Description
There is an ever-present need for managers to justify marketing expenditures to the firm. This can only be done when we can establish a direct link between marketing metrics and future customer value and firm performance.
Journal of Retailing 85 (1, 2009) 95–111
Choosing the Right Metrics to Maximize Pro?tability
and Shareholder Value
J. Andrew Petersen
a,?
, Leigh McAlister
b
, David J. Reibstein
c
,
Russell S. Winer
d
, V. Kumar
e
, Geoff Atkinson
f
a
Kenan-Flagler Business School, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, United States
b
McCombs School of Business, University of Texas at Austin, Austin, TX 78712, United States
c
Wharton School, University of Pennsylvania, Philadelphia, PA 19104, United States
d
Stern School of Business, New York, NY 10012, United States
e
J. Mack Robinson School of Business, Georgia State University, Atlanta, GA 30303, United States
f
Overstock.com, Salt Lake City, UT 84121, United States
Abstract
There is an ever-present need for managers to justify marketing expenditures to the ?rm. This can only be done when we can establish a direct
link between marketing metrics and future customer value and ?rm performance. In this article, we assess the marketing literature with regard to
marketing metrics. Subsequently, we develop a framework that identi?es key metrics that ?rms should focus on that can give a ?rm a better picture
of how they got to where they are now and insights towards how they can continue to grow into the future. We then identify several organizational
challenges that need to be addressed in order for ?rms to build the capabilities of collecting the right data, measuring the right metrics, and linking
those metrics to customer value and ?rm performance. Finally, we offer guidelines for future research with regard to marketing metrics to help
?rms establish successful marketing strategies, measure marketing effectiveness, and justify marketing expenditures to top management.
Published by Elsevier Inc on behalf of New York University.
Keywords: Metrics; Customer lifetime value; Customer equity; Shareholder value; Referral behavior
Introduction
“You can’t manage what you don’t measure.”
- Old Management Adage
In recent years, there has been a signi?cant increase in the
number and type of marketing metrics that managers can use
to measure marketing effectiveness and to develop marketing
strategies with the goal of increasing ?rm performance. The
purpose of these marketing metrics is twofold. First, market-
ing metrics serve to increase marketing’s accountability within
the ?rm and to justify spending valuable ?rm resources on
marketing initiatives to top management (Rust et al. 2004b). Sec-
ond, marketing metrics can help managers and retailers identify
the drivers of future customer and ?rm value and build link-
?
Corresponding author.
E-mail addresses: Andrew [email protected] (J.A. Petersen),
[email protected](L. McAlister), [email protected]
(R.S. Winer), [email protected] (V. Kumar), [email protected]
(G. Atkinson).
ages between marketing strategy and ?nancial outcomes. When
retailers are able to identify the drivers of customer and store
value, managers can then maximize customer and store pro?ts
(Kumar et al. 2006a). The increase in the number of available
marketing metrics has been the result of several different fac-
tors. First, increases in database technology have given ?rms the
ability to collect more information about their own customers
and to an extent some information about competitors and their
competitor’s customers. Second, the advent of new channels
of distribution for products and services, such as the Internet,
has signi?cantly increased the availability and complexity of
marketing metrics. Finally, the identi?cation of new drivers of
customer and ?rm value, for example word-of-mouth and refer-
ral behavior (Hogan et al. 2003; Kumar et al. 2007; Reichheld
2003), has led to an increase in the number of different types of
marketing metrics beyond just measurements of customer value
and return on investment.
However, with the abundance of marketing metrics to choose
from (see Farris et al. (2006) for a detailed list of relevant mar-
keting metrics), the challenge marketing managers and retailers
now face is not whether to use marketing metrics, but instead
0022-4359/$ – see front matter. Published by Elsevier Inc on behalf of New York University.
doi:10.1016/j.jretai.2008.11.004
96 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
how to determine which metrics are the most important metrics
to utilize for a given ?rm. While there is no single or “silver”
metric that can summarize marketing performance (Roberts and
Ambler 2006), too many metrics can just provide clutter to
the marketing metrics dashboard. Thus, the most appropriate
metrics are those that are effective at measuring marketing pro-
ductivity, help managers to develop effective forward-looking
marketing strategies, help predict a customer’s future value to
the ?rm, and help predict the ?rm’s future ?nancial performance.
Thus, to choose the appropriate metrics it is necessary for
managers to answer the following ?ve key questions:
1. What metrics are currently in place in different ?rms?
2. What metrics should be in place in different ?rms?
3. How can managers link strategic actions to these new met-
rics?
4. How do these metrics relate to (a) customer value to the ?rm
and (b) ?rm performance?
5. What are the challenges ?rms will face when migrating to
these new metrics?
We address these ?ve key questions in the following four sec-
tions of this paper. We will ?rst reviewthe key marketing metrics
that exist in the marketing literature and in marketing practice.
Second, we present a framework that will help managers iden-
tify key metrics that should be used by different ?rms. Third, we
discuss the steps ?rms can follow to migrate to these metrics.
We also provide examples of situations in marketing research
Table 1
Discussions on multiple metrics.
Author (year) Title (journal) Finding/contribution
Gupta and Zeithaml (2006) Customer Metrics and Their Impact on Financial
Performance (Mktg. Sci.)
Discussion of links between potential unobservable and
observable metrics that have been shown to impact
?nancial performance. In addition a call for future
research that addresses key controversies in
metric-related research.
Unobservable: Customer Satisfaction, Service Quality,
Loyalty, and Intention to Purchase; Observable:
Acquisition, Retention, Cross-selling, Customer
Lifetime Value, Customer Equity.
Rust et al. (2004a) Measuring Marketing Productivity: Current Knowledge
and Future Directions (JM)
Introduces a broad framework for assessing marketing
productivity based on a review of past and current
research. Suggests new paths for research to evaluate
marketing productivity. Speci?cally deals with what we
know and what we would like to know with respect to
the chain of marketing productivity. The chain includes:
Tactical Actions, Strategies, Customer Impact,
Marketing Assets, Marketing Impact, Market Position,
Financial Impact, Financial Position, Impact on Firm
Value, and Value of the Firm.
Farris et al. (2006) Marketing Metrics: 50+ Metrics Every Executive
Should Master (Book)
List and description of major metrics used in academics
and practice. These metrics are broken down into 9
main categories.
These Categories are: (1) Share of Hearts, Minds, and
Markets, (2) Margins and Pro?ts, (3) Product and
Portfolio Management, (4) Customer Pro?tability, (5)
Sales Force and Channel Management, (6) Pricing
Strategy, (7) Promotion, (8) Advertising Media and Web
Metrics and (9) Marketing and Finance.
Ambler (2003) Marketing and the Bottom Line: The Marketing Metrics
to Pump Up Cash Flow (Book)
A book that ?rst argues that multiple metrics are
necessary – since there are many different approaches to
measuring the same performance.
The main focus is on the value of brand equity – noting
that it is a complex asset.
Kumar (2008a, b) Managing Customer for Pro?t (Book) The author shows how to use Customer Lifetime Value
(CLV) to target customers with higher pro?t potential,
manage and reward existing customers based on their
pro?tability, and invest in high-pro?t customers to
prevent attrition and ensure future pro?tability. The
author introduces customer-centric approaches to
allocating marketing resources for maximum
effectiveness, pitching the right products to the right
customers at the right time, determining when a
customer is likely to leave, and whether to intervene,
managing multichannel shopping, and calculating a
customer’s referral value (CRV).
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 97
and practice that identify how ?rms can overcome certain chal-
lenges related to migrating to new marketing metrics. Finally,
we will provide guidelines for future research in marketing that
will continue to extend our knowledge towards selecting the best
marketing metrics that marketing managers can directly link to
key ?nancial outcomes.
A review of marketing metrics
The use of marketing metrics was the main result of pressures
from top management and shareholders to justify marketing
spending. However, the consequence of this growing need for
quantitative measures of marketing and ?rm performance has
led to a multitude of metrics measuring everything from levels
of customer satisfaction to the number of unique clicks to a spe-
ci?c website. The goal of all of these metrics has been to gain
some quanti?able measure of any number of business goals,
from measuring the effectiveness of marketing campaigns to
proxies of ?rm value. While it is important for marketing man-
agers to know the entire consideration set of available metrics,
that is beyond the scope of this paper. For more complete lists
of multiple metrics it is worthwhile to refer to books and journal
articles that discuss multiple metrics (Ambler 2003; Farris et
al. 2006; Gupta and Zeithaml 2006; Kumar 2008a, b; Rust et al.
2004a,b). Inaddition, see Table 1for a discussionof eachof these
books or articles and the key insights each provides. Instead, we
present an assessment of a key set of seven categories of metrics
found in the marketing literature that are relevant to retailers. In
addition, we also discuss two speci?c types of marketing met-
rics with regard to the ?ve key questions this paper sets out to
answer. These include: (1) backward-looking versus forward-
looking metrics and (2) the customer and brand value metrics
that link directly to ?nancial performance and ?rm value. We
discuss how each of these types of metrics have impacted mar-
keting research and practice in addition to how each of these
types of metrics relates directly to manager’s ability to generate
effective marketing strategies.
Assessment of the literature
In this section, we review the literature in marketing that
develops or discusses key metrics that can be used by retailers.
We break these key metrics in the literature into seven distinct
categories and give examples along with some discussion of
literature in each of these categories. These seven categories
include:
1. Brand value metrics
2. Customer value metrics
3. Word of mouth and referral value metrics
4. Retention and acquisition metrics
5. Cross-buying and up-buying metrics
6. Multi-channel shopping metrics
7. Product return metrics
Each of these seven categories of marketing metrics relevant
to retailers serves two main purposes. First, these retailer mar-
keting metrics can be used for strategic and tactical marketing
campaigns. As an example, a marketing manager can strate-
gically use each customer’s predicted referral value score to
determine whichcustomer totarget next time periodwithreferral
incentives (Kumar et al. 2007). In addition, a marketing man-
ager can strategically use each customer’s predicted value (CLV)
to determine which customers to select for a given marketing
campaign that encourages cross-buying, up-buying, or multi-
channel shopping (Kumar and Petersen 2005). Second, these
retailer marketing metrics can be used for short-term or long-
term goals and predictions. As an example, in the short-term the
goal of the ?rm may be to increase the general awareness of
a given brand. Thus, the ?rm would try to increase the overall
percentage of consumers in the marketplace who are aware of
the brand in the following time period (e.g., month or quarter).
However, a long-termgoal resulting fromcontinuous short-term
increases in brand awareness may be to increase overall brand
equity, a long-term metric.
The questions that still remain – How does a manager link
these different types of metrics to strategic goals and how can
these strategic goals be linked directly to customer pro?tabil-
ity and shareholder value? The following discussion provides
a general sense of the literature in marketing that either identi-
?es speci?c marketing metrics or builds conceptual or empirical
linkages between marketing metrics and customer pro?tability,
shareholder value, or both. We also provide a series of tables
that summarize this information (see Tables 2A–2G). We begin
by discussing the marketing metrics found in each of the seven
categories listed previously.
Table 2A
Examples of brand value and brand equity metrics.
Author (year) Finding/contribution
Keller (1993) The author provides a conceptual model of brand equity from the customer’s perspective.
Simon and Sullivan (1993) A brand’s value is the capitalized value of the pro?ts that result from associating that brand’s
name with speci?c products and services.
Kerin and Sethuraman (1998) The authors discuss the link between brand value and shareholder value. The authors ?nd this
link to be positive, but the functional form of the relationship is concave (decreasing returns)
with respect to the ?rm’s Market to Book Ratio.
Madden et al. (2006) Using the Fama-French method, the authors ?nd empirical evidence that stronger brands
deliver stronger shareholder value with less risk to the ?rm.
Leone et al. (2006) The authors provide a discussion on the link between Brand Equity and Customer Equity. The
authors show that while the literature has been divergent in nature, there are great similarities
between the two.
98 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
Table 2B
Examples of customer value metrics.
Author (year) Finding/contribution
Venkatesan and Kumar (2004) The authors de?ne a customer-level CLV objective function and empirically identify the
behavioral drivers of contribution margin and purchase frequency. In addition, the authors
present an optimal resource allocation strategy that can be used by the ?rm to send the
appropriate marketing communications at the right time to maximize pro?ts.
Rust et al. (2004b) The authors de?ne ?nancial return as a function of the change in Customer Equity as a result
of an incremental increase in spending. The authors ?nd drivers of Customer Equity based on
survey information of customers from the Airline industry. In addition, the authors use a
brand-switching matrix to introduce competitive effects into the model estimation.
Gupta et al. (2004) The authors demonstrate how valuing customers makes it feasible to value ?rms, including
high growth ?rms with negative earnings. They demonstrate their valuation method by using
publicly available data for ?ve ?rms. They ?nd that a 1% improvement in retention, margin,
or acquisition cost improves ?rm value by 5%, 1% and .1%, respectively. They also ?nd that a
1% improvement in retention has almost ?ve times greater impact on ?rm value than a 1%
change in discount rate or cost of capital.
Petersen and Kumar (2008) The authors introduce a new CLV objective function that includes product returns as a
separate part of the equation. The authors show that the newly proposed CLV objective
function provides better ?t, better prediction, lower over-prediction bias, and better
segmentation that previous CLV models. In addition, the authors show the ?rm requires fewer
resources to maximize pro?tability when it takes product returns into account.
