoras
219
.
217
Assuming the platform is open and there are little or no barriers to entry.
218
Diversification can be regarded as developing a separate individual product for an independent
marketing channel (i.e., website) or developing for other platforms. Such a diversification reduces the
risk of lock-in, but does not eliminate remora’s curse because the startup is nevertheless investing its
scarce resources into the chosen platform(s).
219
As the platform owner’s revenues are directly proportional to participants in the platform, it
ensures most prominence to participants that generate most revenue. In the case of indirect
monetization, the revenue can be substituted with impression, clicks, or other types of economic
value. The notable limitation is that the incentive does not always lead to action, which is the case of
the “fair” platform owner, although the stability of such a state always represents a risk for the startup
as it has no means of influencing the platform owner’s strategic choices.
193
4.8 Summary and discussion on dilemmas
The idea for dilemmas was born from reading the material. Initially, it was
discovered that founders identified and named them in their post-mortem sto-
ries. For example, one founder mentioned a “cold start problem”, which was
then also discovered in other cases, although not by the same name.
Table 27 Applicability of dilemmas across platform types
Cold start Lonely user Monetization Remora
Content platform x x x
Social platform x x x x
Exchange platform x x x x
The cold start dilemma is applicable to all platform types considered in this
study; a particular number or amount (i.e. critical mass) of users or content is
needed to evoke willingness to join a platform, whether the platform is based
on content, social, or exchange interaction. Both the cold start and the lonely
user dilemmas are chicken-and-egg problems, and relate, respectively, to con-
tent platforms and social platforms, the latter requiring an active user base or
content to generate growth through network effects. In addition, marketplaces
(i.e., exchange platforms) face liquidity needs. It is negligible whether
liquidity in their context is understood as content (e.g., product listings) or us-
ers (i.e., buyers and sellers).
The cold start dilemma relates to content platforms with interaction such as
content creation and consumption, and also transactions in the context of ex-
change platforms, whereas the lonely user dilemma relates to social platforms
with interaction such as joining the platform; typically, users register or oth-
erwise subscribe as followers. Both, however, aim at user generation (UG)
effects, so that users’ actions lead to a desired response from other users, such
as content contribution, sharing, and invitations
220
.
Moreover, the cold start dilemma can be defined as a problem of one-sided
content platform, when users are homogeneous, or a two-sided problem, when
users are divided into consumers and contributors of content. Similarly, the
lonely user problem can be a problem of similar side critical mass (i.e., friends
or acquaintances are required to join and actively utilize the platform), or a
220
If the users are classified as one group, it is termed a one-sided platform. If they are classified as
two complementing groups, it is termed a two-sided platform. If they are classified as three or more
groups, it is termed a multisided platform.
194
two-sided problem (e.g., men and women finding each other in a dating web-
site). The only platform type that is categorically two-sided is the exchange
platform, which always requires different sides (i.e., buyers or sellers) for in-
teraction to take place.
In terms of implications, it is important to distinguish pure content plat-
forms from social platforms because contributing content can be regarded as
more demanding than engaging in social interaction; thus, different types of
incentive might be required. Then again, for exchange it is important to build
liquidity; a good volume of both sellers and buyers, so that goods are sold at
appropriate prices. The incentives of the platform owner and traders are usu-
ally well aligned as the rewards of exchange platforms tend to be tied to the
volume of transactions taking place in the platform
221
. Finally, social effects
are associated with UG; users, for example, upload videos on YouTube for
others to watch, not primarily to gain economic benefit
222
.
The monetization dilemma and remora’s curse are applicable to all platform
startups; the company needs to be financed which requires direct or indirect
monetization, that is, charging the user for access and/or usage or charging a
third party, most typically advertisers. In a similar vein, it depends on the
user/content acquisition strategy whether the remora model is applied and
therefore applicable, which is possible in all platform types: content platforms
can attempt to source content, social platforms users, and exchange platforms
product listings.
It is typical that attempts to solve one dilemma result in the discovery of
another. This principle is demonstrated in the following figure.
221
However, this is not always the case; eBay takes a commission but the Finnish auction site
Huuto.net only charges for premium services while also monetizing by offering advertising space.
222
Although YouTube offers a partnership program for the most popular content providers.
195
Figure 15 Dilemmas and associated problems
If solving the cold start problem or lonely user problem by offering a free
product [1], the startup faces the monetization dilemma [3]. Therefore, even
successfully building the user base does not guarantee business viability. This
is due to the fundamental difference between a customer and a user; the
former brings in revenue, while the latter brings a cost that needs to be
covered by indirect monetization. There is a discrepancy between the growth
of the user base and growth of revenue that is a consequence of indirect
business models not being perfectly elastic to the growth of user base. This is
implied, for example, in Goldfarb's (2003) model, based on the assumption
Cold start dilemma
Freefying Remora
Remora’s
curse
Monetization
dilemma
r
u
n
w
a
y
Paid product
Freemium
Feature
definition
problem
Problem of active
use
sol ved
Problem of
quality variance
Problem of
free
Illusion of
free
1 2
3
4
6
7
8
11
1
10
9
13
14
17
Lonely user dilemma
Transferability
problem Real-time problem
16 15
196
that users do not provide the revenue directly but it comes from advertisers
223
.
It then follows that users are not worthless while also not being as valuable as
many startup founders would like to think. Based on the author’s analysis,
seeking customers, even at the risk of “scaring away” users who are unwilling
to pay, seems a more recommendable strategy
224
. As a minimum, the startup
should look for ways to diversify indirect monetization instead of being de-
pendent on advertising. Further, the lack of consideration for business viability
also concerns the platform literature; for example, Evans and Schmalensee
(2010, 5) noted “we do not address whether a platform that attains a critical
mass would in fact be profitable; this would require the explicit consideration
of costs and other revenue.”
To solve the cold start dilemma, the startup is tempted to join a platform
with a pre-existing user base [2], anticipating that the barrier for users to join
is lower when they have already committed to the host platform. This comes at
the cost of giving away power (i.e., remora’s curse [3]). Whereas remora’s
curse addresses managing a relationship with the platform owner, the cold
start dilemma relates to becoming the platform owner. Particular problems of a
remora include platform dependence and potential hold-ups. Realization of
remora’s curse, that is, the host platform cutting access [6] to users or content,
will in effect lead the startup back to the cold start problem, but only given
that it has failed to reach a critical mass.
When the startup solves the monetization problem through the freemium
model [7], it is left with a problem of feature definition [8]. In other words,
giving away too many features leads to low conversion from paid to free user,
whereas giving too little away leads to lack of adoption in the first place. An-
other option is paid product [9], although this can lead to a similar problem of
lack of adoption (i.e., cold start). Note that there are two different states for
users’ WTP: positive and negative. For negative WTP, paid products always
result in defection, and there is a problem with free [10] because the startup is
forced to subsidize. However, if WTP is positive, then the startup risks an
illusion of free [11] in which it offers a free product, even though the users
would have been willing to pay.
With regard to users, different problems arise before and after they join a
platform, so that:
223
The assumption can be extended by arguing that the advertising market has its own dynamics,
which means that users in one website are not interchangeable with those in another with regard to
their advertising value. For example, consider Friendster that, when selling advertising space to
American companies, noticed their visitors mostly comprised Filipino consumers (see Chafkin 2007).
224
This line of thinking is based on the idea that not all startups can become category leaders (e.g.,
Facebook) that are able to accumulate hundreds of millions of page views per day, and thus attract
advertisers’ interest.
197
Before joining ? cold start dilemma
After joining ? lonely user dilemma, problem of active use & quality
variance
Theproblem of active use [13] implies that even after solving the cold start
problem, the startup is at risk of losing the achieved critical mass if the users
become inactive. This can cause ‘negative tipping’, which is essentially the
reverse of exponential growth. The problem of quality variance [14] will, in
effect, require the startup to introduce either manual or automatic monitoring
mechanisms. In the ideal user generation (UG) model, it is assumed that the
user base is self-controlling; thus, the platform offers tools such as a reporting
function and recruits some active members as moderators of quality. However,
even if the users are active in keeping misconduct in check, the problem arises
when the low-quality content is not malicious but otherwise not interesting to
other users. For example, consider the case of an indie music portal that failed
due to low-quality bands (Hagiu & Wright 2013). It seems reasonable to as-
sume that, in some cases, the startup needs to incur monitoring and interven-
tion costs to assure that the user-generated content matches the interest of
other users
225
.
Coincidentally, the runway [17] keeps depleting while the startup deter-
mines the problems. If founders are unaware of platform-specific issues, as
many of them were in the sample, it will take them some time to understand
the problem, and then some more time to think of potential solutions. Then,
they might run into additional problems as displayed in Figure 17. In contrast,
by being aware of potential risks, the startup is ablea priori to prepare a range
of solutions for multiple dilemmas at the same time.
Furthermore, relying on UG aggravates the cold start dilemma. Instead of
in-house production or syndication through partners to acquire customers and
content, the startup expects users to play this role. When the process fails, the
startup can find itself looking for “plan B”. However, at this stage, it might be
too late, as exemplified by one startup’s story:
"We modified our technology to be a very flexible and scalable
platform from which we could launch any type of application, for
any client, in any industry. We thought we could position our
solution as helping brands create a comprehensive distributed
touch point strategy by complementing their presences on Face-
book and Twitter with a presence on IM [instant messaging].
The plan was to partner with marketing agencies as well as sell
225
For example, refer to YouTube’s tactics of getting video material from attractive women by
posting on Craigslist (Evans 2009a, 113).
198
directly to clients similar to the approach taken by providers of
custom branded widgets, Facebook apps, and mobile apps. This
strategy eventually produced some great results but it was a case
of too little, too late. When we finally decided to pivot we had al-
ready spent most of the capital raised in our seed round."
The end of the runway signifies failure. In the absence of financial buffers,
the runway might not provide a sufficiently long period of time to solve the
problems. In contrast, venture funding, although providing resources, can lock
in some choices, which prevents a later adaptation (i.e., pivot). Furthermore,
venture funding can impose a situation of “go big or go home”, which might
negate the apparent freedom afforded by the funding
226
.
Finally, there are two specific problems associated with the lonely user di-
lemma in Figure 15. First, thetransferability problem [15], which implies that
a critical mass is not automatically transferable from one context (e.g., loca-
tion, niche market, or demography) to another context (e.g., another city or
user demography). Second, thereal-time problem [16], which implies that, in
particular circumstances, the emergence of a match between parties of a two-
sided platform (i.e., network effects) is dependent on time. An empty chat
room is an example: no matter how many users have registered, if none are
present, their value at timet is zero for the only user.
This also marks how the cold start and lonely user dilemmas differ: content
is static while social interaction is dynamic
227
. Registration does not guarantee
content production (e.g., becoming an active user) and content production
does not necessitate registration or other type of subscription. Therefore, the
root of these two motivational problems differs. Simply put, it is assumed that
users do not generate content for exactly the same reasons that they join a
social network, although there might be an overlap. More precisely, their
behavior can involve spillover effects, as implied in Chapter 3.3.
In sum, this chapter has shown empirical grounding to the chicken-and-egg
problem presented in the platform literature. More importantly, the study has
shown that the problem 1) can take specific forms (i.e., cold start and lonely
user) based on the type of coordination required (e.g., timeliness), and 2) is not
isolated, although some of its potential solutions applied by the failed plat-
forms startups are associated with further dilemmas; for example, the moneti-
zation dilemma and remora’s curse. This is an important finding as most of the
226
This was conceptualized as “Peter Pan’s dilemma”, although is not discussed thoroughly in the
study (see Chapter 4.1).
227
However, content can have different modes of freshness. A good treatment to the topic is given
by Kim and Tse (2011) who study knowledge-sharing markets and argue that there is both knowledge
that expires rapidly and knowledge that remains valid for a long time; although, while the content is
static in both cases, its benefit to the user is dynamic. For example, consider yesterday’s news that is
not so valuable today.
199
literature considers the chicken-and-egg problem in isolation. It is argued here
that potential solutions can aggravate the platform startup’s problems in the
big picture; for example, by denying monetization or making it dependent on
the host platform’s strategic choices. Hence, solving the cold start problem can
come at a significant cost, and thus 3) potential solutions need to be consid-
ered in terms of their impact on cascading dilemmas and problems.
201
5 SOLVING THE DILEMMAS
5.1 Introduction
As described in the method chapter, after several rounds of GT analysis, the
researcher reverted to the data, and coded 1) “what if’” statements from
founders; that is, what they would have done differently; and 2) the attempts
expressed in the post-mortems to solve the problem when it had been identi-
fied. This process was accompanied by interviewing six startup founders. The
proposed solutions are synthesized with the platform literature, and their
strengths and weaknesses discussed. In addition, separate solutions that arose
from the literature are analyzed in terms of their appropriateness. Finally, a
summary is presented.
The solutions here do not relate to pricing, subsidies, or integration into a
larger platform (i.e., remora), as these solutions and their strengths and weak-
nesses have already been discussed in the previous chapters. It is also note-
worthy to mention that in most solutions, sides of the platform are treated sep-
arately. Essentially, if growing each side separately from one another is taken
as a goal, the chicken-and-egg problem transforms into classic marketing
problems: "How to acquire customers?" and "How to build awareness?". This
vastly expands the scope of solutions as an array of marketing tactics (e.g.,
promotion, personal selling, and various means of digital marketing) becomes
available.
Despite this premise, a startup is forced to consider both market sides (i.e.,
sets of customers) to generate any action on the platform as their interdepend-
ence remains, regardless of the applied user-acquisition methods. However,
some observed solutions are now discussed.
5.2 Solutions
5.2.1 Exhibits
The solutions discussed in the following are formulated and given names
based on the post-hoc analysis. Exhibits of these solutions are presented in
Table 28.
202
Table 28 Exhibits from post-hoc analysis
ID Insight
[A] "One of the unseen benefits of the new system was that it enabled us to anonymize, extract, and
aggregate bookmark data. So we dove into that and started looking at what products we might
be able to deliver powered by the “corpus” of what would soon be 100 million bookmarks."
(Agulnick 2010).
"This was no mean engineering feat. We had a very, very large and complex back-end. And
even with this, the quality of the data coming through to the end-user was just not that good.
Too much spam, still. Duplicate posts. Sometimes mis-categorized. Difficulty applying our
reputation algorithms. Not good." (Ehrenberg 2008).
[C] "[The startup] employed a group of talented journalists and community representatives who
sought out and interacted constantly with members of each of our communities to encourage
them to participate." (Potts 2007).
[D] "The main failure of [the startup] was marketing. Dev and I came from PayPal, a strongly viral
product at a company almost hostile to marketing. Our efforts in SEO [search engine optimiza-
tion], SEM [search engine marketing], virality, platforms, PR, and partnerships weren’t terri-
ble, but drawing users to a live event requires constant, skillful work." (Goldenson 2009).
[E] "Like creating content, I no longer think marketing is something smart novices can figure out
part-time. As the Web gets super-saturated, marketing is the difference-maker, and it’s too deep
a skill to leave to amateurs […] Next time we’ll raise enough to hire a marketing expert early."
(Goldenson 2009).
[F] "We struck upon the idea that if we had fifty journalists, and they each cross-promoted each
other to their social networks, then over time we would get more and more people to read each
other’s content." (Biggar 2010).
[G] "Because we were basically calling on friends of friends who ran events to be our customers, we
didn’t learn what event organizers in general wanted or how to acquire them as customers in a
scalable way with the 'private social network product'." (J ohnson & Fraser 2010).
[H] "We could have and should have used the proceeds of the convertible note to get out from under
Facebook’s thumb rather to invest further in the Facebook platform." (Rafer 2009).
"Since the service was our child we were reluctant to make a decision about closing it faster
and limit losses. We’ve been tricking ourselves thinking that everything would be fine while we
couldn’t get the application back to work properly." (Brodzinski 2009).
[J] "[The startup] was designed as one community, but it really was a network of unaffiliated com-
munities. […]. The site was not optimized for that. We should have had more tools for assign-
ment creators to tie their contests to their existing communities." (Powazek 2008).
[K] "It’s very, very difficult to start from scratch in a community and get to critical mass without
help. For a variety of reasons that made sense at the time, [we] chose not to go the media part-
ner route. But as newspapers and broadcasters have become more savvy in the past few months
about their need for hyper-local efforts, it makes more sense for hyper-local entrepreneurs to
hook up with media partners […]." (Potts 2007).
[L] "Figure out the difference between a website, a service, a product, an application and a plat-
form. You need to figure out which one you’re building because what users do with each one
and how you make money is very different. If you answer all of the above, you’ve got a problem
because the answer determines why people use what you’re offering, and it says that your focus
is scattershot. The difference between these is another post entirely, and one I’m probably not
qualified to write yet." (Hemrajani 2010).
[M] "To serve investors and the entire ecosystem who we heard from, we launched CB Insights, a
subscription platform that offers faster, friendlier, comprehensive intelligence about private
companies. It was built after talking to customers this time […]. To serve entrepreneurs, we
remade ChubbyBrain as a place to leverage our data in ways that would benefit them. Our first
tool is the free Funding Discovery Engine (FDE), which emerged directly from the question we
repeatedly got from entrepreneurs to the old ChubbyBrain […]." (The Chubby Team 2010).
203
[N] "I had to deal with, while building BricaBox, why we weren’t modifying an existing Open
Source solution, like WordPress MU. We were a CMS [content management system] at heart,
after all. Next time, I’ll give more consideration to building off and participating in existing
Open Source project." (Westheimer 2008).
[O] "Another reason PlayCafe’s complexity hurt us is that developing good content and technology
simultaneously required too much time. We tried to make each deep and stable — important, we
thought, given our live nature — but we were too slow to iterate in a novelty- and entertain-
ment-based business." (Goldenson 2009).
[P] "The easiest way to avoid chicken-and-egg problems is simply to have a product that is useful
on its own. Del.icio.us, for example – it’s just a bookmark manager that happens to be more
useful as more people use it." (Tang 2008).
[Q] "The way to do it seems to be to make sure you're passionate enough about your own product to
use it yourself (like submitting your own stories to Reddit, or how we all created 3-4 sockpup-
pets at in Asphere and had conversations with ourselves on the forums) and to get out there and
put it in front of lots of other people. It’s that last point where we failed: we just kinda built the
product, launched it, and let it die." (Tang 2008).
The following sections address the treatment of the exhibits and interviews
in alphabetical order.
5.2.2 Advertising
Advertising emerged as a topic in the interviews. Especially in platforms that
serve consumers (i.e., side A, demand) and companies (i.e., side B, supply), it
was perceived as important that the platform owner conduct marketing to at-
tract consumers. Often, founders simplify marketing to mean advertising. Alt-
hough marketing comprises much more than advertising (e.g., Hunt 2002), it
is discussed here as a possible solution.
In general, there are two approaches to advertising:
Strategy A: mass media advertising
Strategy B: targeted, niche advertising
While the former is generally considered too expensive for startups, the
latter seems a more feasible tactic. In fact, both were mentioned by the inter-
viewed founders. One of the founders explained that their local market in Af-
rica has quite low mass media advertising costs, and that if they were able to
acquire funding, they would employ it to drive adoption through mass media.
Another founder mentioned that they only employ targeted, low-cost adver-
tising. Another interviewed founder mentioned, early in the interview, "all
morning we have done Facebook advertising" and revealed that it is the most
cost-effective marketing channel for them. This was interesting, as most
startups placed Google before Facebook.
204
In the discussion, it came apparent that marketing "super platforms" exist
that enable almost anyone connected on the Internet to be reached. In particu-
lar, two alternatives were discussed, Facebook and Google:
· Facebook: enables targeting social network users, based on their
demographic information and preferences (i.e., likes), with organic
and paid messages.
· Google: enables targeting searchers, based on, for example, keywords
used and location, with organic website content and paid ads.
Characteristic to the low-cost advertising approach are testing with small
budgets, carefully calculating the cost of conversion, and attempting to find
keywords or demographic niches that are more likely to attract users to the
platform.
In general, the marketing platforms require very little startup capital, and
enable freedom in managing the budget and also highly advanced targeting
functionalities:
Google AdWords ? search intent
Facebook Ads ? demographics and preferences (i.e., likes)
Somewhat consistent with the earlier division, these two often mentioned
marketing channels are distinct in terms of motive of usage (Google ?
content, and Facebook ? social interaction).
The main limitations of niche advertising are: 1) user acquisition through
linear growth; that is, the more users are wanted, the more it will cost. Often,
this results in issues, as the user acquisition cost is higher than the immediate
or lifetime revenue of the user, which is particularly relevant when the plat-
form offers free access and usage. In such a case, there is incompatibility: paid
user acquisition and free offerings can easily tilt the finances of a company
into a critical state; 2) there are natural limits to the size of a niche: if, for ex-
ample, a niche is based on a particular interest in a particular location, the
growth potential remains limited
228
; and 3) platforms are expected to accumu-
late new users as interaction becomes self-sustainable, and therefore advertis-
ing can be perceived as a kick-off or temporary solution, not one that is sus-
tainable or structural.
As a consequence, the following proposition can be formed:
228
Although this feature is a consequence of small market size, it is nevertheless a limitation.
205
Proposition: Advertising is a successful solution to the cold start dilemma
if it leads to exponential, not linear, growth of user base and interaction.
Other limitations of advertising are resources, know-how, and return on in-
vestment. Startups tend to lack marketing skills and budgets. For example, the
interviewed startup which mentioned Facebook as a cost-effective method of
user acquisition is in close partnership with an advertising agency that pro-
vides them with marketing services on an on-demand basis. Based on other
investigated cases, this is a rare luxury among startups. Return on investment,
also termed ROAS (i.e., return on ad spend), measures how well advertising
investments generated revenue. As noted by one of the interviewed founders,
if paid user acquisition is applied, the platform's revenue potential must be
high
229
.
5.2.3 Aggregation
The tactic applied in [A] was to aggregate data from various sources into a
single "corpus" of content that would be valuable to the platform’s users. Alt-
hough this tactic seems to have some potential, the example of [A] vividly
shows the linkage between the cold start and monetization dilemmas: that
solving the cold start dilemma by aggregating data leaves the question “how to
monetize this data?” unanswered.
First, it is not self-evident that the data, or content, per se are valuable to
users and, second, that they would be willing to pay for accessing it. Further,
as described in , the quality of the aggregated data can become problem-
atic. According to the ideal user generation (UG) model, deploying users to
“clean” the data is a potential solution, although this can be problematic given
that users might want high-quality data but might not necessarily want to pro-
duce/edit it. Thus, the cold start dilemma is effectively not solved by aggrega-
tion unless users consider the provided content suitable for their needs.
5.2.4 Community
If theoretical network effects do not materialize, the ideal UG model fails and
the startup will be in trouble. Consider the case reported in [C]: for the
purpose of kick-off, or community building, the approach seems logical and
sound. However, it simultaneously restricted the startup’s opportunity for
229
CAC (customer acquisition cost) 50) was larger
than the original sample, it was possible to gather useful industry insight that
enhanced the findings.
The discussions with founders outside the sample extended over a period of
more than three years, and involved active participation in startup events in
Finland, Sweden, and the United States. During these events, the researcher
conversed with founders and aimed to develop his strategic thinking, and also
an understanding on the circumstances and rationale of startup decision-mak-
ing.
264
Fourth, comparison with the extant literature helps to confirm findings of
inductive studies (Miles & Huberman 1994). According to Eisenhardt (1989,
544), based on potential conflicts “readers may assume that the results are
incorrect (a challenge to internal validity), or if correct, are idiosyncratic to
the specific cases of the study (a challenge to generalizability).” There were
no major conflicts with the constructs arising from the platform literature; in
fact, the strategic dilemmas elicited from post-mortems can be connected
rather easily to the platform literature. It was found that the literature
confirmed and deepened the analysis by presenting several strategic solutions.
In contrast, the empirical analysis sheds light on the dilemmas’ relationships in
the context of online platforms. The two approaches complement one another.
Fifth, the stories are public and can be accessed by anyone. Other research
might confirm or refute the conclusions made in this study, and “readers may
apply their own standards” (Eisenhardt 1989, 544). Hidden information can be
regarded as a major obstacle in assessing the credibility of research (Sommer
& Sommer 1992). Therefore, public data are an advantage for evaluating the
credibility of this study. General principles of scientific inquiries include repli-
cation (Easley, Madden, & Dunn 2000). For this purpose, and to demonstrate
the logic applied in the analysis, we have provided a coding paradigm that lists
the codes and their meaning. Thus, another researcher can internalize the cod-
ing structure and repeat the study in question. The coding guide can be found
in Appendix 1.
6.5.3 Success with theory
As the study’s purpose is to create a substantive theory, the evaluation of GT
should also consider the extent to which this attempt is successful. Kempster
and Parry (2011) note that a constant concern in GT research has been the ina-
bility to raise the abstraction level, while remaining “stuck” in description.
Although it cannot be stated that strategic problems as a core category explain
all similarities and variation (Kan & Parry 2004), the author maintains that,
according to his interpretations, these problems represent a remarkable pattern
in the data and, combined with the insight of the platform literature, arguably
influence the success or failure of any given platform startup that is compati-
ble with the online-specific typology presented in Chapter 3.
According to Charmaz (1990, 1164), “a theory explicates a phenomena,
specifies concepts which categorize the relevant phenomena, explains rela-
tionships between concepts and provides a framework for making predic-
tions.” Making exact predictions through our framework requires skill. How-
ever, the purpose is indeed to provide such characterization, or a dilemma
265
roadmap. The extent to which founders are able to understand the results of
this study is difficult to know. However, in verbal discussions with them,
identifying these specific dilemmas has intuitively resonated with the 50-or-so
founders with whom the author has conversed.
According to Wagner et al. (2010), an “adequate use” of GT would identify
which approach (i.e., Glaserian or Straussian) the researcher has followed,
mention the specific GT techniques employed, and generate real theory as op-
posed to case descriptions. This study was positioned in the internal GT debate
(see Chapter 2.4), explicated its use of GT techniques (Chapter 2.4), and
aimed to transcend description by conceptualizing “lasting” dilemmas. How-
ever, as a substantive theory, the theory presented here is limited in analytical
generalizability. It depends on the reader to determine its usefulness. Argua-
bly, the study offers more insight for scholars and practitioners familiar with
the strategic problems than for others. This is simply the nature of substantive
theory, the credibility of which is a “joint venture” of the researcher and the
reader (Glaser & Strauss 1965).
According to Strauss and Corbin (1994), a theory is “ready” when there are
no more novel possibilities; that is, a point of theoretical saturation has been
reached. Theoretical saturation occurs when there are no new variations in
terms of codes and their explanation is relevant to the central construct, and
when the imaginable settings and configurations relating to the theory,
emerging from and applied to the data at hand, have been exhausted. A corol-
lary to this method is, however, that GT is never truly ready as, by constant
comparison, it can always be expanded (Glaser 1978). Grounded theory gives
a representation of reality, but this representation is not meant to be decisive as
reality and contexts keep changing
259
.
6.5.4 Saturation
Theoretical saturation can be deemed a criterion for GT studies, although its
existence or absence is difficult to verify (Gasson 2003). Accordingly, data
collection (i.e., theoretical sampling) must continue until saturation. Strauss
and Corbin (1994) refer to this as “category saturation”, implying that the core
category and its subcategories need to be exhaustively covered. Simultane-
ously, theoretical sampling is a technique of verification applied by the re-
searcher (Suddaby 2006). Saturation emerges when there are no more
259
Theory is never absolute because it can never be completely proven; it can falsified (i.e., refuted)
or verified to some extent (Hunt 2002). However, a theory has an indirect relationship to facts that, in
turn, have an indirect relationship to what is termed reality (Meyling 1997).
266
surprises that challenge the emerging coding system (Finch 2002). Suddaby
(2006, 639) states that saturation is signaled by “repetition of information and
confirmation of existing conceptual categories”, which he notes depends not
only on the empirical context but also on the researcher’s experience and ex-
pertise; in grounded theory terms, theoretical sensitivity (Glaser 1978).
Reaching saturation in this study was signaled by three indices. First, the re-
searcher found patterns; several instances of each dilemma were found across
different post-mortems, and thus the data were perceived to be sufficient (refer
to exhibits in Chapter 4). Second, discussion with founders beyond the sample
yielded no significant new insight that would have challenged or increased
understanding beyond the initial findings. Third, familiarization with industry
circumstances supported the existence of the dilemmas but did not yield addi-
tional, platform-specific dilemmas, which was a criterion in the theoretical
integration phase. Although the author evaluates that the findings relating to
the dilemmas are saturated, he does not make the same claim for the solutions
which require more research to be comprehensive. In fact, it seems that more
strategies and tactics to solve the issues emerge constantly; therefore, it is ar-
guable whether a definitely comprehensive description of them is even possi-
ble. Moreover, theory-wise, the biases should be integrated to the dilemmas,
as they seem to increase explanatory power concerning why the founders were
unable to solve the dilemmas. In this sense, the theoretical claims made in this
study do not establish a saturated whole, or a "ready" theory, that cannot be
expanded by further studies. Thus, further studies are required for the above-
mentioned purposes.
A part of the reason for employing multiple criteria to assess research qual-
ity stems from different scientific paradigms and philosophical stances (Kuhn
1970). This study represents acritical realist approach, thereby assuming that
the analyzed material reflects reality (Kempster & Parry 2011). While it has
been argued that classic GT follows a positivist agenda in “discovering” the-
ory (Gasson 2003), Glaser in his later works (see e.g., Glaser 2004 & 2008)
seems to belittle the risk of researcher’s bias in finding this theory
260
. In this
study, it has been deemed important to recognize the fallibility of the human
condition. After all, asystematic distortion in either the data or their interpre-
tation would lead to unhelpful conclusions and be contrary to the purpose of
GT, as embedded in Glaser’s (1971) criterion of “theory that works”.
260
Glaser’s logic is twofold: if an informant’s account is biased, either the bias becomes a social
process to examine (e.g., impression management) or it is irrelevant because it nevertheless influences
the informant’s actions. However, Glaser fails to explain how the researcher concludes whether a
piece of data is biased or not, and how this influences GT’s ability to account for mechanisms of
reality.
267
For example, consider the existence of asystematic bias in the post-mortem
stories. While a relativist might include this as an interpretation of the world
and accept it at face value, it is assumed in this study that a systematic bias
would destroy the applicability of the results. If all interpretations are false,
then correcting for the problems will not work because the problems were de-
fined correctly in the first instance; that is, the proposed mechanism is not
faulty. In other words, there would be some other unidentified mechanism(s)
that account for the actualization of real strategic dilemmas.
Therefore, potential biases need to be considered; not only those arising
from the data but also the researcher, given that the analysis of this study is in
fact an interpretation of interpretations of reality. Wagner et al. (2010) observe
that some researchers have employed Glaser and Strauss’ (1967) unwilling-
ness to address biases as an excuse for their own ignorance, but contend that
excluding discussion on biases is not the correct way to approach credibility. It
is important to acknowledge that interpretation is inherent in GT, although this
does not reduce the credibility of its results. Interpretations influence and
shape reality, and therefore understanding them might lead to results that can
be applied to action or employed to predict actions of others. As Glaser and
Strauss (1965, 9) state: “Not infrequently people successfully stake their
money, reputations and even lives as well as the fate of others upon their in-
terpretations.” There is potentially an unlimited number of biases in any type
of research with human respondents (Tourangeau 1984). The author has tried
to identify the major ones relating to this study, which will be discussed next
in terms of risks relating to data, method, and the researcher.
6.5.5 Risks relating to data
6.5.5.1 Ulterior motives
The “truthfulness” of data is linked in many ways to the motives for ex-
pounding them. Table 34 illustrates founders’ reasons for sharing their stories.
268
Table 34 Reasons for writing post-mortems
Explicit reasons for story Example
Reflection of what happened /
Avoidance of repeating mis-
takes
"The purpose of this postmortem is to thoroughly reflect on what
went wrong, so I, and perhaps others, will not make the same mis-
takes again." (Nowak 2010).
Inspiring other founders, practi-
cal usefulness for other startups
"A year from now this story will either be a testament to our meth-
odology or an embarrassing reminder of all the mistakes we made.
Either way, the hope is that it avoids the polish of hindsight and
will be not only inspirational, but methodically practical to some-
one considering quitting their job." (Lance & Snider 2006).
Therapeutic purpose, addresses
emotions relating to failure
"In the last five years, writing about my failures has been the best
possible therapy […] I could have managed for myself." (Feld
2006).
Introspection, making sense of
the failure experience
"This post-mortem will serve to get things off my chest, organize
my thoughts, get the most out of the experience, and share my ex-
perience with others." (Diaz 2010).
Responding to questions and
third-party interpretations
"I’m […] writing this to be able to point to a single, detailed,
lengthy answer to the inevitable questions I’ll be getting from
friends and colleagues about what happened with [my startup].
Now people can read to their heart’s content." (Diaz 2010).
Stories are generally told for specific audiences. In the case of startup post-
mortems, founders tend to address the stories to other founders (see Table 34).
In this sense, we can refer to knowledge transfer; founders are interested in
helping others to avoid repeating their mistakes. Indeed, one of the most fre-
quent explicit motives to write a post-mortem is to help other founders avoid
common mistakes. The second motive is psychological, and might serve a
therapeutic purpose: by telling their stories, founders are able to reflect on
failure, a stressful experience with which to come to terms. Reflecting on past
failures was also perceived as learning for future startup projects; some found-
ers might encourage potential founders to create a startup.
In addition to reported reasons, there might be implicit reasons for story-
telling. This study can only speculate on the founders’ true motives, which
remain hidden. For example, a founder might engage in strategic behavior
through storytelling, which can hinder the story’s trustworthiness. Social ef-
fects can also take place, as noted by a practitioner: “it’s in nobody’s best in-
terest to call attention to their own bad decisions, and it’s even less wise to
poke fun at the bad decisions of your co-workers, who may be a vital part of
the personal network that will keep you alive after the startup explodes.”
Therefore, founders might “soften” their own part in failures by omitting some
information. Moreover, the story can be written in such a way that is intended
to maintain a professional profile, for example, to impress investors or future
employers (Bansal & Clelland 2004).
269
These points are not necessarily detrimental to the credibility of a story;
secondary motives can underlie narratives, and are only problematic if the
story in question is distorted. For example, the ulterior motive of ‘appearing
experienced’ does not make the story less credible if the facts and interpreta-
tions within it are otherwise objective. In other words, the person making a
claim is to be separated from the claim (cf. argumentum ad hominem).
6.5.5.2 Self-serving bias
Several authors have recognized the risk of self-serving bias in entrepreneurial
studies. The self-serving motives risk distorting the trustworthiness of a story
by omitting personal mistakes and assigning the blame to external as opposed
to internal reasons. For example, Lussier (1996) found in his study that only
five percent of the entrepreneurs surveyed identified poor management as a
failure factor; thereby implying a tendency to blame external factors. Beaver
and J ennings (2005) advise against employing surveys to find truthful ac-
counts for failure as people are more likely to give self-serving responses and
less likely to admit personal fault. Following the attribution theory, Zacharakis
(1999) argues that individuals are more likely to attribute their own failures to
external causes (e.g., recession), and failures of others to internal causes (e.g.,
poor management skills).
In contrast to Lussier's (1996) findings, Mantere, Aula, Schildt, and Vaara
(2013) found that entrepreneurs in their case companies were ready to accept
blame, and attributed less of it to their subordinates than that attributed by
subordinates to the entrepreneurs, indicating a low self-serving bias. However,
they stress that cognitive and emotional processing relating to the failure expe-
rience influence failure narratives; namely, grief recovery and self-justification
(Mantere et al. 2013). There is little that can be done to control such effects.
Nevertheless, a proxy measure of self-attribution is employed in this study.
The measure is constructed so that if a founder in his/her post-mortem explic-
itly attributes failure, at least in part, to his own actions, this is considered self-
attribution.
Exhibits based on this analysis are presented in the following table.
270
Table 35 Examples of self-attribution
Exhibit
[1] "I’d also like to thank our venture-capital investors […] who took a big risk on an unproven
concept and then took a large financial loss when we were unable to successfully execute on
that concept." (Potts 2007).
[2]
"I have a tremendous amount of respect towards everyone that I’ve worked with on this en-
deavor and do not wish to even hint at a “should've”/“would’ve” discussion. What’s done is
done. There is no way to go back. And ultimately it failed under my watch, and that is mine to
bear." (Yaghmour 2010).
[3]
"As co-founder and CEO of [the startup], the buck stops with me and no one else." (Rafer
2009).
[4]
"For one, we stuck with the wrong strategy for too long. I think this was partly because it was
hard to admit the idea wasn’t as good as I originally thought or that we couldn’t make it work.
