Description
The main purpose of this work is to evaluate the long-term effectiveness of a hotel’s chain
loyalty program from a behavioral perspective.
International Journal of Culture, Tourism and Hospitality Research
Exploring effects of hotel chain loyalty program
Pedro Pimpão Antónia Correia J oão Duque J osé Carlos Zorrinho
Article information:
To cite this document:
Pedro Pimpão Antónia Correia J oão Duque J osé Carlos Zorrinho , (2014),"Exploring effects of hotel chain loyalty program",
International J ournal of Culture, Tourism and Hospitality Research, Vol. 8 Iss 4 pp. 375 - 387
Permanent link to this document:http://dx.doi.org/10.1108/IJ CTHR-03-2014-0020
Downloaded on: 24 January 2016, At: 22:25 (PT)
References: this document contains references to 25 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 786 times since 2014*
Users who downloaded this article also downloaded:
Lijia (Karen) Xie, Chih-Chien Chen, (2014),"Hotel loyalty programs: how valuable is valuable enough?", International J ournal
of Contemporary Hospitality Management, Vol. 26 Iss 1 pp. 107-129http://dx.doi.org/10.1108/IJ CHM-08-2012-0145
Mark D. Uncles, Grahame R. Dowling, Kathy Hammond, (2003),"Customer loyalty and customer loyalty programs", J ournal
of Consumer Marketing, Vol. 20 Iss 4 pp. 294-316http://dx.doi.org/10.1108/07363760310483676
Lina Xiong, Ceridwyn King, Clark Hu, (2014),"Where is the love?: Investigating multiple membership and hotel customer
loyalty", International J ournal of Contemporary Hospitality Management, Vol. 26 Iss 4 pp. 572-592http://dx.doi.org/10.1108/
IJ CHM-03-2013-0141
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication
Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Exploring effects of hotel chain loyalty
program
Pedro Pimpão, Antónia Correia, João Duque and José Carlos Zorrinho
Pedro Pimpão is a PhD
Researcher at the ISEG –
Lisbon School of
Economics and
Management, Technical
University of Lisbon,
Lisbon, Portugal.
Antónia Correia is a
Professor at the
European University,
Laureate International
Universities, CEFAGE,
Lisbon, Portugal.
João Duque is Professor
at the ISEG – Lisbon
School of Economics and
Management, Lisbon,
Portugal.
José Carlos Zorrinho is a
Professor at the
University of Évora,
Évora, Portugal.
Abstract
Purpose – The main purpose of this work is to evaluate the long-term effectiveness of a hotel’s chain
loyalty program from a behavioral perspective.
Design/methodology/approach – ADirichlet model was estimated to assess purchase frequency and
hotel choice within one of the biggest hotel chains in Portugal. The sample comprises hotels where a
loyalty program was implemented, with a total of 176,099 reservations. Data were extracted from the
customer relationship management (CRM) systems of the hotel group.
Findings – The results suggest that instead of being loyal to a certain hotel, customers are loyal to the
branded hotel chain. As the hotels are all part of the branded group, this polygamy is not only accepted
but also very welcome.
Research limitations/implications – The level of penetration and purchase frequency of CRM was
measured. Nevertheless, a thorough understanding of these will be critical for the success of this
program.
Practical implications – This research is a step toward assessing hotel chain competitiveness, by
improving and suggesting segmented groups of brands/hotels and to induce cross-selling products
accepting polygamous loyalty as the only way to sustain long-term relationships with customers.
Originality/value – This is one of the few research studies, if not the only one, to assess loyalty with
tangible indicators, such as purchase frequency. Further, the results suggest that loyalty programs are
more effective if multiple options are available and as such, cross-selling is perhaps the only way to ?x
customers.
Keywords CRM, Loyalty program, Customer loyalty, Dirichlet model, Repeat purchase behavior
Paper type Research paper
1. Introduction
Internet marketing, as one of the most powerful technological tools, hugely increases
competition across products and even more for tourism destinations, and it has changed
travelers’ behavior. It has enabled customers to engage directly with hotels, without
intermediaries (Buhalis and Law, 2008). One of the reactions of the industry was to develop
loyalty programs to sustain and enact repeat purchases. These programs, which are
supported on one-to-one relationships, provide accurate knowledge of customers’ needs
and wants (Sharp and Sharp, 1997).
The most common loyalty programs are those of customer relationship management (the
so-called CRM). CRM systems have emerged to focus on customer-centric technologies,
helping the implementation of loyalty programs, as an “info-structure” to support its
interoperability, its personalization and its constant networking (Buhalis, 2003; Buhalis and
Law, 2008). CRM studies have been applied in several industries, in particular in airline
alliances (Boland et al., 2002), although none of these researches rely on Dirichlet
Assumptions. In this sense, loyalty programs could be important in increasing the expected
response from hotels to customers, as the data transmitted through the Internet has rapidly
increased in recent years. However, it is costly to initiate and maintain those loyalty
Received 19 March 2014
Revised 11 July 2014
Accepted 4 August 2014
DOI 10.1108/IJCTHR-03-2014-0020 VOL. 8 NO. 4 2014, pp. 375-387, © Emerald Group Publishing Limited, ISSN 1750-6182 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 375
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
programs because of their long-term commitment (Liu, 2007). The cost-effectiveness of
these programs is very controversial. This controversy has been discussed since
Cunningham (1957), but no one till now has reached a reasonable answer to the main
question that is embedded in CRM philosophy – is it worth spending hundreds of
thousands of euros to implement CRM? How effective might these programs be for
positioning the brands/hotels in customers’ minds? For this discussion, Mägi (2003) argues
that membership in loyalty programs increases a customer’s share of wallet for four out of
seven programs and decreases shares of competitors. This is valid at the chain level and
not at the store level.
The challenge of this research is to contribute to reaching a reasonable answer to these
questions in a speci?c context – one of the best-known Portuguese hotel chains worldwide.
Loyalty is assumed to be a behavioral construct that can be measured in light of purchase
measures (Liu, 2007; Reichheld, 1993; Sharp and Sharp, 1997). Hence, it is assumed that
repeat purchase and loyalty are somewhat connected (Dick and Basu, 1994). From an
empirical point of view, the study provides tools for hotels to assess from a tangible
perspective the effectiveness of loyalty programs. Repeat purchase behaviors are
assessed in a temporal and geographical context. Temporal or strength context is used to
account for loyalty persistence (Dick and Basu, 1994; Liu, 2007), whereas geographical or
differentiation context (Dick and Basu, 1994) comprises the chain effect of this program
across the different kinds of hotels of the group.
This article builds on a dynamic performance behavior Dirichlet model to elucidate market
structure. The model is able to describe the various brand performance patterns, and in
that sense, it also helps explain and estimate them (Ehrenberg et al., 2004). The Dirichlet
model estimates these patterns, where big and small brands/hotels differ greatly in how
many buyers they have, but usually far less in how loyal these buyers are (Ehrenberg et al.,
2004). In this sense, this work blends hotel choice and purchase frequency in a
performance assessment process, monitoring results of brands’/hotels’ performance,
through several key performance indicators. It contributes to hotel brand’s market shares
positioning within the hotel chain, where geographical and type of hotel penetration rates
are estimated.
We ?rst describe the research in a theoretical framework, linking CRM systems with loyalty
programs and a Dirichlet model and then develop and estimate the brand performance
measures (BPM). The results illustrate how a chain effect between different localizations
and types of hotels evolves from the variety integrated effect and heterogeneity among
customers to being a loyalty program which helps maintain the polygamy of its customers
to achieve customer loyalty. The study has important implications for managers who are
charged with allocating resources to improve service bene?ts and customer relationships
over time.
2. Theoretical framework
Management of customer relationships is a key activity for the competitiveness of tourism
organizations (Buhalis and Law, 2008), as well for the industry as a whole (Frow et al.,
2011). CRM constitutes a long-term strategic approach that addresses all aspects of
identifying customers, developing customer insights and building customer relationships
(Boulding et al., 2005; Frow et al., 2011; Lacey and Morgan, 2009; Liu, 2007).
In any CRM system of the hotel industry, one major element is the measurement process
(Boulding et al., 2005). This opens the door to the performance assessment process, one
of the ?ve key processes of CRM identi?ed by Frow et al. (2011). The performance
assessment process ensures that ?rms’ strategic objectives are being delivered in an
appropriate and acceptable way, providing ?rms with the opportunity to gain deeper
insights into their customers for future improvements (Boulding et al., 2005; Frow et al.,
2011). Enabling performance monitoring in a micro-view, such as key performance
PAGE 376 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
indicators, has become an integral part of the CRM system and, speci?cally, loyalty
programs (Frow et al., 2011).
The idea that loyalty programs are an important component of ?rms’ CRM strategy is widely
accepted in marketing and tourism areas (Buhalis and Law, 2008; Hansen et al., 2010;
Lacey and Morgan, 2009; Liu, 2007). Loyalty programs or frequency programs are
considered one of the best ways of supporting a ?rms’ knowledge base by forcing stronger
relationships with customers (Buhalis, 2003; Buhalis and Law, 2008; Hansen et al., 2010).
Traditionally, loyalty programs are identi?ed by their degree of indirect defensive
orientation and their two-stage behavior time nature (Liu, 2007; Sharp and Sharp, 1997).
The indirect defensive orientation means building a closer program with current customers
(Dowling and Uncles, 1997), through points rewards systems or special offers. Clearly, a
?rst short-term effect results in the program points to the time of purchase and second, to
a long-term behavior commitment effect through redeeming points for free rewards and,
consequently, increasing pro?ts (Dowling and Uncles, 1997; Hansen et al., 2010; Liu, 2007;
Sharp and Sharp, 1997).
It is not clear that the long-term nature of loyalty programs ends with customer attrition.
However, loyalty programs, supported in dynamic processes and Internet communications
technology (Buhalis and Law, 2008), such as CRM systems, give rise to service ?rms and
help them to identify possible “defectors,” i.e. ?nd customers who are leaving and trying to
win them back (Czepiel and Rosenberg, 1992; Liu, 2007; Reichheld and Sasser, 1990).
Every hotel’s performance should be effectively measured on how well performance targets
are met (Reichheld, 1993; Reichheld and Sasser, 1990) – the key performance measures.
This is why a customer crisis occurs when the service goes unperformed (Czepiel and
Rosenberg, 1992). Thus, behavior performance measures are actions which individuals
adopt and which change their relationship with their environment. Buhalis and Law (2008)
argue that loyalty programs support and help to promote the customization of tourism
products. Hence, Sharp and Sharp (1997) argue that loyalty programs have an effect on
repeat purchases and that is a ?rst step to understanding the effectiveness of
cause-and-effect relationships in the program’s system.
Taking repeat purchases as the focus of a loyalty program requires an instrument to
measure it and to make this study aware of the effectiveness of the potential of such
programs. In this sense, we followed Ehrenberg et al. (2004) and Sharp and Sharp’s (1997)
works who argue that the repeat purchase effect follows a Dirichlet Distribution.
The Dirichlet model is based on the Stated Preference Theory, i.e. a theory about choice
between competitive entities such as brands/hotels, which assumes that customers are
able to order their preferences across different alternatives (Goodhardt et al., 1984).
Moreover, the Dirichlet model does more than explain and describe hotel choice and
purchase frequency simultaneously. It also includes estimates of them, allowing the use of
various performance patterns over a time-span (Goodhardt et al., 1984; Sharp and Sharp,
1997). However, customer characteristics are not included in this stationary model
because these effects are already incorporated in each BPM, and it is not in?uenced by
previous purchases (Ehrenberg et al., 2004).
Using a single statistical model, such as a Dirichlet model, it is comparatively
uncomplicated to describe the various performance patterns of the loyalty program, and to
specify the distribution of purchases by a population of customers of each of the hotels
(Ehrenberg et al., 2004; Goodhardt et al., 1984). The Dirichlet model is particularly relevant
for elucidating these purchase behaviors, i.e. market structure, and follows a combination
of two probability density functions, the negative binomial distribution (NBD) and the
Dirichlet multinomial distribution.