Brand equity
Many of the metrics that measure brand value or brand equity
stem from the research of Keller (1993) who provides a con-
ceptual model that outlines how to measure brand equity from
the customer’s perspective. Following this, many studies began
to not only measure a customer’s individual brand value or
the ?rm’s brand equity, but also began to link this measure to
both customer equity and shareholder value. In most cases there
have been positive links between increases in brand equity with
increases of customer or ?rm value. For example, using brand
values published in Financial World and Market-to-Book ratios
from publicly traded companies, Kerin and Sethuraman (1998)
found a positive, but decreasing returns, relationship between
brand value and shareholder value. However their sample only
included ?rms that were listed on the “Most Valued Brands” list.
Using a list of brand values from Interbrand and market capi-
talizations from publicly accessible data, Madden et al. (2006)
found evidence that increases in a brand’s strength were related
to increases in shareholder value. This was done by analyzing the
performance of different portfolios of companies – (1) World’s
Most Valued Brands (WMVB), (2) Reduced-Market (RM) –
which contains all those ?rms in the Center for Research in
Securities Market database (except those in WMVB), and (3)
Full-Market (FM) – which contains all ?rms. However, there is
still work to be done in this area. As Leone et al. (2006) points
out, while there are some links between brand equity and cus-
tomer equity, the literature exploring links between brand equity
and customer equity is sparse. This gives a great opportunity
for research to continue to develop methods to link brand and
customer equity.
Customer value
There has been a signi?cant amount of literature that has
set out to develop metrics that measure the value of customers,
whether it is at the individual level in the form of customer life-
time value (CLV) or at the aggregate level (customer equity). Up
Table 2C
Examples of word of mouth and referral value metrics.
Author (year) Finding/contribution
Hogan et al. (2003) The authors empirically show through surveys that losing customers through defection to a
competitor or no longer using a product can have a great impact on the revenue stream across
the life of the product cycle. This is due to the word of mouth effect that is lost when
customers no longer advocate on behalf of the company.
Kumar et al. (2007) The authors present an equation that measures the value of referrals for customers from a
?rm. The authors also show the results of an effective ?eld experiment with a B2C ?rm where
customers were targeted based on their CLV and CRV (Customer Referral Value) scores.
Villanueva et al. (2008) Firms acquire customers through costly but quick investments in advertising and slow but
cheap investments in building word of mouth. The authors ?nd that in the short-term a ?rm
can increase customer equity more using advertising. However, in the long-term the ?rm can
increase customer equity signi?cantly more with investments in word of mouth.
Kumar et al. (2008c) The authors provide a straightforward four-step process for predicting Customer Referral
Value (CRV). Then the authors run a series of three ?eld experiments to determine the most
effective marketing strategy for sending targeted marketing communications to customers
based on CLV and CRV.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 99
Table 2D
Examples of retention and acquisition metrics.
Author (year) Finding/contribution
Thomas (2001) The author shows that customer acquisition and retention are not independent processes.
However, because of data limitations, customer management decisions are frequently based
only on an analysis of acquired customers. This analysis shows that these decisions can be
biased and misleading. The author presents a modeling approach that estimates the length of a
customer’s lifetime and adjusts for this bias.
Verhoef (2003) The author investigates the differential effects of customer relationship perceptions and
relationship marketing instruments on customer retention and customer share development
over time. The results show that affective commitment and loyalty programs that provide
economic incentives positively affect both customer retention and customer share
development, whereas direct mailings in?uence customer share development. However, the
effect of these variables is rather small.
Reinartz et al. (2005) In this research, the authors present a modeling framework for balancing resources between
customer acquisition efforts and customer retention efforts. The key question that the
framework addresses is, “What is the customer pro?tability maximizing balance?” In addition,
they answer questions about how much marketing spending to allocate to customer acquisition
and retention and how to distribute those allocations across communication channels.
Fader et al. (2005) The authors present a new model that links the well-known RFM (recency, frequency,
monetary value) paradigm with customer lifetime value (CLV). The stochastic model,
featuring a Pareto/NBD framework to capture the ?ow of transactions over time and a
gamma-gamma sub-model for dollars per transaction, reveals a number of subtle but
important non-linear associations that would be missed by relying on observed data alone.
to this point, the purpose of measuring CLVand customer equity
has been for optimal customer selection in marketing campaigns
and to measure marketing effectiveness post-campaign. Rust et
al. (2004b) use survey results from consumers located in two
different northeastern US towns to determine drivers of cus-
tomer choice and customer lifetime value. In addition, they are
able to project the return on marketing expenditures for differ-
ent types of campaigns in each of the companies studied and
account for competitive information using a brand switching
matrix. Venkatesan and Kumar (2004) use a sample of B2B
customers from a multinational high-tech ?rm to ?rst deter-
mine the behavioral and demographic drivers of CLV. Then they
determine an optimal resource allocation strategy using genetic
algorithms. The end result is a customer-level resource allo-
cation strategy that maximizes each customer’s lifetime value.
Research in marketing is also beginning to identify linkages
between these metrics and overall ?rmvalue. Gupta et al. (2004)
use information frompublicly traded companies to estimate cus-
tomer equity and ?rm value. They ?nd that as long as a ?rm is
able to project its customer growth pattern and estimate its cur-
rent customer margin that it is feasible to determine customer
equity and overall ?rm value. This is especially important in
situations where ?rms have short histories of transactions, are
involved in a high-growth period, and have a negative cash ?ow
due to early capital investments. Additional research by Kumar
and Shah (forthcoming) was able to ?nd a direct relationship
Table 2E
Examples of cross-buying and up-buying metrics.
Author (year) Finding/contribution
Verhoef et al. (2001) In this article the authors investigate how satisfaction and payment equity affect cross-buying
at a multiservice provider. They also consider its competitors’ performance on these factors.
The results show that the effect of satisfaction differs between customers with lengthy and
short relationships. It also shows that payment equity negatively affects cross-buying for
customers with long relationships. However, if the prices of the supplier are perceived as
fairer than the prices of the competitor, the customers’ probability of cross-buying increases.
Knott et al. (2002) The authors present and evaluate next-product-to-buy (NPTB) models for improving the
effectiveness of cross-selling. The NPTB model reduces the waste of poorly targeted
cross-selling activities by predicting the product each customer would be most likely to buy
next. The ?eld test shows that the NPTB model increases pro?ts compared to a heuristic
approach, and that pro?ts are incremental over and above sales that would have occurred
through other channels.
Kumar et al. (2006b) The authors provide a framework that enables managers to determine the appropriate
product(s) to sell to customers at a given time based on that customer’s history and the history
of other customers at the ?rm.
Kumar et al. (2008a) The authors provide empirical evidence of the antecedents and consequences of crossbuying
in a non-contractual retail setting.
Kumar et al. (forthcoming) The authors provide a framework and methodology for targeting the right customers with the
right products at the right time. They do this by running a ?eld experiment with a company and
comparing the results of a typical sales campaign with one that is driven by their methodology.
100 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
Table 2F
Examples of multi-channel shopping metrics.
Author (year) Finding/contribution
Thomas and Sullivan (2005) This article presents a marketing communications process that uses customer relationship
management ideas for multichannel retailers. The authors describe and then demonstrate the
process with enterprise-level data from a major U.S. retailer with multiple channels.
Kumar and Venkatesan (2005) The authors develop a conceptual framework, which identi?es the customer-level
characteristics and supplier factors that are associated with purchase behavior across multiple
channels. They also propose and empirically show that multichannel shoppers provide
bene?ts as measured by several customer-based metrics.
Venkatesan et al. (2007) The authors explore the drivers of multichannel shopping and the impact of multichannel
shopping on customer pro?tability. The authors provide evidence that multichannel shopping
is associated with higher customer pro?tability.
Pauwels and Neslin (2008) The authors use a multichannel customer management perspective to assess the revenue
impact of adding bricks-and-mortar stores to a ?rm’s already existing repertoire of catalog
and Internet channels. We decompose the revenue impact into customer acquisition,
frequency of orders, returns, and exchanges, and size of orders, returns, and exchanges. The
analysis estimates the net impact of adding the store channel was to increase revenues by
37%. The majority of this increase was due to an improvement in customer retention through
higher purchase frequency.
between CLV and shareholder value. This suggests that if mar-
keting managers can continue to run marketing campaigns to
increase customer value that this will directly lead to increases
in shareholder value. This research is only a start though. There
is still a need to continually improve measures of CLV and to
link CLV to ?nancial outcomes.
Word of mouth and referral value
Ever since Reichheld (2003) suggested that the key to busi-
ness growth lies in the positive word of mouth of a ?rm’s
customers, that is Net Promoter Score, research in marketing
has started to explore this connection between word of mouth
and customer/?rmvalue. Initially, Hogan et al. (2003) were able
to show the value lost over time when a customer disadopts or
defects from a ?rm, using the Bass model for the diffusion of
new products and a Monte Carlo simulation. This lost value was
not only a function of lost purchases, but it was also a function
of the lost word of mouth the customer spread about the prod-
uct causing losses of potential future sales. Even more troubling
to this ?nding is the fact that customers who are acquired via
word of mouth are signi?cantly more pro?table in the long-term
than customers who are acquired via advertising and promotion.
Using data from an Internet ?rm that provided free web hosting
to its registered users, Villanueva et al. (2008) found that cus-
tomer who were acquired using costly, but short-termmarketing
advertisements and promotions give fewer than half the pro?ts
of customers acquired using cheap, but long-term investments
in word of mouth marketing. This makes it even more important
to identify the customers who are valuable with regard to word
of mouth and referral behavior and attempt to retain those cus-
tomers. Additionally, a study by Kumar et al. (2007) using data
froma ?nancial services andtelecommunications ?rmfoundthat
customers with a high CLV are often not the same as customers
with a high customer referral value (CRV), making it is espe-
cially important to know which customers are spreading word
of mouth. Thus, it is critical to not only measure the value of
word of mouth and customer lifetime value, but also to continue
researching ways to link additional metrics such as customer
word of mouth and referral behavior to marketing strategy and
then to ?nancial performance.
Customer retention and acquisition
Increases in customer retention and acquisition are tenets
to successful marketing strategies. However, ?rms need to be
careful not to make decisions about customer acquisition and
customer retention in isolation. Research by Thomas (2001)
Table 2G
Examples of product returns metrics.
Author (year) Finding/contribution
Anderson et al. (forthcoming) The authors show that not including product returns in the demand function creates an
overestimation bias of demand. In addition, the authors empirically show that customers have
an option value for product returns that is measurable.
Petersen and Kumar (forthcoming) The authors use data from a B2C catalog retailer to empirically describe the antecedents and
consequences of product return behavior. The authors ?nd that product returns positively
affect (to a threshold) future purchase behavior – making them necessary, but not evil.
Anderson et al. (2008) The authors run a ?eld experiment with a retail ?rm and empirically show that products
offered on sale have a lower probability of being returned than products purchased at full price.
Petersen and Kumar (2008) The authors integrate product return behavior in the CLV objective function and show that not
including it or including it as a component of buying behavior (i.e., net buying behavior as
products purchased minus products returned) offers signi?cant decreases in predictive
accuracy and optimal resource allocation.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 101
used data from an airline pilot service organization and a
latent-class Tobit modeling framework to show that customer
acquisition and retention are inherently linked. Thus, the ?rm
would never want to only maximize acquisition rates or max-
imize retention rates to maximize pro?tability since customer
retention relies directly on which customers were acquired. This
wouldonlyleadtoacquiringandretainingcustomers whoare not
pro?table in the long term. Instead, it is ideal to maximize pro?ts
from customer lifetime value (CLV) by simultaneously manag-
ing acquisition and retention of customers. Research by Reinartz
et al. (2005) used data from a high-tech B2B ?rm which simul-
taneously modeled acquisition likelihood, relationship duration,
and customer pro?tability. The authors found that it is neces-
sary to quantify trade-off between investments in acquisition
and retention in order to maximize ?rm pro?tability. Once we
can understand the links between customer acquisition and cus-
tomer retention, it is necessary to begin to link these metrics
with some ?nancial outcomes. Recent research by Fader et al.
(2005) used data from CDNOW to link RFM (Recency, Fre-
quency, and Monetary Value) to CLV using iso-value curves.
Once a ?rm is able to start establishing linkages between RFM
and CLV, the next step is to develop marketing strategies that can
continue to use metrics like RFM to increase CLV. Research by
Verhoef (2003) shows how establishing a series of direct mar-
keting campaigns or even loyalty programs that build affective
commitment and potentially lead to increases in future customer
purchase behavior. However, the end result is that while some
linkages have been established between customer acquisition,
customer retention, and CLV, there are still signi?cant opportu-
nities for research in marketing to advance these measures and
linkages.
Cross-buying and up-buying
Cross-buying and up-buying offer ?rms the chance to con-
tinue to increase revenue and pro?t contributions from current
customers, since it has been shown that customers who cross-
buy are more pro?table than customers who do not (Kumar et
al. 2008a). The dif?culty in implementing strategies to effec-
tively increase cross-buying and up-buying is in determining: (1)
which customers are likely to crossbuy, (2) which new products
those customers are likely to purchase, (3) what marketing mes-
sage to send those customers, and (4) when those customers are
likely to crossbuy. Recent research has started to address some of
these issues. First, Kumar et al. (2008a) used data from a major
catalog retailer to identify the drivers and consequences of cus-
tomers who crossbuy in different product categories. Managers
canuse these drivers, suchas average interpurchase time, toiden-
tify the ideal customers within the ?rm’s database that would
be most responsive to cross-selling and up-selling campaigns.