If we had been honest with ourselves earlier on we may have been able to pivot sooner and
have enough capital left to properly execute the new strategy. I believe the biggest mistake I
made as CEO of [the startup] was failing to pivot sooner." (Nowak 2010).
[5]
"A final point that should be made is that this is not an attempt to blame anyone. The
journalists aren’t to blame: we didn’t make a sufficiently good product for them. The developer
isn’t to blame; we tried to hire someone for a startup role who had no interest in startups. No,
the only people to blame is us, and more specifically me, since I was at the helm when it all
went down." (Biggar 2010).
[6]
"When [my co-founder] had to leave the company due to a family illness, I took over as CEO
and led the company without a formal peer for the final two years. All that adds up to me hav-
ing absolutely no one to blame for [the startup’s] failure but myself, and as a result I can’t now
nor could ever be dispassionate in thinking about what happened." (Hedlund 2010).
Self-attribution refers to a founder explicitly attributing failure to self; that
is, taking responsibility rather than blaming external conditions. To be coded,
the founder had to explicitly indicate his/her or the team’s shortcoming in ex-
plaining the failure. There is a variation in interpretation as some founders
sought to attribute reasons to external factors more than others. Self-attribution
was coded in the material, and 41% of cases included references to self-attrib-
ution. Note that this does not indicate that the founder was not attributing
blame to self, only whether or not it was explicitly stated in the post-mortem.
Therefore, it is concluded that, in general, the risk of self-justification con-
siderably altering the stories is negligible. It is important to note that post-
mortems have been made public on the Internet; thus, an intentional “twisting
of facts” would risk the founders losing face and credibility (Krumpal 2011).
Founders are likely to be aware of this and increase their level of candor. The
final interpretation is thus in line with Mantere et al. (2003). As such, founders
in our sample can generally be regarded as candid in their accounts.
271
6.5.5.3 Recall bias
Apart from apparent candidness, there are other types of bias that relate to rec-
ollection and interpretation. Elliot (2005) describes recall bias as forgetting
past events or details in them, leading to deterioration of the data quality.
261
In this study, recall bias, or forgetting important points, might be less rele-
vant because the bias concerns details, not a gestalt (i.e., an organized whole
that is perceived as more than the sum of its parts) of problems experienced
(see Maitlis 2005). It is not likely that the gestalt (i.e., the whole story) would
have been falsely remembered. It is more a problem of interpretation than
memory if a specific dilemma was incorrect.
In the case of this study, all narratives were written within a year of failure.
Although this does not remove the risk of recall bias, the longer the delay in
reporting the event, the greater the likelihood of confusion (Coughlin 1990). A
year might be regarded as too short a period to forget critical details, although
sufficient for distancing one from the immediacy of failure, or grief recovery
(Shepherd & Kuratko 2009). This might, in fact, improve the quality of post-
mortems as the founders have had time to reflect, and are perhaps better able
to place their story into a “bigger picture”.
6.5.5.4 Unintentional false attribution
Moreover, false attribution can be a consequence of sense-making even with-
out being self-serving or memory-distorted (Mantere et al. 2013). This is when
a founder is unable to find the correct reasons, interprets facts incorrectly, or
there is simply no prominent explanation other than the failure was a combi-
nation of many events, some unforeseen and random. In other words, it is a
type of sense-making bias.
The fact that reports were voluntary (i.e., unpaid, not commanded) supports
this perspective as, clearly, founders wanted to share some experiences they
genuinely believed were of interest to others. If interpretations are internally
consistent, there is no reason why “amateur theories” should be perceived as
meaningless before closer examination. In fact, economics has also suffered
from time to time with regard to questioning inaccurate rationality assump-
tions (Kanazawa 1998; Nagel 1963).
However, the sense-making styledoes differ among founders. For instance,
the founder of a location-based startup was highly capable in conceptualizing
specific problems, which indicates good analytical skills, whereas other
261
In psychology, this is defined assystematic error in reproducing past events (Coughlin 1990).
272
founders were less capable of doing so or this capability was not visible in
their stories. Inherently, this results in some incommensurability of data. This
is why superficial stories were omitted from the sample; their analysis was not
sufficiently rich to provide grounds for proper theorization. In our case, the
hindsight perspective is useful as it enables reflection by founders. This prop-
erty was particularly useful in the post-hoc analysis, when the data were re-
analyzed for “what if” thoughts and proposals for solutions. After experienc-
ing failure, founders tend to be more aware of the errors committed during the
startup phase, whereas insights might be more anecdotal if the founders were
interviewed during the experience (Schoenberger 1991). When founders share
their a posteriori insights, they become a rich and valuable research material.
6.5.5.5 Communality
Another risk might be communality between the stories, so that they are not
independent in interpreting the chain of events. More precisely, if the stories
were written after first reading other stories, these other stories might have
influenced how the founders explained their own failures, potentially leading
to a systematic bias in interpretation. It is very difficult to account for such a
bias, as this would necessitate determining whether or not founders read each
other’s stories. There are indications of this behavior occurring; for example,
one founder wrote (Diaz 2010):
"There are many post-mortems from failed startups out there,
mainly because there are a lot of failed startups, and the people
that start them tend to be very introspective and public about
their successes and failures. I’m no different."
For example, data can be biased by the influence of other post-mortem sto-
ries or thought-leaders, so that founders identify challenges selectively, based
on what others have previously noted. As a consequence, they might miss
some points that they otherwise would have noticed. If peer influence is a
source of sharing motivation, the risk is that stories become contaminated by
other stories, so that interpretation follows earlier interpretation. Undoubtedly,
some communality or patterns in the data relate to this bias. For example,
founders employ vocabulary such as “iteration”, “pivots”, and “minimum via-
ble product” to refer to their failures. Some of these terms originate from fa-
mous startup thinkers who have their own approaches to failure. Thus, the
thinking of experts might have influenced the founders’ interpretations. How-
ever, if founders perceive these expert explanations as accurate, they should be
273
treated with a degree of plausibility insofar as many of the experts are also
entrepreneurs
262
. Moreover, the vocabulary utilized by Internet standards has
evolved as a combination of, for example, many startups, experts, and inves-
tors. The collective “slang” therefore cannot be regarded as a bias, but a way
of communicating within a community (Mazrui 1995).
However, communality is only a problem if an interpretation would other-
wise have been different because, due to a desire to adhere to other stories, the
founder has changed his/her interpretation. A positive sign of the lack of this
bias is included in Appendix 3, which represents support for two opposing
perspectives on whether or not to launch early. If the founders were subject to
conformity bias, conflicting perspectives would be less likely to arise. Based
on a strong dichotomy regarding early launch, and dissimilarities of details
presented in the dilemma exhibits, a systematic conformity bias (Moscovici &
Faucheux 1972) is not perceived to be a major issue.
6.5.5.6 Anchoring bias
Anchoring bias can take place if founders are overly focused on one aspect of
the story. When anchoring, individuals focus on one value, or piece of infor-
mation, over other values or information (Bunn 1975). According to Elliott
(2005), the narrative story format is prone to simplifications; that is, empha-
sizing particular information over other information. For startups this might
include overemphasizing one aspect of failure, while neglecting others. When
these other aspects are judged critical by some objective measure, the resulting
account would be distorted.
In other words, founders might highlight one specific point as the over-
arching explanation for the failure, while ignoring other aspects. For example,
they might not have a perfect understanding on the platform model, and there-
fore might not properly interpret the dynamics leading to failure. However, at
the same time, their interpretation might be more valuable as they are not
“contaminated” by platform theory, and utilize their own terminology and
experience to explicate critical aspects. Furthermore, founders are very
knowledgeable on their cases; thus, even if they highlight particular aspects,
they are more likely to have a more comprehensive picture than outsiders.
In assessing the seriousness of this bias, one has to acknowledge that em-
ploying GT, or conducting any research, requires simplification of the
surrounding world’s complexities. In this case, the stories generally included
262
For example, Paul Graham, Eric Ries, and Steven Blank are thought-leaders whose terminology
is widely applied by startup founders in their post-mortems.
274
not only one overarching explanation for failure but several of them, thereby
reducing the impression of a strong anchoring bias.
6.5.5.7 Different interpretations
Founders, customers, investors, and other stakeholder groups can have differ-
ent opinions on what actually happened. However, this study is only based on
founders’ interpretations. Table 36 presents some examples of interpretations,
collected from comments on post-mortem stories, which differ from those
found in the post-mortems.
Table 36 Examples of different interpretations
Source Interpretation
Comments in
Bragiel (2008)
"I checked [the startup] twice […], when they launched and again in a year.
There were zero improvements in that year! Still their blog was filled with crap
like 'look at our brand new and cool office'. I posted a comment asking what's up
and reporting some bugs. The comment was quickly deleted and nobody ever got
back to me. That day I knew they were dead."
"I was probably what you would call an early Chicago adopter of [the startup],
and I thought it was a great idea, and still do. With each succeeding version of
the software, however, the interface seemed to get weirder and weirder until I
couldn’t figure it out anymore."
Comments in
Goldenson (2009)
"As it pertains to critical mass, we (if you remember, The Legion Team) always
found it very frustrating that there were only a certain number of slots available
for each team. In my opinion, it was this limitation that created the bottleneck. "
"Of course, scalability becomes an issue; but it's a critical one to place at the top
of the priority list. Each week the 'ringleader' would have to go back and decide
who was getting invites that day, and it ultimately limited participation. Why?
Because we were all already part of the same community. We were all already on
the same team. We wanted to play with each other, not against each other. It was
us against the world, albeit if only 20 at a time. "
"But with the inability for all of us to come back night after night, interest waned,
and the lure of the Internet equivalent of A.D.D. [attention deficit disorder] took
hold. ('Oh, look! A chicken!')"
"Your words here are nothing short of courageous, and I have the utmost respect
for them. This was just meant to shed some insight from a user perspective on
why it was hard for us to grow as a […] community."
"The way this guy characterizes this is grossly misleading. [The startup] was a
great idea, but extremely poorly executed. They spent $600/$900k of investors’
money. [...] The quality of their video was easily eclipsed by an $800 trip to best
buy as far as technology goes. They'd often have laziness issues on cam, where
the host would just plain forget to have put new/charge the batteries in their mic,
causing long delays. Why the talent was left to do anything requiring responsi-
bility past showing up is befuddling."
275
It seems reasonable to assume that the founders were unable to fully under-
stand the customer experience (i.e., that of outsiders), or perspectives of other
third parties; as noted by Elliot (2005), narratives concern giving meaning to
personal experiences, not those of others. However, the data might be biased
because they only contains single-person interpretations, excluding the per-
spectives of co-founders, customers, investors, and other stakeholders. This
bias could be reduced by interviewing other parties, such as co-founders, cus-
tomers, and even competitors, who “view the focal phenomena from diverse
perspectives” (Eisenhardt & Graebner 2007, 28). Stories were not tested in
this “360 degrees” perspective, because the truthfulness of particular cases
was not perceived as a problem due to storytelling motives expressed by the
founders (see Table 34). Thus, this study regards founders as a reliable source
for identifying strategic challenges, despite the coexistence of alternative inter-
pretations of failure reasons.
6.5.5.8 Reverse survivorship bias
Finally, there might be a risk of a reverse survivorship bias because all cases
were failures. If we consider the existence of survivorship bias (Brown et al.
1992) as only relating to successful cases, “failure bias” is the reverse by only
relating to stories of failure. Technically, the end result might be the same, a
skewed realization of the world. Therefore, it is important to keep in mind that
the context of the observed dilemmas was failure, and failed startups. Poten-
tially, successful startups do not observe the same challenges at all, although
the author finds this difficult to believe. Alternatively, successful startups
might have overcome these dilemmas and encountered even more problems at
later stages, which remain undiscovered in this study. However, given the fact
that there were many analyzed cases, and especially that the focus is on early-
stage online platforms, implicitly assuming that overcoming these dilemmas
leads to later stages where the strategic problems are different, and purpose-
fully not considered in this study, the reverse survivorship bias cannot be re-
garded as a major risk for accomplishing the aim of the study.
Moreover, based on discussions with successful founders and also those
with undecided outcomes, it was discovered that founders beyond the sample
recognized the same issues identified in this study. This leads the author to
believe that the same dilemmas exist more widely in the context of online
startups.
276
6.5.5.9 Final assessment of data
The post-mortems included in this dissertation were written by the founders
(i.e., entrepreneurs) of failed companies, offering a useful insight on self-re-
flected reasons for failure. In other words, the primary analyst is the founder.
The researcher is a secondary analyst who synthesizes stories, aims to find
similarities and differences in them, and increases the level of abstraction
(Goulding 2005).
Many founders seemed talented at conceptualizing their problems; for ex-
ample, the concept of ‘cold start’ is taken from a founder’s terminology. How-
ever, as post-mortems are subjective interpretations of reality, their reliability
needs to be judged case by case and, when necessary, their trustworthiness
questioned.
If the stories are not truth but interpretation, how does this statement affect
their value? First, it is clear that, in human sciences, interpretations of reality
vary according to the observer or storyteller. However, this premise excludes
neither their usefulness nor their credibility as a potential explanation of
events. In fact, theories are similar in that they arepotential explanations, not
observed facts. Therefore, employing interpretations, given that they are
treated critically in the analysis, is a natural method for theory formulation
(Glaser & Strauss 1967).
Therefore, it seems that the correct way to approach post-mortems is to
consider them interpretations, not objective truth. According to Charmaz
(1990), GT is highly compatible with interpretation. For example, consider a
hypothetical startup failure and the research task of determining why it failed.
An accountant might argue that it failed because of a lack of revenue (i.e., a
correct statement); a marketer because they were unable to convert sufficient
customers (i.e., a second correct statement); and the CEO because it was a
recession when they started, and potential customers lacked the money to buy
(i.e., a third correct statement). Essentially, GT is compatible with various
epistemological stances and does not argue that interpretation means no relia-
ble information can be acquired (Charmaz 1990). According to Glaser (2002),
it is more important in GT to argue for usefulness than for accuracy.
Verifying the trustworthiness of the stories is hindered by the researcher’s
non-participation, which would have enabled a deeper understanding on a par-
ticular story (Poon, Swatman, & Swatman 1999). Lacking a better measure,
stories were included that appeared candid, fulfilled the formal criteria, and
contained sufficient detail to be analyzed. However, as noted, they might be
subject to several biases, which have been discussed in the previous
277
subchapters. Essentially, the post-mortems are founders’ interpretations, not
objective or factual statements of the “truth”
263
in a positivistic sense. That
being stated, it is not plausible, considering the often altruistic motives for
writing them (see Table 34, p. 268), that founders would deliberately mislead
readers. Rather, it is plausible that informants were not inventing strategic
business problems but reporting what they actually perceived.
Moreover, some founders seemed aware of biases and expressed a desire to
prevent them affecting the truthfulness of their story. For example, one
founder wrote (Ehrenberg 2008):
"These observations aren’t worth much. But the interpersonal
dynamics, the issues of organizational structure, the need to
change strategy in light of new information, the relationship with
key investors, all of these are very instructive. I will endeavor to
be as honest and candid as possible."
Statements of this type, along with the self-attribution analysis and explicit
motives for story writing, and also the general requirements of GT for data,
lead the research to believe that the material is of adequate quality for achiev-
ing the research purpose. Moreover, it can be concluded that they are suffi-
ciently precise to articulate the key issues of platform startups on the Internet.
6.5.6 Risks relating to method
Grounded theory has been employed in previous studies focusing on platforms
(e.g., Curchod & Neysen 2009; Palka, Pousttchi, and Wiedemann 2009;
Mantere et al. 2013). Curchod and Neysen (2009) employed GT to identify
perceived positive and negative network effects by users of eBay, an exchange
platform. Palka et al. (2009) applied it to identify users’ motives to share viral
messages in a mobile platform. Mantere et al. (2013) analyzed interviews to
build narrative attributions of entrepreneurial failure. Generally, GT is re-
garded as providing a robust method for analyzing qualitative data (Corbin &
Strauss 1990); in particular, employing a systematic approach to build theo-
retical constructs, while remaining rooted in informants’ experiences. Most of
the platform literature remains analytical as it is derived from economics
(Birke 2008; Shy 2011). Qualitative inquiries are therefore necessary to vali-
date and complement analytical models as they provide a high level of depth
(Salomon 1991).
263
Given that failure is a relative concept (Watson & Everett 1999), and its interpretation varies
among the venture’s participants, it is not clear that one truth exists. In contrast, it is more likely that
each party has their own interpretation, and the failure outcome is a combination of events and
conditions.
278
In general, purposive, or judgmental, sampling is regarded as appropriate
when the research seeks to create new concepts or theory, as opposed to test-
ing hypotheses with statistical inference (Eisenhardt & Graebner 2007).
Considering the purpose of this study, selective sampling is not perceived as
an issue. Moreover, secondary data pose no problem for GT. Goulding (2005)
mentions secondary data as one format in her methodological overview on
GT. According to Glaser (2004), “all is data” and the researcher should decide
on the most appropriate data for the study. Post-mortem stories were not only
publicly available, they also enabled a substantial and varietal sample of Web
2.0 startups to be quickly amassed. Variety was perceived to be useful for the
purpose of constant comparison, whereas having multiple informants on one
case would have enabled a thicker and perhaps more accurate description.
However, according to Glaser (2004), accuracy of description is not a major
concern in GT.
The methodology’s influence is a particular risk, meaning that results origi-
nate from the use of a particular method. As GT aims to find relationships
between constructs (Strauss & Corbin 1994), finding a relationship between
individual strategic dilemmas (i.e., categories) is a natural result of applying
the method. However, credibility is not threatened for two reasons. First, there
are connections between the problems in the real world and they are revealed
as grounded in the data; thus, they are not concocted. Second, the connections
were filtered by selective coding, with some being discarded. Applying judg-
ment emphasizes the most important connections; thus the method does not
allow a pre-determined model to emerge.
Charmaz (1990) discusses the merits and challenges of grounded theory in
more detail. In general, Eisenhardt (1989) notes that the resulting theory might
be narrow and limited to a specific context, as opposed to “a grand theory”. As
previously established, grand theory was not the purpose of this study, but ra-
ther a substantive one. In general, the chosen method was seen to provide rigor
in the analytical process, which was perceived to be vague and complicated
before learning the key principles of GT. Therefore, the researcher feels the
method was appropriate and useful. For a description of how grounded theory
was applied in this study, refer to Chapter 2.4.
6.5.7 Risks relating to researcher
In general, the nature of qualitative analysis puts strong emphasis on the re-
searcher’s judgment (Seale 1999). This is also a limitation of this study,
because the data include self-assessment a priori, and are interpreted by the
researcher a posteriori; in other words, interpreting interpretations.
279
In particular, researcher bias becomes an issue in GT through the interpre-
tation mechanism by which the researcher elicits meanings from the data
(Partington 2000). For example, during data collection, the researcher can
influence informants’ responses. This is a particular concern with the inter-
view method. Elliott (2005, 11) elaborates on this concern:
"[W]ithin certain contexts the narrator may be influenced by im-
agined or possible future audiences […]. The very fact that the
conversation is being recorded suggests that it will at least be
listened to at some future time and may also be transcribed and
parts of it translated into a written text."
In this study, stories were collected from the Internet and, considering they
were published in personal blogs targeting other startup founders, it is unlikely
that founders expected them to become part of academic scrutiny. This elimi-
nates the possibility that the researcher might have influenced the reports. The
researcher did not influence in any way the writing of post-mortems, and
therefore data collection was completely unobtrusive from the researcher’s
side.
Theoretical sampling can risk confirmation bias if one purposefully seeks
only positive answers (Nickerson 1998). It was borne in mind that feedback
might not only confirm but also conflict with the researcher’s assumptions.
Hence, it can be argued that contextualization is a double-edged sword: by
increasing domain-specific information, understanding is increased on the
conditions in which startups make strategic decisions. At the same time, con-
textualization can risk going native so that informants’ sense-making is
adopted as truth (Marker 1998); for example, the researcher might start to
overemphasize and/or neglect facts in accordance with founders’ biases. Also,
Charmaz (1990) mentions the risk of going native. This might occur when the
researcher is so immersed in the informants’ reality that he/she loses objectiv-
ity, becomes naïve, and accepts informants’ conclusions at face value. As the
author was not actually a member in any of the observed or analyzed startups,
this risk can be deemed low. Partial immersion, in fact, is beneficial to under-
standing the context; that is, specific circumstances and conditions which in-
fluence actions such as strategic choices. This is implied in theoretical sensi-
tivity, denoting that the researcher is not a tabula rasa but does have a per-
sonal, professional, and theoretical perspective on the studied phenomenon
(Strauss & Corbin 1994).
Regarding analysis, the researcher might be biased in his interpretation;
even if no harmful
264
bias exists, some degree of bias is inherent in interpreta-
tive qualitative analysis (J ohnson 1997). It is important to note that GT is an
264
Harmful as in deliberately excluding facts that would contest the theory under development.
280
interpretative method (Gasson 2003), and therefore particular risks can arise
regarding misinterpretation. However, interpretation is associated with theo-
retical sensitivity (Glaser 1978; Strauss & Corbin 1994). Relating to this
study, the author read a large amount of related information on online business
to increase his understanding on the topic, including, for example, startup-fo-
cused blogs, online community discussions, and books, which is likely to in-
fluence his understanding (J ones & Alony 2011), although hopefully in a way
that leads to more useful conclusions.
Finally, Glaser (2004), who places great faith on a researcher’s ability to
employ data in the correct way, states that:
"GT discovers and conceptualizes the latent patterns of what is
going on. It is always relevant. If a GT is accused as being inter-
pretive, which is probably meaningless, it is a very relevant in-
terpretation."
In other words, Glaser consistently approaches constructivism in his un-
derlying logic that descriptive accuracy is not as relevant as the theory’s abil-
ity to influence action. The way in which subjects perceive the world in turn
influences their action, and so perceptions/interpretation have the potential to
become self-fulfilling prophecies (Merton 1948; see also Charmaz 1990). This
perspective is adopted insofar as to state that whether or not the interpretation
is perfectly correct in description, its implications point the reader to the cor-
rect direction.
In GT, verification is not generally performed by others but by the re-
searcher through engaging in constant comparison of subjects (Glaser &
Strauss 1967) and utilizing his/her curiosity in theoretical sampling, or ac-
quiring new slices of data to challenge the emerging theory. Glaser and
Strauss (1965) place a great deal of trust in the researcher, and their message is
that researchers should trust their findings, achieved through hard work and
immersion in the context of the research subjects. Even if two analysts rarely
reach exactly the same conclusions when analyzing a topic independently
(Glaser & Strauss 1967), both can agree on each other’s works.
Although this might seem paradoxical, it only highlights the non-linear cre-
ative process associated with GT, leading to individuals emphasizing different
angles. For example, this study could have focused on founders’ decision-
making biases, or on the ideal UG model. According to the author, both would
have explained the problematic features of the failed startups. It was simply
the author’s judgment call to focus on strategic dilemmas, which can be re-
garded as no more right or wrong than alternative calls. What can be further
stated on this, however, is that the author believes the few identified dilemmas
can be employed better to explain platform startups’ failures than utilizing a
281
large number of generic failure factors that seem to be randomly scattered in
startups; that is, in the context of platform startups, their relevance is higher.
6.5.8 Generalizability
A question typically posed for any type of research is “how generalizable is
it?” The issue is tackled in this subchapter.
The highest level of abstraction in GT is termed aformal theory, which re-
lates to a phenomenon in general, whereas a substantive theory relates to a
phenomenon in a given context (Glaser 1971). For example, a theory on busi-
ness failure would be considered a formal theory (i.e., general focus), whereas
a theory on startup failure would be a substantive failure (i.e., contextual fo-
cus). To these, Glaser (1971) adds “grand theories” that can cover almost all
types of situation and exist across phenomena; for example, explaining failure
in any type of context could constitute such a theory.
According to Glaser and Strauss (1965), a theory should be judged on its
intended generalizability. In some cases, substantive theory can be regarded as
being equivalent to interim theory (Glaser 1971), or a step in the process to
formal theory. However, it can also be perceived as an independent entity, and
therefore its merits should be judged by considering the context for which it
was devised. It is noteworthy that, in this case, the researcher relies onanalyti-
cal or logical generalization when applying the theoretical conclusions to
other units in the same context (Collingridge & Gantt 2008). This process
should not be confused with statistical inference, that is, generalization from
sample to population, as it is a statistical technique for determining applicabil-
ity (Wagner et al. 2010).
Corbin and Strauss (1990) relate generalizability to a study’s reproducibil-
ity, which resembles the transference criteria discussed earlier. They (ibid.,
250) note that “probably no theory that deals with a […] phenomenon is actu-
ally reproducible insofar as finding new situations or other situations whose
conditions exactly match those of the original study, though many conditions
may be similar.” Consequently, Strauss and Corbin (1994) postulate that there
are spheres of applicability, ranging from an individual case to a local com-
munity to an international setting, and so on, towards a global pattern. In a
similar vein, Urquehart et al. (2010, 364) note that “as the researcher moves
up the level of abstraction, the range and scope of the theory increases.”
By applying these ideas, it can be contemplated that there are spheres of
applicability also for this study, which are presented in the following figure.
282
Figure 18 Spheres of applicability
Therefore, the findings of this study are generalizable, at a minimum, to the
four online platform types that tend to apply UG, at least implicitly, and indi-
rect monetization, including the freemium model in an attempt to internalize
externalities from interaction among platform participants.
Increasing the level of abstraction is generally perceived to increase a the-
ory’s applicability (Strauss & Corbin 1994). Moving away from an online
platform also means that more general theory literature becomes available,
although at the cost of losing empirical context (Glaser 1971). Generally, the
process of generalization aims to lose contextual factors and introduce a gen-
eral logic (i.e., formal theory) concerning why particular relationships hold
across many contexts. Increasing the abstraction therefore removes online-
specificities; for example, different motives of interaction manifested in plat-
form types, or commonly applied UG and indirect monetization.
This way of examining the generalizability implies that strategic problems
do not always exist; that is, if their existence depends on specific conditions,
when some contextual conditions are removed from the picture, the dilemmas
can no longer be identified. In particular, the problems addressed here require
online-specificities, or they might not emerge. For example, if the platform
charges an access fee and customers are willing to pay, there is no monetiza-
tion dilemma. Consistent with the perspective of critical realism (Kempster &
Parry 2011), the dilemmas are contingent upon the context.
This issue can be illustrated by attempting to increase abstraction on the
study’s main model, Figure 15 (p. 195), which shows the relationships be-
tween dilemmas. In Figure 19, the author followed Glaser’s (2008) advice to
abstract from time, place, and people, and only to apply general concepts. In
online
platforms
other firms
other platforms
other contexts
283
this study, a probable or most appropriate way is arguably towards business
failure in general, a formative theory on failure of sorts.
Figure 19 A tentative formal theory
The theory explains failure through strategic problems and biases. It argues
that there are strategic problems requiring solutions for the company to avoid
failure. If these problems are solved, there will be a derivative problem that
requires a solution, and so on. It is argued that solving the problems is pre-
ceded by their identification. This identification is potentially prevented by
bias; because of his or her biases, the founder or manager might be unable to
identify the strategic problems. If this happens, the company will fail. How-
ever, even if the problem is identified, its solutions might be associated with a
bias; the founder or manager is unlikely to think of a solution in the proper
way due to his or her bias. Again, the company will fail. Only by solving all of
the problems and their derivative problems can the company achieve its true
potential in the market. However, it is argued that the true potential can also
equal failure; for example, when there is no true demand.
While being logical, this approach highlights the problem of moving from
substantive to formal theory: it loses the context. Thus, there is now a general
description, but with two emergent issues. First, the theory still needs to be
applied to other contexts to determine how well it fits to them (Charmaz
1990). Second, by moving from the substantive context, the practical applica-
bility to online startups has, to a major extent, now been lost. Indeed, this
seems to contest the critical realist criteria, namely, practical adequacy and
plausibility (Kempster & Parry 2011), as there are no longer grounds to esti-
mate the contexts in which the theory works without reintroducing and study-
ing those contexts. As a consequence, practitioners are less likely to
Failure
Strategic
problem
Identification
yes no
Solution
Derivative
problem
Bias
no
yes
yes
True potential no
284
understand what this means in their context. Thus, generalizability at worst
seems to lead to a double bind of losing both context and relevance. If the
study accomplishes other evaluative criteria for GT, the added usefulness of
generalizing from substantive theory can be regarded as negligible.
This is somewhat in conflict with Corbin and Strauss (1990, 267) who ar-
gue that “[t]he more systematic and widespread the theoretical sampling, the
more conditions and variations that will be discovered, therefore the greater
the generalizability, precision, and predictive capacity of the theory.” It is ar-
gued here, based on the aforementioned logic, that while theoretical sampling
increases the generalizability of a study, it reduces precision, and that this
might decrease, not increase, the predictive capacity of the theory. This con-
flict seems also to apply when examining classic GT. In fact, Glaser (2004)
mentions that the general concepts apply in any domain where they exist, but
that they require modification by constant comparison. Essentially, if they
need to be modified when applied, then they do not apply at their general
level, and are thus not generalizable prior to being employed in action.
The Glaserian and Straussian approaches differ in their understanding on
reproducibility. While Corbin and Strauss (1990, 251) require that “given the
same theoretical perspective of the original researcher and following the same
general rules for data gathering and analysis, plus a similar set of conditions,
another investigator should be able to come up with the same general
scheme.” Glaser (2002) leans more towards variation in ability to conceptual-
ize, and perceives that some are more talented in this than others and that con-
ceptualization differs from description, which is a simpler cognitive process.
In fact, such an idea is implicit in classic GT through the concept of theoretical
sensitivity. If traits such as personal and professional experience (Strauss &
Corbin 1994) influence how well the researcher is able to elicit concepts and
theory from the data, it is unlikely that two persons who vary greatly on these
dimensions would reach exactly the same conclusions.
Corbin and Strauss (1990) fail to explain this discrepancy, unless their “the-
oretical perspective” means the exact same theoretical sensitivity. Given the
interpretative nature of GT, such a match can be considered realistic as two
identical snowflakes. In contrast, the origins of GT lie in the substantive the-
ory with its contextual score (i.e., usefulness): “The invalidation or adjustment
of a theory is only legitimate for those social worlds or structures to which it
is applicable” (Glaser & Strauss 1965, 10). In this study, the author contends
that it would be unlikely for another researcher reproducing the study to draw
exactly the same conclusions. However, in whatever ways they might explain
failure, other researchers would most probably rely on the properties of the
dilemmas. In other words, when reconciling perceptions between the imagi-
nary new research and this study, there should be no fundamental
285
disagreement. Thus, the spirit of the conclusions would remain intact, and the
author assesses that reproducing this study would yield results leading to
similar practical implications and usefulness.
As can be seen in the platform literature, strategic challenges vary by plat-
form type. For example, lessons from online platforms might not be applicable
to platform markets such as the newspaper/media industry, payment cards, or
operating systems examined in other studies (Rysman 2009). For example,
taking the general assumption of network effects prevent switching from the
standards discussion (Katz & Shapiro 1994) would be an overstatement, as
network effects are by definition not decisive in an environment with multiple
coexisting platforms and multihoming, which the Internet as a business envi-
ronment clearly is and proprietary standards are not.
In terms of improving the quality of the substantive theory, Figure 10 (p.
93), which displays the totality of emerging dilemmas and biases, is a strong
candidate as it explains much of the startup process leading to failure. In par-
ticular, biases and bounded rationality in decision-making have been acknowl-
edged since Simon (1956); including cognitive psychology and behavioral
game theory would increase the explanatory power of the substantive theory,
and potentially lead to discoveries that could be formalized and applied across
contexts. Indeed, including biases would improve the explanatory power of
why founders choose particular strategies that can afterwards be considered
destructive, and thus improve the work done here.
In the spirit of substantive theory, the results are limited to Internet startups,
or early-stage technology ventures in online business. Therefore, based on this
study, the results cannot readily be generalized to other types of technology
startup such as life science or clean-tech, other types of startup, platform com-
panies, or firms in general. These entities are likely to have different dynamics
that render the conclusions of this study inapplicable. However, the generali-
zability is extended beyond failed startups to all platform startups on the Inter-
net. There is no reason to believe that successful platform startups have not
faced these issues; indeed, they have solved them. Thus, further theoretical
sampling can be targeted at the strategies of successful platform startups.
Moreover, there are startups withindecisive outcomes and, without doubt, the
greatest share of startups encountered by the author in the various startup
events belong to this group. They can benefit from this study by identifying
potential strategic problems and solutions, and by adopting strategic platform
thinking.
Finally, the author would like to point out that two-sidedness is not an ex-
clusive feature of platforms; in contrast, as a perspective, it can be applied to
examine very many situations. As two-sided dynamics relate not only to plat-
forms, Rysman (2009) refers to two-sided strategies as opposed to two-sided
286
platforms. By further developing his idea, one can speak of platform logic, or
two-sided logic, that is applicable to contexts not normally understood as plat-
forms. For example, events: the better the speakers, the more tickets can be
sold. Relatedly, the higher the speaker fees, the higher the sales; given that the
quality of the speakers is proportional to his or her fee. Another example is
education: the better the teachers are in a given school, the better the caliber of
students wanting to apply. Moreover, two-sided logic applies to retailing,
which was considered distinctly separate from the platform model in Chapter
3.1. Clearly, the selection and variety of merchants affects how likely the end
customer is to buy, and thus two-sided dynamics are present. Internal market-
ing is another example of two-sided logic. The happier the employees, the
better they serve the customers, and thus the happier customers become; vice
versa, the more unhappy or moodier the customer, the worse the motivation of
the employees serving him/her, and so on
265
. The implication for firms is to
understand what each interactive side appreciates, and assess the degree of
critical mass needed to evoke the desired response (i.e., conversion).
6.5.9 Overall assessment of credibility
Regarding credibility, the nature of data means that they cannot be treated as
facts. Risks to credibility comprise informant biases (e.g., recall bias and re-
spondent bias) that can involve consciously or unconsciously ignoring relevant
aspects, researcher bias closely associated with interpretation, and the specific
methodological choices.
In general, a limited number of firms and respondents might result in a one-
sided perspective and prevent generalization in a statistical sense; although, if
judged as a qualitative case study, the number of cases here can considered in
line with such studies. Moreover, there are also benefits associated with the
approach taken in the study and, considering the research purpose, the issue of
“myopia” cannot be regarded as a major issue. First, the founders are highly
knowledgeable on the challenges they faced. Second, stories are independent
accounts, written in different places at different times and interpreting differ-
ent startups, which lends support to identified patterns in the real world. Third,
the researcher has confidence in knowing the startup reality sufficiently well
to make credible claims on it. Finally, the analysis suggests that the founders
were aware of their personal biases, and might even have made them explicit.
Overall, this leads the author to believe that the material and analytical
265
The mentioned chain effect relies on contagion of emotions, but the logic is two-sided as the
welfare of each group is linked to the other.
287
procedures are appropriate to draw credible conclusions. Regarding saturation,
the researcher does not expect that the conclusions would drastically change if
more post-mortems had been included in the study. This has been validated by
discussions with numerous founders beyond the initial sample.
Other strengths arise from the use of GT. By detailing specific techniques,
it enabled a novice PhD student to choose techniques that felt natural and
flexible but still guided the analytical process from start to finish. Grounded
theory is distinct from qualitative data analysis (QDA) methods due to its em-
phasis on theory, not “thick descriptions" or case studies. It also differs from
analytical modeling, which often makes strong assumptions to remain tracta-
ble; GT is not constrained by the same rigor, and can therefore expand to great
lengths while remaining rooted in the relevancy of the data. Truly under-
standing the strengths of the method motivates one to continue, and creates
confidence in one’s analytical abilities. Overall, the author feels confident in
recommending the grounded theory method to studies that aim at conceptual-
ization and theory development.
289
REFERENCES
Aarikka-Stenroos, L. (2011) Reference communication and third actors in the
initiation of business relationships. Turku School of Economics
Series A-6:2001, Turku.
Ackoff, R. (1969) Systems, organizations, and interdisciplinary research. In:
Systems thinking, ed. F. Emery, 330–347. Penguin Books, Balti-
more.
Agarwal, R., – Gort, M. (2001) First-Mover Advantage and the Speed of
Competitive Entry, 1887–1986. Journal of Law and Economics,
Vol. 44 (1), 161–177.
Aggarwal, C. C. – Yu, P. S. (2012) On the network effect in Web 2.0 applica-
tions. Electronic Commerce Research and Applications, Vol. 11
(2), 142–151.
Agulnick, T. (2010) Xmarks is shutting down. , retrieved J une 2012.
Akerlof, G. A. (1970) The Market for “Lemons”: Quality Uncertainty and the
Market Mechanism. The Quarterly Journal of Economics, Vol. 84
(3), 488–500.