Hence, the Dirichlet is a parsimonious model because it only needs four numerical
inputs – b (brand penetration), w (average purchase frequency), B (market penetration)
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 377
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and W (market purchase frequency) (Ehrenberg et al., 2004). These inputs give the
opportunity to estimate several key BPMs (e.g. penetration, frequency of purchase, repeat
buying, 100 per cent loyal) and postulate that each customer has a certain propensity to
buy a given brand, i.e. making a reservation at a given hotel. This propensity is assumed
to be steady for the time being but to differ across heterogeneous customers (Ehrenberg
et al., 2004). Such heterogeneous behavior aggregates to BPMs (Ehrenberg et al., 2004).
Finally, the Dirichlet model also provides benchmarks when analyzing data for another
year, country (geographical extensions) or category (Ehrenberg et al., 2004). These
benchmarks help in exploring marketing issues, such as customer loyalty programs as an
empirical example. The patterns of observed customer purchase behavior associated with
this model and study are the most important information for estimating customer behavior
(Lacey and Morgan, 2009) and are slowly being recognized over the years, followed by
many replications across different hotels, years and countries to develop their
generalizability.
3. Data and methodology
3.1 Dirichlet assumptions
This work considers ?ve assumptions, following Dirichlet Assumptions and in line with Bassi
(2011), Ehrenberg et al. (2004) and Goodhardt et al. (1984). The ?rst aims to specify the
probability vector of the ith customer card making any speci?c combination {r
j
} of
reservations in the j ? 1 [. . .], g types of hotels with the length T (succession of
reservations). It can be modeled by a multinomial distribution with parameters r, p
1
,. . ., p
g
:
P(r
1
, . . ., r
g
) ? r! ?
j?1
g
?
p
j
rj
rj !
?
where, r is the amount of reservations (quantity) made with card i in Year 1.
The second assumption considers that the probabilities p
j
vary among customers’ cards
according to a Dirichlet distribution with parameters ?
1
[. . .], ?
g
, i.e:
f(p
1
, . . ., p
g?1?
?
1
, . . ., ?
g
) ?
?(?
1
? . . .??
g
)
?(?
1
). . . ?(?
g
)
p
1
?
1
?1
. . . p
g?1
?
g?1
?1
(1 ? p
1
? . . .?p
g?1
)?
g?1
Successive reservations by the ith customer card are independent. These ?rst two
assumptions capture the customer heterogeneity for purchase frequency and for hotel
choice (Ehrenberg et al., 2004). As a third assumption, the number of reservations n
i
made
by ith customer card in each of a succession of equal non-overlapping periods of length T,
follows a Poisson distribution with mean u
i
T. The fourth assumption considers a Gamma
distribution with parameters m and K, which characterizes the variance between mean
purchasing rates and each customer card. Assumptions three and four show the
probabilistic incidence of speci?c purchases of the hotel (Ehrenberg et al., 2004). As for the
last assumption, customers’ hotel-choice probabilities and average-purchase-frequencies
are distributed independently (i.e. statistical independence of these two aspects) over the
loyalty program customers.
The model follows a negative-binomial distribution with mean mT and exponent k, from that
of Bassi’s (2011) work: the number of reservations of the hotel category made by all
customers, in a certain time period (Year 1; Year 2); and the number of reservations a
customer makes of each of the g geographical types of hotels in a period of time T. This is
called the NBD–Dirichlet model:
f
k, m, ?
1
,. . .,?
g
(r
1
, . . ., r
g
) ? f(r?m, k)f
?
1
,. . .,?
g
(r
1
, . . ., r
g?
r
1
? . . .?r
g
? r)
?
(k ? r ? 1) !
r!(k ? 1) !
(
k
m ? k
)
k
(
1 ?
k
m ? k
)
r ?(?
1
, . . ., ?
g
)k!
?
( ?
j?1
g
?
j
? r
)
?
j?1
g ?(?
j
? r
j
)
r
j
!?(?
j
)
Finally, to activate the model, g ? 2 reservations need to be estimated: m, k, ?
1
[. . .], ?
g
.
With the g observed per card purchase rates m
j
, the iterative estimation procedure
PAGE 378 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
calculates the hotel category purchase rate as m ? ?
j?1
g
m
j
and equals the theoretical and
observed hotel chain market shares:
?
j
?
J?1
g
?
j
?
m
j
?
J?1
g
m
j
Further, the hotel chain’s market shares must add up to 1, so there are g ? 1 equations to
be solved. Parameter K is calculated by ?tting the NBD model to the distribution of
reservations of the hotel category.
There are two aspects of customer diversity in the Dirichlet model included in this work,
namely, how customers differ from each other in: their purchasing rates, and their hotel
choice preferences (Goodhardt et al., 1984). Thus, the methodology used in this research
follows the Dirichlet model basic lines, which directly or indirectly describe the buying
behavior patterns and also analyze differences in customers’ preferences for the 12 hotels
included in the same chain group.
3.2 Data and estimation procedures
Based on the Dirichlet model, this research rests on a quantitative observed behavior and
con?rms the in?uence of the environment of brands/hotels and customers. Therefore, this
research is based on data drawn from a hotel chain’s loyalty program (CRM systems) and
covers purchases during the ?rst two years of the program, from April 1, 2011 to March 31,
2012 (Year 1), and from April 1, 2012 to March 31, 2013 (Year 2). The loyalty program
membership is free. For each euro spent, the program reverts 10 points for a silver card,
12 points for a gold card and 15 points for a platinum card, and other offers are also
included. For this study, there were 23,817 cards issued; 6,057 cards used; and 7,532
reservations for Year 1 and 50,358 cards issued; 31,701 cards used; and 44,274
reservations for Year 2. This sample comprises domestic and international customers for
the two years of analysis. Table I show that customers come from Portugal (mainly Year 1)
and Europe (mainly Year 2) and are mostly men (70 per cent) of 50 years old, on average.
Traditionally the customers of this hotel chain tend to have three or more short breaks along
the year, in particular, the ones who have a loyalty card.
Bearing in mind that the aim of the research is to assess the repeat purchase behavior, all
the cards with less than two purchases were discarded even because most of them never
used the loyalty card after the ?rst purchase. Data were analyzed using Dirichlet.xls
software (Kearns, 2009), after a previous validation. The Dirichlet model is usually applied
using package software to generate estimates of the brand performance measures
(Rungie, 2003). Hence, panel data provide all the necessary inputs to calculate the
Dirichlet model and to compare observed data to Dirichlet estimations.
In this work, the iterative estimation method adapted from Bassi’s (2011) Technical Note
assumes geographical and chain effects of a long-term nature. The iterative procedures
adopted require only aggregated data as input, i.e. only input values are needed, such as
hotel penetrations b
i
and average purchase rates m
j
in this work. Hence, this study’s
speci?cation derives from Bassi’s (2011) and from Goodhardt et al.’s (1984) works, which
indicate n customers making purchases in a market (loyalty program) with g brands/hotels.
The present work considers g geographical hotels – the ones belonging to the hotel group.
Table I Socio-demographic characteristics for Year 1 and Year 2
Age (years)
Year 1 Year 2
49-61 50-71
Country Portugal 38% Great Britain 25%
Great Britain 22% Portugal 19%
Germany 7% Spain 8%
Gender M – 72% F – 28% M – 70% M – 30%
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 379
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We consider 12 hotels (?ve “Historical and Cultural”; and seven “Resorts”) from a total of 82
hotels, belonging to the same hotel group. The ?ve “Historical and Cultural Hotels” are
located in Portugal (Algarve, Alentejo, Lisbon and the Tagus Valley (here after LTV), Centre
and North regions) and the seven “Resort Hotels” are located in Portugal (Algarve, LTV and
Madeira regions), Brazil, Europe, America and Africa. Table II indicates that at 7 of the 12
hotels in the sample, the majority of customers in Year 2 come froma country other than that
of the hotel.
Further, the number of countries of customers’ origin increased from 67 to 119 from Year 1
to Year 2, due to the promotion efforts to establish the loyalty card among different
nationalities.
4. Results and discussion
The results suggest that this type of hotel chain loyalty program plays a role in customers’
repeat purchases over time, contributing to their loyalty behavior. Hence, results suggest
that this loyalty program has contributed to holding some polygamy across the hotels of this
group, which is a pattern of hotel customers (member of more than one loyalty scheme),
called “polygamous loyalty” by Dowling and Uncles (1997).
In this sense, we illustrate and summarize the common patterns of repeat purchase and
hotel choice characterized in the Dirichlet model in the three subsections that follow. The
?rst gives results for the customer heterogeneity and analyses the parameters of the
Dirichlet model. The patterns observed and analyzed of the seven brand performance
measures (i.e. brand penetration, average purchase frequency, repeat buying, 100 per
cent loyal, buying once, buying ?ve or more times and market share) are given in the
second subsection. The third and last subsection analyzes the nature and degree of variety
on offer in each hotel and the effect of customers’ types of purchases on hotel
segmentation.
4.1 Customer heterogeneity in the hotel chain loyalty program
With data on cards issued/used and reservations in 12 hotels in a hotel chain loyalty
program, the Dirichlet model was estimated using the iterative method. This model uses
observed market penetration and purchase frequency to estimate m and K and observed
brand penetrations and market shares to estimate S.
Table III shows the results of parameters m, k and S for the two years.
Parameter m represents the mean purchase rate, i.e. it re?ects the size of the market
(Ehrenberg et al., 2004). This parameter increased from 2.9 in Year 1 to 3.3 in Year 2,
meaning that market penetration does not vary greatly among customers in the different
Table II Other socio-demographic characteristics for Year 1 and Year 2
Hotel
Region of customers’ majority
Year 1 Year 2
Madeira Resort Hotel Europe Europe
LTV Resort Hotel Portugal Europe
Europe Resort Hotel Europe Europe
Algarve Resort Hotel Portugal Europe
North Historical and Cultural Hotel Portugal Portugal
Centre Historical and Cultural Hotel Portugal Portugal
LTV Historical and Cultural Hotel Brazil Europe
Africa Resort Hotel Portugal Portugal
Alentejo Historical and Cultural Hotel Portugal Brazil
Algarve Historical and Cultural Hotel Portugal Europe
Brazil Resort Hotel Brazil Brazil
America Resort Hotel America America
PAGE 380 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
brands/hotels. This result suggests the extending effect of time of loyalty program, i.e. that
they do not have an immediate effect on customer behavior.
The parameter k re?ects the extent to which overall purchasing offers differ from the mean,
i.e. how often they buy (Stern and Hammond, 2004). Results reveal that parameter k
decreased from 1.921 in Year 1 to 1.540 in Year 2, and this means that purchasing
frequencies vary greatly among customers (Bassi, 2011), mostly in the second year, where
penetration rate is higher. Further, parameter S also decreases in Year 2 (5.4) compared
with Year 1 (7.8), suggesting that purchase probabilities differ greatly among customers
(Bassi, 2011). Overall these results, in accordance with Bassi, 2011, pointed out that these
customers are heterogeneous and this heterogeneity removes the need for differential
strategies (Ehrenberg et al., 2004).
Thus, parameters m, k and S make estimations of the loyalty program behavior by
estimating some BPMs.
4.2 The two-year effects on BPMs
In Table IV, we present the observed values derived from the brand performance
measures, along with their Dirichlet benchmarks (estimated values) for two years. Brand
penetration (bj) is the key factor that changes when sales increase and is one of the most
important BPMs (Ehrenberg et al., 2004). This performance measure indicates the
percentage of customers buying (reservation) at least once in a speci?ed time period for
the total of existing clients (cards issued) in the loyalty program (Sharp and Sharp, 1997).
The average purchase frequency (wj) is the number of purchases per customer
(reservation) in a speci?ed time for the total of cards used in the loyalty program (Sharp and
Sharp, 1997). The 100 per cent loyal BPM is the one where the customer returns to the
same hotel in two equal-length time periods and repeat buying is the BPM where the
customer returns not to the same hotel, but to the same hotel chain.
To examine differences between brands/hotels in the two equal-length time periods
(years), brand performance measures were adopted, the results of which are shown in
Table IV. According to the ?gures shown in Table IV, there are no signi?cant differences
between observed and estimated values, mainly at the second year of the loyalty program,
due to the fact of the ?rst year being the beginning of the loyalty program. Table IV was built
following observed brand penetration (bj) from the highest to the lowest value. All of the 12
hotels analyzed have an increased observed penetration rate from Year 1 to Year 2. Results
also show the connection between observed average purchase frequency (wj) and
observed buying ?ve or more times – higher values of the former correspond to higher
values of the latter (e.g. LTV Resort Hotel with 2.89 of wj corresponding 5.22 per cent of
buying ?ve or more times in Year 1; Africa Resort Hotel with 3.24 of wj corresponding 2.50
per cent of buying ?ve or more times in Year 1; and Brazil Resort Hotel with 2.44 and 2.06
of wj corresponding 3.81 per cent and 3.63 per cent of buying ?ve or more times in Year
1 and Year 2, respectively).