In addition, consequences of crossbuying behavior included
increases in contribution margin per order and number of orders.
Knott et al. (2002) used a Next-Product-to-Buy (NPTB) model
to assist a retail bank with identifying the customers who were
likely to purchase a speci?c loan product. Finally, Kumar et al.
(2006b, 2008d) use data froma B2Bhigh-tech ?rm, a B2C?nan-
cial services ?rm, and a B2Ctelecommunications ?rmprovide a
framework for targeting the right customers, with the right prod-
ucts, at the right time. The results show that when these three
decisions are modeled simultaneously, managers can signi?-
cantly increase their customer targeting accuracy. These studies,
though, only provide results of experiments with a few individ-
ual ?rms. There is still a great opportunity for research to explore
the effects of cross-selling and up-selling strategies on customer
and ?rm pro?tability across different ?rms and industries.
Multi-channel shopping
Where it was less common in the past for ?rms to have a
presence in multiple channels, with the advent of the Internet,
almost all ?rms now have a multichannel presence. The chal-
lenge then is to understand how each of the channels can impact
customer purchase behavior and customer pro?tability. First,
research has shown that customers who shop in multiple chan-
nels are more pro?table than customers who shop in only a
single channel. Using data from a high-tech B2B ?rm, Kumar
and Venkatesan (2005) show that customers who shop across
multiple distribution channels are more likely to score highly on
various customer-based metrics such as revenue and likelihood
to stay active. In addition, Venkatesan et al. (2007) use data from
an apparel retailer to show that customer who purchase across
more distribution channels have a higher future pro?t poten-
tial. However, research is just starting to analyze how ?rms can
develop a strategy to communicate with customers effectively in
different channels. Thomas and Sullivan (2005) use data from a
major US retailer to develop a six-step process of how to man-
age marketing communications with multichannel customers.
Additional research by Pauwels and Neslin (2008) uses data
from a major catalog retailer to quantify the impact of opening
a brick-and-mortar retail store when the only channels the ?rm
previously used was catalog and Internet. However, with the
continuing growth of retailers across many different channels,
several questions still remain. These research questions include
how to effectively migrate customers to different channels or
how to measure the impact of channels where no purchases
occur, for example using the Internet for search and the brick-
and-mortar store for purchase. This leaves ample opportunities
for future research to develop metrics that measure the impact
of multi-channel shopping on customer pro?tability.
Product returns
Up until recently many ?rms have seen product returns only
as a necessary hassle of doing business and a drain on pro?ts.
However, recent research has shown that products returns do
play a major role in the exchange process and customers who do
return a moderate amount of products are, ceteris paribus, the
most pro?table in the future. Petersen and Kumar (forthcoming,
2008) used data from a major US catalog retailer to ?rst show
that customers who return from10 to 15%of purchases purchase
more than customer who return too many or too few products.
The authors also showed that product returns are a key driver in
the computation of CLV and ?rms that do not incorporate prod-
uct returns directly into calculations of CLV will overestimate
CLV and improperly allocate marketing resources. In addition,
Anderson et al. (forthcoming) develop a structural model that
shows that it there is an option value of product returns that
102 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
is measurable, suggesting that omitting product returns from
an estimation of demand creates a bias and that it is possi-
ble to ?nd optimal product return policies for different ?rms.
This is mainly due to the fact that customers who have sat-
isfactory product return experiences tend to purchase more in
the future and have a stronger positive relationship with the
?rm. However, for ?rms to link a product returns metric to cus-
tomer value and ?rm performance more research is needed to
continue to identify the drivers and consequences of product
return behavior. Currently, there are two studies which iden-
tify some of the drivers of product return behavior. Anderson et
al. (2008) use data from a mail order catalog ?rm to compare
varying cross-item and within item attributes and their impact
on demand. One main ?nding was that the lower the price of
the item the less likely the item would be returned. Addition-
ally, Petersen and Kumar (forthcoming) empirically determined
several antecedents of customer product returns, including vari-
ables such as cross-buying and multichannel shopping behavior.
However, the sparse research on customer product return behav-
ior still leaves a signi?cant opportunity for future research to
continue to build product return metrics that can be useful for
managing customers for pro?ts.
As a result of all of this research, there exist many different
metrics for retailers to use to manage their customers. It becomes
increasingly important for managers to understand how each of
these metrics can be used strategically for short-term and long-
term business goals, that is how short-term metrics are linked to
long-term metrics and how both short-term and long-term met-
rics are linked to ?nancial performance. To do so, managers need
to also understand which metrics will provide relevant infor-
mation about the past and the current ?nancial position of the
?rm (backward-looking) and which metrics will help managers
lead ?rms into the future (forward-looking). Next, we provide
some discussion on the differences between backward-looking
and forward-looking metrics and which metrics can be linked to
future ?nancial performance.
Backward-looking versus forward-looking metrics
Many marketing metrics used by ?rms currently are
backward-looking, or at best present-looking, in nature
(Zeithaml et al. 2006). Examples of backward-looking met-
rics include measures of customer satisfaction relating to past
purchase experiences, measures of service quality relating to
past service experiences, and measures of perceived loyalty
that re?ect the customer’s perception of their own behavior up
to the current time period. Many of these backward-looking
metrics, along with several other operational and behavioral
measures, are provided for easy viewing on a periodic basis
(e.g., quarterly, monthly, weekly, daily, and even real-time) for
top managers through marketing dashboards, see Reibstein et
al. (2005) for a detailed description of marketing dashboards.
These backward-looking metrics serve the purpose of helping
marketing managers quantify the effectiveness of past marketing
campaigns that provide a clearer picture of current ?rm perfor-
mance. However, while these metrics can show managers why
the ?rm is at its current state, these metrics have been shown to
offer little to no predictive ability to future customer behavior
or ?rm performance.
More recently much of the academic research has focused
on forward-looking metrics that offer some predictive ability
about future customer behavior or ?rm performance (Kumar
2008b). These forward-looking metrics harness the power of
past customer attitudes and behaviors to try and offer some
predictive capabilities about future customer behavior and ?rm
performance. As a result, many of the backward-looking met-
rics have been used as predictors of future customer behavior
and ?rm performance. However, the results tend to be mediocre
at best. For example, when ?rms measure customer satisfaction,
by the time the data is received and analyzed, it re?ects yester-
day’s perceptions of satisfaction. It also does not include any
information about competitors’ actions or potential customer
prospects. All of these factors cause customer satisfaction to be
a less than adequate predictor of future customer behavior or
?rm performance.
This has led to many new forward-looking metrics, such as
customer lifetime value (CLV). Venkatesan and Kumar (2004)
uses behavioral information about past customer interactions
with the ?rm to predict future customer behavior and customer
value. The authors use variables such as average interpur-
chase time and cross-buying behavior and relate these variables
directly with future purchase frequencies and future contribu-
tion margins. In addition, these ?ndings have led to research in
marketing which has been able to account for competitive inter-
actions with customers. Rust et al. (2004b) use a brand switching
matrix and customer survey information to account for switch-
ing behavior among a set of customers. In an alternative method,
Kumar et al. (2008c) impute each customer’s competitive pur-
chase behavior by analyzing deviations in each customer’s
average interpurchase time. With regard to customer acquisi-
tion, Reinartz et al. (2005) account for a prospect’s likelihood
to buy from a ?rm for the ?rst time by comparing demographic
pro?les of prospects with those of current customers who are
pro?table. The end result of these forward-looking metrics, see
Zeithaml et al. (2006) for a more complete list and discussion,
has been to allow manager to more strategically plan effective
marketing actions and better justify the spending of marketing
resources on both current and potential customers.
Marketing metrics and ?rm value
Many of the metrics used by marketing managers can be
split into to main categories: metrics that measure the value of
brands and metrics that measure the value of customers. Many
of the metrics related to the value of brands were the product
of managers who wanted to quantify the intangible value of
?rms for purposes of mergers and acquisitions. Consequently,
this spurred much research on brand value and brand equity
(Aaker 1991; Keller 1993). Recently, there has been a call for
researchthat links brandequitytocustomer equityand?rmvalue
(Madden et al. 2006). Additional research has also showed how
to link an individual’s brand value to that individual’s customer
value. Kumar et al. (2008b) develop a conceptual framework
that ?rst uses the components of an individual’s brand value and
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 103
Fig. 1. Metrics we need to know.
links this value directly to behavioral outcomes such as purchase
frequency and contribution margin. Then the authors link these
behavioral outcomes with individual-level and ?rm-level ?rm
strategies to enhance CLV and ?rm value. These studies have
shown the importance of brand value metrics in their relation
to customer and ?rm value. It also allows managers to justify
marketing spend on increasing individual brand value and brand
equity since they can directly measure the impact of brand value
on customer and ?rm value.
There have been many different types of metrics that measure
customers, from customer satisfaction to customer behavior.
However, what seems to be a common thread amongst much
of the research on customer metrics is that the end goal is to
relate customer metrics to some form of customer value, for
example CLV or customer equity. There is a signi?cant amount
of research that has been providing frameworks for linking mar-
keting spend to customer value to ?rm value (Srivastava et al.
1998). Perhaps one of the biggest challenges to overcome in this
area is the availability of data on customer and ?rm value. Only
recently has research by Kumar and Shah (forthcoming) been
able to empirically link CLV to shareholder value. The authors
?rst calculate CLV and then link those values to a ?rm’s stock
price over time. This research is an important step for managers
to show that marketing efforts are measurable and are directly
related to ?rm value.
What we need to know
In Fig. 1, we organize the relevant metrics into four groups:
at the customer and store levels for the current and the future.
In addition, while the methods with which one estimates these
metrics at the customer or store level (unit of analysis) and for
the current and future (time period) may differ across the three
key retail channels (online, catalog, and brick-and-mortar) giv-
ing rise to 12 cells (2 unit of analysis ×2 time period ×3 retail
channels), we do not expect that the theoretical constructs them-
selves will differ across channels. Thus, we provide the metrics
and their linkages in a single framework in Fig. 1. We break
down the current metrics into three main categories: (1) transac-
tion information, (2) marketing information, and (3) competitive
information. We then provide some discussion on research that
has started to link these current metrics with future metrics –
both at the customer and store level. Finally, we discuss how
these future metrics are linked directly to ?nancial outcomes for
?rms.
Current value measures at the customer level
Pro?tability, of course, is an important metric with which
to evaluate customers. A number of different approaches have
been suggested to understand the drivers of the current level
104 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
of customer pro?tability. A number of these approaches have
already been covered in the “Assessment of the Literature”.
However, research needs to focus on metrics of pro?tability that
are forward-looking. Research should not use metrics such as
past customer value (PCV) which measures the past pro?t of
a customer. Instead, retailers should consider several key mea-
sures that relate to future pro?tability. One is the popular “RFM”
measures that report the time since the customer’s last purchase
(recency), the frequency with which the customer makes pur-
chases (frequency) and the amount a customer typically spends
(monetary value). Each piece of the RFM measure has been
shown to be a key driver in computing future customer prof-
itability and recent research has also shown how managers can
directly predict CLV by using each of the three inputs of RFM
(Fader et al. 2005).
Similarly, understanding the acquisition rate and the reten-
tion rate can give guidance as to the current numbers of active
customers. In addition, many ?rms use measures of “cost of
acquisition” and “cost of retention” as guidance for which cus-
tomers to acquire and retain. However, research has shown that
it is not necessarily ideal to only focus on trying to acquire cus-
tomers who are inexpensive to acquire and inexpensive to retain
(Reinartz et al. 2005). Instead, it is important to understand that
there are also pro?table customers who are expensive to acquire,
expensive toretain, or both. The keyis determiningwhichfactors
make different customers pro?table to the ?rm. Thus, managers
and retailers instead need to focus on metrics related to acquisi-
tion and retention that focus on acquisition pro?ts and retention
pro?ts. Inaddition, these metrics shouldnot be maximizediniso-
lation. There is an inherent dependence between acquisition and
retention requiring managers to determine spending on acquisi-
tion and retention simultaneously (Reinartz et al. 2005; Thomas
2001).
Past research has used past purchase frequencies to predict
future purchase frequencies. However, it is rare for customers,
especially in non-contractual settings to follow the same past
purchase frequencies over time. Instead research needs to focus
on methods to predict future interpurchase times to help describe
expected customer buying behavior. Regularities in interpur-
chase times have been fruitfully modeled with a generalized
gamma distribution. Allenby et al. (1999) use data from a major
investment brokerage to model inter-trading times of investors.
Additionally, Venkatesan and Kumar (2004) use a generalized
gamma distribution to predict the interpurchase times of the cus-
tomers from a high-tech B2B ?rm. Interpurchase times have
also allowed researchers to impute potential competitive pur-
chase behavior from customers. If a customer tends to follow a
general purchasing pattern, it is possible for statistical models
to uncover deviations in that purchasing pattern and attribute
those deviations to purchases from competitors (Kumar et al.