Albuquerque, P. – Pavlidis, P. – Chatow, U. – Chen, K.-Y. – J amal, Z. (2012)
Evaluating promotional activities in an online two-sided market of
user-generated content. Marketing Science, Vol. 31 (3), 406–432.
Almeida, P. – Dokko, G. – Rosenkopf, L. (2003) Startup size and the mecha-
nisms of external learning: increasing opportunity and decreasing
ability? Research Policy, Vol. 32 (2), 301–315.
Anderson, C. (2009) Free: how today’s smartest businesses profit by giving
something for nothing. Hyperion, New York, NY.
Anderson, S. P. – Gabszewicz, J . J . (2006) The media and advertising: a tale
of two-sided markets. In: Handbook of the Economics of Art and
Culture, Volume 1, eds. K. Arrow – M. Intriligator, 567–614.
Elsevier B. V., Amsterdam.
Anding, M. – Hess, T. (2002) Online Content Syndication-A Critical Analysis
From The Perspective Of Transaction Cost Theory. In: Proceed-
ings of the 10
th
European Conference on Information Systems, ed.
S. Wrycza, 551–563.
290
Argyres, N. S. (1999) The impact of information technology on coordination:
Evidence from the B-2 “Stealth” bomber. Organization Science,
Vol. 10 (2), 162–180.
Arkes, H. R. – Blumer, C. (1985) The psychology of sunk cost. Organiza-
tional Behavior and Human Decision Processes, Vol. 35 (1), 124–
140.
Armstrong, M. (2006) Competition in Two-Sided Markets. RAND Journal of
Economics, Vol. 37 (3), 668–691.
Armstrong, M. – Wright, J . (2007) Two-sided markets, competitive bottle-
necks and exclusive contracts. Economic Theory, Vol. 32 (2),
353–380.
Arndt, J . (1978) How broad should the marketing concept be? Journal of Mar-
keting, Vol. 42 (1), 101–103.
Arroyo-Barrigüete, J . – Ernst, R. – López-Sánchez, J . – Orero-Giménez, A.
(2010) On the identification of critical mass in Internet-based ser-
vices subject to network effects. The Service Industries Journal,
Vol. 30 (5), 643–654.
Audretsch, D. – Acs, Z. (1994) New-firm startups, technology, and macro-
economic fluctuations. Small Business Economics, Vol. 6 (6),
439–449.
Axelrod, R. M. (2006) The evolution of cooperation. Basic Books, New York,
NY.
Azoulay, P. – Shane, S. (2001) Entrepreneurs, Contracts, and the Failure of
Young Firms. Management Science, Vol. 47 (3), 337–358.
Bakos, Y. (1998) The emerging role of electronic marketplaces on the Inter-
net. Communications of the ACM, Vol. 41 (8), 35–42.
Bakos, Y. – Katsamakas, E. (2008) Design and ownership of two-sided net-
works: Implications for Internet platforms. Journal of Manage-
ment Information Systems, Vol. 25 (2), 171–202.
Ball, L. – Mankiw, N. – Romer, D. – Akerlof, G. – Rose, A. – Yellen, J . –
Sims, C. (1988) The new Keynesian economics and the output-in-
flation trade-off. Brookings Papers on Economic Activity, Vol.
1988 (1), 1–82.
Banerjee, A. V. (1992) A simple model of herd behavior. The Quarterly Jour-
nal of Economics, Vol. 107 (3), 797–817.
Banerji, A. – Dutta, B. (2009) Local network externalities and market seg-
mentation. International Journal of Industrial Organization,
27(5), 605–614.
Banks, J . – Deuze, M. (2009) Co-creative labour. International Journal of
Cultural Studies, Vol. 12 (5), 419–431.
291
Bansal, P. – Clelland, I. (2004) Talking trash: Legitimacy, impression man-
agement, and unsystematic risk in the context of the natural envi-
ronment. Academy of Management Journal, Vol. 47 (1), 93–103.
Baum, J . A. C. – Mezias, S. J . (1992) Localized Competition and Organiza-
tional Failure in the Manhattan Hotel Industry, 1898-1990. Ad-
ministrative Science Quarterly, Vol. 37 (4), 580–604.
Bauman, Z. (2010) Hermeneutics and Social Science (Routledge Revivals):
Approaches to Understanding. Taylor & Francis, New York, NY.
Beaver, G. – J ennings, P. (2005) Competitive advantage and entrepreneurial
power: The dark side of entrepreneurship. Journal of Small Busi-
ness and Enterprise Development, Vol. 12 (1), 9–23.
Beck, R. – Beimborn, D. – Weitzel, T. – König, W. (2008) Network effects as
drivers of individual technology adoption: Analyzing adoption
and diffusion of mobile communication services. Information
Systems Frontiers, Vol. 10 (4), 415–429.
Bekkelund, K. (2011) Succeeding with freemium: Exploring why companies
have succeeded and failed with freemium. Innovation and Entre-
preneurship, Specialization Project 4350, Norwegian University
of Science and Technology, Trondheim. , retrieved J une 2013.
Belleflamme, P. – Toulemonde, E. (2004) B2B marketplaces: when it pays to
be small. Presented at the5th ZEW Conference: The Economics of
Information and Communication Technologies, Mannheim, Ger-
many.
Benkler, Y. (2002) Coase’s Penguin, or, Linux and “The Nature of the Firm.”
Yale Law Journal, Vol. 112 (3), 369–446.
Benway, J . P. – Lane, D. M. (1999) Banner blindness: What searching users
notice and do not notice on the World Wide Web. Rice University
Electronic Theses and Dissertations, Rice University, Houston,
TX.
Bergen, M. – Dutta, S. – Walker, O. C., J r. (1992). Agency relationships in
marketing: A review of the implications and applications of
agency and related theories. Journal of Marketing, Vol. 56 (J uly),
1–24.
Berman, R. – Katona, Z. (2012) The role of search engine optimization in
search marketing. Marketing Science, Vol. 32 (4), 644–651.
Berners-Lee, T. (2010) Long Live the Web: A Call for Continued Open
Standards and Neutrality. Scientific American, 26.11.2010.
Besen, S. – Farrell, J . (1994) Choosing How to Compete: Strategies and Tac-
tics in Standardization. Journal of Economic Perspectives, Vol. 8
(2), 117–131.
292
Beuscart, J . – Mellet, K. (2009) Business Models of the Web 2.0: Advertising
or the Tale of Two Stories. Communications and Strategies, Spe-
cial Issue (November 2008), 165-81
Biernacki, P. – Waldorf, D. (1981) Snowball Sampling: Problems and Tech-
niques of Chain Referral Sampling. Sociological Methods &
Research, Vol. 10 (2), 141–163.
Biggar, P. (2010) Why we shut NewsTilt down. , retrieved J une 2012.
Birke, D. (2008). The Economics of Networks - A Survey of the Empirical
Literature. Journal of Economic Surveys, Vol. 23 (4), 762–793.
Blank, S. (2010) Customer Development for Web Startups. Steve Blank.
, retrieved 2.7.2013.
Boudreau, K. (2010) Open platform strategies and innovation: Granting access
vs. devolving control. Management Science, Vol. 56 (10), 1849–
1872.
Boudreau, K. – Hagiu, A. (2009) Platform rules: multi-sided platforms as reg-
ulators. In: Platforms, Markets and Innovation, ed. A. Gawer,
169–191. Edward Elgar Publishing, Cheltenham.
Boudreau, K. – Lakhani, K. R. (2012). How to manage outside innovation.
MIT Sloan Management Review, Vol. 50 (4), 69–76.
Boyle, R. D. – Desai, H. B. (1991) Turnaround strategies for small firms.
Journal of Small Business Management, 29 (3), 33–42.
Bragiel, P. (2008) Anatomy Of A Failure: Lessons Learned.
, retrieved J une 2012.
Briscoe, B. – Odlyzko, A. – Tilly, B. (2006) Metcalfe’s law is wrong. IEEE
Spectrum. , retrieved J une 2013.
Brown, S. J . – Goetzmann, W. – Ibbotson, R. G. – Ross, S. A. (1992) Survi-
vorship bias in performance studies. Review of Financial Studies,
Vol. 5 (4), 553–580.
Brodzinski, P. (2009) Lessons Learned: Startup Failure Part 1.
, retrieved J une 2012.
Bruderl, J . – Schussler, R. (1990) Organizational mortality: The liabilities of
newness and adolescence. Administrative Science Quarterly, Vol.
35 (3), 530–547.
Brunn, P. – J ensen, M. – Skovgaard, J . (2002). e-Marketplaces:: Crafting A
Winning Strategy. European Management Journal, Vol. 20 (3),
286–298.
293
Bunn, D. W. (1975) Anchoring bias in the assessment of subjective probabil-
ity. Operational Research Quarterly, Vol. 26 (2), 449–454.
Burns, M. Blesener, E. (2013) Commenters, we want you back. TechCrunch.
,
retrieved 1.7.2013.
Cable, D. – Shane, S. (1997) A prisoner’s dilemma approach to entrepreneur-
venture capitalist relationships. Academy of Management Review,
22 (1), 142–176.
Caillaud, B. – J ullien, B. (2003) Chicken & egg: Competition among interme-
diation service providers. RAND Journal of Economics, Vol. 34
(2), 309–328.
Calvano, E. – J ullien, B. (2012) Issues in online advertising and competition
policy: A two sided markets perspective. In: Recent Advances in
the Analysis of Competition Policy and Regulation, eds. J . E.
Harrington – Y. Katsoulacos, 179–197. Edward Elgar Publishing,
Cheltenham.
Camerer, C. (2003) Behavioral Game Theory: Experiments in Strategic Inter-
action. Princeton University Press, Princeton, NJ .
Campbell, C. (2012) How much is Zynga paying for new gamers? IGN.
, retrieved 5.7.2013.
Cappel, J . J . – Huang, Z. (2007). A usability analysis of company websites.
Journal of Computer Information Systems, Vol. 48 (1), 117.
Casey, T. R. – Töyli, J . (2012) Dynamics of two-sided platform success and
failure: An analysis of public wireless local area access. Techno-
vation, Vol. 32 (12), 703–716.
Cennamo, C. – Santalo, J . (2013). Platform competition: Strategic trade-offs in
platform markets. Strategic Management Journal, Vol. 34 (11),
1331–1350.
Chafkin, M. (2007) How to kill a great idea! Inc. Magazine. , retrieved J une 2013.
Chakravorti, S. – Roson, R. (2006) Platform competition in two-sided mar-
kets: The case of payment networks. Review of Network Econom-
ics, Vol. 5 (1), 118–143.
Charmaz, K. (1990) “Discovering” chronic illness: Using grounded theory.
Social Science & Medicine, Vol. 30 (11), 1161–1172.
Charmaz, K. (2006) Constructing grounded theory: A practical guide through
qualitative analysis. SAGE Publications, London.
294
Chauhan, S. S. – Proth, J .-M. (2005). Analysis of a supply chain partnership
with revenue sharing. International Journal of Production Eco-
nomics, Vol. 97 (1), 44–51.
Chen, A. (2009) How to create a profitable Freemium startup. @andrewchen.
, retrieved
1.7.2013.
Chen, J . – Esteban, S. – Shum, M. (2008) How much competition is a second-
ary market? Texas A&M University Workshop. , retrieved J une
2013.
Chesbrough, H. (2004) Managing open innovation. Research-Technology
Management, Vol. 47 (1), 23–26.
Chesler, M. (1987) Professionals’ views of the “dangers” of self-help groups.
Center for Research on Social Organization Working Paper No.
345, Ann Arbor, MI.
ChubbyBrain (2011) Analyzing 32 startup failure post-mortems to find the 20
top reasons that startups fail. Chubby Brain.
, retrieved J une 2013.
Church, J . – Gandal, N. (1992) Network effects, software provision, and
standardization. Journal of Industrial Economics, Vol. 40 (1), 85–
103.
Church, J . – Gandal, N. (2004) Platform competition in telecommunications.
Centre for Economic Policy Research No. 4659, London.
Clements, M. T. (2004) Direct and indirect network effects: are they equiva-
lent? International Journal of Industrial Organization, Vol. 22
(5), 633–645.
Clements, M. T. – Ohashi, H. (2005) Indirect network effects and the product
cycle: video games in the U.S., 1994–2002. Journal of Industrial
Economics, Vol. 53 (4), 515–542.
Clemons, E. K. (2009) Business Models for Monetizing Internet Applications
and Web Sites: Experience, Theory, and Predictions. Journal of
Management Information Systems, Vol. 26 (2), 15–41.
Coase, R. H. (1937) The nature of the firm. Economica, Vol. 4 (16), 386–405.
Coch, L. – French, J . (1948) Overcoming resistance to change. Human Rela-
tions, Vol. 1 (4), 512–532.
Cochran, J . J . – Darrat, A. F. – Elkhal, K. (2006) On the bankruptcy of internet
companies: An empirical inquiry. Journal of Business Research,
Vol. 59 (10–11), 1193–1200.
295
Cohen, A. (2003) The perfect store: inside eBay. Back Bay Books; Boston,
MA.
Cohen, W. M. – Klepper, S. (1992) The tradeoff between firm size and diver-
sity in the pursuit of technological progress. Small Business Eco-
nomics, Vol. 4 (1), 1–14.
Collingridge, D. S. – Gantt, E. E. (2008) The quality of qualitative research.
American Journal of Medical Quality, Vol. 23 (5), 389–395.
Constine, J . – Kutler K.-M. (2012) Facebook buys Instagram for $1 billion,
turns budding rival into its standalone photo app. TechCrunch.
, retrieved 30.6.2013.
Corbin, J . – Strauss, A. (1990) Grounded theory research: Procedures, canons,
and evaluative criteria. Qualitative Sociology, Vol. 13 (1), 3–21.
Corbin, J . – Strauss, A. (2008) Basics of Qualitative Research: Techniques
and Procedures for Developing Grounded Theory. SAGE Publi-
cations, Los Angeles, CA.
Coughlin, S. S. (1990) Recall bias in epidemiologic studies. Journal of Clini-
cal Epidemiology, Vol. 43 (1), 87–91.
Cova, B. – Dalli, D. (2009) Working consumers: the next step in marketing
theory? Marketing Theory, Vol. 9 (3), 315–339.
Covey, N. (2010) Changing models a global perspective on paying for content
online. Nielsen Newswire. , retrieved J une 2013.
Creswell, J . W. (2008) Research Design: Qualitative, Quantitative, and Mixed
Methods Approaches (3rd ed.). SAGE Publications, Los Angeles,
CA.
Curchod, C. – Neysen, N. (2009) Disentangling Positive and Negative Exter-
nalities on Two-Sided Markets: the EBay Case. Université
catholique de Louvain Working Paper No. 09/03, Louvain-la-
Neuve.
Cusumano, M. (2010) Technology strategy and management: The evolution of
platform thinking. Communications of the ACM, Vol. 53 (1), 32–
34.
D’aveni, R. A. (1994) Hypercompetition: Managing the dynamics of strategic
maneuvering. The Free Press, New York, NY.
Dagger, D. – O’Connor, A. – Lawless, S. – Walsh, E. – Wade, V. P. (2007)
Service-oriented e-learning platforms: From monolithic systems
to flexible services. Internet Computing, IEEE, Vol. 11 (3), 28–
35.
296
Daugherty, T. – Eastin, M. S. – Bright, L. (2008). Exploring consumer moti-
vations for creating user-generated content. Journal of Interactive
Advertising, Vol. 8 (2), 1–24.
Davis, J . (2009) Revealing the research “hole”of early childhood education for
sustainability: A preliminary survey of the literature. Environ-
mental Education Research, Vol. 15 (2), 227–241.
Developer Economics 2012 (2012) VisionMobile. , retrieved J une 2013.
Diaz, A. (2010) YouCastr - A post-mortem. , retrieved J une 2012.
Diaz Ruiz, C. (2012) Theories of markets: Insights from marketing and the
sociology of markets. The Marketing Review, Vol. 12 (1), 61–77.
Dickens, Z. (2010) Start-up Post-mortem: Contrastream. , retrieved
J une 2012.
Dimitras, A. I. – Zanakis, S. H. – Zopounidis, C. (1996) A survey of business
failures with an emphasis on prediction methods and industrial
applications. European Journal of Operational Research, Vol. 90
(3), 487–513.
Dixon, C. (2010) The bowling pin strategy. , retrieved J une 2013.
Dow, S. – Kulkarni, A. – Bunge, B. – Nguyen, T. – Klemmer, S. – Hartmann,
B. (2011) Shepherding the crowd yields better work. Proceedings
of the ACM Conference on Human Factors in Computing Systems,
eds. D. Tan – G. Fitzpatrick – C. Gutwin –B. Begole – W. A.
Kellogg, 1669-1674. ACM, New York, NY.
Dushnitsky, G. – Klueter, T. (2011) Is There an eBay for Ideas? Insights from
Online Knowledge Marketplaces. European Management Review,
Vol. 8 (1), 17–32.
Dwyer, F. R. – Schurr, P. H. – Oh, S. (1987) Developing buyer-seller relation-
ships. J ournal of Marketing, Vol. 51 (2), 11–27.
Easley, R. W. – Madden, C. S. – Dunn, M. G. (2000) Conducting marketing
science: the role of replication in the research process. Journal of
Business Research, Vol. 48 (1), 83–92.
Economides, N. – Katsamakas, E. (2006) Two-sided competition of proprie-
tary vs. open source technology platforms and the implications for
the software industry. Management Science, Vol. 52 (7), 1057–
1071.
297
Edwards, T. (2013) Facebook targets 76 million fake users in war on bogus
accounts. Business Insider. , retrieved J une 2013.
Ehrenberg, R. (2008) Monitor110: A Post Mortem. , retrieved J une
2012.
Eisenhardt, K. (1989) Building Theories from Case Study Research. The
Academy of Management Review, Vol. 14 (4), 532–550.
Eisenhardt, K. – Graebner, M. E. (2007) Theory Building From Cases: Op-
portunities And Challenges. The Academy of Management Jour-
nal, Vol. 50 (1), 25–32.
Eisenmann, T. – Parker, G. – Van Alstyne, M. (2009) Opening platforms:
how, when and why? In: Platforms, Markets and Innovation, ed.
A. Gawer, 131–162. Edward Elgar Publishing, Cheltenham.
Eisenmann, T. – Parker, G. – Van Alstyne, M. W. (2006) Strategies for two-
sided markets. Harvard Business Review, Vol. 84 (10), 92–101.
Eisenmann, T. – Parker, G. – Van Alstyne, M. (2011) Platform envelopment.
Strategic Management Journal, Vol. 32 (12), 1270–1285.
Elliott, J . (2005) Using narrative in social research: Qualitative and quantita-
tive approaches. SAGE Publications, London.
Elliott, N. – Lazenbatt, A. (2005) How to recognise a “quality” grounded the-
ory research study. The Australian Journal of Advanced Nursing,
Vol. 22 (3), 48–52.
Ellison, N. B. (2007) Social network sites: Definition, history, and scholarship.
Journal of Computer-Mediated Communication, Vol. 13 (1), 210–
230.
Evans, D. S. (2002) The Antitrust Economics of Two-Sided Markets. SSRN
Working Paper No. 332022. Social Science Research Network,
Rochester, NY.
Evans, D. S. (2003) Some Empirical Aspects of Multi-sided Platform Indus-
tries. SSRN Working Paper No. 447981. Social Science Research
Network, Rochester, NY.
Evans, D. S. (2009a) How catalysts ignite: the economics of platform-based
start-ups. In: Platforms, Markets and Innovation, ed. A. Gawer,
99–130. Edward Elgar Publishing Cheltenham.
Evans, D. S. (2009b) Two-Sided Market Definition. SSRN Working Paper No.
1396751. Social Science Research Network, Rochester, NY.
Evans, D. S. – Hagiu, A. – Schmalensee, R. (2006) Invisible Engines: How
Software Platforms Drive Innovation and Transform Industries.
MIT Press, Cambridge, MA.
298
Evans, D. S. – Schmalensee, R. (2010) Failure to Launch: Critical Mass in
Platform Businesses. SSRN Working Paper 1353502. Social Sci-
ence Research Network, Rochester, NY
Facebook (2013) Advertising guidelines. Facebook. , retrieved J une 2013.
Farrell, J . – Klemperer, P. (2007) Coordination and lock-in: Competition with
switching costs and network effects. In: Handbook of industrial
organization, Volume 3, eds. M. Armstrong – R. Porter, 1967–
2072. Elsevier B. D., Amsterdam.
Farrell, J . – Saloner, G. (1985) Standardization, compatibility, and innovation.
The RAND Journal of Economics, Vol. 16 (1), 70–83.
Farrell, J . – Saloner, G. (1986) Installed Base and Compatibility: Innovation,
Product Preannouncements, and Predation. The American Eco-
nomic Review, Vol. 76 (5), 940–955
Fehr, E. – Kirchler, E. – Weichbold, A. – Gächter, S. (1998) When social
norms overpower competition: Gift exchange in experimental la-
bor markets. Journal of Labor Economics, Vol. 16 (2), 324–351.
Feld, B. (2006) Talking About Failure. , retrieved J une 2012.
Festinger, L. (1962) A theory of cognitive dissonance, Volume 2, Stanford
university press, Stanford, CA.
Finch, J . H. (2002) The role of grounded theory in developing economic the-
ory. Journal of Economic Methodology, Vol. 9 (2), 213–234.
Fischer, M. – Albers, S. – Wagner, N. – Frie, M. (2011) Practice prize win-
ner—dynamic marketing budget allocation across countries, prod-
ucts, and marketing activities. Marketing Science, Vol. 30 (4),
568–585.
Fletcher, A. (2007) Predatory pricing in two-sided markets: A brief comment.
Competition Policy International, Vol. 3 (1), 221–224.
Flynn, L. R. – Goldsmith, R. E. – Eastman, J . K. (1996) Opinion leaders and
opinion seekers: two new measurement scales. J ournal of the
Academy of Marketing Science, Vol. 24 (2), 137–147.
Freeman, J . – Carroll, G. R. – Hannan, M. T. (1983) The liability of newness:
Age dependence in organizational death rates. American Socio-
logical Review, Vol. 48 (5), 692–710.
Funk, J . (2006) Early competition in complex network industries: The case of
the mobile internet. Hitotshubashi University Institute of Innova-
tion Research Working Paper, Tokyo, J apan.
Galbreth, M. R. – March, S. T. – Scudder, G. D. – Shor, M. (2003) A game-
theoretic model of e-marketplace participation growth. Journal of
Management Information Systems, Vol. 22 (1), 295–319.
299
Ganley, D. – Lampe, C. (2009) The ties that bind: Social network principles in
online communities. Decision Support Systems, Vol. 47 (3), 266–
274.
Gans, J . S. (2012) Mobile application pricing. Information Economics and
Policy, Vol. 24 (1), 52–59.
Gaskill, L. R. – Van Auken, H. E. – Manning, R. A. (1993) A factor analytic
study of the perceived causes of small business failure. Journal of
Small Business Management, Vol. 31 (4), 18.
Gasson, S. (2003) Rigor in grounded theory research - an interpretive perspec-
tive on generating theory from qualitative field studies. In: Hand-
book for information systems research, eds. M. Whitman – A.
Woszczynski, 79–102. Idea Group Publishing, Hershey, PA.
Gawer, A. (2009) Platform dynamics and strategies: from products to services.
In: Platforms, Markets and Innovation, ed. A. Gawer, 45–76.
Edward Elgar Publishing, Cheltenham.
Gawer, A. – Cusumano, M. (2008) How Companies Become Platform Lead-
ers. MIT Sloan Management Review, Vol. 49 (2), 28–35.
Gawer, A. – Henderson, R. (2007) Platform owner entry and innovation in
complementary markets: Evidence from Intel. Journal of Eco-
nomics & Management Strategy, Vol. 16 (1), 1–34.
Gazé, P. – Vaubourg, A-G. (2011) Electronic platforms and two-sided mar-
kets: A side-switching analysis. The Journal of High Technology
Management Research, Vol. 22 (2), 158–165.
Gilder, G. (1993) Metcalf’s law and legacy. Forbes ASAP, 13.9.1993.
Gillin, P. (2009) The new influencers: A marketer’s guide to the new social
media. Linden Publishing, Freshno, CA.
Glaser, B. G. (1971) Status Passage. Transaction Publishers, Chicago, IL.
Glaser, B. G. (1978) Theoretical sensitivity: advances in the methodology of
grounded theory. Sociology Press, Mill Valley, CA.
Glaser, B. G. (1992) Basics of grounded theory analysis: Emergence vs forc-
ing. Sociology Press, Mill Valley, CA.
Glaser, B. G. (2004) “Naturalist inquiry” and grounded theory. Forum: Qual-
itative Social Research, Vol. 5 (1), 1–14.
Glaser, B. G. (2002) Conceptualization: On theory and theorizing using
grounded theory. International Journal of Qualitative Methods,
Vol. 1 (2), 23–38.
Glaser, B. G. – Holton, J . (2004) Remodeling grounded theory. Forum: Qual-
itative Social Research, Vol. 5 (2), 1–17.
Glaser, B. G. – Strauss, A. L. (1965) Discovery of substantive theory: A basic
strategy underlying qualitative research. American Behavioral
Scientist, Vol. 8 (6), 5–12.
300
Glaser, B. G. – Strauss, A. L. (1967) The discovery of grounded theory: strat-
egies for qualitative research. Transaction Publishers, Piscataway,
NJ .
Golder, P. N. – Tellis, G. J . (1993) Pioneer advantage: Marketing logic or
marketing legend? Journal of Marketing Research, Vol. 30 (2),
158–170.
Goldenson, M. (2009) 10 lessons from a failed startup.
, retrieved J une 2012.
Goldfarb, A. (2003) Advertising, profits, switching costs, and the internet. In:
Proceedings of the 5th international conference on electronic
commerce, eds. N. M. Sadeh – M. J . Dively – R. J . Kauffman – Y.
Labrou – O. Shehory – R. Telang – L. F. Cranor, 266–275. ACM
Press, New York, NY.
Google Inc. Announces Fourth Quarter and Fiscal Year 2012 Results (2013)
Google Investor Relations. , retrieved J une 2013.
Goulding, C. (2005) Grounded theory, ethnography and phenomenology: A
comparative analysis of three qualitative strategies for marketing
research. European Journal of Marketing, Vol. 39 (3/4), 294–308.
Grabner-Kräuter, S. – Kaluscha, E. A. (2003) Empirical research in on-line
trust: A review and critical assessment. International Journal of
Human-Computer Studies, Vol. 58 (6), 783–812.
Gulli, A. – Signorini, A. (2005) The indexable web is more than 11.5 billion
pages. In: Proceedings of WWW '05 Special interest tracks and
posters of the 14th international conference on World Wide Web,
eds. A. Ellis – T. Hagino – F. Douglis – P. Raghavan, 902–903.
ACM Press, New York, NY.
Gunawardana, A. – Meek, C. – Biggs, J . (2008) A quality-based auction for
search ad markets with aggregators. Paper presented at theACM
EC Workshop on Ad Auctions, Chicago, IL.
Hagiu, A. (2006) Pricing and commitment by two-sided platforms. The RAND
Journal of Economics, Vol. 37 (3), 720–737.
Hagiu, A. (2007) Merchant or two-sided platform? Review of Network Eco-
nomics, Vol. 6 (2), 1–19.
Hagiu, A. (2008) Platforms, pricing, commitment and variety in two-sided
markets. Doctoral dissertation, Princeton University, Princeton,
NJ .
Hagiu, A. (2009) Multi-sided platforms: From microfoundations to design and
expansion strategies. Harvard Business School Working Paper,
No. 09–115, Boston, MA.
301
Hagiu, A. and T. Eisenmann (2007) A Staged Solution to the Catch-22. Har-
vard Business Review, Vol. 85 (11).
Hagiu, A. – J ullien, B. (2011) Why do intermediaries divert search? The
RAND Journal of Economics, Vol. 42 (2), 337–362.
Hagiu, A. – Spulber, D. (2012) First-party content and coordination in two-
sided markets. Management Science, Vol. 59 (4), 933–949.
Hagiu, A. – Wright, J . (2011) Multi-sided platforms. Harvard Business School
Working Paper, No. 12-024, Boston, MA.
Hagiu, A. – Wright, J . (2013) Do you really want to be an eBay? Harvard
Business Review, Vol. 91 (3), 102-108.
Hagiu, A. – Yoffie, D. B. (2009) What’s your Google strategy? Harvard Busi-
ness Review, Vol. 87 (4), 74–81.
Halligan, B. – Shah, D. (2009) Inbound marketing: get found using Google,
social media, and blogs. J ohn Wiley & Sons, Hoboken, NJ .
Haltiwanger, J . – J armin, R. – Miranda, J . (2009) Business dynamics statistics
briefing: High growth and failure of young firms. Ewing Marion
Kauffman Foundation, Kansas City, MO.
Hammer, E. (2008) The Last AnNounce(r)ment. , retrieved J une 2012.
Han, K. S. – Noh, M. H. (1999) Critical failure factors that discourage the
growth of electronic commerce. International Journal of Elec-
tronic Commerce, Vol. 4 (2), 25–43.
Hartung, A. (2011) How Facebook beat MySpace. , retrieved 2.7.2013.
Heath, H. – Cowley, S. (2004) Developing a grounded theory approach: a
comparison of Glaser and Strauss. International Journal of Nurs-
ing Studies, Vol. 41 (2), 141–150.
Hedlund, M. (2010) Why Wesabe Lost to Mint. , retrieved J une 2012.
Hemrajani, K. (2010) Post-mortem. , retrieved J une 2012.
Hennig-Thurau, T. – Gwinner, K. P. – Walsh, G. – Gremler, D. D. (2004)
Electronic word-of-mouth via consumer-opinion platforms: What
motivates consumers to articulate themselves on the internet?
Journal of Interactive Marketing, Vol. 18 (1), 38–52.
Henshel, R. L. – J ohnston, W. (1987) The emergence of bandwagon effects: A
theory. The Sociological Quarterly, Vol. 28 (4), 493–511.
302
Herings, P. J . J . – Schinkel, M. P. (2001) World-Wide-Welfare: a micro-eco-
nomic analysis of the new economy. METEOR Working Paper,
No. RM/01/004, Maastricht research school of Economics of
Technology and Organizations, Maastricht.
Hermida, A. – Thurman, N. (2008) A clash of cultures: The integration of
user-generated content within professional journalistic frame-
works at British newspaper websites. Journalism Practice, Vol. 2
(3), 343–356.
Hill, S. (2006) Network-Based Marketing: Identifying Likely Adopters via
Consumer Networks. Statistical Science, Vol. 21 (2), 256–276.
Hinde, S. (2003) Spam: the evolution of a nuisance. Computers & Security,
Vol. 22 (6), 474–478.
Hinz, O. – Skiera, B. – Barrot, C. – Becker, J . U. (2011) Seeding Strategies for
Viral Marketing: An Empirical Comparison. Journal of Market-
ing, Vol. 75 (6), 55–71.
Hofer, C. W. (1980) Turnaround strategies. Journal of Business Strategy, Vol.
1 (1), 19–31.
Holton, J . (2010) The Coding Process and Its Challenges. The Grounded The-
ory Review - An International Journal, Vol. 9 (1), 21–40.
Honjo, Y. (2000) Business failure of new firms: an empirical analysis using a
multiplicative hazards model. International Journal of Industrial
Organization, Vol. 18 (4), 557–574.
Hopkins, C. C. (1968). Scientific advertising. New Line Publishing.
Horowitz, D. – Kamvar, S. D. (2010) The anatomy of a large-scale social
search engine. In: Proceedings of the 19th international confer-
ence on World Wide Web, eds. M. Rappa – P. J ones – J . Freire –S.
Chakraparti, 431–440. ACM Press, New York, NY.
Howe, J . (2006) The rise of crowdsourcing. Wired Magazine, Vol. 14 (6), 1–4.
Hsiao, R-L. (2003) Technology fears: distrust and cultural persistence in elec-
tronic marketplace adoption. The Journal of Strategic Information
Systems, Vol. 12 (3), 169–199.
Hsieh-Yee, I. (2001) Research on Web search behavior. Library & Infor-
mation Science Research, Vol. 23 (2), 167–185.
Huang, P. – Ceccagnoli, M. – Forman, C. – Wu, D. J . (2009). Participation in
a Platform Ecosystem: Appropriability, Competition and Access
to the Installed Base. Working paper 09-14, NET Institute. Avail-
able athttp://papers.ssrn.com/abstract=1480900.
Hunt, S. D. (2002) Foundations of Marketing Theory: Toward a General The-
ory of Marketing. M.E. Sharpe.
303
Hyrynsalmi, S. – Mäkilä, T. – J ärvi, A. – Suominen, A. – Seppänen, M. –
Knuutila, T. (2012) App store, marketplace, play! an analysis of
multi-homing in mobile software ecosystems. In: Proceedings of
the Fourth International Workshop on Software Ecosystems, eds.
S. J ansen – J . Bosch – C. Alves, 59–72. CEUR Workshop Pro-
ceedings, Cambridge, MA.
Iacovou, C. L. – Benbasat, I. – Dexter, A. S. (1995) Electronic data inter-
change and small organizations: adoption and impact of technol-
ogy. MIS Quarterly, Vol. 19 (4), 465–485.
Idu, A. – van de Zande, T. – J ansen, S. (2011) Multi-homing in the apple eco-
system: why and how developers target multiple apple app stores.
In: Proceedings of the International Conference on Management
of Emergent Digital EcoSystems, eds. W. I. Grosky – Y. Badr – R.
Chbeir, 122–128. ACM Press, New York, NY.
Darwell, B. (2013) Facebook’s frictionless sharing mistake. Inside Facebook,
22.1.2013. Insider Facebook. , re-
trieved J une 2013.
Honan, M. (2012) Twitter’s New API Rules Likely Spell the End of Third-
Party Clients. Wired. , retrieved J une 2013.
Isaacson, W. (2011) Steve Jobs. Simon & Schuster, New York, NY.
J anis, I. L. (1982) Groupthink. Houghton Mifflin, Boston, MA.
J ankowski, J . (2013) Increasing website conversions using content repetitions
with different levels of persuasion. In: Intelligent Information and
Database Systems, 5
th
Asian Conference, Proceedings, Part II,
eds. A. Selamat – N. T. Nguyen – H. Haron, 439–448. Springer,
Berlin.
J ansen, S. – Bloemendal, E. (2013) Defining app stores: The role of curated
marketplaces in software ecosystems. In: Software Business. From
Physical Products to Software Services and Solutions. Proceed-
ings of the 4
th
International Conference, ICSOB 2013, eds. G.
Herzwurm – T. Margaria, 195–206. Springer, Berlin.
J ohnson, R. B. (1997) Examining the validity structure of qualitative research.
Education, Vol. 118 (2), 282–292.
J ohnson, R. – Fraser, J . (2010) EventVue post-mortem. , retrieved J une 2012.
J ones, M. – Alony, I. (2011) Guiding the use of grounded theory in doctoral
studies – An example from the Australian film industry. Interna-
tional Journal of Doctoral Studies, Vol. 6, 95–114.
304
J ullien, B. (2005) Two-sided markets and electronic intermediaries. CESifo
Economic Studies, Vol. 51 (2-3), 233–260.
Katila, R. – Rosenberger, J . – Eisenhardt, K. (2008) Swimming with Sharks:
Technology Ventures, Defense Mechanisms and Corporate Rela-
tionships. Administrative Science Quarterly, Vol. 53 (2), 295–332.
Kaiser, U. – Wright, J . (2006) Price structure in two-sided markets: Evidence
from the magazine industry. International Journal of Industrial
Organization, Vol. 24 (1), 1–28.
Kan, M. – Parry, K. W. (2004) Identifying paradox: A grounded theory of
leadership in overcoming resistance to change. The Leadership
Quarterly, Vol. 15 (4), 467–491.
Kan, J . (2006) Actual lessons from Kiko. , retrieved J une 2012.
Kanazawa, S. (1998) In defense of unrealistic assumptions. Sociological The-
ory, Vol. 16 (2), 193–204.
Kane, T. (2010) The importance of startups in job creation and job destruction.
Kauffman Foundation Research Series: Firm Formation and Eco-
nomic Growth. Ewing Marion Kauffman Foundation, Kansas
City, MO.
Karjaluoto, E. (2009) Why your web startup will fail. , retrieved J une 2012.
Kärrberg, P. (2010) A two-sided market approach to value chain dynamics in
telecom services: A study lens for mobile platform innovation and
pricing strategies. In: Proceedings of the 2010 Ninth International
Conference on Mobile Business and 2010 Ninth Global Mobility
Roundtable, ed. J . E. Guerrero, 505–509. IEEE Computer Society,
Conference Publishing Services, Los Alamitos, CA.