Furthermore, we analyze groups of brands/hotels by comparing estimated and observed
values. Results indicate that four types of groups exist:
1. the “small trend” group in almost all BPMs (Alentejo and Algarve Historical and Cultural
Hotels; Africa Resort Hotel);
Table III Parameters m, k and S for Year 1 and Year 2
Parameter Year 1 Year 2
M 2.9 3.3
K 1.921 1.540
S 7.8 5.4
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 381
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
b
l
e
I
V
B
r
a
n
d
p
e
r
f
o
r
m
a
n
c
e
m
e
a
s
u
r
e
s
f
o
r
y
e
a
r
1
a
n
d
y
e
a
r
2
–
c
o
m
p
a
r
a
t
i
v
e
t
a
b
l
e
(
o
b
s
e
r
v
e
d
a
n
d
e
s
t
i
m
a
t
e
d
)
b
j
(
%
)
w
j
M
a
r
k
e
t
s
h
a
r
e
1
0
0
%
L
o
y
a
l
R
e
p
e
a
t
b
u
y
i
n
g
(
%
)
w
p
j
O
n
c
e
(
%
)
5
o
r
m
o
r
e
%
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
Y
e
a
r
1
M
a
d
e
i
r
a
R
e
s
o
r
t
H
o
t
e
l
2
.
0
7
1
.
8
4
1
.
1
1
1
.
2
4
9
.
8
6
2
.
4
0
9
.
4
9
2
.
4
4
2
8
.
3
6
4
.
9
5
8
8
.
9
8
8
2
.
4
1
0
.
3
4
0
.
4
0
L
T
V
R
e
s
o
r
t
H
o
t
e
l
0
.
5
4
1
.
2
7
2
.
8
9
1
.
2
3
2
.
5
9
1
3
.
0
5
9
.
4
3
1
3
.
2
1
2
7
.
9
4
4
.
9
6
2
2
.
1
9
8
2
.
6
9
5
.
2
2
0
.
3
9
E
u
r
o
p
e
R
e
s
o
r
t
H
o
t
e
l
0
.
6
4
1
.
0
3
1
.
9
8
1
.
2
3
3
.
0
6
1
6
.
5
7
9
.
4
1
1
6
.
7
1
2
7
.
7
7
4
.
9
6
3
4
.
1
2
8
2
.
8
1
4
.
0
1
0
.
3
8
A
l
g
a
r
v
e
R
e
s
o
r
t
H
o
t
e
l
1
.
0
9
0
.
9
8
1
.
1
2
1
.
2
3
5
.
1
8
6
.
2
0
9
.
4
1
6
.
2
0
2
7
.
7
3
4
.
9
6
8
4
.
5
8
8
2
.
8
4
0
.
3
3
0
.
3
8
N
o
r
t
h
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
6
8
0
.
5
8
1
.
0
5
1
.
2
3
3
.
2
5
2
.
3
4
9
.
3
7
2
.
9
9
2
7
.
4
4
4
.
9
7
9
3
.
3
8
8
3
.
0
4
0
.
0
0
0
.
3
7
C
e
n
t
r
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
5
3
0
.
5
0
1
.
1
6
1
.
2
3
2
.
5
4
7
.
3
3
9
.
3
6
8
.
8
3
2
7
.
3
8
4
.
9
7
8
0
.
8
3
8
3
.
0
8
0
.
1
7
0
.
3
7
L
T
V
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
3
3
0
.
3
0
1
.
1
0
1
.
2
3
1
.
5
7
2
.
4
2
9
.
3
4
2
.
9
6
2
7
.
2
3
4
.
9
7
9
3
.
5
5
8
3
.
1
8
0
.
0
0
0
.
3
6
A
f
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
1
1
0
.
2
8
3
.
2
4
1
.
2
3
0
.
5
1
1
0
.
0
0
9
.
3
4
1
1
.
6
7
2
7
.
2
2
4
.
9
7
1
9
.
1
7
8
3
.
1
9
2
.
5
0
0
.
3
6
A
l
e
n
t
e
j
o
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
3
2
0
.
2
8
1
.
0
8
1
.
2
3
1
.
5
1
4
.
7
5
9
.
3
4
5
.
5
9
2
7
.
2
2
4
.
9
7
8
8
.
8
3
8
3
.
1
9
0
.
0
0
0
.
3
6
A
l
g
a
r
v
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
2
4
0
.
2
1
1
.
1
2
1
.
2
3
1
.
1
2
6
.
7
7
9
.
3
3
7
.
1
4
2
7
.
1
7
4
.
9
7
8
5
.
3
4
8
3
.
2
2
0
.
0
0
0
.
3
6
B
r
a
z
i
l
R
e
s
o
r
t
H
o
t
e
l
0
.
0
9
0
.
1
9
2
.
4
4
1
.
2
3
0
.
4
4
1
3
.
3
3
9
.
3
3
1
4
.
2
9
2
7
.
1
5
4
.
9
7
2
6
.
6
7
8
3
.
2
4
3
.
8
1
0
.
3
6
A
m
e
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
0
4
0
.
0
4
1
.
2
2
1
.
2
3
0
.
1
9
1
5
.
5
6
9
.
3
2
1
5
.
5
6
2
7
.
0
4
4
.
9
8
6
6
.
6
7
8
3
.
3
1
0
.
0
0
0
.
3
6
Y
e
a
r
2
M
a
d
e
i
r
a
R
e
s
o
r
t
H
o
t
e
l
6
.
6
7
6
.
5
9
1
.
4
0
1
.
4
2
8
.
0
7
1
1
.
7
8
9
.
2
7
1
1
.
9
6
3
9
.
6
1
5
.
6
5
6
4
.
4
5
7
3
.
6
1
1
.
6
1
1
.
5
7
A
l
g
a
r
v
e
R
e
s
o
r
t
H
o
t
e
l
3
.
7
4
3
.
4
1
1
.
2
7
1
.
3
9
4
.
5
2
1
0
.
6
7
8
.
9
6
1
1
.
3
4
3
7
.
6
9
5
.
7
0
6
8
.
1
0
7
5
.
0
9
1
.
1
6
1
.
3
9
L
T
V
R
e
s
o
r
t
H
o
t
e
l
2
.
9
5
3
.
0
3
1
.
4
2
1
.
3
9
3
.
5
7
1
3
.
7
3
8
.
9
2
1
4
.
1
7
3
7
.
4
6
5
.
7
0
5
6
.
9
4
7
5
.
2
7
1
.
8
4
1
.
3
7
N
o
r
t
h
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
2
.
7
2
2
.
7
9
1
.
4
2
1
.
3
8
3
.
2
9
8
.
7
3
8
.
9
0
1
0
.
7
2
3
7
.
3
1
5
.
7
1
7
3
.
5
8
7
5
.
3
8
0
.
3
2
1
.
3
6
E
u
r
o
p
e
R
e
s
o
r
t
H
o
t
e
l
1
.
1
8
2
.
0
1
2
.
3
6
1
.
3
8
1
.
4
2
1
5
.
1
8
8
.
8
3
1
6
.
1
1
3
6
.
8
5
5
.
7
2
2
7
.
5
0
7
5
.
7
3
4
.
7
1
1
.
3
2
C
e
n
t
r
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
8
0
1
.
8
0
1
.
3
8
1
.
3
8
2
.
1
7
9
.
0
6
8
.
8
1
1
1
.
7
5
3
6
.
7
2
5
.
7
2
6
9
.
1
0
7
5
.
8
3
0
.
7
9
1
.
3
1
L
T
V
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
7
5
1
.
5
3
1
.
2
0
1
.
3
7
2
.
1
2
8
.
8
4
8
.
7
8
1
0
.
4
8
3
6
.
5
5
5
.
7
3
7
5
.
1
9
7
5
.
9
5
0
.
3
1
1
.
2
9
A
l
e
n
t
e
j
o
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
2
3
1
.
0
1
1
.
1
2
1
.
3
7
1
.
4
8
5
.
2
1
8
.
7
4
6
.
8
9
3
6
.
2
4
5
.
7
3
8
5
.
3
7
7
6
.
1
9
0
.
0
4
1
.
2
7
A
m
e
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
4
4
0
.
9
0
2
.
7
8
1
.
3
7
0
.
5
3
1
8
.
3
5
8
.
7
3
1
8
.
3
5
3
6
.
1
7
5
.
7
3
1
8
.
6
0
7
6
.
2
4
4
.
8
0
1
.
2
6
A
l
g
a
r
v
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
8
0
0
.
8
0
1
.
3
6
1
.
3
7
0
.
9
7
8
.
9
6
8
.
7
2
1
3
.
5
7
3
6
.
1
1
5
.
7
4
6
6
.
2
1
7
6
.
2
8
0
.
6
8
1
.
2
6
B
r
a
z
i
l
R
e
s
o
r
t
H
o
t
e
l
0
.
3
7
0
.
5
6
2
.
0
6
1
.
3
7
0
.
4
5
1
2
.
4
8
8
.
7
0
1
5
.
0
9
3
5
.
9
7
5
.
7
4
3
5
.
8
5
7
6
.
3
9
3
.
6
3
1
.
2
4
A
f
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
3
6
0
.
4
3
1
.
6
2
1
.
3
6
0
.
4
4
5
.
6
5
8
.
6
9
7
.
1
3
3
5
.
9
0
5
.
7
4
5
4
.
9
8
7
6
.
4
5
1
.
7
8
1
.
2
4
PAGE 382 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2. the “med/big trend” group in almost all BPMs (LTV, Centre and North Historical and
Cultural Hotels; America and LTV Resort Hotels;
3. the “logged hotels” group (Brazil and Europe Resort Hotels); and
4. the “stabilized hotels” group (Madeira and Algarve Resort Hotels).
The “small trend” group is identi?ed when:
the estimated brand penetration is lower than observed brand penetration (Ebj ? Obj);
and
estimated average purchase frequency is higher than observed average purchase
frequency (Ewj ? Owj).
The “med/big trend” group follows the opposite pattern (Bassi, 2011). Although results
indicate that Centre and LTV Historical and Cultural Hotels have an Ebj ? Obj, the fact is
that both increased their market share opposed to Alentejo and Algarve Historical and
Cultural Hotels and Africa Resort Hotel. LTV Resort Hotel and North Historical and Cultural
Hotel tend toward being “stabilized hotels,” such as Madeira and Algarve Resort Hotels.
These last hotels have the highest market shares, although Ebj ? Obj. Finally, it is implied
that the frequency with which customers of a brand/hotel bought the whole category (wpj)
increases slightly from 10 to 13 with decreasing market share. Table IV results indicate that
this is the typical trend which was called “natural monopoly” by McPhee (1963). This effect
is more evident in Madeira and Algarve Resorts, which suggests that these “monopolize”
light category customers (Ehrenberg et al., 2004). In this sense, Figure 1 helps us to compare
these ?ndings with a dynamic analysis through a growth evolution [(Year 2-Year 1)/Year 1] of
brand performance measures (BPM).
Figure 1 was initially designed in terms of brand penetration (b) (y-axis) and later according
to other BPMs, such as brand and market average purchase frequency (w and wpj), 100
per cent loyal, repeat buying, buying once and ?ve or more times (x-axis). Figure 1 analysis
mostly focuses on the effects of correlations/synergies between BPMs.
Figure 1 Brand penetration associations with different purchases repeat patterns for
Year 1 and Year 2 – (observed and estimated)
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 383
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
First, as the brand penetration goes up all the other BPMs identi?ed go up to a certain point
and then go down, except buying once, which goes down more and more. However, there
is an “optimal point” at the Madeira Resort Hotel, where all these BPM’s reach their
maximum of growth evolution, except buying once, which reaches its maximum of decline.
Second, Figure 1 shows that, on the one hand, Algarve, LTV and America Resort Hotels;
North, Centre and LTV Cultural and Historical Hotels follow the same path of Madeira
Resort, a way to reach the “optimal point”. On the other hand, Algarve and Alentejo Cultural
and Historical Hotels, Brazil, Europe and Africa Resort Hotels, have to make an effort to
increase the frequency of purchase and brand penetration, respectively.