2008a,b,c,d, forthcoming). Thus, research needs to continue to
develop methods that can predict future customer buying pat-
terns.
Product returns are a growing problem with obvious impact
on a customer’s value, with the amount of product returns
exceeding $100 Billion annually (Blanchard 2007). However,
research has shown that there can be some pro?table bene?ts to
customer product returns. For example, customers who return
a moderate amount of purchases, 10–15% in total, have been
shown to on average purchase more into the future than any other
customer. Inaddition, customer product returnbehavior has been
linked directly to future customer buying behavior and CLV
(Petersen and Kumar forthcoming, 2008). This is a key example
where a metric commonly used for a different purpose outside
of marketing, in this case in operations management for sup-
ply chain management, can be used by marketing researchers to
establishlinkages betweencurrent andfuture customer behavior.
Metrics that go beyond the retailer’s own database to shed
light on a customer’s spending with competitor retailers allows
the retailer to understand what “share of wallet” it is attracting.
This leaves open the possibility that the retailer might focus on
growing customer expenditure from those customers for whom
the retailer currently has a small share of wallet and a large size
of wallet. Du et al. (2007) use data from a major US bank with
information about its customers’ account balances within and
outside of the bank. The authors ?nd little correlation between
the volume of transaction within the bank and with other banks,
along with the fact that a relatively small percentage of cus-
tomers account for a large proportion of external transactions.
This leaves opportunities for the bank to signi?cantly increase
value from customers with low share of wallet and high size of
wallet. In addition, this analysis can also show that it might not
be fruitful to spend signi?cant marketing resources to encourage
cross-selling and up-selling on highly pro?table customers who
are already spending their entire budget (100% share of wallet)
and instead market to customers with potential – those with low
share of wallet and high size of wallet. However, managers need
to be sure not to just base strategic decisions only on a customer’s
current share of wallet. Instead, research needs to focus on link-
ing share of wallet and size of wallet with future metrics such as
CLV by using share of wallet and size of wallet to identify the
optimal set of customers for marketing campaigns.
Current value metrics at the store level
Revenue continues to be an important metric with which
to evaluate stores. Two different models of store revenue were
presented at the Conference on Customer Experience Manage-
ment in Retailing at Babson College, April 24–26, 2008. Fig. 2
presents the model that Len Schlesinger (former COO of The
Fig. 2. The Limited’s model.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 105
Fig. 3. Overstock’s web analytics triangle.
Limited, nowPresident of Babson College) reported having used
to revitalize The Limited’s brick-and-mortar stores. Breaking
store sales down as described in Fig. 2 gave The Limited clear
guidance about how to motivate store personnel. Traf?c to the
store was largely determined by attributes of the shopping cen-
ter and the store’s position in that shopping center. Retail prices
were determined centrally. So incentives were designed to focus
store personnel on the two shaded boxes: conversion percent-
age and units purchased per transaction. The end result was that
these two metrics could be tied directly to future total store sales,
which in turn could be directly linked to future ?rm pro?ts and
shareholder value. Thus, Schlesinger’s experience suggests that
key brick-and-mortar store level metrics should include traf?c,
conversion percentage, units per transaction and average unit
retail price.
Also at that conference, Geoff Atkinson, VP of Tactical Mar-
keting for Overstock.com(an online lowprice retailer) presented
the model in Fig. 3. Atkinson suggested that many websites
make the mistake of looking primarily at conversion percentage
(as The Limited brick-and-mortar store did), which misses the
importance of average order size. Overstock chooses to focus on
revenue per visit as the primary metric of website performance
since that metric takes both conversion and average order size
into account. Overstock uses these metrics when comparing two
or more variables and tracks the results using Omniture and sev-
eral internal systems. Overstock measures site performance on
these metrics daily and weekly. Trends in these metrics are used
to evaluate website design improvements. Overstock’s expe-
rience suggests that online store level metrics should include
orders, revenue, visits, conversions, average order size and rev-
enue per visit – similar to those in brick-and-mortar and other
of?ine channels (e.g., catalogs).
In addition to the revenue focused metrics presented by The
Limited and Overstock.com, we also need metrics linking store
revenue to store pro?t. Such measures include marketing spend-
ing, pro?t margins, contribution dollars, and awareness of retail
stores and speci?c retail brands. In general if we follow the
hierarchy of effects model (awareness, liking, trial, repeat, loy-
alty), it is thought that raising awareness of the retail store or
a given brand will eventually lead to an initial and potentially
repeat purchase. In addition, if the store has a grasp on its pro?t
from the sales of goods (pro?t margin and contribution dollars)
and its ability to acquire and retain customers (effectiveness of
marketing spend), with the general awareness of the store and
its products/services, then a direct link can be drawn to ?rm
pro?tability and shareholder value.
Future value measures at the customer level
Virtually all measures of the important metric “Customer
Lifetime Value” are built with historical data and hence re?ect
the customer’s expected future value if nothing changes as the
customer moves into the future. But, of course, things will
change and some of those changes are under the control of
marketers. There is need for forward-looking measures of a
customer’s expected lifetime value and those measures should
include the impact of different marketing programs on that life-
time value.
Questioning Reichheld’s (2003) work on a “net promoter
score” (the percentage of surveyed customers who report that
they are willing to recommend this company to a friend) led to
exploring that a customer’s value can be more than the value
of the purchases that customer makes from the company as
observed by Kumar et al. (2007). Speci?cally, Kumar et al.
(2007) de?ne a customer’s referral value or CRV as the value
of the business that a customer brings in minus the marketing
costs that prompted that customer to make a referral. However,
this research is just the beginning for metrics related to word of
mouth and referrals. Research needs to continue to focus on the
drivers of the value of word of mouth and its implications on
customer and ?rm value.
Future value metrics at the store level
Positive “word of mouth”, particularly as spread on the Inter-
net, has been shown to be an important indicator of future
success. Godes and Mayzlin (2004b) linked chat room com-
ments to TV show ratings, Chevalier and Mayzlin (2006) linked
online reviews and ratings to online book sales, and Dhar
and Chang (2007) linked CD sales to blog posts and reviews.
Measures of retailer-related online chat, reviews and blogs are
relevant future value metrics at the store level. Research needs
to continue to work on linking general word of mouth or ‘buzz’
about products to future sales. This way managers can predict
the impact of a word of mouth marketing campaign is likely to
have on future sales to determine the optimal level of investment.
A retailer’s own brand equity, as measured by consumer sur-
veys, should be related to the future value of the ?rm. As noted
by Leone et al. (2006), one could also measure a retailer’s brand
equity as the sum of the customer equity (closely related to an
aggregate measure of customer lifetime value) associated with
each of the retailer’s customers. On a related point, Leone et
al. (2006) point out as well that the value of a manufacturer’s
brand to a retailer is not the same as the value of that brand to
the manufacturer. In particular, the value of a manufacturer’s
brand to a retailer is related to the value, to the retailer, of the
customers who buy that brand fromthe retailer. Thus, going for-
ward research should focus on trying to link the increase of a
?rm or a retailer’s brand value to a corresponding increase in
total sales. Further, it is also important for some ?rms to iden-
tify an individual’s brand value (IBV) and relate that IBV to
106 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
that customer’s lifetime value (Kumar et al. 2008b). Then the
?rm can also make strategic decisions to build brand value and
in turn customer value at the individual level through targeted
marketing campaigns.
Just as we noted that share of wallet is an important cus-
tomer level, current value metric that requires data beyond that
found in a retailer’s CRM database, “growth of the retailer’s
customer base” is a store level, future value metric that requires
data beyond that in a retailer’s CRMdatabase. In order to growa
store’s customer base, the retailer needs information about con-
sumers who are not currently shopping at the store. With this
information a retailer can compute its share of customers across
the industry or in the marketplace and also identify the best
prospects to target in order to grow the customer base. There
are signi?cant challenges to gathering data across all ?rms in an
industry or even to gather panel data from a large sample of cus-
tomers. However, it is still necessary for managers to understand
what factors impact a growth in their customer base – whether it
includes customers that switch from one company to another or
whether it includes new customers who have never purchased in
a given product category. Either way, research has to continue to
develop methods for predicting a ?rm’s growth in its customer
base.
Finally, the ultimate store level, future value metric is a ?nan-
cial outcome: shareholder value or stock price. The market’s
expectations about the future of the ?rm are impounded in a
?rm’s shareholder value or stock price. It would be particularly
useful to develop an understanding of the factors that in?uence a
retailer’s stock price. In other words, what are the current/future
and customer/store level metrics that can be linked to a retailer’s
stock price and in turn how can these metrics be strategically
used to increase a retailer’s stock price? To this point few stud-
ies have been able to build links between marketing metrics
and shareholder value or stock price, with few exceptions (e.g.,
Kumar and Shah forthcoming). However, there are still ample
opportunities to continue to develop these linkages, especially
since it is often necessary for marketers to continue to justify
their resource allocations with empirical evidence showing the
impact of their decisions.
How do we get there?
As noted earlier in this paper, it is critical to develop metrics
in twelve “cells” of activity: at both the customer and store level
and each with implications for both the current and the future
time period. Then, each of those cells must be measured for
the three key channels – online, catalog, and brick-and-mortar.
An immediate question is how do we obtain the data neces-
sary for these measurements? There are immediate challenges
that are faced by both researchers and practitioners when devel-
oping and implementing marketing strategies based on metrics
that are directly linked to customer value and ?rm performance.
First, ?rms need to be able to capture the appropriate customer
and marketing data to even develop measures for different met-
rics. Then, managers and academics alike need to determine
how to operationalize these variables so that they can be used
effectively. This can only be done when the ?rm captures data
at relevant intervals of time. Managers then need to properly
track the incoming data and the effectiveness of marketing pro-
grams to continue to develop new programs. Next, it is crucial
to use this data to build ?eld experiments to test the effective-
ness of these predictions. Finally, it is necessary for these ?rms
to disseminate this knowledge within ?rms to gain buy-in from
not only marketing and sales, but also from top management.
Only then can marketing justify its position within the ?rm. We
discuss each of these points in detail below.
Data capture
Let us assume that we can use the basic data obtained for
both current metrics and for forecasting, extrapolation into the
future, or both. This implies that we need measures from the
three channels at two levels of aggregation. To take measures at
the customer level implies that the ?rmmust have an information
systemin place to systematically track and capture that informa-
tion. Consider the online case. There is a considerable amount
of data generated from online behavior but only purchase data
can be captured longitudinally.
1
Brick-and-mortar channels only
permit longitudinal data collection through devices like super-
market panels which address a very small sample of consumers.
The best data are obtained from catalogs where it is relatively
easy and customary to capture customer-level data. Ultimately,
these data can then be aggregated to the store level. However,
take the instance of a ?rm that desires to implement a CRM
system to capture data about current and potential customers.
Research suggests that about 70% of CRM system implementa-
tions fail in improving the bottom-line. How might a ?rm create
a successful approach to capturing data? Much of the success of
technology adoption in a ?rm comes from a ?rm’s willingness
to develop an incentive structure based on CRM usage and per-
formance and the willingness to initiate, maintain, and terminate
relationships with customers based on the recommendations of
the CRM system (Krasnikov et al. 2008; Reinartz et al. 2004).
Thus, it is important for ?rms to not just begin to collect data
about customers, but to adopt data capture technologies that are
aligned with the incentive structure of the ?rm.
Operationalization
Then once the data is collected, the next task is to identify
howeach of the metrics will be operationalized. In simpler cases,
metrics such as pro?t at the customer level can directly mean
how much pro?t a customer has provided the retailer in a given
time frame. However, many metrics are much more dif?cult to
operationalize. We only need to look at metrics such as customer
satisfaction and customer loyalty to see the issues of metric oper-
ationalization. How should a ?rm operationalize a metric such
as loyalty? We could de?ne loyalty any number of ways. First, in
terms of behavioral loyalty, it could refer to customer tenure, or
the length of time since the customer made a purchase. It could
1
Considerable progress is being made on “behavioral targeting” which per-
mits search behavior across web site visits to be captured, but these data are not
typically added to a customer’s data ?le.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 107
also be de?ned as the number of times a customer has purchased
from a given ?rm. Finally, it could also be a combination of sev-
eral metrics, such as an RFM score. In addition, loyalty could
also be attitudinal. Firms could measure loyalty through sur-
veys as the perception of the ?rm in the eyes of the customer.
Also, other metrics which may seemstraightforward to measure,
such as product returns, can give different implications to man-
agers based on their operationalization. For example, a recent
study showed that three different operationalizations of product
returns – (1) number of product returns, (2) value of product
returns, and (3) ratio of products returned to total products pur-
chased – gave three different conclusions to managers. This is
a major problem when marketing managers try to link market-
ing metrics to ?nancial performance. In different cases, different
variable operationalizations provide vastly different results. This
makes it extremely important for marketers to begin to standard-
ize a single method of measuring each metric and to always use
that method when linking a metric to a ?nancial outcome. This
can increase the validity of the measure and also allow for com-
parisons of metrics across ?rms, across research studies, and
across time.
Measurement interval
Another relevant issue is the measurement or data interval.