Katz, M. L. – Shapiro, C. (1985) Network externalities, competition, and
compatibility. The American Economic Review, Vol. 75 (3), 424–
440.
Katz, M. L. – Shapiro, C. (1986) Technology adoption in the presence of net-
work externalities. Journal of Political Economy, Vol. 94 (4),
822–841.
Katz, M. L. – Shapiro, C. (1994) Systems competition and network effects.
The Journal of Economic Perspectives, Vol. 8 (2), 93–115.
Kaufman, B. E. (2007) The institutional economics of J ohn R. Commons:
complement and substitute for neoclassical economic theory. So-
cio-Economic Review, Vol. 5 (1), 3–45.
Kelly, T. (2009) Zynga and the End of the Beginning. Gamasutra.
, retrieved J une 2013.
305
Kempster, S. – Parry, K. W. (2011) Grounded theory and leadership research:
A critical realist perspective. The Leadership Quarterly, Vol. 22
(1), 106–120.
Kerin, R. A. – Varadarajan, P. R. – Peterson, R. A. (1992) First-mover ad-
vantage: A synthesis, conceptual framework, and research propo-
sitions. The Journal of Marketing, Vol. 56 (4), 33–52.
Kim, K., & Tse, E. (2011). Dynamic competition strategy for online
knowledge-sharing platforms. International Journal of Electronic
Commerce, Vol. 16 (1), 41–76.
Kirkpatrick, D. (2010) The Facebook effect: the inside story of the company
that is connecting the world. Simon & Schuster, New York, NY.
Kittur, A. – Nickerson, J . V. – Bernstein, M. – Gerber, E. – Shaw, A. – Zim-
merman, J . – Lease, M. – Horton, J . (2013) The future of crowd
work. In Proceedings of the 2013 conference on Computer sup-
ported cooperative work, eds. A. Bruckman – S. Counts – C.
Lampe – L. Terveen, 1301–1318. ACM Press, New York, NY.
Klein, B. (1998) Hold-up problem. The New Palgrave Dictionary of Econom-
ics and the Law, 2, 241.
Klein, D. A. A. (2012) The strategic management of intellectual capital.
Routledge, London.
Kopelman, J . (2009) Let’s just add in a little virality. Redeye VC.
, retrieved J une 2013.
Kotha, S. – Vadlamani, B. L. (1995) Assessing generic strategies: an empirical
investigation of two competing typologies in discrete manufac-
turing industries. Strategic Management Journal, Vol. 16 (1), 75–
83.
Kousha, K. – Thelwall, M. (2007) Google Scholar citations and Google
Web/URL citations: A multi-discipline exploratory analysis.
Journal of the American Society for Information Science and
Technology, Vol. 58 (7), 1055–1065.
Kretz, L. (2010) A Happy Failure Story: Lessons Learned From My First
Startup. , retrieved J une 2012.
Kristiansen, E. G. (1998) R&D in the presence of network externalities: Tim-
ing and compatibility. The RAND Journal of Economics, Vol. 29
(3), 531–547.
Krumpal, I. (2011) Determinants of social desirability bias in sensitive sur-
veys: a literature review. Quality & Quantity, Vol. 47 (4), 2025–
2047.
Kuhn, T. S. (1970). The structure of scientific revolutions. Chicago/London.
306
Köster, D. (1999) Wettbewerb in Netzproduktmärkten. Gabler Verlag:
Wiesbaden.
Laffont, J .-J . – Rey, P. – Tirole, J . (1998) Network competition: II. Price dis-
crimination. The RAND Journal of Economics, Vol. 29 (1), 38–56.
Lance, E. – Snider, D. (2006) Boompa.com: Post-mortem. , retrieved J une 2012.
Laudan, L. (1981) Science and hypothesis: Historical essays on scientific
methodology. Cambridge University Press; Cambridge.
Layard, P. R. G. – Layard, R. – Glaister, S. (1994) Cost-benefit analysis.
Cambridge University Press, Cambridge.
Lea, M. – Spears, R. – de Groot, D. (2001) Knowing me, knowing you: Ano-
nymity effects on social identity processes within groups. Person-
ality and Social Psychology Bulletin, Vol. 27 (5), 526–537.
Lee, R. S. (2013) Vertical integration and exclusivity in platform and two-
sided markets. American Economic Review, Vol. 103 (7), 2960–
3000.
Lee, S. C. – Brandyberry, A. A. (2003) The e-tailer’s dilemma. SIGMIS Data-
base, Vol. 34 (2), 10–22.
Lerner, J . – Tirole, J . (2004) The economics of technology sharing: Open
source and beyond. NBER Working Paper No. 10956, National
Bureau of Economic Research, Cambridge, MA.
Li, S. – Liu, Y. – Bandyopadhyay, S. (2010) Network effects in online two-
sided market platforms: A research note. Decision Support Sys-
tems, Vol. 49 (2), 245–249.
Li, Z. – Penard, T. (2013) The role of quantitative and qualitative network ef-
fects in b2b platform competition. Managerial and Decision Eco-
nomics, Vol. 35 (1), 1–19.
Libai, B. – Biyalogorsky, E. – Gerstner, E. (2003) Setting referral fees in af-
filiate marketing. Journal of Service Research, Vol. 5 (4), 303–
315.
Lieberman, M. B. (2005) Did first-mover advantage survive the dot-com
crash. Working Paper, Anderson Graduate School of Manage-
ment, UCLA, Los Angeles, CA.
Lieberman, M. B. – Montgomery, D. B. (1988) First-mover advantages. Stra-
tegic Management J ournal, Vol. 9 (S1), 41–58.
Lincoln, Y. S. – Guba, E. G. (1985) Naturalistic Inquiry (1st edition.). SAGE
Publications, London.
Locke, K. (1996) Rewriting the Discovery of Grounded Theory after 25
Years? Journal of Management Inquiry, Vol. 5 (3), 239–245.
307
Luchetta, G. (2012) Is the Google platform a two-sided market? SSRN
Working Paper No. 2048683. Social Science Research Network,
Rochester, NY.
Lussier, R. N. (1996) Reasons why small businesses fail: and how to avoid
failure. The Entrepreneurial Executive, Vol. 1 (2), 10–17.
Lyles, M. A. (1981) Formulating strategic problems: Empirical analysis and
model development. Strategic Management J ournal, Vol. 2 (1),
61–75.
Lyles, M. A. – Howard, T. (1985) Problem formulation and organizational
decision-making?: Biases and assumptions underlying alternative
models of strategic problem formulation. University of Illinois at
Urbana-Champaign, College of Commerce and Business Admin-
istration, Champaign, IL.
Lyles, M. A. – Howard, T. (1988) Strategic problem formulation: Biases and
assumptions embedded in alternative decision-making models.
Journal of Management Studies, Vol. 25 (2), 131–145.
Lyon, D. W. – Lumpkin, G. – Dess, G. G. (2000) Enhancing entrepreneurial
orientation research: Operationalizing and measuring a key strate-
gic decision making process. Journal of Management, Vol. 26 (5),
1055–1085.
Lyons, K. – Messinger, P. R. – Niu, R. H. – Stroulia, E. (2012) A tale of two
pricing systems for services. Information Systems and E-Business
Management, Vol. 10 (1), 19–42.
Maicas, J . P. – Polo, Y. – Sese, F. J . (2009) The role of (personal) network
effects and switching costs in determining mobile users’ choice.
Journal of Information Technology, Vol. 24 (2), 160–171.
Maitlis, S. (2005) The social processes of organizational sensemaking. Acad-
emy of Management Journal, Vol. 48 (1), 21–49.
Mangold, W. G. – Faulds, D. J . (2009) Social media: The new hybrid element
of the promotion mix. Business Horizons, Vol. 52 (4), 357–365.
Mantere, S. – Aula, P. – Schildt, H. – Vaara, E. (2013) Narrative attributions
of entrepreneurial failure. Journal of Business Venturing, Vol. 28
(4), 459–473.
Mäntymäki, M. (2011) Continuous use and purchasing behaviour in social
virtual worlds. Turku School of Economics Series A-3:2011,
Turku.
Marker, M. (1998) Going native in the academy: Choosing the exotic over the
critical. Anthropology & Education Quarterly, Vol. 29 (4), 473–
480.
308
Martin, P. Y. – Turner, B. A. (1986) Grounded theory and organizational re-
search. The Journal of Applied Behavioral Science, Vol. 22 (2),
141–157.
Maruyama, M. – Ohkita, K. (2011) Platform strategy of video game software
in J apan, 1984-1994. Managerial and Decision Economics, Vol.
32 (2), 105–118.
Mas, I. – Radcliffe, D. (2011) Scaling mobile money. Journal of Payments
Strategy & Systems, Vol. 5 (3), 298–315.
Mason, R. O. – Mitroff, I. I. (1981) Challenging strategic planning assump-
tions: Theory, cases, and techniques. J ohn Wiley & Sons,
Hoboken, NJ .
Mata, J . – Portugal, P. – Guimarães, P. (1995) The survival of new plants:
Start-up conditions and post-entry evolution. International Jour-
nal of Industrial Organization, Vol. 13 (4), 459–481.
Mathur, V. K. (1979) Some unresolved issues in the location theory of the
firm. Journal of Urban Economics, Vol. 6 (3), 299–318.
May, R. (2007) A Startup Idea Postmortem: Proof That Good Ideas Aren’t
Always Good Business. , retrieved J une 2012.
Mazrui, A. (1995) Slang and code switching: the case of Sheng in Kenya. Af-
rikanistiche Arbeitspapiere, 42, 168–179.
McCarthy, P. – Plantholt, B. – Riordan, S. (1981) Success versus survival?: the
dilemma of high technology firms. Massachusetts Institute of
Technology, Cambridge, MA.
McGhee, G. – Marland, G. R. – Atkinson, J . (2007) Grounded theory research:
literature reviewing and reflexivity. Journal of Advanced Nursing,
Vol. 60 (3), 334–342.
McGrady, D. (2008) 7 Reasons Why My Social Music Site Never Took Off.
, retrieved J une 2012.
McIntyre, D. (2011) Network intensity: Assessing the strategic value of in-
stalled base. Journal of Business Strategies, Vol. 28 (2), 177–195.
McIntyre, D. – Subramaniam, M. (2009) Strategy in network industries: a re-
view and research agenda. Journal of Management, Vol. 35 (6),
1494–517.
Mears, P. (1966) Discussion of financial ratios as predictors of failure. Journal
of Accounting Research, Vol. 4, 119–122.
Melin, L. (1985) Strategies in managing turnaround. Long Range Planning,
Vol. 18 (1), 80–86.
309
Merton, R. K. (1948) The self-fulfilling prophecy. The Antioch Review, Vol. 8
(2), 193–210.
Meyling, H. (1997) How to change students ‘conceptions of the epistemology
of science. Science & Education, Vol. 6 (4), 397–416.
Mian, S. Q. – Teixeira, J . – Koskivaara, E. (2011) Open-source software im-
plications in the competitive mobile platforms market. In: Build-
ing the e-World Ecosystem, eds. T. Skersys – R. Butleris – L.
Nemuraite – R. Suomi, 110–128. Springer, Berlin.
Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An ex-
panded sourcebook. Sage, Thousand Oaks, CA.
Milgrom, P. – Roberts, J . (1986) Price and advertising signals of product
quality. The Journal of Political Economy, Vol. 94 (4), 796–821.
Miller, D. (1977) Common syndromes of business failure. Business Horizons,
Vol. 20 (6), 43–53.
Mital, M. – Sarkar, S. (2011) Multihoming behavior of users in social net-
working web sites: a theoretical model. Information Technology &
People, Vol. 24 (4), 378–392.
Moh, T.-S. – Murmann, A. J . (2010) Can you judge a man by his friends? –
Enhancing spammer detection on the Twitter microblogging plat-
form using friends and followers. In: Information Systems, Tech-
nology and Management; Proceedings of the 4th International
Conference, ICISTM 2010, Bangkok, Thailand, eds. S. K. Prasad
– H. M. Vin – S. Sahni – M. P. J aiswal – B. Thipakorn, 210–220.
Springer, Berlin.
Moore, G. A. (1991) Crossing the chasm. Harper Business, New York, NY.
Moscovici, S. – Faucheux, C. (1972) Social influence, conformity bias, and
the study of active minorities. In: Advances in experimental social
psychology, Vol. 6, ed. L. Berkowitz, 149–202. Academic Press,
London.
Mungamuru, B. – Weis, S. (2008) Competition and fraud in online advertising
markets. In: Financial Cryptography and Data Security, Revised
Selected Papers from the 12th International Conference, FC
2008, Cozumel, Mexico, ed. G. Tsudik, 187–191. Springer, Berlin.
Murphy, L. (2011) Classical freemium doesn’t exist at scale. Saas growth
strategies. , retrieved J une 2013.
Nagel, E. (1963) Assumptions in economic theory. The American Economic
Review, Vol. 53 (2), 211–219.
Nash, J . (1950) Equilibrium points in n-person games. Proceedings of the Na-
tional Academy of Sciences, Vol. 36 (1), 48–49.
310
Neilsen, E. H. – Rao, M. V. H. (1987) The strategy-legitimacy nexus: A thick
description. Academy of Management Review, Vol. 12 (3), 523–
533.
Newberry, M. (2010) Don't Drink Your Own Kool-Aid , retrieved J une 2012.
Nickerson, R. S. (1998) Confirmation bias: a ubiquitous phenomenon in many
guises. Review of General Psychology, Vol. 2 (2), 175–220.
Nickinson, P. (2013) Two more casualities of Twitter’s power play. Android
Central. , retrieved 5.7.2013.
Niculescu, M. – Wu, D. J. (2013) Economics of free under perpetual licensing:
Implications for the software industry. Information Systems Re-
search, Vol. 25 (1), 173–199
Nielsen, J . (1998) Web pages must live forever. Nielsen Norman Group.
, retrieved J une 2013.
Noe, T. – Parker, G. (2005) Winner take all: Competition, strategy, and the
structure of returns in the internet economy. Journal of Economics
& Management Strategy, Vol. 14 (1), 141–164.
Nowak, K. (2010) Getting Caught Mid-Pivot: Lessons from the Imercive
Postmortem. , retrieved J une
2012.
O’Reilly, T. (2005) Web 2.0. , retrieved J une 2013.
Ochoa, X. – Duval, E. (2008) Quantitative analysis of user-generated content
on the web. In: Proceedings of First International Workshop Un-
derstanding Web Evolution (WebEvolve '08), eds. D. De Roure
and W. Hall, 19–26, Beijing, China.
Odlyzko, A. – Tilly, B. (2005) A refutation of Metcalfe’s Law and a better
estimate for the value of networks and network interconnections.
, retrieved J une 2013.
Oestreicher-Singer, G. – Zalmanson, L. (2009). Paying for content or paying
for community? The effect of social involvement on subscribing
to media web sites. ISIC 2009 Proceedings, eds. H. Chen – S.
Slaughter. AIS.
Olson, E. L. (2008) The implications of platform sharing on brand value.
Journal of Product & Brand Management, Vol. 17 (4), 244–253.
311
Ordanini, A. – Micelli, S. – Di Maria, E. (2004) Failure and success of B-to-B
exchange business models: A contingent analysis of their perfor-
mance. European Management Journal, Vol. 22 (3), 281–289.
Oreg, S. (2003) Resistance to change: developing an individual differences
measure. Journal of Applied Psychology, Vol. 88 (4), 680–693.
Orton, J . D. (1997) From inductive to iterative grounded theory: Zipping the
gap between process theory and process data. Scandinavian Jour-
nal of Management, Vol. 13 (4), 419–438.
Oxford Dictionary. (2013) dilemma. , retrieved J une 2013.
Packham, G. A. (2002) Competitive advantage and growth: the challenge for
small firms. International Journal of Management and Decision
Making, Vol. 3 (2), 165–179.
Paden, J . (2011) Why conversion rate optimization is guaranteed to improve
SEO results. Digital Relevance. , retrieved J une 2013.
Palka, W. – Pousttchi, K. – Wiedemann, D. G. (2009) Mobile word-of-mouth
– A grounded theory of mobile viral marketing. Journal of Infor-
mation Technology, Vol. 24 (2), 172–185.
Pandit, N. R. (1996) The creation of theory: A recent application of the
grounded theory method. The Qualitative Report, Vol. 2 (4), 1–
14.
Park, N. K. – Mezias, J . M. – Song, J . (2004) A Resource-based view of stra-
tegic alliances and firm value in the electronic marketplace. Jour-
nal of Management, Vol. 30 (1), 7–27.
Parker, G. G. – Van Alstyne, M. W. (2005) Two-sided network effects: A the-
ory of information product design. Management Science, Vol. 51
(10), 1494–1504.
Parker, G. – Van Alstyne, M. (2008) Innovation, openness, and platform con-
trol. SSRN Working Paper No. 1079712. Social Science Research
Network, Rochester, NY.
Parker, G. – Van Alstyne, M. (2010) Two-sided network effects: A theory of
information product design. Management Science, Vol. 51 (10),
1494–1504.
Parr, B. (2011a) What's it like to work at a failing startup? , retrieved J une
2012.
Parr, B. (2011b) Twitter acquires TweetDeck. Mashable. , retrieved J une 2013.
312
Parrish, E. D. – Cassill, N. L. – Oxenham, W. (2006) Niche market strategy in
the textile and apparel industry. Journal of Fashion Marketing
and Management, Vol. 10 (4), 420–432.
Partington, D. (2000) Building grounded theories of management action. Brit-
ish Journal of Management, Vol. 11 (2), 91–102.
Pauly, M. V. (1968) The economics of moral hazard: comment. The American
Economic Review, Vol. 58 (3), 531–537.
Pauwels, K. – Weiss, A. (2008) Moving from free to fee: How online firms
market to change their business model successfully. Journal of
Marketing, Vol. 72 (3), 14–31.
Pawson, R. – Tilley, N. (2009) Realist evaluation. In: Evidence-based Practice
– Modernising the Knowledge Base of Social Work? Eds. H.-U.
Otto – A. Polutta – H. Ziegler, 151–184. Verlag Barbara Budrich,
Germany.
Peitz, M. – Waelbroeck, P. (2006) Why the music industry may gain from free
downloading – the role of sampling. International Journal of In-
dustrial Organization, Vol. 24 (5), 907–913.
Pempek, T. A. – Yermolayeva, Y. A. – Calvert, S. L. (2009) College students’
social networking experiences on Facebook. Journal of Applied
Developmental Psychology, Vol. 30 (3), 227–238.
Perry, C. – J ensen, O. (2001) Approaches to combining induction and deduc-
tion in one research study. In: Australian and New Zealand Mar-
keting Academy Conference, Massey University, Auckland, New
Zealand. , retrieved J une 2013.
Piezunka, H. (2011) Technological platforms; An assessment of the primary
types of technological platforms, their strategic issues and their
linkages to organizational theory. Journal Für Betriebswirtschaft,
Vol. 61 (2-3), 179–226.
Pitkänen, L. – Salminen, J . (2012) Crowdsourcing research to mobile consum-
ers? Emerging themes on videographic data collection. Paper pre-
sented at the Planetary Scientific Research Centre Conference,
Dubai, United Arab Emirates.
Poon, S. – Swatman, P. – Swatman, P. P. (1999) A longitudinal study of ex-
pectations in small business internet commerce. International
Journal of Electronic Commerce, 3 (3), 21–33.
Porter, C. E. (2004) A typology of virtual communities: A multi-disciplinary
foundation for future research. Journal of Computer-Mediated
Communication, Vol. 10 (1).
313
Potts, M. (2007) Backfence: Lessons Learned.http://recoveringjournalist.
typepad.com/recovering_journalist/2007/07/backfence-lesso.html,
retrieved J une 2012.
Powazek, D. (2008) Pixish Closing October 31. , retrieved J une 2012.
Pujol, N. (2010) Freemium: Attributes of an emerging business model. SSRN
Working Paper No. 1718663. Social Science Research Network,
Rochester, NY.
Rafer, S. (2009) Couldery Shouldery. , retrieved J une 2012.
Raivio, Y. – Luukkainen, S. (2011) Mobile networks as a two-sided platform -
Case Open Telco. Journal of Theoretical and Applied Electronic
Commerce Research, Vol. 6 (2), 77–89.
Rao, R. S. – Chandy, R. K. – Prabhu, J . C. (2008) The fruits of legitimacy:
Why some new ventures gain more from innovation than others.
Journal of Marketing, Vol. 72 (4), 58–75.
Rappa, M. (2013) Business models on the Web. , retrieved J une 2013.
Razi, M. A. – Tarn, J . M. – Siddiqui, F. A. (2004) Exploring the failure and
success of DotComs. Information Management & Computer Secu-
rity, Vol. 12 (3), 228–244.
Reeves, C. A. – Bednar, D. A. (1994) Defining quality: alternatives and impli-
cations. Academy of Management Review, Vol. 19 (3), 419–445.
Reisinger, M. (2004) Two-sided markets with negative externalities. Discus-
sion Papers in Economics 2004-27, Ludwig-Maximilians-Univer-
sität, Munich.
Reisinger, M. (2012) Platform competition for advertisers and users in media
markets. International Journal of Industrial Organization, Vol. 30
(2), 243–252.
Ren, S. – Park, J . – van der Schaar, M. (2011) Profit maximization on user-
generated content platforms. In: 49th Annual Allerton Conference
on Communication, Control, and Computing, 135–142. IEEE.
Ries, E. (2008) Three decisions to make on virtual goods. Startup lessons
learned. , retrieved J une 2013.
Ries, E. (2010) Stop lying on stage. Startup lessons learned. , re-
trieved J une 2013.
Ries, E. (2011) The Lean Startup. Penguin Books Ltd, London.
314
Riggins, F. J . (2003) Market segmentation and information development costs
in a two-tiered fee-based and sponsorship-based web site. Journal
of Management Information Systems, Vol. 19 (3), 69–86.
Riordan, M. H. – Williamson, O. E. (1985) Asset specificity and economic
organization. International Journal of Industrial Organization,
Vol. 3 (4), 365–378.
Roberts, E. B. (1990) Evolving toward product and market-orientation: the
early years of technology-based firms. Journal of Product Inno-
vation Management, Vol. 7 (4), 274–287.
Robinson, W. T. – Fornell, C. – Sullivan, M. (1992) Are market pioneers in-
trinsically stronger than later entrants? Strategic Management
Journal, Vol. 13 (8), 609–624.
Rochet, J .-C. – Tirole, J . (2003) Platform competition in two-sided markets.
Journal of the European Economic Association, Vol. 1 (4), 990–
1029.
Rochet, J .-C. – Tirole, J. (2005) Two-sided markets?: A progress report. IDEI
Working Paper No. 275, Institut d’Économie Industrielle, Tou-
louse.
Rogers, E. M. (1995) Diffusion of Innovations. Simon and Schuster, New
York, NY.
Rohlfs, J . (1974) A theory of interdependent demand for a communications
service. The Bell Journal of Economics and Management Science,
Vol. 5 (1), 16–37.
Roseman, S. (2010) Out with the Old business in with the New.
, retrieved J une 2012.
Roson, R. (2005) Two-sided markets: A tentative survey. Review of Network
Economics, Vol. 4 (2), 142–160.
Rumbo, J . D. (2002) Consumer resistance in a world of advertising clutter:
The case of Adbusters. Psychology & Marketing, Vol. 19 (2),
127–148.
Rysman, M. (2004) Competition between networks: A study of the market for
yellow pages. The Review of Economic Studies, Vol. 71 (2), 483–
512.
Rysman, M. (2007) An empirical analysis of payment card usage. The Journal
of Industrial Economics, Vol. 55 (1), 1–36.
Rysman, M. (2009) The economics of two-sided markets. Journal of Eco-
nomic Perspectives, Vol. 23 (3), 125–143.
Saha, R. – Mantena, R. – Tilson, V. (2012) Essays on economics of technol-
ogy-enabled intermediaries. University of Rochester, Rochester,
NY.
315
Salminen, J . (2010) Power of Google: A study on online advertising exchange.
LAP Lambert Academic Publishing, Saarbrücken, Germany.
Salminen, J . – Hytönen, A. (2012) Viral coefficient – Unveiling the Holy Grail
of online marketing. Paper presented at the Conference of Theo-
retical and Applied Aspects of Cybernetics, Kiev, Ukraine.
Salminen, J . – Teixeira, J . (2013) Fool’s gold? Developer dilemmas in a
closed mobile application market platform. Presented at the ICEC
2013 The 15th International Conference on Electronic Commerce,
Turku.
Salomon, G. (1991) Transcending the qualitative-quantitative debate: The an-
alytic and systemic approaches to educational research. Educa-
tional Researcher, Vol. 20 (6), 10–18.
Samanta, S. K. (2009) Impact of the utility and cost on price in a two sided
market, SSRN Working Paper No. 1339620. Social Science Re-
search Network, Rochester, NY.
Sawhney, M. S. (1998) Leveraged high-variety strategies: From portfolio
thinking to platform thinking. Journal of the Academy of Market-
ing Science, Vol. 26 (1), 54–61.
Sawhney, M. – Verona, G. – Prandelli, E. (2005) Collaborating to create: The
Internet as a platform for customer engagement in product inno-
vation. Journal of Interactive Marketing, Vol. 19 (4), 4–17.
Schilling, M. (2009) Protecting or diffusing a technology platform: tradeoffs
in appropriability, network externalities, and architectural control.
In: Platforms, Markets and Innovation, ed. A. Gawer. Edward El-
gar Publishing, Cheltenham.
Schlosser, A. E. – White, T. B. – Lloyd, S. M. (2006) Converting web site
visitors into buyers: How web site investment increases consumer
trusting beliefs and online purchase intentions. Journal of Mar-
keting, Vol. 70 (2), 133–148.
Schmalensee, R. – Evans, D. S. (2007) Industrial organization of markets with
two-sided platforms. Competition Policy International, Vol. 3 (1).
Schmidt, C. (2004) The Analysis of Semi-structured Interviews. In: A Com-
panion to Qualitative Research, eds. U. Flick – E. von Kardoff –
I. Steinke, 253–258. SAGE Publications, London.
Schnettler, S. (2009) A structured overview of 50 years of small-world re-
search. Social Networks, Vol. 31 (3), 165–178.
Schoenberger, E. (1991) The corporate interview as a research method in eco-
nomic geography. The Professional Geographer, Vol. 43 (2),
180–189.
316
Schumpeter, J . A. (1961) The theory of economic development: An inquiry
into profits, capital, credit, interest, and the business cycle, Vol-
ume 55. Transaction Books, New Brunswick, NJ .
Schwenk, C. R. (1984) Cognitive simplification processes in strategic deci-
sion-making. Strategic Management Journal, Vol. 5 (2), 111–128.
Seale, C. (1999) Quality in qualitative research. Qualitative Inquiry, Vol. 5
(4), 465–478.
Semenzin, D. – Meulendijks, E. – Seele, W. – Wagner, C. – Brinkkemper, S.
(2012) Differentiation in freemium: Where does the line lie? In:
Software Business; Proceedings of the Third International Con-
ference, ICSOB 2012, Cambridge, M. A. Cusumano – B. Iyer – N.
Venkatraman, 291–296. Springer, Berlin.
Shampanier, K. – Mazar, N. – Ariely, D. (2007) Zero as a special price: The
true value of free products. Marketing Science, Vol. 26 (6), 742–
757.
Shapiro, C. (1983) Optimal pricing of experience goods. The Bell Journal of
Economics, Vol. 14 (2), 497–507.
Shapiro, C. – Varian, H. (1998) Information rules: A Strategic guide to the
economy. Harvard Business School Press, Boston, MA.
Sharma, S. – Mahajan, V. (1980) Early Warning Indicators of Business Fail-
ure. Journal of Marketing, Vol. 44 (4), 80.
Shaughnessy, H. (2012) Facebook’s 1 billion users: Why the sky is still the
limit. Forbes. , retrieved J une 2013.
Shepherd, D. A. – Kuratko, D. F. (2009) The death of an innovative project:
How grief recovery enhances learning. Business Horizons, Vol. 52
(5), 451–458.
Sheridan, J . (2011) Free as in beer and free as in speech: Public sector infor-
mation after 450 days of open government. Australian Law Li-
brarian, Vol. 19 (4), 251–255.
Shy, O. (2011) A short survey of network economics. Review of Industrial
Organization, Vol. 38 (2), 119–149.
Simon, H. A. (1956) Rational choice and the structure of the environment.
Psychological Review, Vol. 63 (2), 129–138.
Singh, J . V. – Tucker, D. J . – House, R. J . (1986) Organizational legitimacy
and the liability of newness. Administrative Science Quarterly,
Vol. 31 (2), 171–193.
Smith, D. – Menon, S. – Sivakumar, K. (2005) Online peer and editorial rec-
ommendations, trust, and choice in virtual markets. Journal of In-
teractive Marketing, Vol. 19 (3), 15–37.
317
Smith, T. (2008) Transmutable Postmortem. , retrieved J une
2012.
Sommer, R. – Sommer, B. B. (1992) A practical guide to behavioral research.
Oxford University Press, New York, NY.
Soonsawad, P. (2013) Developing a new model for conversion rate optimiza-
tion: A case study. International Journal of Business and Man-
agement, Vol. 8 (10), 41–51.
Spulber, D. F. (2010) Solving the circular conundrum: communication and
coordination in internet markets. Northwestern University Law
Review, Vol. 104 (2), 537–592.
Standing, S. – Standing, C. – Love, P. E. D. (2010) A review of research on e-
marketplaces 1997–2008. Decision Support Systems, Vol. 49 (1),
41–51.
Stanoevska-Slabeva, K. (2002) Toward a community-oriented design of Inter-
net platforms. International Journal of Electronic Commerce,
Vol. 6 (3), 71–95.
Summary of Startup Genome Report Extra: Premature Scaling (2011) Startup
Compass. , retrieved December 2011.
Stinchcombe, A. L. (1965) Social structure and organizations. In: Handbook of
Organizations, ed. J . G. March, 153–193. Rand McNally, Chi-
cago, IL.
Strauss, A. – Corbin, J . (1994) Grounded theory methodology. In: Handbook
of qualitative research, eds. N. K. Denzin – Y. S. Lincoln, 273–
285. Sage, Thousand Oaks, CA.
Stvilia, B. – Twidale, M. B. – Smith, L. C. – Gasser, L. (2008) Information
quality work organization in Wikipedia. Journal of the American
Society for Information Science and Technology, Vol. 59 (6),
983–1001.
Suarez, F. F. (2005) Network effects revisited: the role of strong ties in tech-
nology selection. Academy of Management Journal, Vol. 48 (4),
710–720.
Suddaby, R. (2006) From the editors: What grounded theory is not. Academy
of Management Journal, Vol. 49 (4), 633–642.
Sun, M. – Tse, E. (2007) When does the winner take all in two-sided markets?
Review of Network Economics, Vol. 6 (1), 1–25.
Susarla, A. – Oh, J .-H. – Tan, Y. (2012). Social networks and the diffusion of
user-generated content: evidence from YouTube. Information
Systems Research, Vol. 23 (1), 23–41.
318
Tajirian, A. (2005) Multiple viable domain name marketplaces can co-exist.
DomainMart.
, retrieved J une 2013.
Tang, J . (2008) Postmortem. , retrieved J une 2012.
Tanriverdi, H. – Lee, C.-H. (2008. Within-industry diversification and firm
performance in the presence of network externalities: Evidence
from the software industry. Academy of Management Journal,
Vol. 51 (2), 381–397.
Teece, D. J. (2010) Business models, business strategy and innovation. Long
Range Planning, Vol. 43 (2), 172–194.
Teece, D. J. (2011) Dynamic capabilities: A framework for profitability. Ivey
Business Journal, Vol. 18 (7), 509–533.
The Chubby Team (2010) How to Develop a Product that Nobody Wants: The
Story of ChubbyBrain. , retrieved J une 2012.
Teixeira, J . (2012) Open-source technologies realizing social networks: A
multiple descriptive case-study. In: Proceedings of the 8th IFIP
WG 2.13 International Conference, OSS 2012, Hammamet, Tuni-
sia, September 10-13, 2012, IFIP Advances in Information and
Communication Technology 378/2012, eds. I. Hammouda – B.
Lundell – T. Mikkonen – W. Scacchi, 250–255. Springer, Berlin.
The Entrepreneurial Enlightenment (2012) Startup Compass ,
retrieved J une 2013.
Thelwall, M. (2008) Social networks, gender, and friending: An analysis of
MySpace member profiles. Journal of the American Society for
Information Science and Technology, Vol. 59 (8), 1321–1330.
Thompson, S. A. – Sinha, R. K. (2008) Brand communities and new product
adoption: the influence and limits of oppositional loyalty. Journal
of Marketing, Vol. 72 (6), 65–80.
Tourangeau, R. (1984) Cognitive science and survey methods. In: Cognitive
aspects of survey methodology: Building a bridge between disci-
plines, eds. T. B. J abine – M. L. Straf – J . M. Tanur – R.
Tourangeau, 73–100. National Academy Press, Washington, D.C.
319
Trusov, M. – Bucklin, R. E. – Pauwels, K. H. (2008) Effects of word-of-mouth
versus traditional marketing: Findings from an internet social
networking site. Robert H. Smith School of Business Research
Paper No. RHS 06-065, Robert H. Smith School of Business,
University of Maryland, College Park, MD.
Tversky, A. – Kahneman, D. (1974) J udgment under uncertainty: Heuristics
and biases. Science, Vol. 185 (4157), 1124–1131.
Urquhart, C. – Lehmann, H. – Myers, M. D. (2010) Putting the “theory” back
into grounded theory: guidelines for grounded theory studies in in-
formation systems. Information Systems Journal, Vol. 20 (4),
357–381.
Uyar, A. (2009) Investigation of the accuracy of search engine hit counts.
Journal of Information Science, Vol. 35 (4), 469–480.
Valente, T. W. (1995) Network models of the diffusion of innovations (Quan-
titative Methods in Communication). Hampton Press, New York,
NY.
Van Huyck, J . B. – Battalio, R. C. – Beil, R. O. (1990) Tacit coordination
games, strategic uncertainty, and coordination failure. The Ameri-
can Economic Review, Vol. 80 (1), 234–248.
Varian, H. R. (2003) Innovation, components and complements. MISQ Special
Issue Workshop, University of California, Berkeley, CA.
Venkatesh, V. – Davis, F. D. (2000) A theoretical extension of the technology
acceptance model: four longitudinal field studies. Management
Science, Vol. 46 (2), 186–204.
Viégas, F. B. – Donath, J . (2004) Social network visualization: Can we go be-
yond the graph. Paper presented at the Workshop on Social Net-
works (CSCW'04) , Chicago, IL.
Vesterinen, V. (2010) Cross-Promotion Network Applifier Hit 55 Million Us-
ers In 4 Months. Arctic Startup. , retrieved 4.7.2013.
Wagner, S. M. – Lukassen, P. – Mahlendorf, M. (2010) Misused and missed
use – Grounded theory and objective hermeneutics as methods for
research in industrial marketing. Industrial Marketing Manage-
ment, Vol. 39 (1), 5–15.
Wang, S. – Zheng, S. – Xu, L. – Li, D. – Meng, H. (2008) A literature review
of electronic marketplace research: Themes, theories and an inte-
grative framework. Information Systems Frontiers, Vol. 10 (5),
555–571.
320
Warner, A. (2009) Founder Of Fit Fuel On What You Can Learn From His
Company’s Closure – With Luke Burgis. , retrieved J une 2012.
Warner, A. (2012) Maids In Black: An Entrepreneurial Dilemma (Tech Or
Non-Tech?) – with Rohan Gilkes. , retrieved J une 2012.
Wasserman, N. (2013) The Founder’s Dilemmas: Anticipating and Avoiding
the Pitfalls That Can Sink a Startup. Princeton University Press.
Wasserman, T. – West, C. M. (2006) The business of open source. Paper pre-
sented at the International Conference on Open Source Systems,
Como, Italy.
Watson, J . – Everett, J . (1993) Defining small business failure. International
Small Business Journal, Vol. 11 (3), 35–48.
Watson, J . – Everett, J . (1999) Small business failure rates: Choice of defini-
tion and industry effects. International Small Business Journal,
Vol. 17 (2), 31–47.
Webster, F. E. (1969). New product adoption in industrial markets: a frame-
work for analysis. The Journal of Marketing, Vol. 33 (3), 35–39.
Weitzel, T. – Beimborn, D. – König, W. (2006) A Unified Economic Model of
Standard Diffusion: The Impact of Standardization Cost, Network
Effects, and Network Topology. MIS Quarterly, Vol. 30 (1), 489–
514.
Westheimer, N. (2008) Bricabox: Goodbye world! , retrieved J une 2012.