Finally, Figure 1 indicates a decrease in small brands/hotels, with lower average purchase
frequency, meaning a tendency for their customers to buy less often (Ehrenberg et al.,
2004). This kind of “punished twice” effect just for being small was called “double jeopardy
effect” by McPhee (1963), who explained it as a statistical selection effect. This behavioral
effect, clearly identi?ed in the Africa Resort Hotel, means that the less popular a hotel is, the
less loyal its customers tend to be (Ehrenberg et al., 2004). If a customer buys little from the
?rm, he or she will need to wait a long time for a reward (Liu, 2007). Thus, the customer of
Africa Resort Hotel may not consider the loyalty program relevant. However, if a customer
repeatedly buys, it is because they are truly fond of the program (Stern and Hammond,
2004). In this sense, it is important to analyze these effects on customers’ behaviors and
compare them with the variety of hotels.
4.3 The persistence effect of the types of customers’ purchases and seeking a variety of
hotels
In these ?rst two years of the loyalty program, brands/hotels present differences in
observed and estimated brand performance measures. There are hotels with a higher
observed average purchase frequency than would be expected and a lower observed
penetration than estimated (Africa, Brazil, Europe and LTV Resort Hotels), i.e. selling more
often to the same customer than would be the norm, called excess loyalty by Sharp and
Sharp (1997). However, almost all 12 hotels have increased 100 per cent loyal (BPM),
showing the opposite defended by Sharp and Sharp (1997), which they called “divided
loyalty,” a generalized decrease in 100 per cent loyal (BPM).
Despite the heterogeneity among customers, they are widely expected to fall into relatively
homogeneous and recognizable sub-groupings (Ehrenberg et al., 2004), giving the
opportunity for segmentation between brands/hotels.
Results also show an increase in the repeat purchase item, which could be a non-loyal
attitude toward the recently opening hotels, i.e. an inability to distinguish advantages or that
all competing hotels are seen as similar (Dick and Basu, 1994). However, the reason for the
increase in repeat purchase is likely to be inertia, de?ned as spurious loyalty or true loyalty,
i.e. the customer perceives little differentiation among alternative hotels (Cunningham,
1957; Dick and Basu, 1994).
5. Conclusions and strategic implications
5.1 Conclusions
The Dirichlet model offers feasible methodological paths to explain BPM patterns, as it
allows a combination of purchasing rates and purchasing hotel behavior in a long-term
or/and geographical level. This model considers three parameters (m, k, S), where m
re?ects the market share, k refers to the repeat frequency and S the purchase probabilities.
Thus, the hotels’ positioning, the purchase frequency heterogeneity at geographical and
temporal level, and future choice patterns are underlined by comparing the coef?cients of
these parameters by themed hotel. Empirically, the results suggest that heterogeneity
increases along the two years of the hotel chain loyalty program, as at themed hotels,
voiding the need for segmentation efforts in the execution of the program design and
PAGE 384 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
communication strategies. In fact, customer behavior differs according to localizations (in
Portugal – North, Centre, LTV, Alentejo and Algarve and outside Portugal – America, Africa,
Europe and Brazil) and type of hotel (Resort Hotels or Historical and Cultural Hotels). These
heterogeneities suggest that loyalty programs tend to intensify purchase frequency.
Further, the hotel chain’s loyalty program may result in an increase of penetration mainly at
Historical and Cultural Hotels (low before the program’s implementation) and in more
purchase frequency (low before the program’s implementation) mainly at Resort Hotels.
There is also an increased return buying rate for any hotel and a decrease in the 100 per
cent loyal rate. In this sense, the low permeability between hotels helps to achieve a chain
effect between hotels and therefore helps to plug the polygamy which exists inside
customers.
Changing the fundamental repeat purchase patterns of markets is very dif?cult. However,
it is clearly possible to alter repeat purchase patterns, at least to a small degree, and loyalty
programs are probably the only marketing effort which deliberately focuses on bringing
about such a change. In this work, these changes in patterns by the hotel loyalty program
are achieved through a “chain loyalty” (with different types of hotels and localizations),
through the willingness of the hotel to institute a loyalty program that consists of a
“village-type networking of polygamous customers,” helping to achieve customer loyalty.
5.2 Practical implications
The ?ndings of the study reveal the need and importance of consolidating the hotel group’s
loyalty program to continuously improve the segmented offers.
In this work, results show the level of importance of increasing the number of customers.
Therefore, the data obtained in this work suggested four main groups (“small trend”;
“medium/big trend”; “logged hotels”; and “stabilized hotels”). These groups represent the
hotels’ market share tendency in the Hotel Chain based on the results of the BPM.
Speci?cally, “small trend” is a group containing the smaller hotels (Alentejo and Algarve
Historical and Cultural Hotels; and Africa Resort Hotel) and “medium/big trend” the
opposite (LVT, Centre and North Historical and Cultural Hotels; and America and LVT
Resort Hotel). The “logged” group (Brazil and Europe Resort Hotels) has a huge potential
for growth, but it has not increased their penetration rate enough, which happens with the
“stabilized” group (Madeira and Algarve Resort Hotels).
Furthermore, results also re?ect the need to increase the average amount bought at each
reservation occasion and to induce cross-product buying by existing customers. Due to
these results and based on the four main groups’ classi?cation, we propose reinforcing
communications by creating four segment groups (1 – Algarve, Alentejo, LTV, Centre and
North Historical and Cultural Hotels; 2 – Africa, Europe and Brazil Resort Hotels; 3 –
Algarve, Madeira and America Resort Hotels; 4 – LTV Resort Hotel) to segment suitable
packages of services. In this sense, Groups 1 and 4 need more “heavy buyers,” i.e.
increasing the number of times of returning; and Groups 2 and 3 need more “recent
non-buyers and light buyers,” i.e. more customers. To make this possible, the hotel group
should implement intensive extra-product promotions in Group 1 (e.g. cross-product
coupons and “stay-in coupons” to be discounted on the next reservation), intensive
price-related promotions in Group 2 (e.g. site sales and “on-time coupons” to be
discounted at check-out and after subscription to the loyalty program), some price-related
promotions in Group 3 (e.g. site sales or “on-time coupons”) and some extra-product
promotions in Group 4 (e.g. cross-product coupons or “stay-in coupons”). These four
segmented groups will help justify the importance of a hotel chain’s loyalty program.
5.3 Limitations
This paper adopted the iterative estimation model. The principle of aggregation, i.e.
penetration rate and purchase frequency, however, neither explains behavioral variability
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 385
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
across the 12 hotels nor does it permit prediction of a speci?c behavior in a given hotel. It
is necessary to thoroughly study the frequencies estimated to perform a given behavior. In
addition, this paper only focused on the best hotel in each region/localization,
independently of its dimension. This research is also limited by the time frame; further
research should considered more years and even assessing the impact of these loyalty
programs in other competitors. Further, comparing these results with the ?nancial
achievements of the hotel chain may provide an insightful contribution about the cost
effectiveness of implementing loyalty programmes.
Although this study focused on a behavioral perspective, it does not imply that customer
loyalty is entirely explained. It would help to consider various possible causal sources such
as socio-demographic and attitudinal factors, such as those measuring satisfaction,
commitment, trust and quality relationship (and also the in?uence of the Internet). These
attitudinal ingredients also contribute to understanding customers’ behavior, choices,
concerns and determinants. Finally, conducting a comparative study with other hotel
groups may provide more insights about segmentation procedures.
References
Bassi, F. (2011), “The Dirichlet model: analysis of a market and comparison of estimation procedures,”
Marketing Bulletin, Vol. 22 No. 1, pp. 1-11.
Boland, D. Morrison, D., Morrison, D. and O’Neill, S. (2002), “The future of CRM in the airline industry:
a new paradigm for customer management,” IBM Institute for Business Value, Vol. 3 No. 1, pp. 1-17.
Boulding, W. Staelin, R. Ehret, M. and Johnston, W. (2005), “A customer relationship management
roadmap: what is known, potential pitfalls and where to go,” Journal of Marketing, Vol. 69 No. 4,
pp. 155-166.
Buhalis, D. (2003), Etourism: Information Technology for Strategic Management, Prentice Hall, London.
Buhalis, D. and Law, R. (2008), “Progress in information technology and tourism management: 20
years on and 10 years after the internet – the state of etourism research,” Tourism Management, Vol. 29
No. 4, pp. 609-623.
Cunningham, R. (1957), “Brand loyalty – what, where, how much?,” Harvard Business Review, Vol. 34
No. 1, pp. 116-128.
Czepiel, J. and Rosenberg, L. (1992), “A marketing approach for customer retention,” Journal of
Product and Brand Management, Vol. 1 No. 1, pp. 27-33.
Dick, A. and Basu, K. (1994), “Customer loyalty: toward an integrated conceptual framework,” Journal
of the Academy of Marketing Science, Vol. 22 No. 2, pp. 99-113.
Dowling, G. and Uncles, M. (1997), “Do customer loyalty programs really work?,” Research Brief
No. 1, Centre for Corporate Change at the Australian Graduate School of Management.
Ehrenberg, A. Uncles, M. and Goodhardt, G. (2004), “Understanding brand performance measures:
using dirichlet benchmarks,” Journal of Business Research, Vol. 57 No. 12, pp. 1307-1325.
Frow, P., Payne, A., Wilkinson, I. and Young, L. (2011), “Customer management and CRM: addressing
the dark side,” Journal of Services Marketing, Vol. 25 No. 2, pp. 79-89.
Goodhardt, G., Ehrenberg, A. and Chat?eld, C. (1984), “The Dirichlet: a comprehensive model of
buying behavior,” Journal of the Royal Statistical Society, Vol. 147 No. 5, pp. 621-655.
Hansen, J., Deitz, G. and Morgan, R. (2010), “Taxonomy of service-based loyalty program members,”
Journal of Services Marketing, Vol. 24 No. 4, pp. 271-282.
Kearns, Z. (2009), “Dirichlet.xls,” Marketing Bulletin, Vol. 14, Technical Note 3.
Lacey, R. and Morgan, R. (2009), “Customer advocacy and the impact of B2B loyalty programs,”
Journal of Business and Industrial Marketing, Vol. 24 No. 1, pp. 3-13.
Liu, Y. (2007), “The long-term impact of loyalty programs on consumer purchase behavior and loyalty,”
Journal of Marketing, Vol. 71 No. 4, pp. 19-35.
McPhee, W. (1963), Formal Theories of Mass Behavior, Free Press, New York, NY.
PAGE 386 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Mägi, A. (2003), “Share of wallet in retailing: the effects of customer satisfaction, loyalty cards, and
shopper characteristics,” Journal of Retailing, Vol. 79 No. 2, pp. 97-106.
Reichheld, F. (1993), “Loyalty-based management,” Harvard Business Review, Vol. 71 No. 2,
pp. 64-73.
Reichheld, F. and Sasser, E. (1990), “Zero defections: quality comes to services,” Harvard Business
Review, Vol. 68 No. 5, pp. 105-111.
Rungie, C. (2003), “How to estimate the parameters of the Dirichlet model using likelihood theory in
excel,” Marketing Bulletin, Vol. 14 No. 3, pp. 1-9.
Sharp, B. and Sharp, A. (1997), “Loyalty programs and their impact on repeat-purchase loyalty
patterns,” International Journal of Research in Marketing, Vol. 14 No. 5, pp. 473-486.
Stern, P. and Hammond, K. (2004), “The relationship between customer loyalty and purchase
incidence,” Marketing Letters, Vol. 15 No. 1, pp. 5-19.
Further reading
Bolton, R., Kannan, P. and Bramlet, M. (2000), “Implications of loyalty program membership and
service experiences for customer retention and value,” Journal of the Academy of Marketing Science,
Vol. 28 No. 1, pp. 95-108.
Ehrenberg, A., Goodhardt, G. and Barwise, T. (1990), “Double jeopardy revisited,” Journal of
Marketing, Vol. 54 No. 3, pp. 82-91.
Corresponding author
Pedro Pimpão can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 387
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_274575998.pdf
The main purpose of this work is to evaluate the long-term effectiveness of a hotel’s chain
loyalty program from a behavioral perspective.