The appropriate interval should be matched to the product cat-
egory. For frequently purchased products, weekly measures are
appropriate. While not all customers purchase every week, it is
critical particularly for the store-level measures to have short
measurement intervals. This also permits the analyst to develop
the best measures for the metrics describing future behavior.
For durable goods or other products/services purchased less
frequently, weekly data are probably unnecessary. In any case,
the measurement interval should always be less than the aver-
age interpurchase time in order to account for heterogeneity in
purchasing across consumers. A main issue that arises out of
different measurement intervals occurs when different depart-
ments of a ?rm capture and measure data at different intervals.
For example, take a large multi-national CPG company. It may
be the case that this company measures sales across all stores on
a weekly basis. This company may also measure advertising and
other marketing spend on a monthly basis. Finally, this company
may measure changes in distribution of products on a quarterly
basis. Take another example of a high-tech B2B?rm. This high-
tech B2B ?rm collects information about sales to customers on
a monthly level, but keeps information about marketing expen-
ditures to customers (e.g., sales calls) as they occur. Thus the
question is: Which level of time aggregation should be used for
a longitudinal study? Should the ?rm aggregate the data to the
monthly quarterly level and lose the variability of the weekly or
daily transaction data? The problem here is a loss of variance
in the more frequently measured data due to aggregation. Or,
should the ?rmbring the quarterly or monthly data to the weekly
or daily level by using average weekly or daily values for mar-
keting and distribution data? The problem here is that the ?rm
maymake not gainthe full picture of the purchasingor marketing
cycle by tracking it too frequently. In addition, the correct answer
to both questions may depend on the research question that we
are trying to answer. To avoid this problem in the ?rst place,
?rms need to consider appropriate data measurement intervals
andcapture data across all facets of the business at those intervals
most appropriate for the products and services being sold.
Tracking
A basic level of analysis is tracking the metrics. While this
does not show relationships between metrics, two-dimensional
visual representations (metric over time) can stimulate discus-
sion about the causes of a particular up or down trend. In
addition, tracking is easy to communicate to senior management
and throughout the organization. However, absolutely crucial is
going beyond tracking to develop analyses that link marketing
activities and the metrics. This gives the company the ability
to allocate resources over marketing programs that will pro-
vide the greatest leverage in expanding the particular metric. For
example, if the metric is customer lifetime value in the catalog
channel, linking number of catalogs sent to a customer’s CLV
aids in the determination of the optimal number of catalogs to
mail to each customer. Venkatesan and Kumar (2004) used data
from a high-tech B2B ?rm to track the frequency and timing of
different types of marketingcommunications andtheir impact on
a customer’s decision to purchase. This gives the ?rm an under-
standing of how decisions to allocate resources to customers are
likely to affect a customer’s decision to purchase in the future.
Similarly, if the metric is at the ?rmlevel such as market share or
excess stock price returns, companies can evaluate the impact of
programs such as R&D, branding, and so forth on these metrics.
Measuring alone is insuf?cient – developing models and estab-
lishing relationships between “inputs” and “outputs” is essential.
How can this be done? A recent study which won the Market-
ing Science Practice Prize competition in 2003, exhibits exactly
how a German catalog company successfully tracked its mar-
keting communications over time to optimize catalog sending
to customers (Elsner et al. 2004). The authors used a dynamic
multi-level modeling approach with the elasticities of catalog
mailing along with segmenting customers using RFM scores to
determine optimal catalog mailing practices. This led the catalog
company to go from the 5th to the 2nd market position.
Similarly, it is important to develop relationships between the
metrics themselves to see if there are “suf?cient statistics,” that
is, measures that are highly correlated with others implying that
the dashboard can be smaller. For example, ?rm-level metrics
such as pro?tability and share price should be highly correlated.
In addition, it would be useful to know if a particular metric is
a leading indicator for another. For example, changes in brand
equity should lead changes in sales, market share, or both.
Experimentation
The tracking and linking analyses just described are essential
parts of a metrics program. However, a very useful step beyond
these is to run controlled ?eld experiments where the manager
manipulates some aspect of the marketing program and moni-
tors how it affects a selected metric. Particularly, amenable to
108 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
experimentation are the online and catalog channels. The latter
has been the subject of experimentation for many years, the for-
mer since the inception of the Web-based channel. For example,
a clothing retailer that uses both a catalog and an Internet site
might be interested in seeing the impact of changing the mer-
chandise mix or pricing on sales. The company would randomly
selected customers who either visit the Web site or receive a
catalog and offer the changed merchandise or price and keep
another group as a control. Given the availability of customer-
level data, however, the impact of the policy change on CLV
could also be determined. While this can be done in a brick-and-
mortar channel, it is much more dif?cult to implement given the
need to coordinate over stores and measure the impact at the
customer level (at least for the CLV metric). However, there is
limited research in marketing that deals with direct experimenta-
tion. We ?rst see experimentation in marketing research Eastlack
and Rao (1986) with the Campbell Soup company. However,
since that time, there have been few studies that have conducted
actual experiments with ?rms. An example of a more current
?eld experiment includes Kumar et al. (2006b, 2008d) where the
authors used data from a high-tech B2B ?rm and B2C telecom-
munications and ?nancial services ?rms to guide salesperson
decision-making. While it is more dif?cult to publish studies
that deal with experiments, it is much needed and can take care
of many confounding factors.
Dissemination
From an organizational perspective, it is important that key
metrics be disseminated throughout the company. Employees
should be aware that senior management is very interested in
both company- and customer-level metrics as it not only gives
direction for data collection, but also demonstrates that man-
agement is seriously interested in its customers. There is also
a rationale for “common ground” in that all managers need to
know how their brands are being evaluated. Thus, there is a
serious signaling effect that communicating key metrics has in
the organization. Employees who own stock in the company are
used to tracking the share price daily; they should also get used
to tracking the key metrics at their desks.
Most companies are trying to do the communication through-
out the organization through the use of a dashboard. This allows
senior management the ability to indicate what are the important
measures that all within the organization should be looking at
in common (Pauwels et al. 2008). It has been reported that the
McDonald’s CEO used to have the corporate dashboard strate-
gically placed behind his desk. As such, anyone that came in
to meet with him was looking at the CEO AND the dashboard.
It was a strong message of what was important and what was
being viewed by the top of the organization. Adashboard further
keeps everyone up to date on the status of the brand/?rm. One
could view the dashboard in the form of a tree structure – there
are the key output measures, as discussed previously, and then
the drivers of each of these measures. As one gets further down
the organization it is possible to “drill down” on each of these
measures and understand the causal factors. For example, share-
holder value is a key output measure and CLV being a driver.
Then, the question is what drives CLV? That would be acqui-
sition, purchase amount, purchase frequency, and retention. We
could then look at what marketing actions tend to drive each of
these components. As can be quickly seen, with multiple out-
put measures, and most actions affecting multiple drivers, the
complexity starts to proliferate.
Linkages
Ideally, the linkages between drivers and output variables
need to be well understood, and hopefully, empirically deter-
mined. Unfortunately, these relationships are not all well known.
In part, this is because of the lack of data, and in part, because
they are not all in the same place. In the absence of data driven
relationships, judgmental parameters are necessary, often via
decision calculus. Ironically, it is often the case that managers do
not feel comfortable specifying their own judgment, yet have to
make decisions based on this judgment all the time. If we can get
the judgmental or empirically based relationships established,
then it would be possible to continually experiment and update
the relationships of these linkages. Some companies are very
good at constantly experimenting, while others are not. How-
ever, without continual experimentation, these linkages cannot
be well established or well understood.
The best place to start is by specifying the marketing objec-
tives. Each retailer will have their own objectives. Getting
alignment on the objectives is always a critical, and not nec-
essarily, an easy step. The next step is to establish hypotheses of
what are the drivers. Only then does it make sense to estimate
the relationships. Reibstein et al. (2005), discuss the ?ve steps
in developing such a dashboard which include:
1. Selecting the key metrics
2. Populating the dashboard with data
3. Establishing the relationships between dashboard items
4. Forecasting and “what if” analysis
5. Connecting to ?nancial consequences
In this article we have provided much of the material nec-
essary to help managers develop a marketing dashboard for
their ?rm. We have provided an assessment of the metrics that
exist and those metrics which are necessary to monitor. We have
provided guidance as to how ?rms can overcome the common
challenges of migrating to these new metrics and capturing and
tracking this data over time. Finally, we have shown which met-
rics have been shown to provide linkages to ?nancial outcomes
such as CLV and shareholder value. However, this research still
needs to continue to re?ne the selection and measurement of
marketing metrics and continue to develop linkages between
marketing metrics and ?nancial performance. Only then can
marketing justify its place in the board room.
Guidelines for future research
Ongoing research on marketing metrics will continue to
provide insights to marketing managers as they establish opti-
mal marketing strategies which are directly linked to ?nancial
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 109
outcomes. Thus, we provide a roadmap of much needed research
in the area of marketing metrics. First and foremost, research
needs to continue to establish the linkages between marketing
metrics and ?rm performance. Next, research needs to continue
to focus on the two tenets of marketing – customer acquisi-
tion and customer retention. In addition, marketing is constantly
evolving and managers need to understand how “new” mar-
keting, such as social content, can impact current marketing
practice. Most importantly, the emphasis on these studies should
continue to focus on metrics that can guide future decision-
making and not focus only on the past. Finally, there needs to
be a focus on research that relates directly to retailers, both in
the measurement of retailer brand value and retailer market cap-
italization. Too often research is focused at the ?rm-level giving
retailers few insights about strategic marketing decisions. We
discuss each of these calls for new research below.
Establishing linkages
While we have outline the taxonomy cells for retailer met-
rics, there is much work which remains ahead. The American
Marketing Association ran a conference in Atlanta in July 2008
for the Knowledge Coalition. This ?rst endeavor of its sort was
to establish what we know, in research and in practice, about
the relationships between marketing spending and marketing
results. The result was that much more work still needs to be
done in this area. Too often when a company comes to cut a bud-
get within an organization, marketing is cut off ?rst. Often this
is because marketing is unable to establish empirical linkages
between marketing metrics, such as awareness, and ?nancial
outcomes, such as stock price. Thus, it is crucial that marketing
managers begin to not only use their intuition to drive marketing
decisions, but to also empirically link marketing metrics to ?nan-
cial outcomes. Then it will become easier to justify the need for
marketing resources and also quantify both the negative short-
termand long-termimpact of a reduction in marketing resources.
Customer acquisition and retention
We also know that customer lifetime value is an important
metric for retailers as well as others and that the sum of the
customer lifetime values of all customers is customer equity.
All ?rms strive to grow customer equity over time. This effort
can only happen when resources are spent on trying to retain
current pro?table customers and to acquire new pro?table cus-
tomers. Some research has empirically shown that it is necessary
to balance this acquisition and retention budget (Reinartz et al.
2005; Thomas 2001), however continual work is needed to bet-
ter understand what drives customer acquisition and retention.
Only then can marketing managers leverage the drivers of acqui-
sition and retention to continue to groweach customer’s lifetime
value and overall customer equity.
Social content
A new emerging ?eld, and still not well known, is the impact
of social content, such as product and store reviews, and blogs
(Godes and Mayzlin 2004a,b). Bizrate.com, Shopzilla.com, and
Shopping.com all were established to provide user feedback to
future shoppers about retail shopping sites. There is now the
continual growth of such websites such as Bazaarvoice.comthat
empowers retail customers to help informfuture shoppers. Many
?rms have turned to these forms of “new marketing” to try and
grab a ?rm hold on the emerging trends. However, it is still
unclear how ?rms can learn anything from these new media or
even use these new media in predictive customer behavior and
in turn ?rm pro?ts. The impact of such social networking is
needs to be better understood and provides many opportunities
for future research.
Relating current actions to future actions
The big question facing marketers today is the impact of their
current behavior on future performance (Zeithaml et al. 2006).
It is often hard to use any current metrics to predict future per-
formance. Most measures are short-term, and certainly drawing
the link is easier when actions and results are contemporaneous.
The question we are still in need of better exploration is the long-
termimpact of marketing actions. Thus, future research needs to
begin to identify potential metrics that have the ability to predict
future events (beyond just CLV and customer equity) to some
degree of accuracy so that managers can track these leading indi-
cators to get a better grasp on where the ?rmis heading – or even
any interventions that are necessary to continue to increase ?rm
value.
Retailer brand value
Marketers for years have been focused on brand establish-
ment. Retailers have brands of their own, as well. There has
been very little research that has focused speci?cally on the
establishment of retail brands, whether these are speci?c prod-
ucts with a retailer’s brand name (Kumar and Steenkamp 2007)
or the brand equity of the name of the retail store. One question
is the impact of the brands that a retailer carries and its impact on
the retailer’s brand itself. Future research needs to help quantify
retail and private label brand values so that managers can make
more strategic decisions to increase ?rm value.
Retailer market capitalization
Lastly, the ultimate is drawing the link between marketing
spending and market capitalization. While some work has begun
for manufacturing ?rms, little has been done for speci?cally for
retailers. Given the increased complexity of real estate, and its
?uctuating value, do the same principles hold for retailers as
what has been found to date for manufacturers? This is another
great opportunity for future research to investigate.
So, while progress has been made, there is considerable work
still to be done to better understand the role of retailer metrics.
In summary, this article has identi?ed the directions retailers
and manufacturers can take to not only build a pro?table cus-
tomer database, but also focus on building the brand value/equity
through establishment of superior marketing metrics.