Westland, J . C. (2010). Critical mass and willingness to pay for social net-
works. Electronic Commerce Research and Applications, Vol. 9
(1), 6–19.
Wilcox, J . W. (1971) A simple theory of financial ratios as predictors of fail-
ure. Journal of Accounting Research, Vol. 9 (2), 389–395.
Williamson, O. E. (1981) The economics of organization: The transaction cost
approach. The American Journal of Sociology, Vol. 87 (3), 548–
577.
Wilson, F. (2006) The Freemium Business Model. , retrieved J une 2013.
Wilson, M. (2006) Bubble 2.0 begins to deflate – 9 lessons. , retrieved J une 2012.
Wiltbank, R. – Dew, N. – Read, S. – Sarasvathy, S. D. (2006) What to do
next? The case for non-predictive strategy. Strategic Management
Journal, Vol. 27 (10), 981–998.
321
Wired (2008) Twitter’s business model? Well, ummmm... Wired.
, retrieved 5.7.2013.
Wright, J . (2004) One-sided logic in two-sided markets. Review of Network
Economics, Vol. 3 (1), 1–21.
Wu, R. – Lin, M. (2012) Seller Diversity on a Two-Sided Platform. Research
Collection, School Of Information Systems Working Paper,
Singapore Management University, Singapore.
Wunsch-Vincent, S. – Vickery, G. (2007) Participative web: user-created
content. OECD Directorate for Science, Technology and Industry.
Xia, M. – Huang, Y. – Duan, W. – Whinston, A. B. (2007) To continue shar-
ing or not to continue sharing? – An empirical analysis of user de-
cision in peer-to-peer sharing networks. Information Systems Re-
search, Vol. 23 (1), 247–259.
Xin, M. – Levina, N. (2008) Software-as-a-service model: Elaborating client-
side adoption factors. In: Proceedings of the 29th International
Conference on Information Systems, eds. R. Boland – M.
Limayem – B. Pentland, Paris, France.
Yelle, L. E. (1979) The learning curve: Historical review and comprehensive
survey. Decision Sciences, Vol. 10 (2), 302–328.
Yaghmour, K. (2010) Failing Forward. , retrieved J une 2012.
Yli-Renko, H – J anakiraman, R. (2008) How Customer Portfolio Affects New
Product Development in Technology-Based Entrepreneurial
Firms. Journal of Marketing, Vol. 72 (September), 131–148.
Yoo, C. (2010) Innovations in the Internet’s architecture that challenge the
status quo. Journal of Telecommunications and High Technology
Law, Vol. 8 (1), 79–99.
Zacharakis A, M. G. D. (1999) Differing perceptions of new venture failure: a
matched exploratory study of venture capitalists and entrepre-
neurs. Journal of Small Business Management, Vol. 37 (3), 1–14.
Zeithaml, V. A. (1988) Consumer perceptions of price, quality, and value: a
means-end model and synthesis of evidence. The Journal of Mar-
keting, Vol. 52 (3), 2–22.
Zeldman, J . – Marcotte, E. (2009). Designing with web standards. New Rid-
ers, Berkeley, CA.
Zhang, X. – Zhu, F. (2011) Group Size and Incentives to Contribute: A Natu-
ral Experiment at Chinese Wikipedia. American Economic Re-
view, Vol. 101 (4), 1601–1615.
322
Zhao, Y. – Zhu, Q. (2012) Evaluation on crowdsourcing research: Current
status and future direction. Information Systems Frontiers, April
2012, 1–18.
Zheng, H. – Li, D. – Hou, W. (2011) Task design, motivation, and participa-
tion in crowdsourcing contests. International Journal of Elec-
tronic Commerce, Vol. 15 (4), 57–88.
Zynga Inc. (2013) Zynga reports fourth quarter and full year 2012 financial
results. Nasdaq OMX GlobeNewswire. , retrieved 1.7.2013.
3
2
3
A
P
P
E
N
D
I
X
1
C
O
D
I
N
G
G
U
I
D
E
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
2
_
B
U
S
I
N
E
S
S
M
O
D
E
L
e
c
o
n
o
m
i
c
v
i
a
b
i
l
i
t
y
I
m
p
l
y
i
n
g
l
a
c
k
o
f
e
c
o
n
o
m
i
c
v
i
a
b
i
l
i
t
y
.
T
h
e
l
a
r
g
e
r
l
e
s
s
o
n
h
e
r
e
i
s
t
h
a
t
b
u
s
i
n
e
s
s
i
s
u
l
t
i
m
a
t
e
l
y
d
r
i
v
e
n
b
y
e
c
o
n
o
m
i
c
s
,
a
n
d
e
c
o
n
o
m
i
c
s
a
r
e
d
r
i
v
e
n
b
y
i
n
c
e
n
t
i
v
e
s
.
E
n
t
r
e
p
r
e
n
e
u
r
s
,
a
n
d
i
n
p
a
r
t
i
c
u
l
a
r
w
e
b
e
n
-
t
r
e
p
r
e
n
e
u
r
s
,
a
r
e
o
f
t
e
n
d
r
i
v
e
n
b
y
a
m
y
o
p
i
a
t
h
a
t
s
a
y
s
"
i
f
I
l
i
k
e
i
t
,
s
o
w
i
l
l
e
v
e
r
y
o
n
e
e
l
s
e
.
"
a
g
g
r
e
g
a
t
i
o
n
S
t
r
a
t
e
g
y
t
o
s
o
u
r
c
e
d
a
t
a
f
r
o
m
o
t
h
e
r
w
e
b
s
i
t
e
s
F
i
r
s
t
a
n
d
f
o
r
e
m
o
s
t
,
w
e
’
l
l
u
s
e
t
e
c
h
n
o
l
o
g
y
w
e
’
v
e
b
u
i
l
t
t
o
a
g
g
r
e
g
a
t
e
i
n
f
o
r
m
a
t
i
o
n
a
b
o
u
t
s
t
a
r
t
u
p
s
.
e
x
t
e
n
d
e
d
f
i
r
m
D
e
l
e
g
a
t
i
n
g
t
a
s
k
s
t
o
u
s
e
r
s
.
n
e
t
w
o
r
k
e
f
f
e
c
t
s
R
e
f
e
r
r
i
n
g
t
o
n
e
t
w
o
r
k
e
f
f
e
c
t
s
o
r
n
e
t
w
o
r
k
v
a
l
u
e
.
W
e
s
t
r
u
c
k
u
p
o
n
t
h
e
i
d
e
a
t
h
a
t
i
f
w
e
h
a
d
f
i
f
t
y
j
o
u
r
n
a
l
i
s
t
s
,
a
n
d
t
h
e
y
e
a
c
h
c
r
o
s
s
-
p
r
o
m
o
t
e
d
e
a
c
h
o
t
h
e
r
t
o
t
h
e
i
r
s
o
c
i
a
l
n
e
t
w
o
r
k
s
,
t
h
e
n
o
v
e
r
t
i
m
e
w
e
w
o
u
l
d
g
e
t
m
o
r
e
a
n
d
m
o
r
e
p
e
o
p
l
e
t
o
r
e
a
d
e
a
c
h
o
t
h
e
r
’
s
c
o
n
t
e
n
t
.
p
l
a
t
f
o
r
m
F
o
u
n
d
e
r
s
d
e
s
c
r
i
b
i
n
g
p
l
a
t
f
o
r
m
b
u
s
i
n
e
s
s
m
o
d
e
l
.
F
o
r
t
h
o
s
e
u
n
f
a
m
i
l
i
a
r
w
i
t
h
i
m
e
r
c
i
v
e
,
w
e
w
e
r
e
a
s
o
c
i
a
l
m
e
d
i
a
m
a
r
k
e
t
i
n
g
c
o
m
p
a
n
y
t
h
a
t
p
r
o
v
i
d
e
d
b
r
a
n
d
e
d
i
n
s
t
a
n
t
m
e
s
s
a
g
i
n
g
a
p
p
l
i
c
a
t
i
o
n
s
a
s
n
e
w
t
o
u
c
h
p
o
i
n
t
s
f
o
r
c
o
n
s
u
m
e
r
e
n
g
a
g
e
m
e
n
t
.
u
s
e
r
g
e
n
e
r
a
t
i
o
n
c
o
m
m
u
n
i
t
y
s
t
r
a
t
e
g
y
F
o
u
n
d
e
r
t
a
l
k
s
o
n
a
c
o
m
m
u
n
i
t
y
o
f
u
s
e
r
s
.
T
h
e
a
u
d
i
e
n
c
e
t
o
o
k
r
e
s
p
o
n
s
i
b
i
l
i
t
y
f
o
r
w
h
a
t
w
e
n
t
o
n
a
t
e
a
c
h
l
o
c
a
l
[
s
t
a
r
t
u
p
]
s
i
t
e
a
n
d
d
e
b
a
t
e
d
l
o
c
a
l
i
s
s
u
e
s
i
n
a
c
i
v
i
l
m
a
n
n
e
r
b
e
c
a
u
s
e
i
t
w
a
s
a
b
o
u
t
t
h
e
i
r
c
o
m
m
u
-
n
i
t
y
.
A
n
d
a
s
i
n
t
h
e
i
r
p
h
y
s
i
c
a
l
c
o
m
m
u
n
i
t
y
,
t
h
e
y
w
e
r
e
p
r
o
u
d
o
f
i
t
a
n
d
t
o
o
k
c
a
r
e
o
f
i
t
.
u
g
c
,
c
a
r
e
f
u
l
f
o
r
n
o
t
t
a
k
i
n
g
a
d
v
a
n
t
a
g
e
E
x
p
l
o
i
t
i
n
g
u
s
e
r
s
w
i
l
l
r
e
s
u
l
t
i
n
r
e
s
i
s
t
a
n
c
e
.
S
i
n
c
e
w
e
w
e
r
e
n
’
t
m
a
k
i
n
g
a
n
y
m
o
n
e
y
f
o
r
t
h
e
m
,
a
n
d
i
t
a
p
p
e
a
r
e
d
t
o
t
h
e
m
–
c
o
r
-
r
e
c
t
l
y
–
t
h
a
t
n
o
o
n
e
w
a
s
e
v
e
n
r
e
a
d
i
n
g
t
h
e
i
r
c
o
n
t
e
n
t
,
t
h
e
r
e
w
a
s
n
o
e
a
r
t
h
l
y
r
e
a
s
o
n
t
h
e
y
w
o
u
l
d
k
e
e
p
w
r
i
t
i
n
g
“
f
o
r
u
s
”
.
s
e
a
r
c
h
e
x
t
e
r
n
a
l
i
t
i
e
s
F
o
u
n
d
e
r
s
t
a
l
k
s
o
n
s
e
a
r
c
h
e
n
g
i
n
e
b
e
n
e
f
i
t
s
.
F
o
r
y
e
a
r
s
t
h
e
r
e
w
e
r
e
c
e
r
t
a
i
n
q
u
e
r
i
e
s
a
b
o
u
t
c
e
m
e
n
t
p
r
o
d
u
c
t
i
o
n
a
n
d
u
n
e
m
p
l
o
y
-
m
e
n
t
w
h
e
r
e
w
e
w
e
r
e
t
h
e
f
i
r
s
t
o
r
s
e
c
o
n
d
l
i
n
k
s
.
T
h
e
c
o
r
o
l
l
a
r
y
t
o
t
h
i
s
r
u
l
e
i
s
g
o
o
d
323 323
3
2
4
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
g
o
o
g
l
e
f
o
o
[
w
e
b
s
i
t
e
t
r
a
f
f
i
c
]
w
o
n
’
t
s
a
v
e
y
o
u
.
Y
o
u
n
e
e
d
t
h
a
t
t
r
a
f
f
i
c
t
o
t
r
a
n
s
l
a
t
e
i
n
t
o
a
c
o
m
m
u
n
i
t
y
.
A
v
i
s
i
t
i
s
n
o
t
a
n
i
n
t
e
r
e
s
t
i
n
g
s
t
a
t
i
s
t
i
c
e
s
p
e
c
i
a
l
l
y
i
n
a
b
u
s
i
n
e
s
s
t
h
a
t
r
e
q
u
i
r
e
s
t
h
e
c
o
m
m
u
n
i
t
y
t
o
p
r
o
d
u
c
e
c
o
n
t
e
n
t
.
u
s
e
r
g
e
n
e
r
a
t
i
o
n
,
i
n
c
e
n
t
i
v
e
s
I
n
c
e
n
t
i
v
e
s
t
o
m
o
t
i
v
a
t
e
c
r
o
w
p
a
r
t
i
c
i
p
a
t
i
o
n
.
A
n
d
w
h
e
n
w
e
o
f
f
e
r
e
d
c
o
f
f
e
e
c
a
r
d
s
i
n
e
x
c
h
a
n
g
e
f
o
r
p
o
s
t
s
,
[
s
t
a
r
t
u
p
]
u
s
e
r
s
a
c
t
u
-
a
l
l
y
r
e
b
e
l
l
e
d
.
T
h
e
y
t
h
o
u
g
h
t
w
e
w
e
r
e
m
a
k
i
n
g
t
h
e
s
i
t
e
t
o
o
c
o
m
m
e
r
c
i
a
l
,
e
v
e
n
t
h
o
u
g
h
a
d
s
d
i
d
n
'
t
s
e
e
m
t
o
b
o
t
h
e
r
t
h
e
m
a
t
a
l
l
[
…
]
b
e
c
a
u
s
e
l
o
c
a
l
a
d
s
a
r
e
l
o
c
a
l
i
n
f
o
r
m
a
t
i
o
n
.
v
i
r
a
l
s
t
r
a
t
e
g
y
I
d
e
a
l
p
r
o
d
u
c
t
d
i
f
f
u
s
i
o
n
e
i
t
h
e
r
t
h
r
o
u
g
h
“
b
u
i
l
t
-
i
n
”
v
i
r
a
l
f
e
a
t
u
r
e
s
o
r
f
r
i
c
t
i
o
n
l
e
s
s
s
h
a
r
i
n
g
.
A
n
e
x
c
e
p
t
i
o
n
i
s
i
n
h
e
r
e
n
t
l
y
v
i
r
a
l
i
d
e
a
s
,
e
s
p
e
c
i
a
l
l
y
o
n
e
-
t
o
-
m
a
n
y
v
i
r
a
l
i
t
y
,
w
h
e
r
e
n
o
r
m
a
l
u
s
e
o
f
y
o
u
r
p
r
o
d
u
c
t
r
e
a
c
h
e
s
n
e
w
u
s
e
r
s
,
n
o
t
“
w
o
r
d
-
o
f
-
m
o
u
t
h
”
v
i
r
a
l
t
h
a
t
r
e
q
u
i
r
e
s
u
s
e
r
s
t
o
a
d
v
o
c
a
t
e
y
o
u
.
W
i
t
h
i
n
h
e
r
e
n
t
v
i
r
a
l
i
t
y
,
a
b
a
r
e
l
y
a
d
e
q
u
a
t
e
p
r
o
d
-
u
c
t
m
i
g
h
t
s
u
f
f
i
c
e
,
t
h
o
u
g
h
e
v
e
n
t
h
e
n
m
a
r
k
e
t
i
n
g
s
h
o
u
l
d
a
c
c
e
l
e
r
a
t
e
g
r
o
w
t
h
.
2
_
F
U
N
D
I
N
G
[
r
e
l
a
t
e
s
t
o
P
e
t
e
r
P
a
n
’
s
d
i
l
e
m
m
a
]
m
o
n
e
y
e
x
p
e
n
d
i
t
u
r
e
O
v
e
r
l
y
f
a
s
t
s
p
e
n
d
i
n
g
o
f
f
u
n
d
s
.
[
T
h
e
s
t
a
r
t
u
p
]
w
a
s
f
e
d
-
b
u
t
-
a
n
e
m
i
c
a
n
d
s
l
o
w
l
y
r
o
a
s
t
i
n
g
i
t
s
c
a
s
h
.
D
o
i
t
a
g
a
i
n
a
n
d
I
’
l
l
c
o
n
c
e
n
t
r
a
t
e
t
h
o
s
e
c
o
s
t
s
a
t
t
h
e
b
e
g
i
n
n
i
n
g
.
W
e
w
o
u
l
d
h
a
v
e
h
a
d
t
w
i
c
e
t
h
e
p
r
o
d
u
c
t
i
n
2
/
3
o
f
t
h
e
t
i
m
e
.
m
o
n
e
y
,
e
x
c
e
s
s
G
e
t
t
i
n
g
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
S
o
t
h
e
n
e
t
e
f
f
e
c
t
w
a
s
t
h
a
t
i
t
e
n
a
b
l
e
d
t
o
r
a
i
s
e
a
l
l
t
h
i
s
m
o
n
e
y
t
h
a
t
k
e
p
t
u
s
f
a
r
f
r
o
m
t
h
e
c
u
s
t
o
m
e
r
.
T
r
u
t
h
b
e
t
o
l
d
,
w
e
w
e
r
e
p
r
o
b
a
b
l
y
a
f
r
a
i
d
o
f
c
u
s
t
o
m
e
r
s
a
t
t
h
i
s
p
o
i
n
t
b
e
c
a
u
s
e
w
e
d
i
d
n
’
t
w
a
n
t
t
o
d
i
s
a
p
p
o
i
n
t
t
h
e
m
o
r
l
o
o
k
b
a
d
.
m
o
n
e
y
,
i
r
r
e
l
e
v
a
n
t
F
o
u
n
d
e
r
i
n
d
i
c
a
t
e
s
f
u
n
d
i
n
g
i
s
n
o
t
c
r
i
t
i
c
a
l
.
U
n
l
e
s
s
[
t
h
e
s
t
a
r
t
u
p
]
r
e
a
c
h
e
d
a
n
e
x
p
a
n
s
i
o
n
s
t
a
g
e
,
w
h
e
r
e
t
h
e
s
a
l
e
s
m
o
d
e
l
w
a
s
k
n
o
w
n
a
n
d
r
e
p
e
a
t
a
b
l
e
,
I
’
m
e
v
e
r
-
m
o
r
e
c
o
n
v
i
n
c
e
d
t
h
a
t
r
a
i
s
i
n
g
i
n
s
t
i
t
u
t
i
o
n
a
l
c
a
p
i
-
t
a
l
w
o
u
l
d
n
o
t
h
a
v
e
i
n
c
r
e
a
s
e
d
o
u
r
c
h
a
n
c
e
s
o
f
c
r
e
a
t
i
n
g
v
a
l
u
e
f
o
r
t
h
e
e
x
i
s
t
i
n
g
s
h
a
r
e
h
o
l
d
e
r
s
.
T
h
e
c
o
m
p
a
n
y
w
o
u
l
d
h
a
v
e
s
u
r
v
i
v
e
d
l
o
n
g
e
r
,
b
u
t
t
h
a
t
’
s
n
o
t
t
h
e
g
o
a
l
.
m
o
n
e
y
,
l
a
c
k
L
a
c
k
o
f
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
W
h
e
n
y
o
u
g
o
i
t
a
l
o
n
e
,
y
o
u
’
v
e
g
o
t
t
h
e
f
r
e
e
d
o
m
t
o
d
o
w
h
a
t
e
v
e
r
y
o
u
w
a
n
t
,
b
u
t
i
t
’
s
h
a
r
d
t
o
s
t
a
y
f
o
c
u
s
e
d
w
h
e
n
t
h
e
p
r
o
j
e
c
t
i
s
a
l
w
a
y
s
c
o
m
p
e
t
i
n
g
f
o
r
y
o
u
r
a
t
t
e
n
t
i
o
n
w
i
t
h
o
t
h
e
r
,
u
s
u
a
l
l
y
p
a
y
i
n
g
,
w
o
r
k
.
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
,
m
o
n
e
y
M
o
n
e
y
w
a
s
i
n
d
i
c
a
t
e
d
a
s
a
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
.
T
h
e
s
i
n
g
l
e
b
i
g
g
e
s
t
r
e
a
s
o
n
w
e
a
r
e
c
l
o
s
i
n
g
d
o
w
n
(
a
c
o
m
m
o
n
o
n
e
)
i
s
r
u
n
n
i
n
g
o
u
t
o
f
c
a
s
h
.
D
e
s
p
i
t
e
p
u
t
t
i
n
g
t
h
e
c
o
m
p
a
n
y
i
n
a
n
E
X
T
R
E
M
E
L
Y
l
e
a
n
p
o
s
i
t
i
o
n
,
g
e
n
e
r
-
a
t
i
n
g
r
e
v
e
n
u
e
,
a
n
d
h
o
l
d
i
n
g
o
u
t
a
s
l
o
n
g
a
s
w
e
c
o
u
l
d
,
w
e
d
i
d
n
'
t
h
a
v
e
t
h
e
c
a
s
h
t
o
324 324
3
2
5
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
k
e
e
p
g
o
i
n
g
.
r
e
a
s
o
n
f
o
r
f
a
i
l
u
r
e
,
s
p
e
n
t
t
i
m
e
o
n
f
u
n
d
i
n
g
S
p
e
n
d
i
n
g
t
i
m
e
l
o
o
k
i
n
g
f
o
r
f
u
n
d
i
n
g
h
a
d
a
n
e
g
a
t
i
v
e
i
m
p
a
c
t
.
W
e
s
p
e
n
t
6
m
o
n
t
h
s
r
a
i
s
i
n
g
o
u
r
a
n
g
e
l
r
o
u
n
d
,
a
n
d
f
i
n
a
l
l
y
c
l
o
s
e
d
i
t
a
f
t
e
r
l
a
u
n
c
h
i
n
g
o
u
r
p
u
b
l
i
c
b
e
t
a
i
n
F
e
b
r
u
a
r
y
2
0
0
8
.
W
e
r
a
i
s
e
d
l
e
s
s
t
h
a
n
w
e
s
h
o
u
l
d
h
a
v
e
(
C
h
r
i
s
D
i
x
o
n
n
o
t
e
s
t
h
a
t
t
h
e
w
o
r
s
t
t
h
i
n
g
a
s
e
e
d
-
s
t
a
g
e
c
o
m
p
a
n
y
c
a
n
d
o
i
s
r
a
i
s
e
t
o
o
l
i
t
t
l
e
m
o
n
e
y
a
n
d
o
n
l
y
r
e
a
c
h
p
a
r
t
w
a
y
t
o
a
m
i
l
e
s
t
o
n
e
,
w
h
i
c
h
i
s
g
r
e
a
t
a
d
v
i
c
e
)
.
W
e
s
p
e
n
t
3
m
o
n
t
h
s
g
e
t
t
i
n
g
r
e
a
d
y
f
o
r
r
a
i
s
i
n
g
o
u
r
n
e
x
t
r
o
u
n
d
,
t
h
e
n
6
m
o
n
t
h
s
b
e
a
t
i
n
g
o
u
r
h
e
a
d
a
g
a
i
n
s
t
a
w
a
l
l
t
r
y
i
n
g
t
o
r
a
i
s
e
m
o
n
e
y
d
u
r
i
n
g
t
h
e
f
i
n
a
n
c
i
a
l
c
o
l
l
a
p
s
e
.
2
_
M
A
R
K
E
T
b
a
r
r
i
e
r
s
t
o
a
d
o
p
t
i
o
n
R
e
a
s
o
n
s
w
h
y
u
s
e
r
s
w
o
u
l
d
n
o
t
a
d
o
p
t
t
h
e
p
l
a
t
f
o
r
m
.
A
f
t
e
r
e
v
e
n
t
u
a
l
l
y
d
i
s
c
o
v
e
r
i
n
g
t
h
a
t
s
e
t
u
p
a
n
d
c
o
n
f
i
g
u
r
a
t
i
o
n
w
a
s
o
u
r
p
r
i
m
a
r
y
a
d
o
p
t
i
o
n
p
r
o
b
l
e
m
(
a
n
d
a
f
t
e
r
t
r
y
i
n
g
a
n
d
f
a
i
l
i
n
g
t
o
i
m
p
l
e
m
e
n
t
v
a
r
i
o
u
s
s
t
r
a
t
e
g
i
e
s
t
o
m
a
k
e
i
t
s
i
m
p
l
e
a
n
d
e
a
s
y
)
,
w
e
t
r
i
e
d
t
o
m
o
v
e
t
o
t
h
e
o
t
h
e
r
e
n
d
o
f
t
h
e
s
p
e
c
t
r
u
m
.
c
h
a
n
g
i
n
g
b
e
h
a
v
i
o
r
P
e
r
c
e
i
v
e
d
d
i
f
f
i
c
u
l
t
y
o
f
c
h
a
n
g
i
n
g
b
e
h
a
v
i
o
r
.
I
b
e
g
a
n
t
o
r
e
a
l
i
z
e
t
h
a
t
e
x
p
e
r
t
s
i
n
m
o
s
t
f
i
e
l
d
s
a
r
e
u
s
u
a
l
l
y
p
r
e
t
t
y
b
u
s
y
p
e
o
p
l
e
,
a
n
d
w
i
t
h
t
h
e
e
x
c
e
p
t
i
o
n
o
f
t
e
c
h
n
i
c
a
l
f
o
l
k
s
(
w
h
o
s
p
e
n
d
t
r
e
m
e
n
d
o
u
s
a
m
o
u
n
t
s
o
f
t
i
m
e
i
n
f
r
o
n
t
o
f
c
o
m
p
u
t
e
r
s
)
,
m
o
s
t
o
t
h
e
r
p
e
o
p
l
e
d
o
n
'
t
l
i
k
e
s
i
t
t
i
n
g
i
n
f
r
o
n
t
o
f
o
n
e
,
a
n
d
p
r
o
b
a
b
l
y
w
o
n
'
t
a
n
s
w
e
r
q
u
e
s
t
i
o
n
s
f
o
r
a
f
e
w
m
e
a
s
l
y
b
u
c
k
s
.
c
o
m
p
e
t
i
t
i
o
n
W
h
y
c
o
m
p
e
t
i
n
g
p
l
a
t
f
o
r
m
s
a
r
e
s
u
p
e
-
r
i
o
r
/
i
n
f
e
r
i
o
r
.
W
e
w
e
r
e
o
f
f
e
r
i
n
g
i
n
f
o
r
m
a
t
i
o
n
o
n
g
r
e
a
t
a
l
b
u
m
s
a
n
d
c
o
m
m
u
n
i
t
y
v
o
t
i
n
g
.
B
u
t
o
t
h
e
r
s
i
t
e
s
l
i
k
e
L
a
s
t
.
f
m
a
n
d
H
y
p
e
M
a
c
h
i
n
e
w
e
r
e
o
f
f
e
r
i
n
g
t
h
e
a
c
t
u
a
l
m
u
s
i
c
.
T
h
a
t
w
a
s
a
c
o
m
p
e
t
i
t
i
v
e
a
d
v
a
n
t
a
g
e
t
h
a
t
’
s
h
a
r
d
t
o
b
e
a
t
a
n
d
w
e
l
a
c
k
e
d
a
s
i
g
n
i
f
i
c
a
n
t
u
s
e
r
b
a
s
e
t
o
c
o
n
v
i
n
c
e
e
n
o
u
g
h
p
e
o
p
l
e
l
a
c
k
o
f
h
i
s
t
o
r
y
,
t
r
u
s
t
R
e
p
o
r
t
e
d
c
a
u
s
e
f
o
r
m
i
s
t
r
u
s
t
b
y
c
u
s
t
o
m
e
r
s
.
I
n
c
o
m
m
u
n
i
t
y
-
g
e
n
e
r
a
t
e
d
m
e
d
i
a
,
t
r
u
s
t
i
s
e
v
e
r
y
t
h
i
n
g
.
W
h
e
n
y
o
u
a
s
k
f
o
r
s
u
b
m
i
s
-
s
i
o
n
s
,
c
o
n
t
r
i
b
u
t
o
r
s
g
o
t
h
r
o
u
g
h
a
n
i
n
s
t
a
n
t
i
n
t
e
r
n
a
l
c
a
l
c
u
l
a
t
i
o
n
:
“
D
o
I
t
r
u
s
t
t
h
e
s
e
p
e
o
p
l
e
w
i
t
h
m
y
w
o
r
k
?
”
W
h
e
n
y
o
u
r
s
i
t
e
i
s
b
r
a
n
d
n
e
w
,
y
o
u
’
v
e
g
o
t
n
o
r
e
c
o
r
d
t
o
r
e
l
y
o
n
.
A
n
d
w
i
t
h
m
o
r
e
s
h
a
d
y
“
u
s
e
r
-
g
e
n
e
r
a
t
e
d
c
o
n
t
e
n
t
”
s
c
h
e
m
e
s
p
o
p
p
i
n
g
u
p
e
v
e
r
y
d
a
y
,
p
e
o
p
l
e
h
a
v
e
t
h
e
i
r
d
e
f
e
n
s
e
s
u
p
(
a
s
w
e
l
l
t
h
e
y
s
h
o
u
l
d
)
.
m
a
r
k
e
t
,
s
m
a
l
l
m
a
r
k
e
t
s
y
n
-
d
r
o
m
e
C
o
n
c
l
u
s
i
o
n
o
f
t
h
e
f
o
u
n
d
e
r
t
h
a
t
t
h
e
m
a
r
k
e
t
i
s
t
o
o
s
m
a
l
l
v
i
s
-
à
-
v
i
s
a
m
b
i
t
i
o
n
s
.
G
i
v
e
n
t
h
e
i
n
d
u
s
t
r
y
t
r
e
n
d
s
w
e
w
e
r
e
s
e
e
i
n
g
,
w
e
j
u
s
t
d
i
d
n
'
t
s
e
e
a
l
i
g
h
t
a
t
t
h
e
e
n
d
o
f
t
h
e
t
u
n
n
e
l
w
o
r
t
h
s
u
r
v
i
v
i
n
g
f
o
r
.
T
h
i
s
m
a
r
k
e
t
m
a
y
g
e
t
s
l
i
g
h
t
l
y
b
e
t
t
e
r
,
b
u
t
i
t
'
s
n
o
t
g
o
i
n
g
t
o
b
e
a
b
i
g
c
o
m
p
a
n
y
.
2
_
M
A
R
K
E
T
I
N
G
b
r
a
n
d
i
n
g
B
r
a
n
d
i
n
g
t
h
e
p
l
a
t
f
o
r
m
.
O
u
r
n
a
m
e
,
C
h
u
b
b
y
B
r
a
i
n
,
i
s
m
e
m
o
r
a
b
l
e
,
a
b
i
t
p
l
a
y
f
u
l
a
n
d
a
t
o
u
c
h
i
r
r
e
v
e
r
e
n
t
.
325 325
3
2
6
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
W
e
,
h
o
w
e
v
e
r
,
g
e
n
e
r
a
l
l
y
p
u
t
o
u
t
p
e
r
s
o
n
a
l
i
t
y
-
l
e
s
s
,
f
a
c
t
-
b
a
s
e
d
a
n
a
l
y
s
e
s
t
o
s
h
o
w
h
o
w
g
o
o
d
o
u
r
d
a
t
a
w
a
s
(
t
h
e
r
e
w
e
r
e
s
o
m
e
e
x
c
e
p
t
i
o
n
s
)
.
I
n
s
t
e
a
d
o
f
m
a
k
i
n
g
o
u
r
n
a
m
e
a
n
d
c
o
n
t
e
n
t
/
a
p
p
r
o
a
c
h
c
o
n
s
i
s
t
e
n
t
a
n
d
h
e
n
c
e
a
n
a
s
s
e
t
,
o
u
r
b
r
a
n
d
i
n
g
m
u
l
l
e
t
c
o
n
f
u
s
e
d
p
e
o
p
l
e
.
c
o
m
m
u
n
i
c
a
t
i
n
g
p
r
o
d
u
c
t
,
p
o
s
i
-
t
i
o
n
i
n
g
P
o
s
i
t
i
o
n
i
n
g
t
h
e
p
l
a
t
f
o
r
m
a
g
a
i
n
s
t
c
o
m
p
e
t
i
-
t
i
o
n
.
W
e
d
i
d
n
’
t
d
e
s
c
r
i
b
e
w
h
a
t
w
e
w
a
n
t
e
d
t
o
d
o
c
l
e
a
r
l
y
e
n
o
u
g
h
.
W
h
e
n
I
t
o
l
d
p
e
o
p
l
e
t
h
e
i
d
e
a
i
n
p
e
r
s
o
n
,
t
h
e
y
a
l
w
a
y
s
r
e
a
l
l
y
l
i
k
e
d
i
t
.
B
u
t
w
h
e
n
t
h
e
y
c
a
m
e
t
o
t
h
e
s
i
t
e
,
t
h
e
y
d
i
d
n
’
t
g
e
t
i
t
.
h
i
g
h
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
c
o
s
t
E
x
p
r
e
s
s
e
d
c
o
n
c
e
r
n
o
v
e
r
c
o
s
t
l
y
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
.
W
e
n
e
v
e
r
s
e
r
i
o
u
s
l
y
f
i
g
u
r
e
d
o
u
t
c
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
,
a
n
d
h
a
d
t
r
o
u
b
l
e
g
r
o
w
i
n
g
t
h
r
o
u
g
h
o
u
t
o
u
r
v
a
r
i
o
u
s
p
i
v
o
t
s
.
C
u
s
t
o
m
e
r
a
c
q
u
i
s
i
t
i
o
n
i
s
h
a
r
d
a
n
d
m
o
r
e
e
x
p
e
n
-
s
i
v
e
t
h
a
n
m
o
s
t
p
e
o
p
l
e
r
e
a
l
i
z
e
w
h
e
n
s
t
a
r
t
i
n
g
a
c
o
m
p
a
n
y
.
p
e
r
s
o
n
a
l
s
e
l
l
i
n
g
P
e
r
s
o
n
a
l
s
e
l
l
i
n
g
a
s
a
m
e
a
n
s
t
o
d
r
i
v
e
p
l
a
t
-
f
o
r
m
p
a
r
t
i
c
i
p
a
t
i
o
n
.
T
h
e
p
l
a
n
w
a
s
t
o
p
a
r
t
n
e
r
w
i
t
h
m
a
r
k
e
t
i
n
g
a
g
e
n
c
i
e
s
a
s
w
e
l
l
a
s
s
e
l
l
d
i
r
e
c
t
l
y
t
o
c
l
i
-
e
n
t
s
s
i
m
i
l
a
r
t
o
t
h
e
a
p
p
r
o
a
c
h
t
a
k
e
n
b
y
p
r
o
v
i
d
e
r
s
o
f
c
u
s
t
o
m
b
r
a
n
d
e
d
w
i
d
g
e
t
s
,
F
a
c
e
b
o
o
k
a
p
p
s
,
a
n
d
m
o
b
i
l
e
a
p
p
s
.
T
h
i
s
s
t
r
a
t
e
g
y
e
v
e
n
t
u
a
l
l
y
p
r
o
d
u
c
e
d
s
o
m
e
g
r
e
a
t
r
e
s
u
l
t
s
b
u
t
i
t
w
a
s
a
c
a
s
e
o
f
t
o
o
l
i
t
t
l
e
,
t
o
o
l
a
t
e
.
2
_
P
R
O
D
U
C
T
t
e
c
h
n
o
l
o
g
y
c
h
o
i
c
e
H
o
w
t
e
c
h
n
o
l
o
g
y
c
h
o
i
c
e
a
f
f
e
c
t
e
d
p
l
a
t
f
o
r
m
d
e
v
e
l
o
p
m
e
n
t
.
I
r
e
a
s
o
n
e
d
t
h
a
t
t
h
e
p
l
a
t
f
o
r
m
w
a
s
t
h
e
c
o
r
e
o
f
o
u
r
t
e
c
h
n
o
l
o
g
y
,
a
n
d
w
e
w
e
r
e
a
t
e
c
h
n
o
l
o
g
y
c
o
m
p
a
n
y
,
a
n
d
s
m
a
r
t
t
e
c
h
n
o
l
o
g
y
c
o
m
p
a
n
i
e
s
n
e
e
d
e
d
t
h
e
f
l
e
x
i
b
i
l
i
t
y
t
h
a
t
c
o
m
e
s
f
r
o
m
w
r
i
t
i
n
g
t
h
e
c
o
r
e
o
f
t
h
e
i
r
p
l
a
t
f
o
r
m
t
h
e
m
s
e
l
v
e
s
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
i
s
c
o
u
l
d
o
n
l
y
b
e
c
o
n
s
i
d
e
r
e
d
p
r
e
m
a
t
u
r
e
o
p
t
i
m
i
z
a
t
i
o
n
.
v
a
l
i
d
a
t
i
n
g
i
d
e
a
,
t
e
c
h
n
o
l
o
g
y
E
n
s
u
r
i
n
g
o
r
n
o
t
e
n
s
u
r
i
n
g
t
h
a
t
t
h
e
r
e
i
s
d
e
-
m
a
n
d
f
o
r
t
h
e
p
r
o
d
u
c
t
.