International Journal of Culture, Tourism and Hospitality Research
Exploring effects of hotel chain loyalty program
Pedro Pimpão Antónia Correia J oão Duque J osé Carlos Zorrinho
Article information:
To cite this document:
Pedro Pimpão Antónia Correia J oão Duque J osé Carlos Zorrinho , (2014),"Exploring effects of hotel chain loyalty program",
International J ournal of Culture, Tourism and Hospitality Research, Vol. 8 Iss 4 pp. 375 - 387
Permanent link to this document:http://dx.doi.org/10.1108/IJ CTHR-03-2014-0020
Downloaded on: 24 January 2016, At: 22:25 (PT)
References: this document contains references to 25 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 786 times since 2014*
Users who downloaded this article also downloaded:
Lijia (Karen) Xie, Chih-Chien Chen, (2014),"Hotel loyalty programs: how valuable is valuable enough?", International J ournal
of Contemporary Hospitality Management, Vol. 26 Iss 1 pp. 107-129http://dx.doi.org/10.1108/IJ CHM-08-2012-0145
Mark D. Uncles, Grahame R. Dowling, Kathy Hammond, (2003),"Customer loyalty and customer loyalty programs", J ournal
of Consumer Marketing, Vol. 20 Iss 4 pp. 294-316http://dx.doi.org/10.1108/07363760310483676
Lina Xiong, Ceridwyn King, Clark Hu, (2014),"Where is the love?: Investigating multiple membership and hotel customer
loyalty", International J ournal of Contemporary Hospitality Management, Vol. 26 Iss 4 pp. 572-592http://dx.doi.org/10.1108/
IJ CHM-03-2013-0141
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication
Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Exploring effects of hotel chain loyalty
program
Pedro Pimpão, Antónia Correia, João Duque and José Carlos Zorrinho
Pedro Pimpão is a PhD
Researcher at the ISEG –
Lisbon School of
Economics and
Management, Technical
University of Lisbon,
Lisbon, Portugal.
Antónia Correia is a
Professor at the
European University,
Laureate International
Universities, CEFAGE,
Lisbon, Portugal.
João Duque is Professor
at the ISEG – Lisbon
School of Economics and
Management, Lisbon,
Portugal.
José Carlos Zorrinho is a
Professor at the
University of Évora,
Évora, Portugal.
Abstract
Purpose – The main purpose of this work is to evaluate the long-term effectiveness of a hotel’s chain
loyalty program from a behavioral perspective.
Design/methodology/approach – ADirichlet model was estimated to assess purchase frequency and
hotel choice within one of the biggest hotel chains in Portugal. The sample comprises hotels where a
loyalty program was implemented, with a total of 176,099 reservations. Data were extracted from the
customer relationship management (CRM) systems of the hotel group.
Findings – The results suggest that instead of being loyal to a certain hotel, customers are loyal to the
branded hotel chain. As the hotels are all part of the branded group, this polygamy is not only accepted
but also very welcome.
Research limitations/implications – The level of penetration and purchase frequency of CRM was
measured. Nevertheless, a thorough understanding of these will be critical for the success of this
program.
Practical implications – This research is a step toward assessing hotel chain competitiveness, by
improving and suggesting segmented groups of brands/hotels and to induce cross-selling products
accepting polygamous loyalty as the only way to sustain long-term relationships with customers.
Originality/value – This is one of the few research studies, if not the only one, to assess loyalty with
tangible indicators, such as purchase frequency. Further, the results suggest that loyalty programs are
more effective if multiple options are available and as such, cross-selling is perhaps the only way to ?x
customers.
Keywords CRM, Loyalty program, Customer loyalty, Dirichlet model, Repeat purchase behavior
Paper type Research paper
1. Introduction
Internet marketing, as one of the most powerful technological tools, hugely increases
competition across products and even more for tourism destinations, and it has changed
travelers’ behavior. It has enabled customers to engage directly with hotels, without
intermediaries (Buhalis and Law, 2008). One of the reactions of the industry was to develop
loyalty programs to sustain and enact repeat purchases. These programs, which are
supported on one-to-one relationships, provide accurate knowledge of customers’ needs
and wants (Sharp and Sharp, 1997).
The most common loyalty programs are those of customer relationship management (the
so-called CRM). CRM systems have emerged to focus on customer-centric technologies,
helping the implementation of loyalty programs, as an “info-structure” to support its
interoperability, its personalization and its constant networking (Buhalis, 2003; Buhalis and
Law, 2008). CRM studies have been applied in several industries, in particular in airline
alliances (Boland et al., 2002), although none of these researches rely on Dirichlet
Assumptions. In this sense, loyalty programs could be important in increasing the expected
response from hotels to customers, as the data transmitted through the Internet has rapidly
increased in recent years. However, it is costly to initiate and maintain those loyalty
Received 19 March 2014
Revised 11 July 2014
Accepted 4 August 2014
DOI 10.1108/IJCTHR-03-2014-0020 VOL. 8 NO. 4 2014, pp. 375-387, © Emerald Group Publishing Limited, ISSN 1750-6182 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 375
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
programs because of their long-term commitment (Liu, 2007). The cost-effectiveness of
these programs is very controversial. This controversy has been discussed since
Cunningham (1957), but no one till now has reached a reasonable answer to the main
question that is embedded in CRM philosophy – is it worth spending hundreds of
thousands of euros to implement CRM? How effective might these programs be for
positioning the brands/hotels in customers’ minds? For this discussion, Mägi (2003) argues
that membership in loyalty programs increases a customer’s share of wallet for four out of
seven programs and decreases shares of competitors. This is valid at the chain level and
not at the store level.
The challenge of this research is to contribute to reaching a reasonable answer to these
questions in a speci?c context – one of the best-known Portuguese hotel chains worldwide.
Loyalty is assumed to be a behavioral construct that can be measured in light of purchase
measures (Liu, 2007; Reichheld, 1993; Sharp and Sharp, 1997). Hence, it is assumed that
repeat purchase and loyalty are somewhat connected (Dick and Basu, 1994). From an
empirical point of view, the study provides tools for hotels to assess from a tangible
perspective the effectiveness of loyalty programs. Repeat purchase behaviors are
assessed in a temporal and geographical context. Temporal or strength context is used to
account for loyalty persistence (Dick and Basu, 1994; Liu, 2007), whereas geographical or
differentiation context (Dick and Basu, 1994) comprises the chain effect of this program
across the different kinds of hotels of the group.
This article builds on a dynamic performance behavior Dirichlet model to elucidate market
structure. The model is able to describe the various brand performance patterns, and in
that sense, it also helps explain and estimate them (Ehrenberg et al., 2004). The Dirichlet
model estimates these patterns, where big and small brands/hotels differ greatly in how
many buyers they have, but usually far less in how loyal these buyers are (Ehrenberg et al.,
2004). In this sense, this work blends hotel choice and purchase frequency in a
performance assessment process, monitoring results of brands’/hotels’ performance,
through several key performance indicators. It contributes to hotel brand’s market shares
positioning within the hotel chain, where geographical and type of hotel penetration rates
are estimated.
We ?rst describe the research in a theoretical framework, linking CRM systems with loyalty
programs and a Dirichlet model and then develop and estimate the brand performance
measures (BPM). The results illustrate how a chain effect between different localizations
and types of hotels evolves from the variety integrated effect and heterogeneity among
customers to being a loyalty program which helps maintain the polygamy of its customers
to achieve customer loyalty. The study has important implications for managers who are
charged with allocating resources to improve service bene?ts and customer relationships
over time.
2. Theoretical framework
Management of customer relationships is a key activity for the competitiveness of tourism
organizations (Buhalis and Law, 2008), as well for the industry as a whole (Frow et al.,
2011). CRM constitutes a long-term strategic approach that addresses all aspects of
identifying customers, developing customer insights and building customer relationships
(Boulding et al., 2005; Frow et al., 2011; Lacey and Morgan, 2009; Liu, 2007).
In any CRM system of the hotel industry, one major element is the measurement process
(Boulding et al., 2005). This opens the door to the performance assessment process, one
of the ?ve key processes of CRM identi?ed by Frow et al. (2011). The performance
assessment process ensures that ?rms’ strategic objectives are being delivered in an
appropriate and acceptable way, providing ?rms with the opportunity to gain deeper
insights into their customers for future improvements (Boulding et al., 2005; Frow et al.,
2011). Enabling performance monitoring in a micro-view, such as key performance
PAGE 376 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
indicators, has become an integral part of the CRM system and, speci?cally, loyalty
programs (Frow et al., 2011).
The idea that loyalty programs are an important component of ?rms’ CRM strategy is widely
accepted in marketing and tourism areas (Buhalis and Law, 2008; Hansen et al., 2010;
Lacey and Morgan, 2009; Liu, 2007). Loyalty programs or frequency programs are
considered one of the best ways of supporting a ?rms’ knowledge base by forcing stronger
relationships with customers (Buhalis, 2003; Buhalis and Law, 2008; Hansen et al., 2010).
Traditionally, loyalty programs are identi?ed by their degree of indirect defensive
orientation and their two-stage behavior time nature (Liu, 2007; Sharp and Sharp, 1997).
The indirect defensive orientation means building a closer program with current customers
(Dowling and Uncles, 1997), through points rewards systems or special offers. Clearly, a
?rst short-term effect results in the program points to the time of purchase and second, to
a long-term behavior commitment effect through redeeming points for free rewards and,
consequently, increasing pro?ts (Dowling and Uncles, 1997; Hansen et al., 2010; Liu, 2007;
Sharp and Sharp, 1997).
It is not clear that the long-term nature of loyalty programs ends with customer attrition.
However, loyalty programs, supported in dynamic processes and Internet communications
technology (Buhalis and Law, 2008), such as CRM systems, give rise to service ?rms and
help them to identify possible “defectors,” i.e. ?nd customers who are leaving and trying to
win them back (Czepiel and Rosenberg, 1992; Liu, 2007; Reichheld and Sasser, 1990).
Every hotel’s performance should be effectively measured on how well performance targets
are met (Reichheld, 1993; Reichheld and Sasser, 1990) – the key performance measures.
This is why a customer crisis occurs when the service goes unperformed (Czepiel and
Rosenberg, 1992). Thus, behavior performance measures are actions which individuals
adopt and which change their relationship with their environment. Buhalis and Law (2008)
argue that loyalty programs support and help to promote the customization of tourism
products. Hence, Sharp and Sharp (1997) argue that loyalty programs have an effect on
repeat purchases and that is a ?rst step to understanding the effectiveness of
cause-and-effect relationships in the program’s system.
Taking repeat purchases as the focus of a loyalty program requires an instrument to
measure it and to make this study aware of the effectiveness of the potential of such
programs. In this sense, we followed Ehrenberg et al. (2004) and Sharp and Sharp’s (1997)
works who argue that the repeat purchase effect follows a Dirichlet Distribution.
The Dirichlet model is based on the Stated Preference Theory, i.e. a theory about choice
between competitive entities such as brands/hotels, which assumes that customers are
able to order their preferences across different alternatives (Goodhardt et al., 1984).
Moreover, the Dirichlet model does more than explain and describe hotel choice and
purchase frequency simultaneously. It also includes estimates of them, allowing the use of
various performance patterns over a time-span (Goodhardt et al., 1984; Sharp and Sharp,
1997). However, customer characteristics are not included in this stationary model
because these effects are already incorporated in each BPM, and it is not in?uenced by
previous purchases (Ehrenberg et al., 2004).
Using a single statistical model, such as a Dirichlet model, it is comparatively
uncomplicated to describe the various performance patterns of the loyalty program, and to
specify the distribution of purchases by a population of customers of each of the hotels
(Ehrenberg et al., 2004; Goodhardt et al., 1984). The Dirichlet model is particularly relevant
for elucidating these purchase behaviors, i.e. market structure, and follows a combination
of two probability density functions, the negative binomial distribution (NBD) and the
Dirichlet multinomial distribution.
Hence, the Dirichlet is a parsimonious model because it only needs four numerical
inputs – b (brand penetration), w (average purchase frequency), B (market penetration)
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 377
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and W (market purchase frequency) (Ehrenberg et al., 2004). These inputs give the
opportunity to estimate several key BPMs (e.g. penetration, frequency of purchase, repeat
buying, 100 per cent loyal) and postulate that each customer has a certain propensity to
buy a given brand, i.e. making a reservation at a given hotel. This propensity is assumed
to be steady for the time being but to differ across heterogeneous customers (Ehrenberg
et al., 2004). Such heterogeneous behavior aggregates to BPMs (Ehrenberg et al., 2004).