110 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
References
Aaker, David A. (1991), “Managing Brand Equity,” New York: Free Press.
Allenby, Greg M., Robert P. Leone and Lichung Jen (1999), “A Dynamic Model
of Purchase Timing with Application to Direct Marketing,” Journal of Amer-
ican Statistical Association, 94 (June), 365–74.
Ambler, Tim (2003), “Marketing Metrics,” second ed. Prentice-Hall.
Anderson, Eric T., Karsten Hansen, and Duncan Simester (forthcoming), “The
Option Value of Returns: Theory and Empirical Evidence,” Marketing Sci-
ence.
and Lei K. Wang (2008), “Howare Demand and Returns
Related?,” Working Paper, Kellogg School of Management.
Blanchard, David (2007), “Supply Chains Also Work in Reverse,” in Industry
Week.
Chevalier, Judith A. and Dina Mayzlin (2006), “The Effect of Word of Mouth on
Sales: Online Book Reviews,” Journal of Marketing Research, 43 (August),
345–54.
Dhar, Vasant and Elaine Chang (2007), “Does Chatter Matter? The Impact of
User-Generated Content on Music Sales,” Working Paper, New York Uni-
versity.
Du, Rex Yuxing, Wagner A. Kamakura and Carl F. Mela (2007), “Size and Share
of Customer Wallet,” Journal of Marketing, 71 (April), 94–113.
Eastlack, Joseph O. Jr. and Ambar G. Rao (1986), “Advertising Experiments at
the Campbell Soup Company,” Marketing Science, 8 (1), 57–71.
Elsner, Ralf, Manfred Krafft and Arnd Huchzermeier (2004), “Optimizing
Rhenania’s Direct Marketing Business Through Dynamic Multilevel Mod-
eling (DMLM) in a Multicatalog-Brand Environment,” Marketing Science,
23 (2), 192–206.
Fader, Peter S., Bruce G.S. Hardie and Ka Lok Lee (2005), “RFM and CLV:
Using Iso-Value Curves for Customer Base Analysis,” Journal of Marketing
Research, 42 (November), 415–30.
Farris, Paul W., Neil T. Bendle, Phillip E. Pfeifer and David J. Reibstein (2006),
“Marketing Metrics: 50+ Metrics Every Executive Should Master,” 3rd ed.
Upper Saddle River, NJ: Wharton School Publishing.
Godes, David and Dina Mayzlin (2004a), “Firm-Created Word-of-Mouth Com-
munication: AField-Based Quasi-Experiment,” Working Paper, Yale School
of Management.
(2004b), “Using Online Conversations to Study Word of
Mouth Communication,” Marketing Science, 23 (4), 545–60.
Gupta, Sunil, Donald R. Lehmann and Jennifer Ames Stuart (2004), “Valuing
Customers,” Journal of Marketing Research, 41 (1), 7–18.
and Valarie A. Zeithaml (2006), “Customer Metrics
and Their Impact on Financial Performance,” Marketing Science, 25 (6),
718–39.
Hogan, John E., Katherine N. Lemon and Barak Libai (2003), “What is the true
value of a lost customer?,” Journal of Service Research, 5 (3), 196–208.
Keller, Kevin Lane (1993), “Conceptualizing, Measuring, and Managing
Customer-Based Brand Equity,” Journal of Marketing, 57 (1), 1–22.
Kerin, Roger A. and Raj Sethuraman (1998), “Exploring the Brand Value-
Shareholder Value Nexus for Consumer Goods Companies,” Journal of the
Academy of Marketing Science, 26 (4), 260–73.
Knott, Aaron I., Andrew Hayes and Scott A. Neslin (2002), “Next-Product-
to-Buy Models for Cross-Selling Applications,” Journal of Interactive
Marketing, 16 (3), 59–75.
Krasnikov, Alexander, Satish Jayachandran, and V. Kumar (2008), “The Impact
of CRM Technology on Cost and Pro?t Ef?ciencies: Evidence from US
Commercial Banking Industry,” Working Paper, University of South Car-
olina.
Kumar, Nirmalya and Jan-Benedict E.M. Steenkamp (2007), “Brands Under
Attack fromPrivate Labels,” in Private Label Strategy Boston, MA: Harvard
Business School Press.
Kumar, V. (2008a), “Managing Customers for Pro?t,” Upper Saddle River, NJ:
Wharton School Publishing.
(2008b), “Customer Lifetime Value: The Path to Prof-
itability,” The Netherlands: NOW Publishers, Inc.
, Morris George and Joseph Pancras (2008a), “Crossbuy-
ing in Retailing: Drivers and Consequences,” Journal of Retailing, 84 (1),
15–27.
, Anita Man Luo, and Vithala R. Rao (2008b), “Linking
an Individual’s Brand Value to the Customer Lifetime Value: An Integrated
Framework,” Working Paper, Georgia State University.
Kumar, V. and Andrew Petersen (2005), “Using a Customer-Level Market-
ing Strategy to Enhance Firm Performance: A Review of Theoretical and
Empirical Evidence,” Journal of the Academy of Marketing Science, 33 (4),
504–19.
and Robert P. Leone (2007), “How Valuable is Word of
Mouth?,” Harvard Business Review, 85 (10), 139–46.
(2008c), “Exploring the Effectiveness of Marketing
Communications on Customer Buying and Referral Behavior,” Working
Paper, Georgia State University.
Kumar, V. and Denish Shah (forthcoming), “Expanding the Role of Marketing:
From Customer Equity to Market Capitalization,” Journal of Marketing.
and Rajkumar Venkatesan (2006a), “Managing Retailer
Pro?tability—One Customer at a Time!,” Journal of Retailing, 82 (4),
277–94.
and Rajkumar Venkatesan (2005), “Who are Multi-
channel Shoppers and how do they Perform?: Correlates of Multichannel
Shopping Behavior,” Journal of Interactive Marketing, 19 (2), 44–62.
, Timothy R. Bohling, and Denise Beckman (forthcom-
ing), “The Power of CLV: Managing Customer Value at IBM,” Marketing
Science.
, Rajkumar Venkatesan and Werner Reinartz (2006b),
“Knowing What to Sell, When, and to Whom,” Harvard Business Review,
84 (3), 131–7.
(2008d), “Performance Implications of Adopting a
Customer-Focused Sales Campaign,” Journal of Marketing, 72 (September),
50–68.
Leone, Robert P., Vithala R. Rao, Kevin Lane Keller, Anita Man Luo, Leigh
McAlister and Rajendra K. Srivastava (2006), “Linking Brand Equity to
Customer Equity,” Journal of Service Research, 9 (2), 125–38.
Madden, Thomas J., Frank Fehle and Susan Fournier (2006), “Brands Matter:
An Empirical Demonstration of the Creation of Shareholder Value Through
Branding,” Journal of the Academy of Marketing Science, 34 (2), 224–35.
Pauwels, Koen, Tim Ambler, Bruce Clark, Pat LaPointe, David J. Reibstein,
Bernd Skiera, Berend Wierenga and Thorsten Wiesel (2008), “Dashboards &
Marketing: Why, What, How and Which Research is Needed?,” Cambridge,
MA: Marketing Science Institute.
and Scott A. Neslin (2008), “Building with Bricks and
Mortar: The Revenue Impact of Opening Physical Stores in a Multichannel
Environment,” Report 08-101, MarketingScience Institute, Cambridge, MA.
Petersen, J. Andrewand V. Kumar (forthcoming), “Are Product Returns a Neces-
sary Evil? The Antecedents and Consequences of Product Returns,” Journal
of Marketing.
(2008), “CLV and Optimal Resource Allocation: The
In?uence of Marketing, Buying and Product Returns,” Working Paper, Uni-
versity of North Carolina at Chapel Hill.
Reibstein, DavidJ., YogeshJoshi, DavidNortonandPaul W. Farris (2005), “Mar-
keting Dashboards: A Decision Support System for Assessing Marketing
Productivity,” in Presentation at the Marketing Science Conference.
Reichheld, Frederick F. (2003), “The One Number You Need to Grow,” Harvard
Business Review, 81 (12), 46–54.
Reinartz, Werner, Manfred Krafft and Wayne D. Hoyer (2004), “The Customer
Relationship Management Process: Its Measurement and Impact on Perfor-
mance,” Journal of Marketing Research, 41 (August), 293–305.
, Jacquelyn S. Thomas and V. Kumar (2005), “Balancing
Acquisition and Retention Resources to Maximize Customer Pro?tability,”
Journal of Marketing, 69 (1), 63–79.
Roberts, John and Tim Ambler (2006), “Beware the Silver Metric: Marketing
Performance Measurement Has to Be Multidimensional,” Report 06-113,
Marketing Science Institute, Cambridge, MA.
Rust, Roland T., Tim Ambler, Gregory S. Carpenter, V. Kumar and Rajendra K.
Srivastava (2004a), “MeasuringMarketingProductivity: Current Knowledge
and Future Directions,” Journal of Marketing, 68 (4), 76–89.
, Katherine N. Lemon and Valarie A. Zeithaml (2004b),
“ReturnonMarketing: UsingCustomer EquitytoFocus MarketingStrategy,”
Journal of Marketing, 68 (1), 109–27.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 111
Simon, Carol J. and Mary W. Sullivan (1993), “The Measurement and Determi-
nants of Brand Equity: A Financial Approach,” Marketing Science, 12 (1),
28–52.
Srivastava, Rajendra K., TasadduqA. Shervani andLiamFahey(1998), “Market-
based assets and shareholder value: A framework for analysis,” Journal of
Marketing, 62 (1), 2–18.
Thomas, Jacquelyn S. (2001), “A methodology for linking customer acquisition
to customer retention,” JMR, Journal of Marketing Research, 38 (2), 262–8.
and Ursula Y. Sullivan (2005), “Managing Marketing
Communications with Multichannel Customers,” Journal of Marketing, 69
(October), 239–51.
Venkatesan D Rajkumar and V. Kumar (2004), “A Customer Lifetime Value
Framework for Customer Selection and Resource Allocation Strategy,” Jour-
nal of Marketing, 68 (4), 106–25.
Venkatesan, Rajkumar, V. Kumar and Nalini Ravishanker (2007), “Multichannel
Shopping: Causes and Consequences,” Journal of Marketing, 71 (April),
114–32.
Verhoef, Peter C. (2003), “Understanding the Effect of Customer Relationship
Management Efforts on Customer Retention and Customer Share Develop-
ment,” Journal of Marketing, 67 (4), 30–45.
, Philip Hans Franses and Janny C. Hoekstra (2001),
“The Impact of Satisfaction and Payment Equity on Cross-buying: A
Dynamic Model for a Multi-service Provider,” Journal of Retailing, 77 (3),
359–78.
Villanueva, Julian, Shijin Yoo and Dominique M. Hanssens (2008), “The
Impact of Marketing-Induced vs. Word-of-Mouth Customer Acquisition
on Customer Equity Growth,” Journal of Marketing Research, 45 (1),
48–59.
Zeithaml, Valarie A., Ruth N. Bolton, John Deighton, Timothy L. Keiningham,
N. Katherine, J. Lemon and Andrew Petersen (2006), “Forward Looking
Focus: Can Firms Have Adaptive Foresight?,” Journal of Service Research,
9 (2), 168–83.
doc_582821112.pdf
There is an ever-present need for managers to justify marketing expenditures to the firm. This can only be done when we can establish a direct link between marketing metrics and future customer value and firm performance.
Journal of Retailing 85 (1, 2009) 95–111
Choosing the Right Metrics to Maximize Pro?tability
and Shareholder Value
J. Andrew Petersen
a,?
, Leigh McAlister
b
, David J. Reibstein
c
,
Russell S. Winer
d
, V. Kumar
e
, Geoff Atkinson
f
a
Kenan-Flagler Business School, The University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, United States
b
McCombs School of Business, University of Texas at Austin, Austin, TX 78712, United States
c
Wharton School, University of Pennsylvania, Philadelphia, PA 19104, United States
d
Stern School of Business, New York, NY 10012, United States
e
J. Mack Robinson School of Business, Georgia State University, Atlanta, GA 30303, United States
f
Overstock.com, Salt Lake City, UT 84121, United States
Abstract
There is an ever-present need for managers to justify marketing expenditures to the ?rm. This can only be done when we can establish a direct
link between marketing metrics and future customer value and ?rm performance. In this article, we assess the marketing literature with regard to
marketing metrics. Subsequently, we develop a framework that identi?es key metrics that ?rms should focus on that can give a ?rm a better picture
of how they got to where they are now and insights towards how they can continue to grow into the future. We then identify several organizational
challenges that need to be addressed in order for ?rms to build the capabilities of collecting the right data, measuring the right metrics, and linking
those metrics to customer value and ?rm performance. Finally, we offer guidelines for future research with regard to marketing metrics to help
?rms establish successful marketing strategies, measure marketing effectiveness, and justify marketing expenditures to top management.
Published by Elsevier Inc on behalf of New York University.
Keywords: Metrics; Customer lifetime value; Customer equity; Shareholder value; Referral behavior
Introduction
“You can’t manage what you don’t measure.”