A
n
o
t
h
e
r
c
h
o
i
c
e
w
a
s
t
h
e
d
e
c
i
s
i
o
n
t
o
s
t
a
r
t
w
i
t
h
a
r
c
a
d
e
g
a
m
e
s
,
w
h
i
c
h
w
a
s
d
o
n
e
v
e
r
y
e
a
r
l
y
o
n
(
e
v
e
n
b
e
f
o
r
e
I
j
o
i
n
e
d
,
I
t
h
i
n
k
)
w
i
t
h
o
u
t
m
u
c
h
t
h
o
u
g
h
t
a
t
a
l
l
.
I
t
w
a
s
j
u
s
t
t
h
e
f
i
r
s
t
t
h
i
n
g
t
h
a
t
c
a
m
e
t
o
m
i
n
d
w
h
e
n
w
e
t
h
o
u
g
h
t
o
f
c
a
s
u
a
l
g
a
m
i
n
g
,
s
o
w
e
s
a
i
d
"
L
e
t
'
s
s
t
a
r
t
w
i
t
h
t
h
a
t
,
a
n
d
t
h
e
n
e
x
p
a
n
d
i
n
t
o
o
t
h
e
r
a
r
c
h
e
t
y
p
e
s
w
h
e
n
w
e
h
a
v
e
m
o
r
e
e
x
p
e
r
i
e
n
c
e
.
"
l
a
u
n
c
h
r
e
l
a
t
e
s
t
o
p
i
o
n
e
e
r
’
s
d
i
l
e
m
m
a
f
o
l
l
o
w
e
r
s
a
d
v
a
n
t
a
g
e
s
E
x
a
m
p
l
e
s
o
f
l
a
t
e
-
m
o
v
e
r
’
s
b
e
n
e
f
i
t
s
.
W
e
s
a
b
e
l
a
u
n
c
h
e
d
a
b
o
u
t
1
0
m
o
n
t
h
s
b
e
f
o
r
e
M
i
n
t
.
M
o
r
e
t
h
e
s
h
a
m
e
t
h
a
t
w
e
d
i
d
n
'
t
c
a
p
i
t
a
l
i
z
e
o
n
t
h
a
t
e
a
r
l
y
l
e
a
d
.
T
h
e
r
e
'
s
a
l
o
t
t
o
b
e
s
a
i
d
f
o
r
n
o
t
r
u
s
h
i
n
g
t
o
m
a
r
k
e
t
,
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
i
s
t
a
k
e
s
t
h
e
f
i
r
s
t
e
n
t
r
a
n
t
s
m
a
k
e
.
S
h
i
p
p
i
n
g
a
"
m
i
n
i
m
u
m
v
i
a
b
l
e
p
r
o
d
u
c
t
"
i
m
m
e
d
i
a
t
e
l
y
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
a
r
k
e
t
d
i
r
e
c
t
l
y
m
a
k
e
s
g
o
o
d
s
e
n
s
e
t
o
m
e
,
b
u
t
e
n
g
a
g
i
n
g
w
i
t
h
a
n
d
s
u
p
p
o
r
t
i
n
g
u
s
e
r
s
i
s
a
n
y
t
h
i
n
g
b
u
t
f
r
e
e
.
326 326
3
2
7
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
l
a
u
n
c
h
,
b
e
n
e
f
i
t
s
o
f
e
a
r
l
y
E
x
a
m
p
l
e
s
o
f
e
a
r
l
y
-
m
o
v
e
r
’
s
b
e
n
e
f
i
t
s
.
W
e
t
a
l
k
e
d
a
b
o
u
t
“
r
e
l
e
a
s
e
e
a
r
l
y
/
r
e
l
e
a
s
e
o
f
t
e
n
,
”
b
u
t
w
e
r
e
s
c
a
r
e
d
o
f
l
o
o
k
i
n
g
l
i
k
e
i
d
i
o
t
s
i
n
f
r
o
n
t
o
f
m
a
j
o
r
W
a
l
l
S
t
r
e
e
t
a
n
d
h
e
d
g
e
f
u
n
d
c
l
i
e
n
t
s
.
I
s
i
t
b
e
t
t
e
r
t
o
w
a
i
t
a
b
i
t
b
e
f
o
r
e
r
e
l
e
a
s
i
n
g
t
o
h
a
v
e
a
m
o
r
e
c
o
m
p
e
l
l
i
n
g
p
r
o
d
u
c
t
o
r
t
o
b
e
g
i
n
g
e
t
t
i
n
g
f
e
e
d
-
b
a
c
k
o
n
a
l
e
s
s
i
m
p
r
e
s
s
i
v
e
o
f
f
e
r
i
n
g
?
W
e
c
h
o
s
e
#
1
;
i
n
r
e
t
r
o
s
p
e
c
t
I
t
h
i
n
k
w
e
s
h
o
u
l
d
h
a
v
e
c
h
o
s
e
n
#
2
.
B
y
c
h
o
o
s
i
n
g
t
o
w
a
i
t
w
e
l
o
s
t
o
u
r
i
n
t
i
m
a
c
y
w
i
t
h
t
h
e
c
u
s
t
o
m
e
r
(
M
i
s
t
a
k
e
#
5
a
g
a
i
n
)
,
f
a
l
l
i
n
g
i
n
t
o
t
h
e
c
l
a
s
s
i
c
(
a
s
a
“
g
r
e
e
n
”
e
n
t
r
e
p
r
e
n
e
u
r
I
d
i
d
n
’
t
k
n
o
w
t
h
i
s
,
b
u
t
a
s
a
s
e
a
s
o
n
e
d
f
o
u
r
-
y
e
a
r
v
e
n
t
u
r
e
i
n
v
e
s
t
o
r
I
k
n
o
w
t
h
i
s
n
o
w
)
t
r
a
p
o
f
p
u
r
s
u
i
n
g
a
“
s
c
i
e
n
c
e
p
r
o
j
e
c
t
,
”
n
o
t
b
u
i
l
d
i
n
g
a
c
o
m
m
e
r
c
i
a
l
l
y
s
a
l
a
b
l
e
p
r
o
d
u
c
t
.
l
a
u
n
c
h
,
d
e
l
a
y
e
d
E
x
a
m
p
l
e
s
o
f
l
a
t
e
-
l
a
u
n
c
h
p
r
o
b
l
e
m
s
.
A
t
s
o
m
e
p
o
i
n
t
,
w
e
r
e
a
c
h
e
d
a
n
a
l
m
o
s
t
g
o
-
l
i
v
e
p
r
o
d
u
c
t
.
B
u
t
a
f
t
e
r
s
h
o
w
i
n
g
i
t
t
o
a
f
e
w
V
C
s
,
w
e
r
e
a
l
i
z
e
d
w
e
m
i
s
s
e
d
s
o
m
e
f
e
a
t
u
r
e
s
t
o
m
a
k
e
i
t
t
r
u
l
y
s
o
c
i
a
l
.
I
n
s
t
e
a
d
o
f
r
e
l
e
a
s
i
n
g
t
h
e
b
e
t
a
,
w
e
d
e
c
i
d
e
d
t
o
c
r
e
a
t
e
t
h
e
v
2
o
f
t
h
e
p
r
o
d
u
c
t
.
W
r
o
n
g
.
W
e
w
i
l
l
n
e
v
e
r
f
i
n
i
s
h
t
h
i
s
v
2
(
l
a
c
k
o
f
m
o
n
e
y
,
s
l
o
w
l
y
d
y
i
n
g
m
o
t
i
v
a
t
i
o
n
)
.
l
a
u
n
c
h
,
d
i
s
a
d
v
a
n
t
a
g
e
o
f
e
a
r
l
y
E
x
a
m
p
l
e
s
o
f
e
a
r
l
y
-
l
a
u
n
c
h
p
r
o
b
l
e
m
s
.
A
f
t
e
r
o
u
r
f
u
n
d
i
n
g
,
i
n
s
t
e
a
d
o
f
i
n
v
e
s
t
i
n
g
i
n
t
h
e
p
r
o
d
u
c
t
f
e
a
t
u
r
e
s
t
o
d
e
l
i
v
e
r
o
n
E
v
e
n
t
V
u
e
d
i
r
e
c
t
l
y
d
r
i
v
i
n
g
p
e
o
p
l
e
t
o
e
v
e
n
t
s
(
h
i
n
d
s
i
g
h
t
:
m
i
s
t
a
k
e
)
,
w
e
i
m
m
e
d
i
a
t
e
l
y
w
e
n
t
o
u
t
a
n
d
t
r
i
e
d
t
o
s
e
l
l
t
h
e
s
o
c
i
a
l
n
e
t
w
o
r
k
t
o
o
l
t
o
c
o
n
f
e
r
e
n
c
e
o
r
g
a
n
i
z
e
r
s
.
2
_
T
E
A
M
h
o
b
b
y
,
h
a
l
f
-
h
e
a
r
t
e
d
e
f
f
o
r
t
I
n
d
i
c
a
t
i
o
n
o
f
n
o
n
-
s
e
r
i
o
u
s
b
u
s
i
n
e
s
s
.
I
t
t
u
r
n
e
d
s
o
m
e
t
h
i
n
g
I
l
i
k
e
d
d
o
i
n
g
i
n
t
o
a
c
h
o
r
e
m
a
i
n
l
y
b
e
c
a
u
s
e
a
t
t
h
e
s
a
m
e
t
i
m
e
I
w
a
s
b
u
s
y
m
a
r
k
e
t
i
n
g
t
h
e
s
i
t
e
,
r
e
d
e
s
i
g
n
i
n
g
i
t
,
a
n
d
a
t
t
e
n
d
i
n
g
c
l
a
s
s
e
s
.
(
I
w
a
s
i
n
c
o
l
l
e
g
e
f
o
r
b
u
s
i
n
e
s
s
f
o
r
1
8
h
o
u
r
s
o
f
t
h
e
w
e
e
k
)
.
i
n
t
e
r
n
a
l
i
z
a
t
i
o
n
I
n
d
i
c
a
t
e
s
p
r
o
f
o
u
n
d
u
n
d
e
r
s
t
a
n
d
i
n
g
o
f
b
u
s
i
-
n
e
s
s
l
o
g
i
c
.
B
u
t
n
o
m
a
t
t
e
r
h
o
w
m
a
n
y
t
i
m
e
s
y
o
u
r
e
a
d
o
r
h
e
a
r
c
e
r
t
a
i
n
l
e
s
s
o
n
s
,
y
o
u
h
a
v
e
t
o
d
o
i
t
y
o
u
r
s
e
l
f
t
o
r
e
a
l
l
y
u
n
d
e
r
s
t
a
n
d
.
l
a
c
k
o
f
f
o
c
u
s
M
e
n
t
i
o
n
e
d
l
a
c
k
o
f
f
o
c
u
s
(
r
e
l
a
t
e
s
t
o
p
i
v
o
t
d
i
l
e
m
m
a
)
.
W
e
m
a
d
e
t
h
e
p
r
o
d
u
c
t
m
u
c
h
t
o
o
b
r
o
a
d
s
u
c
h
t
h
a
t
i
t
d
i
d
a
l
i
t
t
l
e
b
i
t
o
f
e
v
e
r
y
t
h
i
n
g
.
U
n
f
o
r
t
u
n
a
t
e
l
y
,
i
t
d
i
d
n
’
t
d
o
a
n
y
t
h
i
n
g
e
x
c
e
p
t
i
o
n
a
l
l
y
w
e
l
l
.
A
s
a
n
e
x
a
m
p
l
e
,
w
h
e
n
w
e
u
s
e
d
i
t
t
o
r
e
p
l
a
c
e
o
u
r
p
h
p
B
B
m
e
s
s
a
g
e
b
o
a
r
d
i
n
s
t
a
n
c
e
,
o
u
r
c
o
m
m
u
n
i
t
y
w
a
s
i
n
a
n
u
p
r
o
a
r
b
e
c
a
u
s
e
i
t
l
a
c
k
e
d
m
a
n
y
o
f
s
p
e
c
i
a
l
i
z
e
d
f
e
a
t
u
r
e
s
a
m
a
t
u
r
e
f
o
r
u
m
o
f
f
e
r
s
(
e
.
g
.
n
o
t
i
f
i
c
a
t
i
o
n
s
a
n
d
h
a
n
d
l
i
n
g
c
o
d
e
e
x
a
m
p
l
e
s
i
n
a
c
o
m
m
e
n
t
)
.
l
e
a
r
n
i
n
g
g
a
i
n
s
M
e
n
t
i
o
n
e
d
l
e
a
r
n
i
n
g
b
e
n
e
f
i
t
s
d
u
e
t
o
s
t
a
r
t
u
p
(
r
e
l
a
t
e
s
t
o
p
l
a
t
f
o
r
m
s
p
e
c
i
f
i
c
i
t
y
)
.
A
s
a
m
u
s
i
c
s
i
t
e
i
t
h
a
s
h
a
s
n
’
t
e
x
a
c
t
l
y
“
f
a
i
l
e
d
”
a
t
t
h
e
m
o
m
e
n
t
,
i
t
s
t
i
l
l
p
u
l
l
s
i
n
a
r
o
u
n
d
3
0
0
0
p
e
o
p
l
e
a
m
o
n
t
h
m
a
i
n
l
y
f
r
o
m
s
e
a
r
c
h
e
n
g
i
n
e
s
.
I
n
o
w
c
o
n
s
i
d
e
r
i
t
a
h
o
b
b
y
s
i
t
e
a
n
d
I
’
m
l
o
o
k
i
n
g
f
o
r
w
a
r
d
t
o
a
p
p
l
y
i
n
g
w
h
a
t
I
l
e
a
r
n
e
d
w
i
t
h
m
y
n
e
x
t
s
t
a
r
t
-
u
p
,
I
n
t
e
g
r
a
t
e
.
327 327
3
2
8
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
m
a
r
k
e
t
m
e
m
o
r
y
R
e
f
e
r
r
i
n
g
t
o
s
i
m
u
l
t
a
n
e
o
u
s
s
i
m
i
l
a
r
s
t
r
a
t
e
g
i
c
d
e
c
i
s
i
o
n
s
t
h
a
t
a
r
e
b
a
s
e
d
o
n
c
o
m
m
o
n
i
n
f
o
r
-
m
a
t
i
o
n
(
r
e
l
a
t
e
s
t
o
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
)
.
A
s
T
w
i
t
t
e
r
r
e
c
e
i
v
e
d
m
o
r
e
a
t
t
e
n
t
i
o
n
,
c
l
o
n
e
s
s
h
o
w
e
d
u
p
e
v
e
r
y
w
h
e
r
e
.
M
y
d
e
c
i
s
i
o
n
w
a
s
n
o
t
t
o
c
o
m
p
e
t
e
w
i
t
h
a
n
e
w
a
n
d
a
l
r
e
a
d
y
c
r
o
w
d
e
d
s
p
a
c
e
,
b
u
t
t
o
f
i
n
d
a
d
i
f
f
e
r
-
e
n
t
a
n
g
l
e
.
I
d
e
c
i
d
e
d
t
o
f
o
c
u
s
o
n
s
c
a
l
a
b
l
e
t
e
c
h
n
o
l
o
g
y
,
m
a
k
i
n
g
a
b
e
t
t
h
a
t
o
n
c
e
T
w
i
t
t
e
r
a
n
d
o
t
h
e
r
s
w
i
l
l
g
e
t
p
o
p
u
l
a
r
,
t
h
e
r
e
w
i
l
l
b
e
a
r
e
a
l
n
e
e
d
f
o
r
t
h
a
t
k
i
n
d
o
f
t
e
c
h
n
o
l
o
g
y
b
y
o
t
h
e
r
p
l
a
y
e
r
s
.
m
o
t
i
v
e
s
f
o
r
s
h
a
r
i
n
g
,
w
h
y
t
e
l
l
s
t
o
r
y
M
o
t
i
v
e
f
o
r
w
r
i
t
i
n
g
p
o
s
t
-
m
o
r
t
e
m
.
I
r
e
a
d
p
o
s
t
s
o
n
H
a
c
k
e
r
N
e
w
s
a
b
o
u
t
“
W
h
y
m
y
s
t
a
r
t
-
u
p
f
a
i
l
e
d
”
a
n
d
f
o
u
n
d
t
h
e
m
i
n
t
e
r
e
s
t
i
n
g
b
e
c
a
u
s
e
i
t
’
s
a
r
a
w
s
t
a
r
t
-
u
p
s
t
o
r
y
n
o
t
t
h
e
f
a
i
r
y
t
a
l
e
s
t
h
a
t
m
a
g
a
z
i
n
e
s
l
i
k
e
t
o
w
r
i
t
e
a
b
o
u
t
.
S
o
h
e
r
e
i
s
m
y
s
h
o
t
a
t
i
t
.
s
e
c
o
n
d
o
p
i
n
i
o
n
,
i
n
t
e
r
p
r
e
t
a
t
i
o
n
O
t
h
e
r
c
o
m
m
e
n
t
s
o
n
t
h
e
s
t
a
r
t
u
p
i
n
a
d
d
i
t
i
o
n
t
o
t
h
o
s
e
o
f
t
h
e
f
o
u
n
d
e
r
.
I
w
a
s
p
r
o
b
a
b
l
y
w
h
a
t
y
o
u
w
o
u
l
d
c
a
l
l
a
n
e
a
r
l
y
C
h
i
c
a
g
o
a
d
o
p
t
e
r
o
f
M
e
e
t
r
o
,
a
n
d
I
t
h
o
u
g
h
t
i
t
w
a
s
a
g
r
e
a
t
i
d
e
a
,
a
n
d
s
t
i
l
l
d
o
.
W
i
t
h
e
a
c
h
s
u
c
c
e
e
d
i
n
g
v
e
r
s
i
o
n
o
f
t
h
e
s
o
f
t
w
a
r
e
,
h
o
w
e
v
e
r
,
t
h
e
i
n
t
e
r
f
a
c
e
s
e
e
m
e
d
t
o
g
e
t
w
e
i
r
d
e
r
a
n
d
w
e
i
r
d
e
r
u
n
t
i
l
I
c
o
u
l
d
n
’
t
f
i
g
u
r
e
i
t
o
u
t
a
n
y
m
o
r
e
.
s
e
l
f
-
a
t
t
r
i
b
u
t
i
o
n
F
o
u
n
d
e
r
a
s
s
i
g
n
i
n
g
b
l
a
m
e
o
f
f
a
i
l
u
r
e
t
o
s
e
l
f
.
A
s
c
o
-
f
o
u
n
d
e
r
a
n
d
C
E
O
o
f
L
o
o
k
e
r
y
,
t
h
e
b
u
c
k
s
t
o
p
s
w
i
t
h
m
e
a
n
d
n
o
o
n
e
e
l
s
e
.
s
t
a
r
t
u
p
f
a
t
i
g
u
e
S
e
n
t
i
m
e
n
t
o
f
f
r
u
s
t
r
a
t
i
o
n
o
v
e
r
m
a
n
y
h
a
r
d
-
s
h
i
p
s
(
r
e
l
a
t
e
s
t
o
‘
r
u
n
w
a
y
’
)
W
e
c
o
u
l
d
h
a
v
e
g
o
n
e
a
b
o
u
t
t
r
y
i
n
g
t
o
f
i
x
M
e
e
t
r
o
b
u
t
t
h
e
t
e
a
m
w
a
s
j
u
s
t
r
e
a
d
y
t
o
m
o
v
e
o
n
.
3
_
T
H
E
O
R
Y
A
N
D
C
O
N
-
S
T
R
U
C
T
S
[
t
h
i
s
s
e
c
t
i
o
n
i
n
c
l
u
d
e
s
d
i
l
e
m
m
a
s
]
d
i
l
e
m
m
a
s
c
o
l
d
s
t
a
r
t
I
n
a
b
i
l
i
t
y
t
o
g
e
t
c
o
n
t
e
n
t
w
i
t
h
o
u
t
u
s
e
r
s
.
U
n
d
e
r
e
s
t
i
m
a
t
e
d
t
h
e
“
C
o
l
d
S
t
a
r
t
”
p
r
o
b
l
e
m
,
I
r
e
a
d
t
h
i
s
a
r
t
i
c
l
e
b
y
B
o
k
a
d
o
S
o
c
i
a
l
D
e
s
i
g
n
w
h
i
c
h
t
a
l
k
s
a
b
o
u
t
a
b
i
g
i
s
s
u
e
y
o
u
f
a
c
e
w
i
t
h
a
s
o
c
i
a
l
s
i
t
e
,
e
s
p
e
c
i
a
l
l
y
w
h
e
n
i
t
r
e
l
i
e
s
o
n
u
s
e
r
-
g
e
n
e
r
a
t
e
d
c
o
n
t
e
n
t
.
T
h
e
v
a
l
u
e
y
o
u
p
r
o
v
i
d
e
t
o
y
o
u
r
u
s
e
r
s
c
e
n
t
e
r
s
a
r
o
u
n
d
t
h
e
c
o
n
t
e
n
t
o
n
t
h
e
s
i
t
e
,
s
o
t
o
b
u
i
l
d
a
u
s
e
r
-
b
a
s
e
y
o
u
n
e
e
d
a
l
o
t
o
f
c
o
n
t
e
n
t
c
r
e
a
t
e
d
b
y
t
h
e
f
i
r
s
t
u
s
e
r
s
t
o
k
i
c
k
-
o
f
f
t
h
e
c
o
m
m
u
n
i
t
y
.
l
o
n
e
l
y
u
s
e
r
I
n
a
b
i
l
i
t
y
t
o
g
e
t
u
s
e
r
s
w
i
t
h
o
u
t
o
t
h
e
r
u
s
e
r
s
.
[
I
]
f
s
o
m
e
o
n
e
w
a
s
n
’
t
o
n
l
i
n
e
w
h
e
n
y
o
u
w
e
r
e
o
n
l
i
n
e
,
t
h
e
y
w
e
r
e
n
o
g
o
o
d
t
o
y
o
u
.
W
h
i
l
e
t
h
e
r
e
a
l
-
t
i
m
e
c
h
a
t
a
s
p
e
c
t
o
f
t
h
e
a
p
p
l
i
c
a
t
i
o
n
m
a
d
e
f
o
r
s
o
m
e
r
e
a
l
l
y
s
e
r
e
n
-
d
i
p
i
t
o
u
s
m
e
e
t
i
n
g
s
,
i
t
a
l
s
o
m
a
d
e
i
t
h
a
r
d
e
r
f
o
r
p
e
o
p
l
e
t
o
g
a
u
g
e
t
h
e
a
c
t
i
v
i
t
y
o
f
t
h
e
i
r
c
o
m
m
u
n
i
t
i
e
s
,
e
s
p
e
c
i
a
l
l
y
i
f
t
h
e
y
l
o
g
g
e
d
i
n
a
t
o
d
d
h
o
u
r
s
,
p
e
o
p
l
e
w
e
r
e
s
e
t
a
s
a
w
a
y
,
e
t
c
.
m
o
n
e
t
i
z
a
t
i
o
n
I
n
a
b
i
l
i
t
y
t
o
c
h
a
r
g
e
m
o
n
e
y
a
n
d
g
e
t
u
s
e
r
s
.
F
o
r
f
o
u
r
y
e
a
r
s
w
e
h
a
v
e
o
f
f
e
r
e
d
t
h
e
s
y
n
c
h
r
o
n
i
z
a
t
i
o
n
s
e
r
v
i
c
e
f
o
r
n
o
c
h
a
r
g
e
,
328 328
3
2
9
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
p
r
e
d
i
c
a
t
e
d
o
n
t
h
e
h
y
p
o
t
h
e
s
i
s
t
h
a
t
a
b
u
s
i
n
e
s
s
m
o
d
e
l
w
o
u
l
d
e
m
e
r
g
e
t
o
s
u
p
p
o
r
t
t
h
e
f
r
e
e
s
e
r
v
i
c
e
.
W
i
t
h
t
h
a
t
i
n
v
e
s
t
m
e
n
t
t
h
e
s
i
s
t
h
w
a
r
t
e
d
,
t
h
e
r
e
i
s
n
o
w
a
y
t
o
p
a
y
e
x
-
p
e
n
s
e
s
,
p
r
i
m
a
r
i
l
y
s
a
l
a
r
y
a
n
d
h
o
s
t
i
n
g
c
o
s
t
s
.
W
i
t
h
o
u
t
t
h
e
r
e
s
o
u
r
c
e
s
t
o
k
e
e
p
t
h
e
s
e
r
v
i
c
e
g
o
i
n
g
,
w
e
m
u
s
t
s
h
u
t
i
t
d
o
w
n
.
p
e
t
e
r
p
a
n
I
n
a
b
i
l
i
t
y
t
o
g
r
o
w
w
i
t
h
o
u
t
f
u
n
d
i
n
g
.
W
h
e
n
y
o
u
’
v
e
g
o
t
m
o
n
e
y
,
y
o
u
c
a
n
t
a
k
e
m
o
r
e
t
i
m
e
t
o
d
o
i
t
r
i
g
h
t
,
b
u
t
y
o
u
’
v
e
g
o
t
h
i
g
h
e
r
e
x
p
e
c
t
a
t
i
o
n
s
f
o
r
r
e
t
u
r
n
s
,
a
n
d
a
w
h
o
l
e
h
o
s
t
o
f
o
t
h
e
r
c
o
m
p
l
i
c
a
t
i
o
n
s
.
W
h
e
n
y
o
u
g
o
i
t
a
l
o
n
e
,
y
o
u
’
v
e
g
o
t
t
h
e
f
r
e
e
d
o
m
t
o
d
o
w
h
a
t
e
v
e
r
y
o
u
w
a
n
t
,
b
u
t
i
t
’
s
h
a
r
d
t
o
s
t
a
y
f
o
c
u
s
e
d
w
h
e
n
t
h
e
p
r
o
j
e
c
t
i
s
a
l
w
a
y
s
c
o
m
p
e
t
i
n
g
f
o
r
y
o
u
r
a
t
t
e
n
t
i
o
n
w
i
t
h
o
t
h
e
r
,
u
s
u
a
l
l
y
p
a
y
i
n
g
,
w
o
r
k
.
p
i
v
o
t
I
m
p
o
s
s
i
b
l
e
t
o
c
h
a
n
g
e
t
h
e
c
o
u
r
s
e
o
f
b
u
s
i
n
e
s
s
b
u
t
a
l
s
o
t
o
k
e
e
p
i
t
.
[
T
]
h
e
m
o
s
t
s
i
g
n
i
f
i
c
a
n
t
c
h
a
n
g
e
,
w
h
i
c
h
I
d
i
d
n
o
t
a
p
p
r
e
c
i
a
t
e
u
n
t
i
l
i
t
w
a
s
t
o
o
l
a
t
e
,
w
a
s
t
h
e
f
u
n
d
a
m
e
n
t
a
l
s
h
i
f
t
i
n
t
h
e
n
a
t
u
r
e
o
f
t
h
e
s
t
a
r
t
u
p
.
I
m
o
v
e
d
f
r
o
m
b
u
i
l
d
i
n
g
a
c
o
n
s
u
m
e
r
-
f
a
c
i
n
g
w
e
b
s
i
t
e
t
o
a
c
o
m
b
i
n
a
t
i
o
n
o
f
i
n
f
r
a
s
t
r
u
c
t
u
r
e
p
r
o
v
i
d
e
r
a
n
d
s
o
f
t
-
w
a
r
e
d
e
v
e
l
o
p
e
r
w
h
i
c
h
m
a
d
e
t
h
e
o
v
e
r
h
e
a
d
c
o
s
t
m
u
c
h
h
i
g
h
e
r
,
t
i
m
e
t
o
m
a
r
k
e
t
l
o
n
g
e
r
,
a
n
d
r
e
s
o
u
r
c
e
s
m
u
c
h
m
o
r
e
d
i
f
f
i
c
u
l
t
t
o
f
i
n
d
.
r
e
m
o
r
a
s
c
u
r
s
e
J
o
i
n
i
n
g
a
p
l
a
t
f
o
r
m
r
e
t
u
r
n
s
u
s
e
r
s
b
u
t
g
i
v
e
s
a
w
a
y
p
o
w
e
r
.
W
e
e
x
p
o
s
e
d
o
u
r
s
e
l
v
e
s
t
o
a
h
u
g
e
s
i
n
g
l
e
p
o
i
n
t
o
f
f
a
i
l
u
r
e
c
a
l
l
e
d
F
a
c
e
b
o
o
k
.
I
’
v
e
r
a
n
t
e
d
f
o
r
y
e
a
r
s
a
b
o
u
t
h
o
w
b
a
d
a
n
i
d
e
a
i
t
i
s
f
o
r
s
t
a
r
t
u
p
s
t
o
b
e
m
o
b
i
l
e
-
c
a
r
r
i
e
r
d
e
p
e
n
d
e
n
t
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
e
r
e
i
s
n
o
d
i
f
f
e
r
e
n
c
e
b
e
t
w
e
e
n
V
e
r
i
z
o
n
W
i
r
e
l
e
s
s
a
n
d
F
a
c
e
b
o
o
k
i
n
t
h
i
s
c
o
n
t
e
x
t
.
i
n
d
u
s
t
r
y
-
s
p
e
c
i
f
i
c
p
r
o
b
l
e
m
A
p
r
o
b
l
e
m
t
h
a
t
i
s
s
p
e
c
i
f
i
c
t
o
a
m
a
r
k
e
t
o
r
i
n
d
u
s
t
r
y
.
I
k
n
o
w
t
h
e
r
e
’
s
a
p
l
a
c
e
f
o
r
a
w
i
s
d
o
m
o
f
c
r
o
w
d
s
a
p
p
r
o
a
c
h
i
n
t
h
e
p
u
b
l
i
s
h
i
n
g
w
o
r
l
d
,
b
u
t
p
u
b
l
i
s
h
e
r
s
a
n
d
a
r
t
i
s
t
s
a
r
e
s
t
i
l
l
f
i
g
u
r
i
n
g
i
t
o
u
t
.
F
o
r
n
o
w
,
i
t
s
e
e
m
s
l
i
k
e
a
d
i
r
e
c
t
s
u
b
m
i
s
s
i
o
n
m
o
d
e
l
i
s
w
h
a
t
p
e
o
p
l
e
a
r
e
m
o
s
t
c
o
m
f
o
r
t
a
b
l
e
w
i
t
h
.
p
r
o
b
l
e
m
o
f
a
c
t
i
v
e
u
s
e
K
e
e
p
i
n
g
u
s
e
r
s
a
c
t
i
v
e
a
f
t
e
r
r
e
g
i
s
t
e
r
i
n
g
.
L
o
o
k
c
a
r
e
f
u
l
l
y
a
t
m
o
s
t
h
y
p
e
r
-
l
o
c
a
l
s
i
t
e
s
a
n
d
s
e
e
j
u
s
t
h
o
w
m
u
c
h
p
o
s
t
i
n
g
i
s
r
e
a
l
l
y
b
e
i
n
g
d
o
n
e
,
e
s
p
e
c
i
a
l
l
y
b
y
m
e
m
b
e
r
s
o
f
t
h
e
c
o
m
m
u
n
i
t
y
a
s
o
p
p
o
s
e
d
t
o
t
h
e
s
i
t
e
s
'
o
p
e
r
a
t
o
r
s
.
A
n
y
b
o
d
y
w
h
o
h
a
s
r
u
n
a
h
y
p
e
r
-
l
o
c
a
l
s
i
t
e
w
i
l
l
t
e
l
l
y
o
u
t
h
a
t
i
t
t
a
k
e
s
a
c
o
u
p
l
e
o
f
y
e
a
r
s
j
u
s
t
t
o
g
e
t
t
o
a
p
o
i
n
t
w
h
e
r
e
y
o
u
'
v
e
t
r
u
l
y
g
o
t
a
v
i
b
r
a
n
t
o
n
l
i
n
e
c
o
m
m
u
n
i
t
y
.
i
l
l
u
s
i
o
n
s
,
f
a
l
l
a
c
i
e
s
a
n
d
b
i
a
s
e
s
b
u
i
l
d
i
t
a
n
d
t
h
e
y
w
i
l
l
c
o
m
e
E
x
p
e
c
t
i
n
g
t
h
e
p
r
o
d
u
c
t
t
o
“
s
e
l
l
i
t
s
e
l
f
”
.
W
e
f
e
l
l
i
n
t
o
t
h
e
“
b
u
i
l
d
i
t
a
n
d
t
h
e
y
w
i
l
l
”
c
o
m
e
s
c
h
o
o
l
o
f
t
h
o
u
g
h
t
(
a
l
t
h
o
u
g
h
e
v
e
n
w
h
e
n
t
h
e
y
c
a
m
e
,
w
e
s
t
i
l
l
w
e
r
e
n
’
t
i
n
g
o
o
d
s
h
a
p
e
)
.
U
s
e
r
s
d
i
d
n
’
t
r
e
v
i
e
w
b
e
c
a
u
s
e
t
h
e
r
e
w
a
s
n
o
e
n
l
i
g
h
t
e
n
e
d
s
e
l
f
-
i
n
t
e
r
e
s
t
f
o
r
t
h
e
m
t
o
d
o
s
o
.
N
o
b
o
d
y
w
a
n
t
e
d
t
o
e
d
i
t
o
u
r
d
a
t
a
f
o
r
t
h
e
s
a
m
e
r
e
a
s
o
n
.
P
e
o
p
l
e
s
u
b
m
i
t
t
e
d
c
o
m
p
a
n
i
e
s
t
o
o
u
r
d
a
t
a
b
a
s
e
329 329
3
3
0
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
m
a
i
n
l
y
l
o
o
k
i
n
g
f
o
r
a
l
i
n
k
.
d
o
g
f
o
o
d
b
l
i
n
d
n
e
s
s
R
e
f
u
s
i
n
g
t
o
a
d
m
i
t
f
l
a
w
s
i
n
t
h
e
p
r
o
d
u
c
t
.
T
h
e
o
r
i
g
i
n
a
l
C
h
u
b
b
y
B
r
a
i
n
w
a
s
a
h
a
p
p
y
m
a
r
r
i
a
g
e
o
f
o
u
r
i
n
t
e
r
e
s
t
s
w
r
a
p
p
e
d
i
n
c
u
s
t
o
m
e
r
i
n
d
i
f
f
e
r
e
n
c
e
.
W
e
n
e
v
e
r
t
a
l
k
e
d
t
o
u
s
e
r
s
/
c
u
s
t
o
m
e
r
s
t
o
s
e
e
i
f
t
h
i
s
w
o
u
l
d
b
e
o
f
i
n
t
e
r
e
s
t
t
o
t
h
e
m
.
A
n
d
s
o
c
u
s
t
o
m
e
r
s
u
l
t
i
m
a
t
e
l
y
t
o
l
d
u
s
t
h
r
o
u
g
h
t
h
e
i
r
i
n
d
i
f
-
f
e
r
e
n
c
e
t
h
a
t
t
h
e
y
d
i
d
n
’
t
c
a
r
e
a
b
o
u
t
o
u
r
i
d
e
a
.
c
o
n
f
i
r
m
a
t
i
o
n
b
i
a
s
S
e
e
k
i
n
g
p
o
s
i
t
i
v
e
f
e
e
d
b
a
c
k
a
n
d
i
g
n
o
r
i
n
g
n
e
g
a
t
i
v
e
o
n
e
.
M
a
n
w
a
s
I
s
t
u
p
i
d
.
W
h
e
n
p
e
o
p
l
e
a
s
k
e
d
t
o
s
i
g
n
u
p
w
i
t
h
o
u
t
i
t
,
I
t
o
l
d
t
h
e
m
n
o
.
W
h
e
n
t
h
e
p
e
o
p
l
e
w
h
o
d
i
d
s
i
g
n
u
p
w
e
r
e
w
o
r
r
i
e
d
a
b
o
u
t
t
h
i
n
g
s
b
e
i
n
g
p
o
s
t
e
d
t
o
t
h
e
i
r
w
a
l
l
s
,
I
d
i
d
n
’
t
u
n
d
e
r
s
t
a
n
d
t
h
e
p
r
o
b
l
e
m
.
W
h
e
n
r
e
a
d
e
r
s
s
a
i
d
t
h
e
y
w
o
u
l
d
n
’
t
s
i
g
n
u
p
t
o
c
o
m
m
e
n
t
,
I
t
h
o
u
g
h
t
i
t
w
a
s
j
u
s
t
a
s
m
a
l
l
m
i
n
o
r
i
t
y
.
p
e
r
f
e
c
t
p
r
o
d
u
c
t
f
a
l
l
a
c
y
R
e
f
u
s
a
l
t
o
l
a
u
n
c
h
u
n
t
i
l
t
h
e
p
r
o
d
u
c
t
i
s
p
e
r
-
f
e
c
t
(
w
h
i
c
h
i
s
t
h
e
n
n
e
v
e
r
i
s
)
.