Finally, the Dirichlet model also provides benchmarks when analyzing data for another
year, country (geographical extensions) or category (Ehrenberg et al., 2004). These
benchmarks help in exploring marketing issues, such as customer loyalty programs as an
empirical example. The patterns of observed customer purchase behavior associated with
this model and study are the most important information for estimating customer behavior
(Lacey and Morgan, 2009) and are slowly being recognized over the years, followed by
many replications across different hotels, years and countries to develop their
generalizability.
3. Data and methodology
3.1 Dirichlet assumptions
This work considers ?ve assumptions, following Dirichlet Assumptions and in line with Bassi
(2011), Ehrenberg et al. (2004) and Goodhardt et al. (1984). The ?rst aims to specify the
probability vector of the ith customer card making any speci?c combination {r
j
} of
reservations in the j ? 1 [. . .], g types of hotels with the length T (succession of
reservations). It can be modeled by a multinomial distribution with parameters r, p
1
,. . ., p
g
:
P(r
1
, . . ., r
g
) ? r! ?
j?1
g
?
p
j
rj
rj !
?
where, r is the amount of reservations (quantity) made with card i in Year 1.
The second assumption considers that the probabilities p
j
vary among customers’ cards
according to a Dirichlet distribution with parameters ?
1
[. . .], ?
g
, i.e:
f(p
1
, . . ., p
g?1?
?
1
, . . ., ?
g
) ?
?(?
1
? . . .??
g
)
?(?
1
). . . ?(?
g
)
p
1
?
1
?1
. . . p
g?1
?
g?1
?1
(1 ? p
1
? . . .?p
g?1
)?
g?1
Successive reservations by the ith customer card are independent. These ?rst two
assumptions capture the customer heterogeneity for purchase frequency and for hotel
choice (Ehrenberg et al., 2004). As a third assumption, the number of reservations n
i
made
by ith customer card in each of a succession of equal non-overlapping periods of length T,
follows a Poisson distribution with mean u
i
T. The fourth assumption considers a Gamma
distribution with parameters m and K, which characterizes the variance between mean
purchasing rates and each customer card. Assumptions three and four show the
probabilistic incidence of speci?c purchases of the hotel (Ehrenberg et al., 2004). As for the
last assumption, customers’ hotel-choice probabilities and average-purchase-frequencies
are distributed independently (i.e. statistical independence of these two aspects) over the
loyalty program customers.
The model follows a negative-binomial distribution with mean mT and exponent k, from that
of Bassi’s (2011) work: the number of reservations of the hotel category made by all
customers, in a certain time period (Year 1; Year 2); and the number of reservations a
customer makes of each of the g geographical types of hotels in a period of time T. This is
called the NBD–Dirichlet model:
f
k, m, ?
1
,. . .,?
g
(r
1
, . . ., r
g
) ? f(r?m, k)f
?
1
,. . .,?
g
(r
1
, . . ., r
g?
r
1
? . . .?r
g
? r)
?
(k ? r ? 1) !
r!(k ? 1) !
(
k
m ? k
)
k
(
1 ?
k
m ? k
)
r ?(?
1
, . . ., ?
g
)k!
?
( ?
j?1
g
?
j
? r
)
?
j?1
g ?(?
j
? r
j
)
r
j
!?(?
j
)
Finally, to activate the model, g ? 2 reservations need to be estimated: m, k, ?
1
[. . .], ?
g
.
With the g observed per card purchase rates m
j
, the iterative estimation procedure
PAGE 378 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
calculates the hotel category purchase rate as m ? ?
j?1
g
m
j
and equals the theoretical and
observed hotel chain market shares:
?
j
?
J?1
g
?
j
?
m
j
?
J?1
g
m
j
Further, the hotel chain’s market shares must add up to 1, so there are g ? 1 equations to
be solved. Parameter K is calculated by ?tting the NBD model to the distribution of
reservations of the hotel category.
There are two aspects of customer diversity in the Dirichlet model included in this work,
namely, how customers differ from each other in: their purchasing rates, and their hotel
choice preferences (Goodhardt et al., 1984). Thus, the methodology used in this research
follows the Dirichlet model basic lines, which directly or indirectly describe the buying
behavior patterns and also analyze differences in customers’ preferences for the 12 hotels
included in the same chain group.
3.2 Data and estimation procedures
Based on the Dirichlet model, this research rests on a quantitative observed behavior and
con?rms the in?uence of the environment of brands/hotels and customers. Therefore, this
research is based on data drawn from a hotel chain’s loyalty program (CRM systems) and
covers purchases during the ?rst two years of the program, from April 1, 2011 to March 31,
2012 (Year 1), and from April 1, 2012 to March 31, 2013 (Year 2). The loyalty program
membership is free. For each euro spent, the program reverts 10 points for a silver card,
12 points for a gold card and 15 points for a platinum card, and other offers are also
included. For this study, there were 23,817 cards issued; 6,057 cards used; and 7,532
reservations for Year 1 and 50,358 cards issued; 31,701 cards used; and 44,274
reservations for Year 2. This sample comprises domestic and international customers for
the two years of analysis. Table I show that customers come from Portugal (mainly Year 1)
and Europe (mainly Year 2) and are mostly men (70 per cent) of 50 years old, on average.
Traditionally the customers of this hotel chain tend to have three or more short breaks along
the year, in particular, the ones who have a loyalty card.
Bearing in mind that the aim of the research is to assess the repeat purchase behavior, all
the cards with less than two purchases were discarded even because most of them never
used the loyalty card after the ?rst purchase. Data were analyzed using Dirichlet.xls
software (Kearns, 2009), after a previous validation. The Dirichlet model is usually applied
using package software to generate estimates of the brand performance measures
(Rungie, 2003). Hence, panel data provide all the necessary inputs to calculate the
Dirichlet model and to compare observed data to Dirichlet estimations.
In this work, the iterative estimation method adapted from Bassi’s (2011) Technical Note
assumes geographical and chain effects of a long-term nature. The iterative procedures
adopted require only aggregated data as input, i.e. only input values are needed, such as
hotel penetrations b
i
and average purchase rates m
j
in this work. Hence, this study’s
speci?cation derives from Bassi’s (2011) and from Goodhardt et al.’s (1984) works, which
indicate n customers making purchases in a market (loyalty program) with g brands/hotels.
The present work considers g geographical hotels – the ones belonging to the hotel group.
Table I Socio-demographic characteristics for Year 1 and Year 2
Age (years)
Year 1 Year 2
49-61 50-71
Country Portugal 38% Great Britain 25%
Great Britain 22% Portugal 19%
Germany 7% Spain 8%
Gender M – 72% F – 28% M – 70% M – 30%
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 379
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We consider 12 hotels (?ve “Historical and Cultural”; and seven “Resorts”) from a total of 82
hotels, belonging to the same hotel group. The ?ve “Historical and Cultural Hotels” are
located in Portugal (Algarve, Alentejo, Lisbon and the Tagus Valley (here after LTV), Centre
and North regions) and the seven “Resort Hotels” are located in Portugal (Algarve, LTV and
Madeira regions), Brazil, Europe, America and Africa. Table II indicates that at 7 of the 12
hotels in the sample, the majority of customers in Year 2 come froma country other than that
of the hotel.
Further, the number of countries of customers’ origin increased from 67 to 119 from Year 1
to Year 2, due to the promotion efforts to establish the loyalty card among different
nationalities.
4. Results and discussion
The results suggest that this type of hotel chain loyalty program plays a role in customers’
repeat purchases over time, contributing to their loyalty behavior. Hence, results suggest
that this loyalty program has contributed to holding some polygamy across the hotels of this
group, which is a pattern of hotel customers (member of more than one loyalty scheme),
called “polygamous loyalty” by Dowling and Uncles (1997).
In this sense, we illustrate and summarize the common patterns of repeat purchase and
hotel choice characterized in the Dirichlet model in the three subsections that follow. The
?rst gives results for the customer heterogeneity and analyses the parameters of the
Dirichlet model. The patterns observed and analyzed of the seven brand performance
measures (i.e. brand penetration, average purchase frequency, repeat buying, 100 per
cent loyal, buying once, buying ?ve or more times and market share) are given in the
second subsection. The third and last subsection analyzes the nature and degree of variety
on offer in each hotel and the effect of customers’ types of purchases on hotel
segmentation.
4.1 Customer heterogeneity in the hotel chain loyalty program
With data on cards issued/used and reservations in 12 hotels in a hotel chain loyalty
program, the Dirichlet model was estimated using the iterative method. This model uses
observed market penetration and purchase frequency to estimate m and K and observed
brand penetrations and market shares to estimate S.
Table III shows the results of parameters m, k and S for the two years.
Parameter m represents the mean purchase rate, i.e. it re?ects the size of the market
(Ehrenberg et al., 2004). This parameter increased from 2.9 in Year 1 to 3.3 in Year 2,
meaning that market penetration does not vary greatly among customers in the different
Table II Other socio-demographic characteristics for Year 1 and Year 2
Hotel
Region of customers’ majority
Year 1 Year 2
Madeira Resort Hotel Europe Europe
LTV Resort Hotel Portugal Europe
Europe Resort Hotel Europe Europe
Algarve Resort Hotel Portugal Europe
North Historical and Cultural Hotel Portugal Portugal
Centre Historical and Cultural Hotel Portugal Portugal
LTV Historical and Cultural Hotel Brazil Europe
Africa Resort Hotel Portugal Portugal
Alentejo Historical and Cultural Hotel Portugal Brazil
Algarve Historical and Cultural Hotel Portugal Europe
Brazil Resort Hotel Brazil Brazil
America Resort Hotel America America
PAGE 380 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
brands/hotels. This result suggests the extending effect of time of loyalty program, i.e. that
they do not have an immediate effect on customer behavior.
The parameter k re?ects the extent to which overall purchasing offers differ from the mean,
i.e. how often they buy (Stern and Hammond, 2004). Results reveal that parameter k
decreased from 1.921 in Year 1 to 1.540 in Year 2, and this means that purchasing
frequencies vary greatly among customers (Bassi, 2011), mostly in the second year, where
penetration rate is higher. Further, parameter S also decreases in Year 2 (5.4) compared
with Year 1 (7.8), suggesting that purchase probabilities differ greatly among customers
(Bassi, 2011). Overall these results, in accordance with Bassi, 2011, pointed out that these
customers are heterogeneous and this heterogeneity removes the need for differential
strategies (Ehrenberg et al., 2004).
Thus, parameters m, k and S make estimations of the loyalty program behavior by
estimating some BPMs.
4.2 The two-year effects on BPMs
In Table IV, we present the observed values derived from the brand performance
measures, along with their Dirichlet benchmarks (estimated values) for two years. Brand
penetration (bj) is the key factor that changes when sales increase and is one of the most
important BPMs (Ehrenberg et al., 2004). This performance measure indicates the
percentage of customers buying (reservation) at least once in a speci?ed time period for
the total of existing clients (cards issued) in the loyalty program (Sharp and Sharp, 1997).
The average purchase frequency (wj) is the number of purchases per customer
(reservation) in a speci?ed time for the total of cards used in the loyalty program (Sharp and
Sharp, 1997). The 100 per cent loyal BPM is the one where the customer returns to the
same hotel in two equal-length time periods and repeat buying is the BPM where the
customer returns not to the same hotel, but to the same hotel chain.
To examine differences between brands/hotels in the two equal-length time periods
(years), brand performance measures were adopted, the results of which are shown in
Table IV. According to the ?gures shown in Table IV, there are no signi?cant differences
between observed and estimated values, mainly at the second year of the loyalty program,
due to the fact of the ?rst year being the beginning of the loyalty program. Table IV was built
following observed brand penetration (bj) from the highest to the lowest value. All of the 12
hotels analyzed have an increased observed penetration rate from Year 1 to Year 2. Results
also show the connection between observed average purchase frequency (wj) and
observed buying ?ve or more times – higher values of the former correspond to higher
values of the latter (e.g. LTV Resort Hotel with 2.89 of wj corresponding 5.22 per cent of
buying ?ve or more times in Year 1; Africa Resort Hotel with 3.24 of wj corresponding 2.50
per cent of buying ?ve or more times in Year 1; and Brazil Resort Hotel with 2.44 and 2.06
of wj corresponding 3.81 per cent and 3.63 per cent of buying ?ve or more times in Year
1 and Year 2, respectively).