- Old Management Adage
In recent years, there has been a signi?cant increase in the
number and type of marketing metrics that managers can use
to measure marketing effectiveness and to develop marketing
strategies with the goal of increasing ?rm performance. The
purpose of these marketing metrics is twofold. First, market-
ing metrics serve to increase marketing’s accountability within
the ?rm and to justify spending valuable ?rm resources on
marketing initiatives to top management (Rust et al. 2004b). Sec-
ond, marketing metrics can help managers and retailers identify
the drivers of future customer and ?rm value and build link-
?
Corresponding author.
E-mail addresses: Andrew [email protected] (J.A. Petersen),
[email protected](L. McAlister), [email protected]
(R.S. Winer), [email protected] (V. Kumar), [email protected]
(G. Atkinson).
ages between marketing strategy and ?nancial outcomes. When
retailers are able to identify the drivers of customer and store
value, managers can then maximize customer and store pro?ts
(Kumar et al. 2006a). The increase in the number of available
marketing metrics has been the result of several different fac-
tors. First, increases in database technology have given ?rms the
ability to collect more information about their own customers
and to an extent some information about competitors and their
competitor’s customers. Second, the advent of new channels
of distribution for products and services, such as the Internet,
has signi?cantly increased the availability and complexity of
marketing metrics. Finally, the identi?cation of new drivers of
customer and ?rm value, for example word-of-mouth and refer-
ral behavior (Hogan et al. 2003; Kumar et al. 2007; Reichheld
2003), has led to an increase in the number of different types of
marketing metrics beyond just measurements of customer value
and return on investment.
However, with the abundance of marketing metrics to choose
from (see Farris et al. (2006) for a detailed list of relevant mar-
keting metrics), the challenge marketing managers and retailers
now face is not whether to use marketing metrics, but instead
0022-4359/$ – see front matter. Published by Elsevier Inc on behalf of New York University.
doi:10.1016/j.jretai.2008.11.004
96 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
how to determine which metrics are the most important metrics
to utilize for a given ?rm. While there is no single or “silver”
metric that can summarize marketing performance (Roberts and
Ambler 2006), too many metrics can just provide clutter to
the marketing metrics dashboard. Thus, the most appropriate
metrics are those that are effective at measuring marketing pro-
ductivity, help managers to develop effective forward-looking
marketing strategies, help predict a customer’s future value to
the ?rm, and help predict the ?rm’s future ?nancial performance.
Thus, to choose the appropriate metrics it is necessary for
managers to answer the following ?ve key questions:
1. What metrics are currently in place in different ?rms?
2. What metrics should be in place in different ?rms?
3. How can managers link strategic actions to these new met-
rics?
4. How do these metrics relate to (a) customer value to the ?rm
and (b) ?rm performance?
5. What are the challenges ?rms will face when migrating to
these new metrics?
We address these ?ve key questions in the following four sec-
tions of this paper. We will ?rst reviewthe key marketing metrics
that exist in the marketing literature and in marketing practice.
Second, we present a framework that will help managers iden-
tify key metrics that should be used by different ?rms. Third, we
discuss the steps ?rms can follow to migrate to these metrics.
We also provide examples of situations in marketing research
Table 1
Discussions on multiple metrics.
Author (year) Title (journal) Finding/contribution
Gupta and Zeithaml (2006) Customer Metrics and Their Impact on Financial
Performance (Mktg. Sci.)
Discussion of links between potential unobservable and
observable metrics that have been shown to impact
?nancial performance. In addition a call for future
research that addresses key controversies in
metric-related research.
Unobservable: Customer Satisfaction, Service Quality,
Loyalty, and Intention to Purchase; Observable:
Acquisition, Retention, Cross-selling, Customer
Lifetime Value, Customer Equity.
Rust et al. (2004a) Measuring Marketing Productivity: Current Knowledge
and Future Directions (JM)
Introduces a broad framework for assessing marketing
productivity based on a review of past and current
research. Suggests new paths for research to evaluate
marketing productivity. Speci?cally deals with what we
know and what we would like to know with respect to
the chain of marketing productivity. The chain includes:
Tactical Actions, Strategies, Customer Impact,
Marketing Assets, Marketing Impact, Market Position,
Financial Impact, Financial Position, Impact on Firm
Value, and Value of the Firm.
Farris et al. (2006) Marketing Metrics: 50+ Metrics Every Executive
Should Master (Book)
List and description of major metrics used in academics
and practice. These metrics are broken down into 9
main categories.
These Categories are: (1) Share of Hearts, Minds, and
Markets, (2) Margins and Pro?ts, (3) Product and
Portfolio Management, (4) Customer Pro?tability, (5)
Sales Force and Channel Management, (6) Pricing
Strategy, (7) Promotion, (8) Advertising Media and Web
Metrics and (9) Marketing and Finance.
Ambler (2003) Marketing and the Bottom Line: The Marketing Metrics
to Pump Up Cash Flow (Book)
A book that ?rst argues that multiple metrics are
necessary – since there are many different approaches to
measuring the same performance.
The main focus is on the value of brand equity – noting
that it is a complex asset.
Kumar (2008a, b) Managing Customer for Pro?t (Book) The author shows how to use Customer Lifetime Value
(CLV) to target customers with higher pro?t potential,
manage and reward existing customers based on their
pro?tability, and invest in high-pro?t customers to
prevent attrition and ensure future pro?tability. The
author introduces customer-centric approaches to
allocating marketing resources for maximum
effectiveness, pitching the right products to the right
customers at the right time, determining when a
customer is likely to leave, and whether to intervene,
managing multichannel shopping, and calculating a
customer’s referral value (CRV).
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 97
and practice that identify how ?rms can overcome certain chal-
lenges related to migrating to new marketing metrics. Finally,
we will provide guidelines for future research in marketing that
will continue to extend our knowledge towards selecting the best
marketing metrics that marketing managers can directly link to
key ?nancial outcomes.
A review of marketing metrics
The use of marketing metrics was the main result of pressures
from top management and shareholders to justify marketing
spending. However, the consequence of this growing need for
quantitative measures of marketing and ?rm performance has
led to a multitude of metrics measuring everything from levels
of customer satisfaction to the number of unique clicks to a spe-
ci?c website. The goal of all of these metrics has been to gain
some quanti?able measure of any number of business goals,
from measuring the effectiveness of marketing campaigns to
proxies of ?rm value. While it is important for marketing man-
agers to know the entire consideration set of available metrics,
that is beyond the scope of this paper. For more complete lists
of multiple metrics it is worthwhile to refer to books and journal
articles that discuss multiple metrics (Ambler 2003; Farris et
al. 2006; Gupta and Zeithaml 2006; Kumar 2008a, b; Rust et al.
2004a,b). Inaddition, see Table 1for a discussionof eachof these
books or articles and the key insights each provides. Instead, we
present an assessment of a key set of seven categories of metrics
found in the marketing literature that are relevant to retailers. In
addition, we also discuss two speci?c types of marketing met-
rics with regard to the ?ve key questions this paper sets out to
answer. These include: (1) backward-looking versus forward-
looking metrics and (2) the customer and brand value metrics
that link directly to ?nancial performance and ?rm value. We
discuss how each of these types of metrics have impacted mar-
keting research and practice in addition to how each of these
types of metrics relates directly to manager’s ability to generate
effective marketing strategies.
Assessment of the literature
In this section, we review the literature in marketing that
develops or discusses key metrics that can be used by retailers.
We break these key metrics in the literature into seven distinct
categories and give examples along with some discussion of
literature in each of these categories. These seven categories
include:
1. Brand value metrics
2. Customer value metrics
3. Word of mouth and referral value metrics
4. Retention and acquisition metrics
5. Cross-buying and up-buying metrics
6. Multi-channel shopping metrics
7. Product return metrics
Each of these seven categories of marketing metrics relevant
to retailers serves two main purposes. First, these retailer mar-
keting metrics can be used for strategic and tactical marketing
campaigns. As an example, a marketing manager can strate-
gically use each customer’s predicted referral value score to
determine whichcustomer totarget next time periodwithreferral
incentives (Kumar et al. 2007). In addition, a marketing man-
ager can strategically use each customer’s predicted value (CLV)
to determine which customers to select for a given marketing
campaign that encourages cross-buying, up-buying, or multi-
channel shopping (Kumar and Petersen 2005). Second, these
retailer marketing metrics can be used for short-term or long-
term goals and predictions. As an example, in the short-term the
goal of the ?rm may be to increase the general awareness of
a given brand. Thus, the ?rm would try to increase the overall
percentage of consumers in the marketplace who are aware of
the brand in the following time period (e.g., month or quarter).
However, a long-termgoal resulting fromcontinuous short-term
increases in brand awareness may be to increase overall brand
equity, a long-term metric.
The questions that still remain – How does a manager link
these different types of metrics to strategic goals and how can
these strategic goals be linked directly to customer pro?tabil-
ity and shareholder value? The following discussion provides
a general sense of the literature in marketing that either identi-
?es speci?c marketing metrics or builds conceptual or empirical
linkages between marketing metrics and customer pro?tability,
shareholder value, or both. We also provide a series of tables
that summarize this information (see Tables 2A–2G). We begin
by discussing the marketing metrics found in each of the seven
categories listed previously.
Table 2A
Examples of brand value and brand equity metrics.
Author (year) Finding/contribution
Keller (1993) The author provides a conceptual model of brand equity from the customer’s perspective.
Simon and Sullivan (1993) A brand’s value is the capitalized value of the pro?ts that result from associating that brand’s
name with speci?c products and services.
Kerin and Sethuraman (1998) The authors discuss the link between brand value and shareholder value. The authors ?nd this
link to be positive, but the functional form of the relationship is concave (decreasing returns)
with respect to the ?rm’s Market to Book Ratio.
Madden et al. (2006) Using the Fama-French method, the authors ?nd empirical evidence that stronger brands
deliver stronger shareholder value with less risk to the ?rm.
Leone et al. (2006) The authors provide a discussion on the link between Brand Equity and Customer Equity. The
authors show that while the literature has been divergent in nature, there are great similarities
between the two.
98 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
Table 2B
Examples of customer value metrics.
Author (year) Finding/contribution
Venkatesan and Kumar (2004) The authors de?ne a customer-level CLV objective function and empirically identify the
behavioral drivers of contribution margin and purchase frequency. In addition, the authors
present an optimal resource allocation strategy that can be used by the ?rm to send the
appropriate marketing communications at the right time to maximize pro?ts.
Rust et al. (2004b) The authors de?ne ?nancial return as a function of the change in Customer Equity as a result
of an incremental increase in spending. The authors ?nd drivers of Customer Equity based on
survey information of customers from the Airline industry. In addition, the authors use a
brand-switching matrix to introduce competitive effects into the model estimation.
Gupta et al. (2004) The authors demonstrate how valuing customers makes it feasible to value ?rms, including
high growth ?rms with negative earnings. They demonstrate their valuation method by using
publicly available data for ?ve ?rms. They ?nd that a 1% improvement in retention, margin,
or acquisition cost improves ?rm value by 5%, 1% and .1%, respectively. They also ?nd that a
1% improvement in retention has almost ?ve times greater impact on ?rm value than a 1%
change in discount rate or cost of capital.
Petersen and Kumar (2008) The authors introduce a new CLV objective function that includes product returns as a
separate part of the equation. The authors show that the newly proposed CLV objective
function provides better ?t, better prediction, lower over-prediction bias, and better
segmentation that previous CLV models. In addition, the authors show the ?rm requires fewer
resources to maximize pro?tability when it takes product returns into account.
Brand equity
Many of the metrics that measure brand value or brand equity
stem from the research of Keller (1993) who provides a con-
ceptual model that outlines how to measure brand equity from
the customer’s perspective. Following this, many studies began
to not only measure a customer’s individual brand value or
the ?rm’s brand equity, but also began to link this measure to
both customer equity and shareholder value. In most cases there
have been positive links between increases in brand equity with
increases of customer or ?rm value. For example, using brand
values published in Financial World and Market-to-Book ratios
from publicly traded companies, Kerin and Sethuraman (1998)
found a positive, but decreasing returns, relationship between
brand value and shareholder value. However their sample only
included ?rms that were listed on the “Most Valued Brands” list.
Using a list of brand values from Interbrand and market capi-
talizations from publicly accessible data, Madden et al. (2006)
found evidence that increases in a brand’s strength were related
to increases in shareholder value. This was done by analyzing the
performance of different portfolios of companies – (1) World’s
Most Valued Brands (WMVB), (2) Reduced-Market (RM) –
which contains all those ?rms in the Center for Research in
Securities Market database (except those in WMVB), and (3)
Full-Market (FM) – which contains all ?rms. However, there is
still work to be done in this area. As Leone et al. (2006) points
out, while there are some links between brand equity and cus-
tomer equity, the literature exploring links between brand equity
and customer equity is sparse. This gives a great opportunity
for research to continue to develop methods to link brand and
customer equity.
Customer value
There has been a signi?cant amount of literature that has
set out to develop metrics that measure the value of customers,
whether it is at the individual level in the form of customer life-
time value (CLV) or at the aggregate level (customer equity). Up
Table 2C
Examples of word of mouth and referral value metrics.
Author (year) Finding/contribution
Hogan et al. (2003) The authors empirically show through surveys that losing customers through defection to a
competitor or no longer using a product can have a great impact on the revenue stream across
the life of the product cycle. This is due to the word of mouth effect that is lost when
customers no longer advocate on behalf of the company.