A
n
d
t
h
e
s
e
b
a
d
b
e
h
a
v
i
o
r
s
w
e
r
e
r
e
i
n
f
o
r
c
e
d
b
y
a
n
u
n
p
l
a
n
n
e
d
e
v
e
n
t
t
h
a
t
s
h
a
r
p
l
y
i
m
p
a
c
t
e
d
o
u
r
p
s
y
c
h
e
:
b
e
i
n
g
o
n
t
h
e
f
r
o
n
t
p
a
g
e
o
f
t
h
e
F
i
n
a
n
c
i
a
l
T
i
m
e
s
.
I
t
i
s
h
a
r
d
t
o
c
a
l
l
i
t
a
m
i
s
t
a
k
e
s
i
n
c
e
w
e
d
i
d
n
’
t
s
e
e
k
t
o
g
e
t
s
u
c
h
e
x
p
o
s
u
r
e
,
b
u
t
I
p
u
t
i
t
d
o
w
n
a
s
M
i
s
t
a
k
e
#
3
.
T
o
b
e
h
o
n
e
s
t
,
t
h
i
s
s
i
n
g
l
e
f
a
c
t
w
a
s
a
v
e
r
y
m
e
a
n
i
n
g
f
u
l
f
a
c
t
o
r
i
n
o
u
r
f
a
i
l
u
r
e
.
I
t
r
a
i
s
e
d
t
h
e
l
e
v
e
l
o
f
e
x
p
e
c
t
a
t
i
o
n
s
s
o
h
i
g
h
t
h
a
t
i
t
m
a
d
e
u
s
r
e
l
u
c
t
a
n
t
t
o
r
e
l
e
a
s
e
a
n
y
t
h
i
n
g
t
h
a
t
w
a
s
n
’
t
e
a
r
t
h
-
s
h
a
t
t
e
r
i
n
g
.
i
l
l
u
s
i
o
n
o
f
f
r
e
e
E
x
p
e
c
t
i
n
g
a
p
r
i
o
r
i
t
h
a
t
c
u
s
t
o
m
e
r
s
w
o
u
l
d
n
o
t
p
a
y
.
V
i
d
e
o
p
r
o
d
u
c
e
r
s
a
r
e
a
f
r
a
i
d
o
f
c
h
a
r
g
i
n
g
f
o
r
c
o
n
t
e
n
t
,
b
e
c
a
u
s
e
t
h
e
y
d
o
n
'
t
t
h
i
n
k
p
e
o
p
l
e
w
i
l
l
p
a
y
.
A
n
d
t
h
e
y
'
r
e
l
a
r
g
e
l
y
r
i
g
h
t
.
C
o
n
s
u
m
e
r
s
s
t
i
l
l
d
o
n
'
t
l
i
k
e
p
a
y
i
n
g
f
o
r
s
t
u
f
f
,
p
e
r
i
o
d
.
i
t
w
o
n
’
t
h
a
p
p
e
n
t
o
m
e
T
e
n
d
e
n
c
y
t
o
e
x
p
e
c
t
t
h
a
t
r
i
s
k
w
i
l
l
n
o
t
r
e
a
l
i
z
e
i
n
o
w
n
s
t
a
r
t
u
p
.
W
e
s
a
w
t
h
a
t
t
h
i
s
d
i
d
n
’
t
y
e
t
e
x
i
s
t
i
n
a
n
e
a
s
y
t
o
u
s
e
f
o
r
m
a
t
.
W
e
s
a
w
o
t
h
e
r
s
t
h
a
t
h
a
d
t
r
i
e
d
a
n
d
w
e
f
i
g
u
r
e
d
t
h
e
y
h
a
d
f
a
i
l
e
d
b
e
c
a
u
s
e
t
h
e
y
d
i
d
a
b
a
d
j
o
b
a
n
d
w
e
c
o
u
l
d
d
o
b
e
t
t
e
r
.
T
h
e
l
e
s
s
o
n
t
o
l
e
a
r
n
h
e
r
e
i
s
I
f
t
h
e
r
e
a
r
e
b
u
r
n
i
n
g
h
u
s
k
s
o
f
c
a
r
s
a
l
o
n
g
t
h
e
r
o
a
d
w
a
y
,
y
o
u
s
h
o
u
l
d
p
r
o
b
a
b
l
y
a
s
s
u
m
e
t
h
a
t
d
r
i
v
i
n
g
t
h
e
r
e
i
s
d
a
n
g
e
r
-
o
u
s
.
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
E
m
p
l
o
y
i
n
g
f
a
m
o
u
s
s
u
c
c
e
s
s
s
t
o
r
i
e
s
a
s
a
m
o
d
e
l
f
o
r
d
e
c
i
s
i
o
n
s
r
e
g
a
r
d
i
n
g
o
n
e
’
s
o
w
n
s
t
a
r
t
u
p
.
[
W
]
e
t
h
o
u
g
h
t
“
l
e
t
’
s
t
a
k
e
e
l
e
m
e
n
t
s
f
r
o
m
s
i
t
e
s
w
e
l
i
k
e
a
n
d
t
w
e
a
k
t
h
e
m
”
a
n
d
w
e
’
l
l
g
e
t
t
h
e
s
a
m
e
m
a
g
i
c
a
l
e
f
f
e
c
t
s
o
n
o
u
r
s
i
t
e
t
h
a
t
t
h
e
y
’
v
e
g
o
t
t
e
n
.
W
r
o
n
g
.
F
e
a
t
u
r
e
s
d
o
n
’
t
w
o
r
k
i
n
a
v
a
c
u
u
m
.
T
h
e
y
w
o
r
k
b
e
c
a
u
s
e
y
o
u
t
a
k
e
t
i
m
e
t
o
u
n
d
e
r
s
t
a
n
d
y
o
u
r
c
u
s
t
o
m
e
r
a
n
d
t
h
e
n
b
u
i
l
d
f
e
a
t
u
r
e
s
t
o
a
c
c
o
m
m
o
d
a
t
e
t
h
e
m
.
c
o
p
y
b
u
s
i
n
e
s
s
m
o
d
e
l
t
o
n
e
w
c
o
n
t
e
x
t
E
x
p
l
i
c
i
t
a
m
b
i
t
i
o
n
t
o
c
o
p
y
e
l
e
m
e
n
t
s
o
f
a
n
-
o
t
h
e
r
b
u
s
i
n
e
s
s
m
o
d
e
l
i
n
t
o
a
n
e
w
i
n
d
u
s
t
r
y
.
T
h
e
i
d
e
a
i
t
s
e
l
f
f
o
r
C
o
n
t
r
a
s
t
r
e
a
m
w
a
s
n
’
t
t
o
o
i
n
n
o
v
a
t
i
v
e
o
r
o
r
i
g
i
n
a
l
.
I
t
w
a
s
i
n
-
s
p
i
r
e
d
b
y
D
i
g
g
.
c
o
m
w
h
i
c
h
h
a
d
a
p
p
l
i
e
d
t
h
e
s
a
m
e
m
o
d
e
l
t
o
n
e
w
s
a
n
d
a
r
t
i
c
l
e
s
o
n
t
h
e
W
e
b
.
I
s
t
i
l
l
b
e
l
i
e
v
e
a
p
p
l
y
i
n
g
t
h
i
s
m
o
d
e
l
t
o
m
u
s
i
c
i
s
i
n
t
e
r
e
s
t
i
n
g
a
n
d
u
s
e
f
u
l
,
b
u
t
t
h
e
r
e
w
e
r
e
s
o
m
a
n
y
o
t
h
e
r
m
e
-
t
o
o
D
i
g
g
s
i
t
e
s
.
330 330
3
3
1
N
a
m
e
G
u
i
d
e
E
x
a
m
p
l
e
s
u
n
k
c
o
s
t
f
a
l
l
a
c
y
R
e
f
u
s
i
n
g
t
o
m
a
k
e
c
h
a
n
g
e
s
d
u
e
t
o
t
i
m
e
s
p
e
n
t
.
S
i
n
c
e
t
h
e
s
e
r
v
i
c
e
w
a
s
o
u
r
c
h
i
l
d
w
e
w
e
r
e
r
e
l
u
c
t
a
n
t
t
o
m
a
k
e
a
d
e
c
i
s
i
o
n
a
b
o
u
t
c
l
o
s
i
n
g
i
t
f
a
s
t
e
r
a
n
d
l
i
m
i
t
l
o
s
s
e
s
.
W
e
’
v
e
b
e
e
n
t
r
i
c
k
i
n
g
o
u
r
s
e
l
v
e
s
t
h
i
n
k
i
n
g
t
h
a
t
e
v
e
r
y
t
h
i
n
g
w
o
u
l
d
b
e
f
i
n
e
w
h
i
l
e
w
e
c
o
u
l
d
n
’
t
g
e
t
t
h
e
a
p
p
l
i
c
a
t
i
o
n
b
a
c
k
t
o
w
o
r
k
p
r
o
p
e
r
l
y
.
t
e
c
h
n
o
l
o
g
y
b
i
a
s
T
e
n
d
e
n
c
y
t
o
b
e
l
i
e
v
e
t
h
a
t
t
e
c
h
n
o
l
o
g
y
w
i
l
l
s
o
l
v
e
b
u
s
i
n
e
s
s
p
r
o
b
l
e
m
s
.
I
n
s
t
e
a
d
o
f
h
a
v
i
n
g
p
r
o
d
u
c
t
m
a
n
a
g
e
m
e
n
t
a
s
t
h
e
a
d
v
o
c
a
t
e
f
o
r
t
h
e
c
u
s
t
o
m
e
r
a
n
d
t
h
e
p
r
o
d
u
c
t
e
v
a
n
g
e
l
i
s
t
,
w
e
h
a
d
t
e
c
h
n
o
l
o
g
y
r
u
n
n
i
n
g
t
h
e
s
h
o
w
i
n
a
v
a
c
u
u
m
.
H
u
g
e
m
i
s
t
a
k
e
.
T
h
i
s
a
l
l
o
w
e
d
u
s
t
o
p
e
r
p
e
t
u
a
t
e
t
h
e
s
c
i
e
n
c
e
p
r
o
j
e
c
t
f
o
r
m
u
c
h
,
m
u
c
h
l
o
n
g
e
r
t
h
a
n
w
e
s
h
o
u
l
d
h
a
v
e
.
r
a
c
e
t
o
b
o
t
t
o
m
R
e
l
a
t
e
s
t
o
c
o
m
p
e
t
i
t
i
v
e
p
r
e
s
s
u
r
e
t
o
o
f
f
e
r
a
f
r
e
e
p
r
o
d
u
c
t
.
W
e
a
l
s
o
c
o
n
s
i
d
e
r
e
d
r
e
f
o
c
u
s
i
n
g
X
m
a
r
k
s
a
s
a
f
r
e
e
m
i
u
m
s
y
n
c
b
u
s
i
n
e
s
s
,
b
u
t
t
h
e
p
r
o
s
p
e
c
t
s
t
h
e
r
e
a
r
e
g
r
i
m
t
o
o
:
w
i
t
h
t
h
e
e
m
e
r
g
e
n
c
e
o
f
c
o
m
p
e
t
e
n
t
s
y
n
c
f
e
a
t
u
r
e
s
b
u
i
l
t
i
n
t
o
M
o
z
i
l
l
a
F
i
r
e
f
o
x
a
n
d
G
o
o
g
l
e
C
h
r
o
m
e
,
i
t
’
s
h
a
r
d
t
o
s
e
e
u
s
e
r
s
p
a
y
i
n
g
f
o
r
a
s
e
r
v
i
c
e
t
h
a
t
t
h
e
y
c
a
n
n
o
w
g
e
t
f
o
r
f
r
e
e
.
r
u
n
w
a
y
I
n
d
i
c
a
t
e
s
t
i
m
e
r
u
n
n
i
n
g
o
u
t
.
B
i
g
c
o
r
p
o
r
a
t
i
o
n
s
c
a
n
a
f
f
o
r
d
t
o
s
p
e
n
d
a
b
i
l
l
i
o
n
d
o
l
l
a
r
s
a
n
d
w
a
i
t
1
0
y
e
a
r
s
,
t
h
o
u
g
h
,
w
h
i
l
e
s
t
a
r
t
u
p
s
u
s
u
a
l
l
y
c
a
n
o
n
l
y
a
f
f
o
r
d
a
f
e
w
t
e
n
s
o
f
t
h
o
u
s
a
n
d
s
a
n
d
a
b
o
u
t
6
m
o
n
t
h
s
.
Y
o
u
h
a
v
e
a
b
o
u
t
z
e
r
o
c
h
a
n
c
e
o
f
b
u
i
l
d
i
n
g
a
n
d
g
e
t
t
i
n
g
p
e
o
p
l
e
t
o
a
d
o
p
t
a
p
l
a
t
f
o
r
m
i
n
6
m
o
n
t
h
s
.
p
a
t
h
d
e
p
e
n
d
e
n
c
e
A
l
t
e
r
n
a
t
i
v
e
e
x
p
l
a
n
a
t
i
o
n
f
o
r
r
e
f
u
s
a
l
t
o
m
a
k
e
c
h
a
n
g
e
s
(
r
e
l
a
t
e
s
t
o
‘
s
u
n
k
c
o
s
t
f
a
l
l
a
c
y
’
)
.
T
h
e
i
n
i
t
i
a
l
i
d
e
a
d
o
e
s
c
h
a
n
g
e
,
a
n
d
i
t
'
s
a
l
m
o
s
t
c
e
r
t
a
i
n
l
y
w
r
o
n
g
.
T
h
e
t
h
i
n
g
i
s
,
t
h
e
i
n
i
t
i
a
l
i
d
e
a
d
e
t
e
r
m
i
n
e
s
h
o
w
t
h
e
i
n
i
t
i
a
l
i
d
e
a
w
i
l
l
c
h
a
n
g
e
,
w
h
i
c
h
i
s
c
r
u
c
i
a
l
t
o
a
l
l
t
h
e
e
x
e
c
u
t
i
o
n
t
h
a
t
f
o
l
l
o
w
s
i
t
.
A
n
d
t
h
e
d
e
v
i
l
i
s
o
f
t
e
n
i
n
t
h
e
d
e
t
a
i
l
s
:
l
i
t
t
l
e
d
e
c
i
s
i
o
n
s
m
a
d
e
e
a
r
l
y
o
n
c
a
n
c
r
u
c
i
a
l
l
y
i
m
p
a
c
t
b
i
g
d
e
c
i
s
i
o
n
s
m
a
d
e
l
a
t
e
r
.
t
r
a
n
s
f
e
r
e
n
c
e
p
r
o
b
l
e
m
R
e
a
l
i
z
a
t
i
o
n
t
h
a
t
f
a
m
o
u
s
e
x
a
m
p
l
e
s
d
o
n
o
t
a
p
p
l
y
i
n
a
l
l
c
a
s
e
s
(
r
e
l
a
t
e
s
t
o
‘
r
e
f
e
r
e
n
c
e
p
o
i
n
t
b
i
a
s
’
)
.
A
m
a
z
o
n
p
r
o
v
e
d
b
e
y
o
n
d
a
n
y
d
o
u
b
t
t
h
a
t
t
h
i
s
i
s
a
v
a
l
i
d
a
n
d
p
o
w
e
r
f
u
l
b
u
s
i
n
e
s
s
m
o
d
e
l
,
b
u
t
v
e
r
y
f
e
w
c
o
m
p
a
n
i
e
s
c
a
n
a
f
f
o
r
d
t
o
b
u
i
l
d
s
u
c
h
a
s
e
r
v
i
c
e
a
n
d
t
h
e
n
w
a
i
t
f
o
r
p
e
o
p
l
e
t
o
c
o
m
e
.
I
t
a
l
s
o
h
e
l
p
s
t
h
a
t
A
m
a
z
o
n
d
i
d
n
o
t
s
t
a
r
t
w
i
t
h
t
h
e
g
o
a
l
o
f
p
r
o
v
i
d
i
n
g
t
h
e
s
e
s
e
r
v
i
c
e
s
,
b
u
t
i
d
e
n
t
i
f
i
e
d
a
n
o
p
p
o
r
t
u
n
i
t
y
o
n
c
e
t
h
e
i
r
o
w
n
r
e
t
a
i
l
i
n
f
r
a
s
t
r
u
c
t
u
r
e
w
a
s
i
n
p
l
a
c
e
.
N
B
.
S
o
m
e
c
o
d
e
s
a
r
e
o
m
i
t
t
e
d
f
o
r
p
a
r
s
i
m
o
n
y
.
331 331
332
APPENDIX 2 IS THE COLD START DILEMMA
REALLY A DILEMMA?
Strictly speaking, any chicken-and-egg dilemma is not a strategic dilemma in
the sense of a contradictory decision-making situation, as it does not involve a
strategic choice, but a dependency. In its base form, it is a rather a question of
“how to attract the first party to the platform?” rather than “which one of the
two negative outcomes should I choose?” However, it is possible to formulate
it as a dilemma that will satisfy the definition, without assuming too much
context and therefore losing generality. This is achieved by introducing two
conflicting conditions:
1. If the startup provides the content, user generation effects will not
take place as the startup has provided the content.
2. If the startup does not provide the content, user generation effects
will not take place as there is no content.
Hence, the strategic action would be to create content or not, and both out-
comes would result in a “cold platform”.
Behind these conditions, there is the premise that users only create content
because other users have previously created content; for example, Wikipedia
only exists because it is Encyclopedia Britannica. Note that the problem was
framed so that there is a dependence on user generation (UG): that is, that us-
ers will take charge of content production in the long run. This excludes first-
party content platforms, such as Spotify that utilizes a music library provided
by the record labels, whereas an indie music platform needs to attract user-
generated music from indie artists. This restriction is based on the theoretical
model of ideal UG (see Chapter 3.4).
A second alternative would be to frame the dilemma in the question:
“Which party should the startup focus first?”, whereby focus on A would ne-
glect B which is required, and vice versa. This, however, would introduce
contextual factors unless strong assumptions were made concerning the feasi-
bility of parties, the applied monetization model, and also the propensity to
produce and generate content. For example, clearly the startup should focus on
those parties who are most likely to generate content; however, this is not a
strategic problem but, rather, one of identification (i.e., how they are to be
found).
Furthermore, the cold start dilemma, as defined here, is irrelevant if UG
does not lead to the virtuous cycle: What happens if user-generated content
does not lead to the creation of new content?
3
3
3
A
P
P
E
N
D
I
X
3
S
U
P
P
O
R
T
F
O
R
E
A
R
L
Y
A
N
D
L
A
T
E
L
A
U
N
C
H
E
S
T
o
o
e
a
r
l
y
T
o
o
l
a
t
e
1
.
"
C
u
t
e
h
a
c
k
s
c
a
n
c
o
s
t
y
o
u
t
i
m
e
.
T
a
k
e
t
h
e
t
i
m
e
t
o
d
o
t
h
i
n
g
s
r
i
g
h
t
f
r
o
m
t
h
e
b
e
g
i
n
n
i
n
g
.
"
(
W
i
l
s
o
n
2
0
0
6
)
.
"
T
h
e
b
i
g
g
e
s
t
f
a
l
l
o
u
t
o
f
b
u
i
l
d
i
n
g
o
u
r
p
l
a
t
f
o
r
m
o
u
r
s
e
l
v
e
s
w
a
s
t
h
a
t
w
e
c
o
u
l
d
n
’
t
b
u
i
l
d
q
u
i
c
k
l
y
e
n
o
u
g
h
.
W
h
e
n
y
o
u
r
o
l
l
y
o
u
r
o
w
n
i
n
f
r
a
s
t
r
u
c
t
u
r
e
,
e
v
e
r
y
t
h
i
n
g
t
a
k
e
s
t
i
m
e
,
m
o
r
e
t
i
m
e
t
h
a
n
y
o
u
c
a
n
a
f
f
o
r
d
.
A
n
d
w
e
h
a
d
p
r
o
m
i
s
e
d
t
h
e
j
o
u
r
n
a
l
i
s
t
s
t
h
a
t
w
e
w
o
u
l
d
v
e
r
y
q
u
i
c
k
l
y
b
u
i
l
d
a
l
a
r
g
e
l
i
s
t
o
f
f
e
a
t
u
r
e
s
,
n
o
n
e
o
f
w
h
i
c
h
w
e
r
e
p
r
o
d
u
c
e
d
n
e
a
r
l
y
q
u
i
c
k
l
y
e
n
o
u
g
h
.
T
h
i
s
w
a
s
t
h
e
m
a
j
o
r
c
a
u
s
e
o
f
d
i
s
i
l
l
u
s
i
o
n
m
e
n
t
–
w
e
o
v
e
r
-
p
r
o
m
i
s
e
d
a
n
d
u
n
d
e
r
-
d
e
l
i
v
e
r
e
d
[
…
]
.
"
(
B
i
g
g
a
r
2
0
1
0
)
.
2
.
"
T
h
e
r
e
’
s
a
l
o
t
t
o
b
e
s
a
i
d
f
o
r
n
o
t
r
u
s
h
i
n
g
t
o
m
a
r
k
e
t
,
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
i
s
t
a
k
e
s
t
h
e
f
i
r
s
t
e
n
t
r
a
n
t
s
m
a
k
e
.
S
h
i
p
p
i
n
g
a
'
m
i
n
i
m
u
m
v
i
a
b
l
e
p
r
o
d
u
c
t
'
i
m
m
e
d
i
a
t
e
l
y
a
n
d
l
e
a
r
n
i
n
g
f
r
o
m
t
h
e
m
a
r
k
e
t
d
i
r
e
c
t
l
y
m
a
k
e
s
g
o
o
d
s
e
n
s
e
t
o
m
e
,
b
u
t
e
n
g
a
g
i
n
g
w
i
t
h
a
n
d
s
u
p
p
o
r
t
i
n
g
u
s
e
r
s
i
s
a
n
y
t
h
i
n
g
b
u
t
f
r
e
e
.
O
b
s
e
r
-
v
a
t
i
o
n
c
a
n
b
e
c
h
e
a
p
e
r
.
M
i
n
t
(
a
n
d
s
o
m
e
o
t
h
e
r
s
)
d
i
d
w
e
l
l
b
y
s
e
e
i
n
g
w
h
e
r
e
w
e
s
c
r
e
w
e
d
u
p
,
a
n
d
w
a
i
t
i
n
g
t
o
l
a
u
n
c
h
u
n
t
i
l
t
h
e
y
h
a
d
a
b
e
t
t
e
r
a
p
p
r
o
a
c
h
.
"
(
H
e
d
l
u
n
d
2
0
1
0
)
.
"
E
v
e
n
t
h
o
u
g
h
I
w
a
s
v
e
r
y
c
a
r
e
f
u
l
a
b
o
u
t
s
c
o
p
e
a
n
d
o
u
r
f
o
c
u
s
o
n
p
r
o
d
u
c
t
,
I
u
n
d
e
r
e
s
-
t
i
m
a
t
e
d
t
h
e
t
i
m
e
i
t
w
o
u
l
d
t
a
k
e
t
o
g
e
t
t
o
a
V
1
a
n
d
t
h
e
n
i
n
j
e
c
t
i
t
i
n
t
o
t
h
e
m
a
r
k
e
t
.
"
(
S
m
i
t
h
2
0
0
8
)
.
3
.
"
A
f
t
e
r
o
u
r
f
u
n
d
i
n
g
,
i
n
s
t
e
a
d
o
f
i
n
v
e
s
t
i
n
g
i
n
t
h
e
p
r
o
d
u
c
t
f
e
a
t
u
r
e
s
t
o
d
e
l
i
v
e
r
o
n
E
v
e
n
t
V
u
e
d
i
r
e
c
t
l
y
d
r
i
v
i
n
g
p
e
o
p
l
e
t
o
e
v
e
n
t
s
(
h
i
n
d
s
i
g
h
t
:
m
i
s
t
a
k
e
)
,
w
e
i
m
m
e
d
i
a
t
e
l
y
w
e
n
t
o
u
t
a
n
d
t
r
i
e
d
t
o
s
e
l
l
t
h
e
s
o
c
i
a
l
n
e
t
w
o
r
k
t
o
o
l
t
o
c
o
n
f
e
r
-
e
n
c
e
o
r
g
a
n
i
z
e
r
s
.
"
(
J
o
h
n
s
o
n
&
F
r
a
s
e
r
2
0
1
0
)
.
"
W
e
t
r
i
e
d
t
o
b
u
i
l
d
t
h
e
u
l
t
i
m
a
t
e
A
J
A
X
c
a
l
e
n
d
a
r
a
l
l
a
t
o
n
c
e
.
I
t
t
o
o
k
a
l
o
n
g
t
i
m
e
.
W
e
c
o
u
l
d
h
a
v
e
d
o
n
e
i
t
p
i
e
c
e
b
y
p
i
e
c
e
.
"
(
K
a
n
2
0
0
6
)
.
4
.
"
[
T
]
r
i
e
d
t
o
b
u
i
l
d
a
s
a
l
e
s
e
f
f
o
r
t
t
o
o
e
a
r
l
y
,
w
i
t
h
t
o
o
w
e
a
k
o
f
a
p
r
o
d
u
c
t
a
f
t
e
r
i
n
i
t
i
a
l
f
i
n
a
n
c
i
n
g
[
…
]
.
"
(
J
o
h
n
s
o
n
&
F
r
a
s
e
r
2
0
1
0
)
.
"
T
o
o
m
u
c
h
m
o
n
e
y
i
s
l
i
k
e
t
o
o
m
u
c
h
t
i
m
e
;
w
o
r
k
e
x
p
a
n
d
s
t
o
f
i
l
l
t
h
e
t
i
m
e
a
l
l
o
t
t
e
d
,
a
n
d
w
a
y
s
t
o
s
p
e
n
d
m
o
n
e
y
m
u
l
t
i
p
l
y
w
h
e
n
a
b
u
n
d
a
n
t
f
i
n
a
n
c
i
a
l
r
e
s
o
u
r
c
e
s
a
r
e
a
v
a
i
l
a
b
l
e
.
B
y
b
e
i
n
g
s
i
m
p
l
y
t
o
o
g
o
o
d
a
t
r
a
i
s
i
n
g
m
o
n
e
y
,
i
t
e
n
a
b
l
e
d
u
s
t
o
p
e
r
p
e
t
u
a
t
e
p
o
o
r
o
r
g
a
n
i
z
a
t
i
o
n
a
l
s
t
r
u
c
t
u
r
e
a
n
d
s
u
b
o
p
t
i
m
a
l
s
t
r
a
t
e
g
i
c
d
e
c
i
s
i
o
n
s
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
5
.
"
W
e
l
a
u
n
c
h
e
d
t
o
o
f
a
s
t
.
I
’
m
a
b
i
g
b
e
l
i
e
v
e
r
i
n
l
a
u
n
c
h
f
a
s
t
,
g
e
t
f
e
e
d
b
a
c
k
,
m
a
k
e
c
h
a
n
g
e
s
.
W
e
l
a
u
n
c
h
e
d
f
a
s
t
,
g
o
t
m
o
r
e
f
e
e
d
b
a
c
k
t
h
a
n
w
e
c
o
u
l
d
h
a
n
-
d
l
e
,
a
n
d
f
a
i
l
e
d
t
o
m
a
k
e
c
h
a
n
g
e
s
.
Y
o
u
c
a
n
’
t
e
x
p
e
c
t
p
e
o
p
l
e
t
o
w
a
i
t
f
o
r
y
o
u
t
o
g
e
t
i
t
r
i
g
h
t
.
"
(
P
o
w
a
z
e
k
2
0
0
8
)
.
"
W
e
t
a
l
k
e
d
a
b
o
u
t
'
r
e
l
e
a
s
e
e
a
r
l
y
/
r
e
l
e
a
s
e
o
f
t
e
n
,
'
b
u
t
w
e
r
e
s
c
a
r
e
d
o
f
l
o
o
k
i
n
g
l
i
k
e
i
d
i
o
t
s
i
n
f
r
o
n
t
o
f
m
a
j
o
r
W
a
l
l
S
t
r
e
e
t
a
n
d
h
e
d
g
e
f
u
n
d
c
l
i
e
n
t
s
.
[
…
]
B
y
c
h
o
o
s
i
n
g
t
o
w
a
i
t
w
e
l
o
s
t
o
u
r
i
n
t
i
m
a
c
y
w
i
t
h
t
h
e
c
u
s
t
o
m
e
r
[
…
]
f
a
l
l
i
n
g
i
n
t
o
t
h
e
c
l
a
s
s
i
c
[
…
]
t
r
a
p
o
f
p
u
r
s
u
i
n
g
a
'
s
c
i
e
n
c
e
p
r
o
j
e
c
t
,
'
n
o
t
b
u
i
l
d
i
n
g
a
c
o
m
m
e
r
c
i
a
l
l
y
s
a
l
a
b
l
e
p
r
o
d
u
c
t
.
"
333 333
3
3
4
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
6
.
"
T
h
a
t
o
l
d
c
l
i
c
h
é
a
b
o
u
t
n
o
t
g
e
t
t
i
n
g
a
s
e
c
o
n
d
c
h
a
n
c
e
t
o
m
a
k
e
a
f
i
r
s
t
i
m
p
r
e
s
s
i
o
n
?
C
o
r
n
y
b
u
t
t
r
u
e
.
W
h
e
n
y
o
u
s
t
u
m
b
l
e
o
u
t
o
f
t
h
e
g
a
t
e
,
i
t
c
a
n
b
e
h
a
r
d
t
o
r
e
g
a
i
n
y
o
u
r
f
o
o
t
i
n
g
.
W
e
s
h
o
u
l
d
h
a
v
e
d
o
n
e
m
o
r
e
t
e
s
t
i
n
g
o
f
b
o
t
h
t
h
e
c
o
r
e
i
d
e
a
a
n
d
t
h
e
s
i
t
e
i
t
s
e
l
f
.
"
(
P
o
w
a
z
e
k
2
0
0
8
)
.
"
A
n
d
t
h
e
s
e
b
a
d
b
e
h
a
v
i
o
r
s
w
e
r
e
r
e
i
n
f
o
r
c
e
d
b
y
a
n
u
n
p
l
a
n
n
e
d
e
v
e
n
t
t
h
a
t
s
h
a
r
p
l
y
i
m
p
a
c
t
e
d
o
u
r
p
s
y
c
h
e
:
b
e
i
n
g
o
n
t
h
e
f
r
o
n
t
p
a
g
e
o
f
t
h
e
F
i
n
a
n
c
i
a
l
T
i
m
e
s
.
I
t
i
s
h
a
r
d
t
o
c
a
l
l
i
t
a
m
i
s
t
a
k
e
s
i
n
c
e
w
e
d
i
d
n
’
t
s
e
e
k
t
o
g
e
t
s
u
c
h
e
x
p
o
s
u
r
e
,
b
u
t
I
p
u
t
i
t
d
o
w
n
a
s
M
i
s
t
a
k
e
#
3
.
T
o
b
e
h
o
n
e
s
t
,
t
h
i
s
s
i
n
g
l
e
f
a
c
t
w
a
s
a
v
e
r
y
m
e
a
n
i
n
g
f
u
l
f
a
c
t
o
r
i
n
o
u
r
f
a
i
l
u
r
e
.
I
t
r
a
i
s
e
d
t
h
e
l
e
v
e
l
o
f
e
x
p
e
c
t
a
t
i
o
n
s
s
o
h
i
g
h
t
h
a
t
i
t
m
a
d
e
u
s
r
e
l
u
c
t
a
n
t
t
o
r
e
l
e
a
s
e
a
n
y
t
h
i
n
g
t
h
a
t
w
a
s
n
’
t
e
a
r
t
h
-
s
h
a
t
t
e
r
i
n
g
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
7
.
"
L
o
o
k
e
r
y
’
s
P
r
o
f
i
l
e
S
a
a
S
/
u
n
i
v
e
r
s
a
l
c
o
o
k
i
e
m
e
c
h
a
n
i
s
m
i
s
f
a
r
m
o
r
e
e
c
o
n
o
m
i
c
a
n
d
e
f
f
e
c
t
i
v
e
t
h
a
n
c
o
o
k
i
e
e
x
c
h
a
n
g
e
s
y
s
t
e
m
s
i
n
a
w
o
r
l
d
w
h
e
r
e
a
d
m
e
d
i
a
a
n
d
t
a
r
g
e
t
i
n
g
d
a
t
a
a
r
e
s
e
p
a
r
a
t
e
c
o
m
m
o
d
i
t
i
e
s
.
T
h
a
t
w
o
r
l
d
i
s
a
y
e
a
r
o
r
m
o
r
e
i
n
t
h
e
f
u
t
u
r
e
.
"
(
R
a
f
e
r
2
0
0
9
)
.
"
I
t
t
o
o
k
u
s
2
4
m
o
n
t
h
s
t
o
f
a
i
l
.
2
y
e
a
r
s
.
T
h
a
t
’
s
a
l
i
f
e
t
i
m
e
i
n
w
e
b
s
t
a
n
d
a
r
d
s
.
W
e
s
h
o
u
l
d
h
a
v
e
l
a
u
n
c
h
e
d
s
o
m
e
t
h
i
n
g
(
e
v
e
n
u
g
l
y
)
i
n
t
h
e
f
i
r
s
t
6
m
o
n
t
h
s
,
a
n
d
p
u
l
l
e
d
t
h
e
p
l
u
g
a
f
t
e
r
1
2
m
o
n
t
h
s
.
D
o
n
’
t
w
a
s
t
e
t
i
m
e
a
n
d
m
o
n
e
y
o
n
a
l
o
n
g
l
a
s
t
i
n
g
p
r
o
j
e
c
t
i
f
y
o
u
d
o
n
’
t
s
e
e
t
h
e
e
n
d
o
f
t
h
e
t
u
n
n
e
l
.
"
(
K
r
e
t
z
2
0
1
0
)
.
8
.
"
[
O
]
u
r
i
n
i
t
i
a
l
a
r
c
h
e
t
y
p
e
c
h
o
i
c
e
b
e
c
a
m
e
t
h
e
b
a
s
i
s
f
o
r
o
u
r
p
r
o
d
u
c
t
.
M
e
a
n
-
w
h
i
l
e
,
m
o
s
t
o
f
t
h
e
g
r
o
w
t
h
i
n
t
h
e
c
a
s
u
a
l
g
a
m
i
n
g
i
n
d
u
s
t
r
y
h
a
s
b
e
e
n
i
n
p
u
z
-
z
l
e
&
w
o
r
d
g
a
m
e
s
,
s
o
w
e
w
e
r
e
p
o
s
i
t
i
o
n
e
d
w
r
o
n
g
f
r
o
m
t
h
e
s
t
a
r
t
.
"
(
T
a
n
g
2
0
0
8
)
.
"
T
h
e
t
e
c
h
n
o
l
o
g
y
d
e
c
i
s
i
o
n
t
o
u
s
e
C
+
+
w
a
s
w
e
l
l
f
o
u
n
d
e
d
.
H
o
w
e
v
e
r
,
i
t
f
a
i
l
e
d
t
o
t
a
k
e
i
n
t
o
a
c
c
o
u
n
t
,
o
r
a
t
l
e
a
s
t
p
r
e
p
a
r
e
f
o
r
t
h
e
i
n
e
v
i
t
a
b
l
e
r
e
a
l
i
t
y
o
f
b
e
i
n
g
u
n
a
b
l
e
t
o
f
i
n
d
p
e
o
p
l
e
t
o
j
o
i
n
m
e
.
A
n
y
m
a
j
o
r
w
e
b
c
o
m
p
a
n
y
r
e
a
c
h
i
n
g
m
a
s
s
i
v
e
s
c
a
l
e
e
v
e
n
t
u
a
l
l
y
t
u
r
n
s
t
o
C
/
C
+
+
f
o
r
p
e
r
f
o
r
m
a
n
c
e
.
G
i
v
e
n
t
h
e
f
a
c
t
t
h
a
t
N
o
u
n
c
e
r
h
a
d
n
o
c
o
m
p
e
t
i
t
i
o
n
a
t
a
l
l
w
h
e
n
t
h
e
i
d
e
a
w
a
s
f
i
r
s
t
c
o
n
c
e
i
v
e
d
[
.
.
.
]
,
I
d
e
c
i
d
e
d
t
o
t
a
k
e
a
l
i
t
t
l
e
b
i
t
l
o
n
g
e
r
b
e
f
o
r
e
l
a
u
n
c
h
i
n
g
t
o
b
u
i
l
d
a
s
o
l
i
d
f
o
u
n
d
a
t
i
o
n
.