Furthermore, we analyze groups of brands/hotels by comparing estimated and observed
values. Results indicate that four types of groups exist:
1. the “small trend” group in almost all BPMs (Alentejo and Algarve Historical and Cultural
Hotels; Africa Resort Hotel);
Table III Parameters m, k and S for Year 1 and Year 2
Parameter Year 1 Year 2
M 2.9 3.3
K 1.921 1.540
S 7.8 5.4
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 381
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
b
l
e
I
V
B
r
a
n
d
p
e
r
f
o
r
m
a
n
c
e
m
e
a
s
u
r
e
s
f
o
r
y
e
a
r
1
a
n
d
y
e
a
r
2
–
c
o
m
p
a
r
a
t
i
v
e
t
a
b
l
e
(
o
b
s
e
r
v
e
d
a
n
d
e
s
t
i
m
a
t
e
d
)
b
j
(
%
)
w
j
M
a
r
k
e
t
s
h
a
r
e
1
0
0
%
L
o
y
a
l
R
e
p
e
a
t
b
u
y
i
n
g
(
%
)
w
p
j
O
n
c
e
(
%
)
5
o
r
m
o
r
e
%
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
O
b
s
e
r
v
e
d
(
%
)
E
s
t
i
m
a
t
e
d
(
%
)
Y
e
a
r
1
M
a
d
e
i
r
a
R
e
s
o
r
t
H
o
t
e
l
2
.
0
7
1
.
8
4
1
.
1
1
1
.
2
4
9
.
8
6
2
.
4
0
9
.
4
9
2
.
4
4
2
8
.
3
6
4
.
9
5
8
8
.
9
8
8
2
.
4
1
0
.
3
4
0
.
4
0
L
T
V
R
e
s
o
r
t
H
o
t
e
l
0
.
5
4
1
.
2
7
2
.
8
9
1
.
2
3
2
.
5
9
1
3
.
0
5
9
.
4
3
1
3
.
2
1
2
7
.
9
4
4
.
9
6
2
2
.
1
9
8
2
.
6
9
5
.
2
2
0
.
3
9
E
u
r
o
p
e
R
e
s
o
r
t
H
o
t
e
l
0
.
6
4
1
.
0
3
1
.
9
8
1
.
2
3
3
.
0
6
1
6
.
5
7
9
.
4
1
1
6
.
7
1
2
7
.
7
7
4
.
9
6
3
4
.
1
2
8
2
.
8
1
4
.
0
1
0
.
3
8
A
l
g
a
r
v
e
R
e
s
o
r
t
H
o
t
e
l
1
.
0
9
0
.
9
8
1
.
1
2
1
.
2
3
5
.
1
8
6
.
2
0
9
.
4
1
6
.
2
0
2
7
.
7
3
4
.
9
6
8
4
.
5
8
8
2
.
8
4
0
.
3
3
0
.
3
8
N
o
r
t
h
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
6
8
0
.
5
8
1
.
0
5
1
.
2
3
3
.
2
5
2
.
3
4
9
.
3
7
2
.
9
9
2
7
.
4
4
4
.
9
7
9
3
.
3
8
8
3
.
0
4
0
.
0
0
0
.
3
7
C
e
n
t
r
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
5
3
0
.
5
0
1
.
1
6
1
.
2
3
2
.
5
4
7
.
3
3
9
.
3
6
8
.
8
3
2
7
.
3
8
4
.
9
7
8
0
.
8
3
8
3
.
0
8
0
.
1
7
0
.
3
7
L
T
V
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
3
3
0
.
3
0
1
.
1
0
1
.
2
3
1
.
5
7
2
.
4
2
9
.
3
4
2
.
9
6
2
7
.
2
3
4
.
9
7
9
3
.
5
5
8
3
.
1
8
0
.
0
0
0
.
3
6
A
f
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
1
1
0
.
2
8
3
.
2
4
1
.
2
3
0
.
5
1
1
0
.
0
0
9
.
3
4
1
1
.
6
7
2
7
.
2
2
4
.
9
7
1
9
.
1
7
8
3
.
1
9
2
.
5
0
0
.
3
6
A
l
e
n
t
e
j
o
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
3
2
0
.
2
8
1
.
0
8
1
.
2
3
1
.
5
1
4
.
7
5
9
.
3
4
5
.
5
9
2
7
.
2
2
4
.
9
7
8
8
.
8
3
8
3
.
1
9
0
.
0
0
0
.
3
6
A
l
g
a
r
v
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
2
4
0
.
2
1
1
.
1
2
1
.
2
3
1
.
1
2
6
.
7
7
9
.
3
3
7
.
1
4
2
7
.
1
7
4
.
9
7
8
5
.
3
4
8
3
.
2
2
0
.
0
0
0
.
3
6
B
r
a
z
i
l
R
e
s
o
r
t
H
o
t
e
l
0
.
0
9
0
.
1
9
2
.
4
4
1
.
2
3
0
.
4
4
1
3
.
3
3
9
.
3
3
1
4
.
2
9
2
7
.
1
5
4
.
9
7
2
6
.
6
7
8
3
.
2
4
3
.
8
1
0
.
3
6
A
m
e
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
0
4
0
.
0
4
1
.
2
2
1
.
2
3
0
.
1
9
1
5
.
5
6
9
.
3
2
1
5
.
5
6
2
7
.
0
4
4
.
9
8
6
6
.
6
7
8
3
.
3
1
0
.
0
0
0
.
3
6
Y
e
a
r
2
M
a
d
e
i
r
a
R
e
s
o
r
t
H
o
t
e
l
6
.
6
7
6
.
5
9
1
.
4
0
1
.
4
2
8
.
0
7
1
1
.
7
8
9
.
2
7
1
1
.
9
6
3
9
.
6
1
5
.
6
5
6
4
.
4
5
7
3
.
6
1
1
.
6
1
1
.
5
7
A
l
g
a
r
v
e
R
e
s
o
r
t
H
o
t
e
l
3
.
7
4
3
.
4
1
1
.
2
7
1
.
3
9
4
.
5
2
1
0
.
6
7
8
.
9
6
1
1
.
3
4
3
7
.
6
9
5
.
7
0
6
8
.
1
0
7
5
.
0
9
1
.
1
6
1
.
3
9
L
T
V
R
e
s
o
r
t
H
o
t
e
l
2
.
9
5
3
.
0
3
1
.
4
2
1
.
3
9
3
.
5
7
1
3
.
7
3
8
.
9
2
1
4
.
1
7
3
7
.
4
6
5
.
7
0
5
6
.
9
4
7
5
.
2
7
1
.
8
4
1
.
3
7
N
o
r
t
h
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
2
.
7
2
2
.
7
9
1
.
4
2
1
.
3
8
3
.
2
9
8
.
7
3
8
.
9
0
1
0
.
7
2
3
7
.
3
1
5
.
7
1
7
3
.
5
8
7
5
.
3
8
0
.
3
2
1
.
3
6
E
u
r
o
p
e
R
e
s
o
r
t
H
o
t
e
l
1
.
1
8
2
.
0
1
2
.
3
6
1
.
3
8
1
.
4
2
1
5
.
1
8
8
.
8
3
1
6
.
1
1
3
6
.
8
5
5
.
7
2
2
7
.
5
0
7
5
.
7
3
4
.
7
1
1
.
3
2
C
e
n
t
r
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
8
0
1
.
8
0
1
.
3
8
1
.
3
8
2
.
1
7
9
.
0
6
8
.
8
1
1
1
.
7
5
3
6
.
7
2
5
.
7
2
6
9
.
1
0
7
5
.
8
3
0
.
7
9
1
.
3
1
L
T
V
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
7
5
1
.
5
3
1
.
2
0
1
.
3
7
2
.
1
2
8
.
8
4
8
.
7
8
1
0
.
4
8
3
6
.
5
5
5
.
7
3
7
5
.
1
9
7
5
.
9
5
0
.
3
1
1
.
2
9
A
l
e
n
t
e
j
o
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
1
.
2
3
1
.
0
1
1
.
1
2
1
.
3
7
1
.
4
8
5
.
2
1
8
.
7
4
6
.
8
9
3
6
.
2
4
5
.
7
3
8
5
.
3
7
7
6
.
1
9
0
.
0
4
1
.
2
7
A
m
e
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
4
4
0
.
9
0
2
.
7
8
1
.
3
7
0
.
5
3
1
8
.
3
5
8
.
7
3
1
8
.
3
5
3
6
.
1
7
5
.
7
3
1
8
.
6
0
7
6
.
2
4
4
.
8
0
1
.
2
6
A
l
g
a
r
v
e
H
i
s
t
o
r
i
c
a
l
a
n
d
C
u
l
t
u
r
a
l
H
o
t
e
l
0
.
8
0
0
.
8
0
1
.
3
6
1
.
3
7
0
.
9
7
8
.
9
6
8
.
7
2
1
3
.
5
7
3
6
.
1
1
5
.
7
4
6
6
.
2
1
7
6
.
2
8
0
.
6
8
1
.
2
6
B
r
a
z
i
l
R
e
s
o
r
t
H
o
t
e
l
0
.
3
7
0
.
5
6
2
.
0
6
1
.
3
7
0
.
4
5
1
2
.
4
8
8
.
7
0
1
5
.
0
9
3
5
.
9
7
5
.
7
4
3
5
.
8
5
7
6
.
3
9
3
.
6
3
1
.
2
4
A
f
r
i
c
a
R
e
s
o
r
t
H
o
t
e
l
0
.
3
6
0
.
4
3
1
.
6
2
1
.
3
6
0
.
4
4
5
.
6
5
8
.
6
9
7
.
1
3
3
5
.
9
0
5
.
7
4
5
4
.
9
8
7
6
.
4
5
1
.
7
8
1
.
2
4
PAGE 382 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2. the “med/big trend” group in almost all BPMs (LTV, Centre and North Historical and
Cultural Hotels; America and LTV Resort Hotels;
3. the “logged hotels” group (Brazil and Europe Resort Hotels); and
4. the “stabilized hotels” group (Madeira and Algarve Resort Hotels).
The “small trend” group is identi?ed when:
the estimated brand penetration is lower than observed brand penetration (Ebj ? Obj);
and
estimated average purchase frequency is higher than observed average purchase
frequency (Ewj ? Owj).
The “med/big trend” group follows the opposite pattern (Bassi, 2011). Although results
indicate that Centre and LTV Historical and Cultural Hotels have an Ebj ? Obj, the fact is
that both increased their market share opposed to Alentejo and Algarve Historical and
Cultural Hotels and Africa Resort Hotel. LTV Resort Hotel and North Historical and Cultural
Hotel tend toward being “stabilized hotels,” such as Madeira and Algarve Resort Hotels.
These last hotels have the highest market shares, although Ebj ? Obj. Finally, it is implied
that the frequency with which customers of a brand/hotel bought the whole category (wpj)
increases slightly from 10 to 13 with decreasing market share. Table IV results indicate that
this is the typical trend which was called “natural monopoly” by McPhee (1963). This effect
is more evident in Madeira and Algarve Resorts, which suggests that these “monopolize”
light category customers (Ehrenberg et al., 2004). In this sense, Figure 1 helps us to compare
these ?ndings with a dynamic analysis through a growth evolution [(Year 2-Year 1)/Year 1] of
brand performance measures (BPM).
Figure 1 was initially designed in terms of brand penetration (b) (y-axis) and later according
to other BPMs, such as brand and market average purchase frequency (w and wpj), 100
per cent loyal, repeat buying, buying once and ?ve or more times (x-axis). Figure 1 analysis
mostly focuses on the effects of correlations/synergies between BPMs.
Figure 1 Brand penetration associations with different purchases repeat patterns for
Year 1 and Year 2 – (observed and estimated)
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 383
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
First, as the brand penetration goes up all the other BPMs identi?ed go up to a certain point
and then go down, except buying once, which goes down more and more. However, there
is an “optimal point” at the Madeira Resort Hotel, where all these BPM’s reach their
maximum of growth evolution, except buying once, which reaches its maximum of decline.
Second, Figure 1 shows that, on the one hand, Algarve, LTV and America Resort Hotels;
North, Centre and LTV Cultural and Historical Hotels follow the same path of Madeira
Resort, a way to reach the “optimal point”. On the other hand, Algarve and Alentejo Cultural
and Historical Hotels, Brazil, Europe and Africa Resort Hotels, have to make an effort to
increase the frequency of purchase and brand penetration, respectively.