T
h
i
s
,
o
f
c
o
u
r
s
e
,
t
u
r
n
e
d
u
p
a
s
a
b
i
g
m
i
s
t
a
k
e
–
b
u
t
i
t
i
s
f
a
r
f
r
o
m
c
l
e
a
r
t
h
a
t
I
w
o
u
l
d
h
a
v
e
b
e
e
n
f
i
r
s
t
t
o
m
a
r
k
e
t
e
i
t
h
e
r
w
a
y
.
"
(
H
a
m
m
e
r
2
0
0
8
)
.
9
.
"
A
s
a
p
i
o
n
e
e
r
i
n
t
h
e
u
s
e
r
-
g
e
n
e
r
a
t
e
d
,
h
y
p
e
r
-
l
o
c
a
l
f
i
e
l
d
,
B
a
c
k
f
e
n
c
e
h
o
p
e
f
u
l
l
y
w
i
l
l
p
a
v
e
t
h
e
w
a
y
f
o
r
m
a
n
y
o
t
h
e
r
e
f
f
o
r
t
s
t
o
c
r
e
a
t
e
l
o
c
a
l
l
y
f
o
c
u
s
e
d
o
n
l
i
n
e
c
o
m
m
u
n
i
t
i
e
s
t
h
a
t
u
l
t
i
m
a
t
e
l
y
w
i
l
l
b
e
c
o
m
e
p
r
o
f
i
t
a
b
l
e
b
u
s
i
n
e
s
s
e
s
.
"
(
P
o
t
t
s
2
0
0
7
)
.
"
[
T
]
h
e
s
p
a
m
i
s
s
u
e
[
…
]
p
r
o
m
p
t
e
d
u
s
t
o
n
o
t
p
u
s
h
v
e
r
s
i
o
n
1
.
0
,
i
n
s
t
e
a
d
w
a
n
t
i
n
g
t
o
s
e
e
i
f
w
e
c
o
u
l
d
c
o
m
e
u
p
w
i
t
h
a
m
o
r
e
p
o
w
e
r
f
u
l
r
e
l
e
a
s
e
u
s
i
n
g
N
L
P
t
o
m
a
r
k
t
h
e
k
i
c
k
-
o
f
f
.
I
n
r
e
t
r
o
s
p
e
c
t
,
t
h
i
s
w
a
s
a
b
i
g
m
i
s
t
a
k
e
[
…
]
W
e
s
h
o
u
l
d
h
a
v
e
g
o
t
t
e
n
i
t
o
u
t
t
h
e
r
e
,
b
e
e
n
k
i
c
k
e
d
i
n
t
h
e
h
e
a
d
b
y
t
o
u
g
h
c
u
s
t
o
m
e
r
s
,
a
n
d
i
t
e
r
a
t
e
d
l
i
k
e
c
r
a
z
y
t
o
a
d
d
r
e
s
s
t
h
e
i
r
n
e
e
d
s
.
"
(
E
h
r
e
n
b
e
r
g
2
0
0
8
)
.
1
0
.
"
A
t
s
o
m
e
p
o
i
n
t
,
w
e
r
e
a
c
h
e
d
a
n
a
l
m
o
s
t
g
o
-
l
i
v
e
p
r
o
d
u
c
t
.
B
u
t
a
f
t
e
r
s
h
o
w
i
n
g
i
t
t
o
a
f
e
w
V
C
s
,
w
e
r
e
a
l
i
z
e
d
w
e
m
i
s
s
e
d
s
o
m
e
f
e
a
t
u
r
e
s
t
o
m
a
k
e
i
t
t
r
u
l
y
s
o
c
i
a
l
.
I
n
s
t
e
a
d
o
f
r
e
l
e
a
s
i
n
g
t
h
e
b
e
t
a
,
w
e
d
e
c
i
d
e
d
t
o
c
r
e
a
t
e
t
h
e
v
2
o
f
t
h
e
p
r
o
d
u
c
t
.
W
r
o
n
g
.
W
e
w
i
l
l
n
e
v
e
r
f
i
n
i
s
h
t
h
i
s
v
2
(
l
a
c
k
o
f
m
o
n
e
y
,
s
l
o
w
l
y
d
y
i
n
g
m
o
t
i
v
a
t
i
o
n
)
.
"
(
K
r
e
t
z
2
0
1
0
)
.
1
1
.
"
[
T
]
h
e
c
o
s
t
o
f
b
u
i
l
d
i
n
g
s
c
a
l
a
b
i
l
i
t
y
a
h
e
a
d
o
f
t
i
m
e
i
s
e
x
t
r
e
m
e
l
y
h
i
g
h
,
a
n
d
f
o
r
m
o
s
t
s
t
a
r
t
u
p
i
s
c
o
s
t
p
r
o
h
i
b
i
t
i
v
e
.
T
h
e
d
e
c
i
s
i
o
n
t
o
t
u
r
n
N
o
u
n
c
e
r
f
r
o
m
a
c
o
n
s
u
m
e
r
s
e
r
v
i
c
e
t
o
a
n
i
n
f
r
a
s
t
r
u
c
t
u
r
e
p
r
o
v
i
d
e
r
f
o
r
c
e
d
a
s
i
g
n
i
f
i
c
a
n
t
e
a
r
l
y
i
n
v
e
s
t
m
e
n
t
t
h
a
t
h
a
d
v
e
r
y
l
i
t
t
l
e
c
h
a
n
g
e
o
f
m
a
t
e
r
i
a
l
i
z
i
n
g
.
"
(
H
a
m
m
e
r
2
0
0
8
)
.
334 334
3
3
5
A
P
P
E
N
D
I
X
4
S
T
R
A
U
S
S
I
A
N
E
V
A
L
U
A
T
I
O
N
O
F
C
R
E
D
I
B
I
L
I
T
Y
C
r
i
t
e
r
i
a
(
C
o
r
b
i
n
&
S
t
r
a
u
s
s
1
9
9
0
)
S
e
l
f
-
e
v
a
l
u
a
t
i
o
n
R
e
s
e
a
r
c
h
p
r
o
c
e
s
s
C
1
:
H
o
w
w
a
s
t
h
e
o
r
i
g
i
n
a
l
s
a
m
p
l
e
s
e
l
e
c
t
e
d
?
T
h
e
s
a
m
p
l
e
w
a
s
s
e
l
e
c
t
e
d
b
a
s
e
d
o
n
i
t
s
d
e
g
r
e
e
o
f
i
n
t
e
r
e
s
t
;
p
o
s
t
-
m
o
r
t
e
m
s
t
o
r
i
e
s
s
e
e
m
e
d
l
i
k
e
a
r
i
c
h
s
o
u
r
c
e
o
f
i
n
f
o
r
m
a
t
i
o
n
.
C
2
:
W
h
a
t
m
a
j
o
r
c
a
t
e
g
o
r
i
e
s
e
m
e
r
g
e
d
?
I
n
i
t
i
a
l
l
y
,
t
h
e
r
e
s
e
a
r
c
h
w
a
s
g
u
i
d
e
d
b
y
t
h
e
i
n
t
e
n
t
t
o
i
d
e
n
t
i
f
y
f
a
i
l
u
r
e
f
a
c
t
o
r
s
.
E
m
e
r
g
i
n
g
c
a
t
e
g
o
r
i
e
s
f
o
l
l
o
w
e
d
t
h
e
f
u
n
c
t
i
o
n
s
w
i
t
h
w
h
i
c
h
t
h
e
s
e
f
a
c
t
o
r
s
w
e
r
e
a
s
s
o
c
i
a
t
e
d
;
f
o
r
e
x
a
m
p
l
e
,
M
a
r
k
e
t
i
n
g
,
P
r
o
d
u
c
t
,
a
n
d
F
i
n
a
n
c
e
.
C
3
:
W
h
a
t
w
e
r
e
s
o
m
e
o
f
t
h
e
e
v
e
n
t
s
,
i
n
c
i
d
e
n
t
s
,
a
c
t
i
o
n
s
,
a
n
d
s
o
o
n
t
h
a
t
,
a
s
i
n
d
i
c
a
t
o
r
s
,
p
o
i
n
t
e
d
t
o
s
o
m
e
o
f
t
h
e
s
e
m
a
j
o
r
c
a
t
e
g
o
-
r
i
e
s
?
T
h
e
f
o
u
n
d
e
r
s
’
s
e
n
s
e
-
m
a
k
i
n
g
i
m
p
l
i
e
d
t
h
e
r
e
a
r
e
m
a
n
y
r
e
a
s
o
n
s
f
o
r
f
a
i
l
u
r
e
;
t
h
e
g
r
o
u
p
i
n
g
s
e
e
m
e
d
t
o
n
a
t
u
r
a
l
l
y
e
v
o
l
v
e
i
n
t
o
t
h
e
s
e
f
u
n
c
t
i
o
n
a
l
a
r
e
a
s
.
C
4
:
O
n
t
h
e
b
a
s
i
s
o
f
w
h
i
c
h
c
a
t
e
g
o
r
i
e
s
d
i
d
t
h
e
o
r
e
t
i
c
a
l
s
a
m
p
l
i
n
g
p
r
o
c
e
e
d
?
T
h
a
t
i
s
,
h
o
w
d
i
d
t
h
e
o
r
e
t
i
c
a
l
f
o
r
m
u
l
a
t
i
o
n
s
g
u
i
d
e
s
o
m
e
o
f
t
h
e
d
a
t
a
c
o
l
l
e
c
t
i
o
n
?
T
h
i
s
w
a
s
b
a
s
e
d
o
n
t
h
e
c
o
r
e
c
a
t
e
g
o
r
y
,
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
.
T
h
e
y
w
e
r
e
p
r
e
s
e
n
t
e
d
t
o
d
i
f
f
e
r
e
n
t
f
o
u
n
d
e
r
s
o
n
v
a
r
i
o
u
s
o
c
c
a
s
i
o
n
s
.
T
h
e
s
e
w
e
r
e
n
o
t
i
n
t
e
r
v
i
e
w
s
,
b
u
t
c
a
s
u
a
l
s
t
a
r
t
u
p
e
v
e
n
t
s
.
A
f
t
e
r
t
h
e
t
h
e
o
r
e
t
i
c
a
l
s
a
m
p
l
i
n
g
w
a
s
d
o
n
e
,
h
o
w
r
e
p
r
e
s
e
n
t
a
t
i
v
e
d
i
d
t
h
e
s
e
c
a
t
e
g
o
r
i
e
s
p
r
o
v
e
t
o
b
e
?
T
h
e
r
e
w
e
r
e
n
o
i
n
d
i
c
e
s
o
f
n
e
w
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
p
e
r
t
a
i
n
i
n
g
t
o
t
h
e
s
p
e
c
i
f
i
c
t
y
p
e
o
f
f
i
r
m
s
i
n
w
h
i
c
h
t
h
i
s
s
t
u
d
y
w
a
s
i
n
t
e
r
e
s
t
e
d
.
C
5
:
W
h
a
t
w
e
r
e
s
o
m
e
o
f
t
h
e
h
y
p
o
t
h
e
s
e
s
p
e
r
t
a
i
n
i
n
g
t
o
c
o
n
c
e
p
-
t
u
a
l
r
e
l
a
t
i
o
n
s
,
t
h
a
t
i
s
,
a
m
o
n
g
c
a
t
e
g
o
r
i
e
s
,
a
n
d
o
n
w
h
a
t
g
r
o
u
n
d
s
w
e
r
e
t
h
e
y
f
o
r
m
u
l
a
t
e
d
a
n
d
t
e
s
t
e
d
?
F
o
r
e
x
a
m
p
l
e
,
t
h
e
r
e
l
a
t
i
o
n
s
h
i
p
b
e
t
w
e
e
n
t
h
e
c
o
l
d
s
t
a
r
t
a
n
d
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
s
.
A
f
t
e
r
a
x
i
a
l
c
o
d
i
n
g
,
i
t
s
e
e
m
e
d
c
l
e
a
r
t
h
a
t
s
t
a
r
t
u
p
s
w
h
i
c
h
t
r
y
t
o
c
o
m
p
e
n
s
a
t
e
a
l
a
c
k
o
f
u
s
e
r
s
b
y
f
r
e
e
p
r
o
d
u
c
t
n
o
t
o
n
l
y
f
a
i
l
t
o
s
o
l
v
e
i
t
,
b
u
t
r
u
n
i
n
t
o
a
d
d
i
t
i
o
n
a
l
t
r
o
u
b
l
e
.
T
h
r
o
u
g
h
c
o
m
p
a
r
i
s
o
n
w
i
t
h
o
t
h
e
r
c
a
s
e
s
,
t
h
i
s
s
e
e
m
e
d
t
o
b
e
a
p
a
t
t
e
r
n
i
n
t
h
e
m
a
t
e
r
i
a
l
.
C
6
:
W
e
r
e
t
h
e
r
e
i
n
s
t
a
n
c
e
s
w
h
e
n
h
y
p
o
t
h
e
s
e
s
d
i
d
n
o
t
h
o
l
d
u
p
a
g
a
i
n
s
t
w
h
a
t
w
a
s
a
c
t
u
a
l
l
y
s
e
e
n
?
H
o
w
w
e
r
e
t
h
e
s
e
d
i
s
c
r
e
p
a
n
-
c
i
e
s
a
c
c
o
u
n
t
e
d
f
o
r
?
H
o
w
d
i
d
t
h
e
y
a
f
f
e
c
t
t
h
e
h
y
p
o
t
h
e
s
e
s
?
N
o
n
e
c
o
m
e
t
o
m
i
n
d
;
t
h
i
s
a
s
p
e
c
t
o
f
a
n
a
l
y
s
i
s
c
o
u
l
d
a
r
g
u
a
b
l
y
h
a
v
e
b
e
e
n
i
m
p
r
o
v
e
d
b
y
m
o
r
e
a
c
-
t
i
v
e
l
y
d
e
v
e
l
o
p
i
n
g
h
y
p
o
t
h
e
s
e
s
.
C
7
:
H
o
w
a
n
d
w
h
y
w
a
s
t
h
e
c
o
r
e
c
a
t
e
g
o
r
y
s
e
l
e
c
t
e
d
?
W
a
s
t
h
i
s
s
e
l
e
c
t
i
o
n
s
u
d
d
e
n
o
r
g
r
a
d
u
a
l
,
d
i
f
f
i
c
u
l
t
o
r
e
a
s
y
?
O
n
w
h
a
t
g
r
o
u
n
d
s
w
e
r
e
t
h
e
f
i
n
a
l
a
n
a
l
y
t
i
c
a
l
d
e
c
i
s
i
o
n
s
m
a
d
e
?
T
h
e
s
e
l
e
c
t
i
o
n
w
a
s
g
r
a
d
u
a
l
,
b
a
s
e
d
o
n
s
e
v
e
r
a
l
c
o
d
i
n
g
p
h
a
s
e
s
:
f
a
i
l
u
r
e
à
f
a
i
l
u
r
e
f
a
c
t
o
r
s
à
s
t
r
a
t
e
g
i
c
p
r
o
b
l
e
m
s
a
n
d
b
i
a
s
e
s
à
d
i
l
e
m
m
a
s
.
O
u
t
o
f
a
l
l
t
h
e
c
h
o
i
c
e
s
,
d
i
l
e
m
m
a
s
s
e
e
m
e
d
t
o
m
o
s
t
i
n
f
l
u
e
n
c
e
f
a
i
l
u
r
e
.
T
h
e
s
e
l
e
c
t
i
o
n
w
a
s
n
o
t
e
a
s
y
o
r
d
i
f
f
i
c
u
l
t
,
i
t
f
e
l
t
n
a
t
u
r
a
l
a
f
t
e
r
r
e
a
d
i
n
g
t
h
e
m
a
t
e
r
i
a
l
m
a
n
y
t
i
m
e
s
.
T
h
e
c
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
b
y
f
o
u
n
d
e
r
s
a
l
s
o
p
o
i
n
t
e
d
t
o
t
h
i
s
d
i
r
e
c
t
i
o
n
.
335 335
3
3
6
E
m
p
i
r
i
c
a
l
g
r
o
u
n
d
i
n
g
o
f
f
i
n
d
i
n
g
s
C
1
:
A
r
e
c
o
n
c
e
p
t
s
g
e
n
e
r
a
t
e
d
?
T
h
e
s
t
u
d
y
g
e
n
e
r
a
t
e
s
n
e
w
c
o
n
s
t
r
u
c
t
s
,
s
u
c
h
a
s
t
h
e
c
o
l
d
s
t
a
r
t
d
i
l
e
m
m
a
,
t
h
e
l
o
n
e
l
y
u
s
e
r
d
i
l
e
m
m
a
,
r
e
m
o
r
a
’
s
c
u
r
s
e
,
a
n
d
t
h
e
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
.
N
e
w
c
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
s
a
l
s
o
r
e
l
a
t
e
t
o
o
n
l
i
n
e
p
l
a
t
f
o
r
m
t
y
p
e
s
a
n
d
t
h
e
‘
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
’
.
T
h
e
t
e
r
m
i
n
o
l
o
g
y
i
s
p
a
r
t
i
a
l
l
y
d
e
r
i
v
e
d
f
r
o
m
t
h
e
f
o
u
n
d
e
r
s
a
n
d
i
n
d
u
s
t
r
y
,
t
h
e
r
e
f
o
r
e
b
e
i
n
g
g
r
o
u
n
d
e
d
.
C
2
:
A
r
e
t
h
e
c
o
n
c
e
p
t
s
s
y
s
t
e
m
a
t
i
c
a
l
l
y
r
e
l
a
t
e
d
?
T
h
e
c
o
n
c
e
p
t
s
a
r
e
r
e
l
a
t
e
d
a
s
a
n
t
e
c
e
d
e
n
t
s
(
e
.
g
.
,
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
)
,
o
u
t
c
o
m
e
s
(
e
.
g
.
,
c
o
l
d
s
t
a
r
t
d
i
l
e
m
m
a
)
a
n
d
s
o
l
u
t
i
o
n
s
(
e
.
g
.
,
f
r
e
e
f
y
i
n
g
)
.
T
h
e
r
e
h
a
s
b
e
e
n
a
n
a
t
t
e
m
p
t
t
o
l
i
n
k
a
n
d
e
x
-
a
m
i
n
e
t
h
e
m
a
s
a
w
h
o
l
e
.
D
u
r
i
n
g
t
h
e
c
o
d
i
n
g
p
r
o
c
e
s
s
,
t
h
e
a
u
t
h
o
r
a
i
m
e
d
t
o
c
o
d
e
r
e
l
a
t
i
o
n
s
h
i
p
s
s
y
s
-
t
e
m
a
t
i
c
a
l
l
y
w
i
t
h
t
h
e
h
e
l
p
o
f
q
u
a
l
i
t
a
t
i
v
e
a
n
a
l
y
s
i
s
s
o
f
t
w
a
r
e
(
i
.
e
.
,
N
V
i
v
o
)
.
C
3
:
A
r
e
t
h
e
r
e
m
a
n
y
c
o
n
c
e
p
t
u
a
l
l
i
n
k
a
g
e
s
,
a
n
d
a
r
e
t
h
e
c
a
t
e
g
o
-
r
i
e
s
w
e
l
l
d
e
v
e
l
o
p
e
d
?
D
o
t
h
e
y
h
a
v
e
c
o
n
c
e
p
t
u
a
l
d
e
n
s
i
t
y
?
T
h
e
l
i
n
k
a
g
e
s
a
r
e
f
o
r
m
e
d
b
e
t
w
e
e
n
d
i
l
e
m
m
a
s
a
n
d
t
h
e
i
r
s
o
l
u
t
i
o
n
s
.
T
h
e
r
e
i
s
a
l
s
o
w
o
r
k
d
o
n
e
t
o
l
i
n
k
d
i
l
e
m
m
a
s
w
i
t
h
f
o
u
n
d
e
r
s
’
b
i
a
s
e
s
,
b
u
t
t
h
a
t
i
s
o
m
i
t
t
e
d
f
r
o
m
t
h
i
s
r
e
p
o
r
t
a
s
t
h
e
r
e
s
e
a
r
c
h
e
r
d
e
c
i
d
e
d
g
o
f
o
r
a
d
e
e
p
e
r
a
n
a
l
y
s
i
s
o
f
d
i
l
e
m
m
a
s
i
n
e
x
c
h
a
n
g
e
f
o
r
r
a
n
g
e
,
o
r
e
x
p
l
a
n
a
t
o
r
y
p
o
w
e
r
.
M
o
r
e
o
v
e
r
,
t
h
e
r
e
p
o
r
t
i
n
c
l
u
d
e
s
s
e
v
e
r
a
l
l
i
n
k
a
g
e
s
f
r
o
m
t
h
e
a
n
a
l
y
s
i
s
o
f
d
i
l
e
m
m
a
s
t
o
t
h
e
i
r
o
n
l
i
n
e
s
p
e
c
i
f
i
c
i
t
i
e
s
,
n
a
m
e
l
y
t
y
p
o
l
o
g
y
a
n
d
i
d
e
a
l
u
s
e
r
g
e
n
e
r
a
t
i
o
n
m
o
d
e
l
.
T
h
e
s
e
f
o
r
t
i
f
y
t
h
e
s
u
b
s
t
a
n
t
i
v
e
n
a
t
u
r
e
o
f
t
h
e
t
h
e
o
r
e
t
i
c
a
l
o
f
f
e
r
i
n
g
.
F
i
n
a
l
l
y
,
a
p
a
r
a
d
i
g
m
m
o
d
e
l
w
a
s
n
o
t
f
o
l
l
o
w
e
d
;
t
h
u
s
,
i
n
t
h
e
S
t
r
a
u
s
s
i
a
n
s
e
n
s
e
,
a
d
e
n
s
i
t
y
o
f
c
a
t
e
g
o
r
i
e
s
m
i
g
h
t
b
e
l
a
c
k
i
n
g
.
C
4
:
I
s
t
h
e
r
e
m
u
c
h
v
a
r
i
a
t
i
o
n
b
u
i
l
t
i
n
t
o
t
h
e
t
h
e
o
r
y
?
V
a
r
i
a
t
i
o
n
i
n
t
h
e
d
a
t
a
w
a
s
c
o
n
s
i
d
e
r
a
b
l
e
d
u
e
t
o
t
h
e
v
a
r
i
e
t
y
o
f
p
r
o
b
l
e
m
s
r
e
p
o
r
t
e
d
b
y
t
h
e
s
t
a
r
t
u
p
s
.
G
i
v
e
n
t
h
e
r
e
l
a
t
i
v
e
l
y
f
e
w
n
u
m
b
e
r
o
f
d
i
l
e
m
m
a
s
,
t
h
e
a
u
t
h
o
r
f
e
e
l
s
h
e
c
h
o
s
e
t
h
o
s
e
t
h
a
t
e
x
p
l
a
i
n
m
o
s
t
o
f
t
h
e
p
r
o
b
l
e
m
s
i
n
t
e
r
m
s
o
f
r
e
l
e
v
a
n
c
e
f
o
r
t
h
e
p
l
a
t
f
o
r
m
m
o
d
e
l
,
c
r
i
t
i
c
a
l
i
t
y
i
n
t
e
r
m
s
o
f
f
a
i
l
-
u
r
e
o
u
t
c
o
m
e
,
a
n
d
p
a
t
t
e
r
n
s
e
m
e
r
g
i
n
g
f
r
o
m
t
h
e
d
a
t
a
.
T
h
u
s
,
t
h
e
a
u
t
h
o
r
f
e
e
l
s
t
h
e
e
n
d
r
e
s
u
l
t
e
s
t
a
b
-
l
i
s
h
e
s
p
a
r
s
i
m
o
n
y
a
n
d
s
c
o
p
e
.
H
o
w
e
v
e
r
,
d
u
e
t
o
e
x
c
l
u
s
i
o
n
o
f
n
o
n
-
p
l
a
t
f
o
r
m
-
s
p
e
c
i
f
i
c
s
t
a
r
t
u
p
d
i
-
l
e
m
m
a
s
a
n
d
f
o
u
n
d
e
r
s
’
b
i
a
s
e
s
,
t
h
e
t
h
e
o
r
y
h
a
s
l
e
s
s
d
i
m
e
n
s
i
o
n
s
a
n
d
e
x
p
l
a
n
a
t
o
r
y
p
o
w
e
r
t
h
a
n
i
t
w
o
u
l
d
h
a
v
e
i
f
t
h
e
s
e
c
a
t
e
g
o
r
i
e
s
w
e
r
e
i
n
c
l
u
d
e
d
.
T
h
e
a
u
t
h
o
r
s
t
i
l
l
b
e
l
i
e
v
e
s
t
h
e
i
r
e
x
c
l
u
s
i
o
n
w
a
s
a
c
o
r
r
e
c
t
c
h
o
i
c
e
,
g
i
v
e
n
t
h
a
t
1
)
t
h
e
e
n
s
u
i
n
g
c
o
m
p
l
e
x
i
t
y
w
o
u
l
d
h
a
v
e
e
x
t
e
n
d
e
d
b
e
y
o
n
d
t
h
e
s
c
o
p
e
o
f
o
n
e
s
t
u
d
y
;
a
n
d
2
)
f
o
r
t
h
e
o
r
e
t
i
c
a
l
i
n
t
e
g
r
a
t
i
o
n
,
i
t
h
a
d
b
e
e
n
n
e
c
e
s
s
a
r
y
t
o
e
x
a
m
i
n
e
t
o
o
m
a
n
y
e
x
t
a
n
t
t
h
e
o
r
i
e
s
,
l
e
a
v
i
n
g
t
h
e
w
o
r
k
’
s
t
h
e
o
r
e
t
i
c
a
l
i
n
t
e
g
r
a
t
i
o
n
f
r
a
g
m
e
n
t
e
d
.
T
h
e
s
e
c
a
t
e
g
o
r
i
e
s
a
r
e
b
e
t
t
e
r
l
e
f
t
f
o
r
f
u
r
t
h
e
r
s
t
u
d
i
e
s
.
C
5
:
A
r
e
t
h
e
b
r
o
a
d
e
r
c
o
n
d
i
t
i
o
n
s
t
h
a
t
a
f
f
e
c
t
t
h
e
p
h
e
n
o
m
e
n
o
n
u
n
d
e
r
s
t
u
d
y
b
u
i
l
t
i
n
t
o
i
t
s
e
x
p
l
a
n
a
t
i
o
n
?
A
n
a
t
t
e
m
p
t
h
a
s
b
e
e
n
m
a
d
e
t
o
e
n
a
b
l
e
t
h
e
r
e
a
d
e
r
t
o
u
n
d
e
r
s
t
a
n
d
h
o
w
W
e
b
2
.
0
s
t
a
r
t
u
p
s
,
i
n
w
h
i
c
h
t
h
i
s
s
t
u
d
y
i
s
i
n
t
e
r
e
s
t
e
d
,
d
i
f
f
e
r
f
r
o
m
e
a
r
l
i
e
r
d
o
t
c
o
m
c
o
m
p
a
n
i
e
s
.
M
o
r
e
o
v
e
r
,
t
h
e
g
e
n
e
r
a
l
r
o
l
e
o
f
c
o
m
p
e
t
i
t
i
o
n
i
s
s
p
e
c
u
l
a
t
e
d
s
o
t
h
a
t
i
t
i
s
s
e
e
n
a
s
a
p
o
t
e
n
t
i
a
l
r
e
a
s
o
n
f
o
r
p
l
a
t
f
o
r
m
s
t
a
r
t
u
p
s
o
n
t
h
e
I
n
t
e
r
n
e
t
a
p
p
l
y
i
n
g
i
n
d
i
r
e
c
t
m
o
n
e
t
i
z
a
t
i
o
n
(
i
.
e
.
,
o
f
f
e
r
f
r
e
e
p
r
o
d
u
c
t
s
)
.
T
h
e
s
e
c
o
n
c
l
u
s
i
o
n
s
a
r
e
n
o
t
d
e
f
i
n
i
t
e
,
a
l
t
h
o
u
g
h
t
h
e
y
s
h
o
w
a
n
a
t
t
e
m
p
t
t
o
u
n
d
e
r
s
t
a
n
d
t
h
e
b
i
g
g
e
r
p
i
c
t
u
r
e
.
336 336
3
3
7
C
6
:
H
a
s
“
p
r
o
c
e
s
s
”
b
e
e
n
t
a
k
e
n
i
n
t
o
a
c
c
o
u
n
t
?
P
r
o
c
e
s
s
h
a
s
b
e
e
n
i
m
p
l
i
c
i
t
l
y
t
a
k
e
n
i
n
t
o
a
c
c
o
u
n
t
s
o
t
h
a
t
,
g
i
v
e
n
t
h
e
i
r
t
y
p
e
a
n
d
s
p
e
c
i
f
i
c
i
t
i
e
s
(
i
.
e
.
,
u
s
e
r
g
e
n
e
r
a
t
i
o
n
a
n
d
i
n
d
i
r
e
c
t
m
o
n
e
t
i
z
a
t
i
o
n
d
i
l
e
m
m
a
s
)
a
r
e
s
e
e
n
a
s
m
e
c
h
a
n
i
s
m
s
a
f
f
e
c
t
i
n
g
p
l
a
t
-
f
o
r
m
s
i
n
t
h
e
o
n
l
i
n
e
c
o
n
t
e
x
t
,
l
e
a
d
i
n
g
t
o
t
h
e
o
u
t
c
o
m
e
o
f
f
a
i
l
u
r
e
.
C
7
:
D
o
t
h
e
t
h
e
o
r
e
t
i
c
a
l
f
i
n
d
i
n
g
s
s
e
e
m
s
i
g
n
i
f
i
c
a
n
t
a
n
d
t
o
w
h
a
t
e
x
t
e
n
t
?
T
h
e
o
r
e
t
i
c
a
l
f
i
n
d
i
n
g
s
o
f
t
h
i
s
s
t
u
d
y
a
r
e
l
e
s
s
s
i
g
n
i
f
i
c
a
n
t
t
h
a
n
i
t
s
i
n
t
e
n
d
e
d
m
a
n
a
g
e
r
i
a
l
u
s
e
f
u
l
n
e
s
s
,
p
a
r
t
i
a
l
l
y
d
u
e
t
o
i
t
s
s
u
b
s
t
a
n
t
i
v
e
n
a
t
u
r
e
.
H
o
w
e
v
e
r
,
i
t
d
o
e
s
e
x
p
a
n
d
t
h
e
c
o
n
c
e
p
t
o
f
t
h
e
-
a
n
d
-
e
g
g
p
r
o
b
l
e
m
i
n
t
h
e
o
n
l
i
n
e
p
l
a
t
f
o
r
m
c
o
n
t
e
x
t
;
t
h
i
s
m
i
g
h
t
e
n
c
o
u
r
a
g
e
o
t
h
e
r
t
h
e
o
r
i
s
t
s
t
o
r
e
c
o
n
s
i
d
e
r
t
h
e
n
a
t
u
r
e
o
f
t
h
e
p
r
o
b
l
e
m
i
n
t
h
e
i
r
p
a
r
t
i
c
u
l
a
r
s
e
t
t
i
n
g
s
.
I
t
a
l
s
o
o
f
f
e
r
s
a
t
y
p
o
l
o
g
y
a
n
d
c
r
i
t
e
r
i
a
f
o
r
t
y
-
p
o
l
o
g
y
o
f
p
l
a
t
f
o
r
m
s
t
h
a
t
i
s
m
e
a
n
i
n
g
f
u
l
,
c
o
n
s
i
d
e
r
i
n
g
t
h
e
i
r
p
r
o
b
l
e
m
s
i
n
r
e
l
a
t
i
o
n
t
o
a
d
o
p
t
i
o
n
a
n
d
a
c
t
i
v
e
u
s
e
.
T
h
i
s
s
t
u
d
y
i
s
l
i
m
i
t
e
d
i
n
t
h
a
t
i
t
d
o
e
s
n
o
t
p
r
e
s
e
n
t
a
“
g
e
n
e
r
a
l
t
h
e
o
r
y
o
f
p
l
a
t
f
o
r
m
s
”
;
h
o
w
e
v
e
r
,
s
u
c
h
a
n
a
t
t
e
m
p
t
w
a
s
n
e
v
e
r
i
t
s
p
u
r
p
o
s
e
i
n
t
h
e
f
i
r
s
t
p
l
a
c
e
.
N
e
v
e
r
t
h
e
l
e
s
s
,
t
h
e
s
t
u
d
y
d
o
e
s
o
f
f
e
r
g
o
o
d
g
r
o
u
n
d
s
f
o
r
c
o
n
t
i
n
u
i
n
g
t
h
e
r
e
s
e
a
r
c
h
o
f
o
n
l
i
n
e
p
l
a
t
f
o
r
m
s
.
337 337
THE FOLLOWING PUBLICATIONS HAVE BEEN RELEASED SINCE 2013
IN TURKU SCHOOL OF ECONOMICS PUBLICATION SERIES A
A-1:2013 Hanna Pitkänen
Theorizing formal and informal feedback practices in
management accounting through three dimensions
A-2:2013 Samppa Suoniemi
The impact of CRM system development on CRM acceptance
A-3:2013 Kirsi Lainema
Managerial interaction – Discussion practices in management
meetings
A-4:2013 Sueila Pedrozo
Consumption, youth and new media: The debate on social issues
in Brazil
A-5:2013 Jani Merikivi
Still believing in virtual worlds: A decomposed approach
A-6:2013 Sanna-Mari Renfors
Myyjän toiminnan laatu kuluttajapalvelujen
myyntikohtaamisessa – Ostajan näkökulma myyjän
suoritusarviointiin
A-7:2013 Maria Höyssä
Where science meets its use – Exploring the emergence of
practical relevance of scientific knowledge in the regional
context
A-8:2013 Karri Rantasila
Measuring logistics costs – Designing a general model for
assessing macro logistics costs in a global context with empirical
evidence from the manufacturing and trading industries
A-9:2013 Taina Eriksson
Dynamic capability of value net management in technology-
based international SMEs
A-10:2013 Jarkko Heinonen
Kunnan yritysilmapiirin vaikutus yritystoiminnan kehittymiseen
A-11:2013 Pekka Matomäki
On two-sided controls of a linear diffusion
A-12:2013 Valtteri Kaartemo
Network development process of international new ventures in
internet-enabled markets: Service ecosystems approach
A-13:2013 Emmi Martikainen
Essays on the demand for information goods
A-14:2013 Elina Pelto
Spillover effects of foreign entry on local firms and business
networks in Russia – A Case study on Fazer Bakeries in St.
Petersburg
A-15:2013 Anna-Maija Kohijoki
ONKO KAUPPA KAUKANA?
Päivittäistavarakaupan palvelujen saavutettavuus Turun seudulla
– Ikääntyvien kuluttajien näkökulma
A-1:2014 Kirsi-Mari Kallio
”Ketä kiinnostaa tuottaa tutkintoja ja julkaisuja
liukuhihnaperiaatteella…?”
– Suoritusmittauksen vaikutukset tulosohjattujen yliopistojen
tutkimus- ja opetushenkilökunnan työhön
A-2:2014 Marika Parvinen
Taiteen ja liiketoiminnan välinen jännite ja sen vaikutus
organisaation ohjaukseen – Case-tutkimus taiteellisen
organisaation kokonaisohjauksesta
A-3:2014 Terhi Tevameri
Matriisirakenteen omaksuminen sairaalaorganisaatioissa
– Rakenteeseen päätyminen, organisaatiosuunnittelu ja
toimintalogiikan hyväksyminen
A-4:2014 Tomi Solakivi
The connection between supply chain practices and firm
performance – Evidence from multiple surveys and financial
reporting data
A-5:2014 Salla-Tuulia Siivonen
“Holding all the cards”
The associations between management accounting,
strategy and strategic change
A-6:2014 Sirpa Hänti
Markkinointi arvon muodostamisen prosessina ja sen yhteys
yrittäjyyden mahdollisuusprosessiin
– Tapaustutkimus kuuden yrityksen alkutaipaleelta
A-7:2014 Kimmo Laakso
Management of major accidents
– Communication challenges and solutions in the preparedness
and response phases for both authorities and companies
A-8:2014 Piia Haavisto
Discussion forums
– From idea creation to incremental innovations. Focus on heart-
rate monitors
A-9:2014 Sini Jokiniemi
"Once again I gained so much"
– Understanding the value of business-to-business sales
interactions from an individual viewpoint
A-10:2014 Xiaoyu Xu
Understanding online game players’ post-adoption behavior: an
investigation of social network games in
A-11:2014 Helena Rusanen
Resource access and creation in networks for service innovation
A-12:2014 Joni Salminen
Startup dilemmas - Strategic problems of early-stage platforms
on the Internet
All the publications can be ordered from
KY-Dealing Oy
Rehtorinpellonkatu 3
20500 Turku, Finland
Phone +358-2-333 9422
E-mail: [email protected]
doc_187405272.pdf