Finally, Figure 1 indicates a decrease in small brands/hotels, with lower average purchase
frequency, meaning a tendency for their customers to buy less often (Ehrenberg et al.,
2004). This kind of “punished twice” effect just for being small was called “double jeopardy
effect” by McPhee (1963), who explained it as a statistical selection effect. This behavioral
effect, clearly identi?ed in the Africa Resort Hotel, means that the less popular a hotel is, the
less loyal its customers tend to be (Ehrenberg et al., 2004). If a customer buys little from the
?rm, he or she will need to wait a long time for a reward (Liu, 2007). Thus, the customer of
Africa Resort Hotel may not consider the loyalty program relevant. However, if a customer
repeatedly buys, it is because they are truly fond of the program (Stern and Hammond,
2004). In this sense, it is important to analyze these effects on customers’ behaviors and
compare them with the variety of hotels.
4.3 The persistence effect of the types of customers’ purchases and seeking a variety of
hotels
In these ?rst two years of the loyalty program, brands/hotels present differences in
observed and estimated brand performance measures. There are hotels with a higher
observed average purchase frequency than would be expected and a lower observed
penetration than estimated (Africa, Brazil, Europe and LTV Resort Hotels), i.e. selling more
often to the same customer than would be the norm, called excess loyalty by Sharp and
Sharp (1997). However, almost all 12 hotels have increased 100 per cent loyal (BPM),
showing the opposite defended by Sharp and Sharp (1997), which they called “divided
loyalty,” a generalized decrease in 100 per cent loyal (BPM).
Despite the heterogeneity among customers, they are widely expected to fall into relatively
homogeneous and recognizable sub-groupings (Ehrenberg et al., 2004), giving the
opportunity for segmentation between brands/hotels.
Results also show an increase in the repeat purchase item, which could be a non-loyal
attitude toward the recently opening hotels, i.e. an inability to distinguish advantages or that
all competing hotels are seen as similar (Dick and Basu, 1994). However, the reason for the
increase in repeat purchase is likely to be inertia, de?ned as spurious loyalty or true loyalty,
i.e. the customer perceives little differentiation among alternative hotels (Cunningham,
1957; Dick and Basu, 1994).
5. Conclusions and strategic implications
5.1 Conclusions
The Dirichlet model offers feasible methodological paths to explain BPM patterns, as it
allows a combination of purchasing rates and purchasing hotel behavior in a long-term
or/and geographical level. This model considers three parameters (m, k, S), where m
re?ects the market share, k refers to the repeat frequency and S the purchase probabilities.
Thus, the hotels’ positioning, the purchase frequency heterogeneity at geographical and
temporal level, and future choice patterns are underlined by comparing the coef?cients of
these parameters by themed hotel. Empirically, the results suggest that heterogeneity
increases along the two years of the hotel chain loyalty program, as at themed hotels,
voiding the need for segmentation efforts in the execution of the program design and
PAGE 384 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
communication strategies. In fact, customer behavior differs according to localizations (in
Portugal – North, Centre, LTV, Alentejo and Algarve and outside Portugal – America, Africa,
Europe and Brazil) and type of hotel (Resort Hotels or Historical and Cultural Hotels). These
heterogeneities suggest that loyalty programs tend to intensify purchase frequency.
Further, the hotel chain’s loyalty program may result in an increase of penetration mainly at
Historical and Cultural Hotels (low before the program’s implementation) and in more
purchase frequency (low before the program’s implementation) mainly at Resort Hotels.
There is also an increased return buying rate for any hotel and a decrease in the 100 per
cent loyal rate. In this sense, the low permeability between hotels helps to achieve a chain
effect between hotels and therefore helps to plug the polygamy which exists inside
customers.
Changing the fundamental repeat purchase patterns of markets is very dif?cult. However,
it is clearly possible to alter repeat purchase patterns, at least to a small degree, and loyalty
programs are probably the only marketing effort which deliberately focuses on bringing
about such a change. In this work, these changes in patterns by the hotel loyalty program
are achieved through a “chain loyalty” (with different types of hotels and localizations),
through the willingness of the hotel to institute a loyalty program that consists of a
“village-type networking of polygamous customers,” helping to achieve customer loyalty.
5.2 Practical implications
The ?ndings of the study reveal the need and importance of consolidating the hotel group’s
loyalty program to continuously improve the segmented offers.
In this work, results show the level of importance of increasing the number of customers.
Therefore, the data obtained in this work suggested four main groups (“small trend”;
“medium/big trend”; “logged hotels”; and “stabilized hotels”). These groups represent the
hotels’ market share tendency in the Hotel Chain based on the results of the BPM.
Speci?cally, “small trend” is a group containing the smaller hotels (Alentejo and Algarve
Historical and Cultural Hotels; and Africa Resort Hotel) and “medium/big trend” the
opposite (LVT, Centre and North Historical and Cultural Hotels; and America and LVT
Resort Hotel). The “logged” group (Brazil and Europe Resort Hotels) has a huge potential
for growth, but it has not increased their penetration rate enough, which happens with the
“stabilized” group (Madeira and Algarve Resort Hotels).
Furthermore, results also re?ect the need to increase the average amount bought at each
reservation occasion and to induce cross-product buying by existing customers. Due to
these results and based on the four main groups’ classi?cation, we propose reinforcing
communications by creating four segment groups (1 – Algarve, Alentejo, LTV, Centre and
North Historical and Cultural Hotels; 2 – Africa, Europe and Brazil Resort Hotels; 3 –
Algarve, Madeira and America Resort Hotels; 4 – LTV Resort Hotel) to segment suitable
packages of services. In this sense, Groups 1 and 4 need more “heavy buyers,” i.e.
increasing the number of times of returning; and Groups 2 and 3 need more “recent
non-buyers and light buyers,” i.e. more customers. To make this possible, the hotel group
should implement intensive extra-product promotions in Group 1 (e.g. cross-product
coupons and “stay-in coupons” to be discounted on the next reservation), intensive
price-related promotions in Group 2 (e.g. site sales and “on-time coupons” to be
discounted at check-out and after subscription to the loyalty program), some price-related
promotions in Group 3 (e.g. site sales or “on-time coupons”) and some extra-product
promotions in Group 4 (e.g. cross-product coupons or “stay-in coupons”). These four
segmented groups will help justify the importance of a hotel chain’s loyalty program.
5.3 Limitations
This paper adopted the iterative estimation model. The principle of aggregation, i.e.
penetration rate and purchase frequency, however, neither explains behavioral variability
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 385
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
across the 12 hotels nor does it permit prediction of a speci?c behavior in a given hotel. It
is necessary to thoroughly study the frequencies estimated to perform a given behavior. In
addition, this paper only focused on the best hotel in each region/localization,
independently of its dimension. This research is also limited by the time frame; further
research should considered more years and even assessing the impact of these loyalty
programs in other competitors. Further, comparing these results with the ?nancial
achievements of the hotel chain may provide an insightful contribution about the cost
effectiveness of implementing loyalty programmes.
Although this study focused on a behavioral perspective, it does not imply that customer
loyalty is entirely explained. It would help to consider various possible causal sources such
as socio-demographic and attitudinal factors, such as those measuring satisfaction,
commitment, trust and quality relationship (and also the in?uence of the Internet). These
attitudinal ingredients also contribute to understanding customers’ behavior, choices,
concerns and determinants. Finally, conducting a comparative study with other hotel
groups may provide more insights about segmentation procedures.
References
Bassi, F. (2011), “The Dirichlet model: analysis of a market and comparison of estimation procedures,”
Marketing Bulletin, Vol. 22 No. 1, pp. 1-11.
Boland, D. Morrison, D., Morrison, D. and O’Neill, S. (2002), “The future of CRM in the airline industry:
a new paradigm for customer management,” IBM Institute for Business Value, Vol. 3 No. 1, pp. 1-17.
Boulding, W. Staelin, R. Ehret, M. and Johnston, W. (2005), “A customer relationship management
roadmap: what is known, potential pitfalls and where to go,” Journal of Marketing, Vol. 69 No. 4,
pp. 155-166.
Buhalis, D. (2003), Etourism: Information Technology for Strategic Management, Prentice Hall, London.
Buhalis, D. and Law, R. (2008), “Progress in information technology and tourism management: 20
years on and 10 years after the internet – the state of etourism research,” Tourism Management, Vol. 29
No. 4, pp. 609-623.
Cunningham, R. (1957), “Brand loyalty – what, where, how much?,” Harvard Business Review, Vol. 34
No. 1, pp. 116-128.
Czepiel, J. and Rosenberg, L. (1992), “A marketing approach for customer retention,” Journal of
Product and Brand Management, Vol. 1 No. 1, pp. 27-33.
Dick, A. and Basu, K. (1994), “Customer loyalty: toward an integrated conceptual framework,” Journal
of the Academy of Marketing Science, Vol. 22 No. 2, pp. 99-113.
Dowling, G. and Uncles, M. (1997), “Do customer loyalty programs really work?,” Research Brief
No. 1, Centre for Corporate Change at the Australian Graduate School of Management.
Ehrenberg, A. Uncles, M. and Goodhardt, G. (2004), “Understanding brand performance measures:
using dirichlet benchmarks,” Journal of Business Research, Vol. 57 No. 12, pp. 1307-1325.
Frow, P., Payne, A., Wilkinson, I. and Young, L. (2011), “Customer management and CRM: addressing
the dark side,” Journal of Services Marketing, Vol. 25 No. 2, pp. 79-89.
Goodhardt, G., Ehrenberg, A. and Chat?eld, C. (1984), “The Dirichlet: a comprehensive model of
buying behavior,” Journal of the Royal Statistical Society, Vol. 147 No. 5, pp. 621-655.
Hansen, J., Deitz, G. and Morgan, R. (2010), “Taxonomy of service-based loyalty program members,”
Journal of Services Marketing, Vol. 24 No. 4, pp. 271-282.
Kearns, Z. (2009), “Dirichlet.xls,” Marketing Bulletin, Vol. 14, Technical Note 3.
Lacey, R. and Morgan, R. (2009), “Customer advocacy and the impact of B2B loyalty programs,”
Journal of Business and Industrial Marketing, Vol. 24 No. 1, pp. 3-13.
Liu, Y. (2007), “The long-term impact of loyalty programs on consumer purchase behavior and loyalty,”
Journal of Marketing, Vol. 71 No. 4, pp. 19-35.
McPhee, W. (1963), Formal Theories of Mass Behavior, Free Press, New York, NY.
PAGE 386 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH VOL. 8 NO. 4 2014
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Mägi, A. (2003), “Share of wallet in retailing: the effects of customer satisfaction, loyalty cards, and
shopper characteristics,” Journal of Retailing, Vol. 79 No. 2, pp. 97-106.
Reichheld, F. (1993), “Loyalty-based management,” Harvard Business Review, Vol. 71 No. 2,
pp. 64-73.
Reichheld, F. and Sasser, E. (1990), “Zero defections: quality comes to services,” Harvard Business
Review, Vol. 68 No. 5, pp. 105-111.
Rungie, C. (2003), “How to estimate the parameters of the Dirichlet model using likelihood theory in
excel,” Marketing Bulletin, Vol. 14 No. 3, pp. 1-9.
Sharp, B. and Sharp, A. (1997), “Loyalty programs and their impact on repeat-purchase loyalty
patterns,” International Journal of Research in Marketing, Vol. 14 No. 5, pp. 473-486.
Stern, P. and Hammond, K. (2004), “The relationship between customer loyalty and purchase
incidence,” Marketing Letters, Vol. 15 No. 1, pp. 5-19.
Further reading
Bolton, R., Kannan, P. and Bramlet, M. (2000), “Implications of loyalty program membership and
service experiences for customer retention and value,” Journal of the Academy of Marketing Science,
Vol. 28 No. 1, pp. 95-108.
Ehrenberg, A., Goodhardt, G. and Barwise, T. (1990), “Double jeopardy revisited,” Journal of
Marketing, Vol. 54 No. 3, pp. 82-91.
Corresponding author
Pedro Pimpão can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
VOL. 8 NO. 4 2014 INTERNATIONAL JOURNAL OF CULTURE, TOURISM AND HOSPITALITY RESEARCH PAGE 387
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
2
:
2
5
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_274575998.pdf