Kumar et al. (2007) The authors present an equation that measures the value of referrals for customers from a
?rm. The authors also show the results of an effective ?eld experiment with a B2C ?rm where
customers were targeted based on their CLV and CRV (Customer Referral Value) scores.
Villanueva et al. (2008) Firms acquire customers through costly but quick investments in advertising and slow but
cheap investments in building word of mouth. The authors ?nd that in the short-term a ?rm
can increase customer equity more using advertising. However, in the long-term the ?rm can
increase customer equity signi?cantly more with investments in word of mouth.
Kumar et al. (2008c) The authors provide a straightforward four-step process for predicting Customer Referral
Value (CRV). Then the authors run a series of three ?eld experiments to determine the most
effective marketing strategy for sending targeted marketing communications to customers
based on CLV and CRV.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 99
Table 2D
Examples of retention and acquisition metrics.
Author (year) Finding/contribution
Thomas (2001) The author shows that customer acquisition and retention are not independent processes.
However, because of data limitations, customer management decisions are frequently based
only on an analysis of acquired customers. This analysis shows that these decisions can be
biased and misleading. The author presents a modeling approach that estimates the length of a
customer’s lifetime and adjusts for this bias.
Verhoef (2003) The author investigates the differential effects of customer relationship perceptions and
relationship marketing instruments on customer retention and customer share development
over time. The results show that affective commitment and loyalty programs that provide
economic incentives positively affect both customer retention and customer share
development, whereas direct mailings in?uence customer share development. However, the
effect of these variables is rather small.
Reinartz et al. (2005) In this research, the authors present a modeling framework for balancing resources between
customer acquisition efforts and customer retention efforts. The key question that the
framework addresses is, “What is the customer pro?tability maximizing balance?” In addition,
they answer questions about how much marketing spending to allocate to customer acquisition
and retention and how to distribute those allocations across communication channels.
Fader et al. (2005) The authors present a new model that links the well-known RFM (recency, frequency,
monetary value) paradigm with customer lifetime value (CLV). The stochastic model,
featuring a Pareto/NBD framework to capture the ?ow of transactions over time and a
gamma-gamma sub-model for dollars per transaction, reveals a number of subtle but
important non-linear associations that would be missed by relying on observed data alone.
to this point, the purpose of measuring CLVand customer equity
has been for optimal customer selection in marketing campaigns
and to measure marketing effectiveness post-campaign. Rust et
al. (2004b) use survey results from consumers located in two
different northeastern US towns to determine drivers of cus-
tomer choice and customer lifetime value. In addition, they are
able to project the return on marketing expenditures for differ-
ent types of campaigns in each of the companies studied and
account for competitive information using a brand switching
matrix. Venkatesan and Kumar (2004) use a sample of B2B
customers from a multinational high-tech ?rm to ?rst deter-
mine the behavioral and demographic drivers of CLV. Then they
determine an optimal resource allocation strategy using genetic
algorithms. The end result is a customer-level resource allo-
cation strategy that maximizes each customer’s lifetime value.
Research in marketing is also beginning to identify linkages
between these metrics and overall ?rmvalue. Gupta et al. (2004)
use information frompublicly traded companies to estimate cus-
tomer equity and ?rm value. They ?nd that as long as a ?rm is
able to project its customer growth pattern and estimate its cur-
rent customer margin that it is feasible to determine customer
equity and overall ?rm value. This is especially important in
situations where ?rms have short histories of transactions, are
involved in a high-growth period, and have a negative cash ?ow
due to early capital investments. Additional research by Kumar
and Shah (forthcoming) was able to ?nd a direct relationship
Table 2E
Examples of cross-buying and up-buying metrics.
Author (year) Finding/contribution
Verhoef et al. (2001) In this article the authors investigate how satisfaction and payment equity affect cross-buying
at a multiservice provider. They also consider its competitors’ performance on these factors.
The results show that the effect of satisfaction differs between customers with lengthy and
short relationships. It also shows that payment equity negatively affects cross-buying for
customers with long relationships. However, if the prices of the supplier are perceived as
fairer than the prices of the competitor, the customers’ probability of cross-buying increases.
Knott et al. (2002) The authors present and evaluate next-product-to-buy (NPTB) models for improving the
effectiveness of cross-selling. The NPTB model reduces the waste of poorly targeted
cross-selling activities by predicting the product each customer would be most likely to buy
next. The ?eld test shows that the NPTB model increases pro?ts compared to a heuristic
approach, and that pro?ts are incremental over and above sales that would have occurred
through other channels.
Kumar et al. (2006b) The authors provide a framework that enables managers to determine the appropriate
product(s) to sell to customers at a given time based on that customer’s history and the history
of other customers at the ?rm.
Kumar et al. (2008a) The authors provide empirical evidence of the antecedents and consequences of crossbuying
in a non-contractual retail setting.
Kumar et al. (forthcoming) The authors provide a framework and methodology for targeting the right customers with the
right products at the right time. They do this by running a ?eld experiment with a company and
comparing the results of a typical sales campaign with one that is driven by their methodology.
100 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
Table 2F
Examples of multi-channel shopping metrics.
Author (year) Finding/contribution
Thomas and Sullivan (2005) This article presents a marketing communications process that uses customer relationship
management ideas for multichannel retailers. The authors describe and then demonstrate the
process with enterprise-level data from a major U.S. retailer with multiple channels.
Kumar and Venkatesan (2005) The authors develop a conceptual framework, which identi?es the customer-level
characteristics and supplier factors that are associated with purchase behavior across multiple
channels. They also propose and empirically show that multichannel shoppers provide
bene?ts as measured by several customer-based metrics.
Venkatesan et al. (2007) The authors explore the drivers of multichannel shopping and the impact of multichannel
shopping on customer pro?tability. The authors provide evidence that multichannel shopping
is associated with higher customer pro?tability.
Pauwels and Neslin (2008) The authors use a multichannel customer management perspective to assess the revenue
impact of adding bricks-and-mortar stores to a ?rm’s already existing repertoire of catalog
and Internet channels. We decompose the revenue impact into customer acquisition,
frequency of orders, returns, and exchanges, and size of orders, returns, and exchanges. The
analysis estimates the net impact of adding the store channel was to increase revenues by
37%. The majority of this increase was due to an improvement in customer retention through
higher purchase frequency.
between CLV and shareholder value. This suggests that if mar-
keting managers can continue to run marketing campaigns to
increase customer value that this will directly lead to increases
in shareholder value. This research is only a start though. There
is still a need to continually improve measures of CLV and to
link CLV to ?nancial outcomes.
Word of mouth and referral value
Ever since Reichheld (2003) suggested that the key to busi-
ness growth lies in the positive word of mouth of a ?rm’s
customers, that is Net Promoter Score, research in marketing
has started to explore this connection between word of mouth
and customer/?rmvalue. Initially, Hogan et al. (2003) were able
to show the value lost over time when a customer disadopts or
defects from a ?rm, using the Bass model for the diffusion of
new products and a Monte Carlo simulation. This lost value was
not only a function of lost purchases, but it was also a function
of the lost word of mouth the customer spread about the prod-
uct causing losses of potential future sales. Even more troubling
to this ?nding is the fact that customers who are acquired via
word of mouth are signi?cantly more pro?table in the long-term
than customers who are acquired via advertising and promotion.
Using data from an Internet ?rm that provided free web hosting
to its registered users, Villanueva et al. (2008) found that cus-
tomer who were acquired using costly, but short-termmarketing
advertisements and promotions give fewer than half the pro?ts
of customers acquired using cheap, but long-term investments
in word of mouth marketing. This makes it even more important
to identify the customers who are valuable with regard to word
of mouth and referral behavior and attempt to retain those cus-
tomers. Additionally, a study by Kumar et al. (2007) using data
froma ?nancial services andtelecommunications ?rmfoundthat
customers with a high CLV are often not the same as customers
with a high customer referral value (CRV), making it is espe-
cially important to know which customers are spreading word
of mouth. Thus, it is critical to not only measure the value of
word of mouth and customer lifetime value, but also to continue
researching ways to link additional metrics such as customer
word of mouth and referral behavior to marketing strategy and
then to ?nancial performance.
Customer retention and acquisition
Increases in customer retention and acquisition are tenets
to successful marketing strategies. However, ?rms need to be
careful not to make decisions about customer acquisition and
customer retention in isolation. Research by Thomas (2001)
Table 2G
Examples of product returns metrics.
Author (year) Finding/contribution
Anderson et al. (forthcoming) The authors show that not including product returns in the demand function creates an
overestimation bias of demand. In addition, the authors empirically show that customers have
an option value for product returns that is measurable.
Petersen and Kumar (forthcoming) The authors use data from a B2C catalog retailer to empirically describe the antecedents and
consequences of product return behavior. The authors ?nd that product returns positively
affect (to a threshold) future purchase behavior – making them necessary, but not evil.
Anderson et al. (2008) The authors run a ?eld experiment with a retail ?rm and empirically show that products
offered on sale have a lower probability of being returned than products purchased at full price.
Petersen and Kumar (2008) The authors integrate product return behavior in the CLV objective function and show that not
including it or including it as a component of buying behavior (i.e., net buying behavior as
products purchased minus products returned) offers signi?cant decreases in predictive
accuracy and optimal resource allocation.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 101
used data from an airline pilot service organization and a
latent-class Tobit modeling framework to show that customer
acquisition and retention are inherently linked. Thus, the ?rm
would never want to only maximize acquisition rates or max-
imize retention rates to maximize pro?tability since customer
retention relies directly on which customers were acquired. This
wouldonlyleadtoacquiringandretainingcustomers whoare not
pro?table in the long term. Instead, it is ideal to maximize pro?ts
from customer lifetime value (CLV) by simultaneously manag-
ing acquisition and retention of customers. Research by Reinartz
et al. (2005) used data from a high-tech B2B ?rm which simul-
taneously modeled acquisition likelihood, relationship duration,
and customer pro?tability. The authors found that it is neces-
sary to quantify trade-off between investments in acquisition
and retention in order to maximize ?rm pro?tability. Once we
can understand the links between customer acquisition and cus-
tomer retention, it is necessary to begin to link these metrics
with some ?nancial outcomes. Recent research by Fader et al.
(2005) used data from CDNOW to link RFM (Recency, Fre-
quency, and Monetary Value) to CLV using iso-value curves.
Once a ?rm is able to start establishing linkages between RFM
and CLV, the next step is to develop marketing strategies that can
continue to use metrics like RFM to increase CLV. Research by
Verhoef (2003) shows how establishing a series of direct mar-
keting campaigns or even loyalty programs that build affective
commitment and potentially lead to increases in future customer
purchase behavior. However, the end result is that while some
linkages have been established between customer acquisition,
customer retention, and CLV, there are still signi?cant opportu-
nities for research in marketing to advance these measures and
linkages.
Cross-buying and up-buying
Cross-buying and up-buying offer ?rms the chance to con-
tinue to increase revenue and pro?t contributions from current
customers, since it has been shown that customers who cross-
buy are more pro?table than customers who do not (Kumar et
al. 2008a). The dif?culty in implementing strategies to effec-
tively increase cross-buying and up-buying is in determining: (1)
which customers are likely to crossbuy, (2) which new products
those customers are likely to purchase, (3) what marketing mes-
sage to send those customers, and (4) when those customers are
likely to crossbuy. Recent research has started to address some of
these issues. First, Kumar et al. (2008a) used data from a major
catalog retailer to identify the drivers and consequences of cus-
tomers who crossbuy in different product categories. Managers
canuse these drivers, suchas average interpurchase time, toiden-
tify the ideal customers within the ?rm’s database that would
be most responsive to cross-selling and up-selling campaigns.
In addition, consequences of crossbuying behavior included
increases in contribution margin per order and number of orders.
Knott et al. (2002) used a Next-Product-to-Buy (NPTB) model
to assist a retail bank with identifying the customers who were
likely to purchase a speci?c loan product. Finally, Kumar et al.
(2006b, 2008d) use data froma B2Bhigh-tech ?rm, a B2C?nan-
cial services ?rm, and a B2Ctelecommunications ?rmprovide a
framework for targeting the right customers, with the right prod-
ucts, at the right time. The results show that when these three
decisions are modeled simultaneously, managers can signi?-
cantly increase their customer targeting accuracy. These studies,
though, only provide results of experiments with a few individ-
ual ?rms. There is still a great opportunity for research to explore
the effects of cross-selling and up-selling strategies on customer
and ?rm pro?tability across different ?rms and industries.
Multi-channel shopping
Where it was less common in the past for ?rms to have a
presence in multiple channels, with the advent of the Internet,
almost all ?rms now have a multichannel presence. The chal-
lenge then is to understand how each of the channels can impact
customer purchase behavior and customer pro?tability. First,
research has shown that customers who shop in multiple chan-
nels are more pro?table than customers who shop in only a
single channel. Using data from a high-tech B2B ?rm, Kumar
and Venkatesan (2005) show that customers who shop across
multiple distribution channels are more likely to score highly on
various customer-based metrics such as revenue and likelihood
to stay active. In addition, Venkatesan et al. (2007) use data from
an apparel retailer to show that customer who purchase across
more distribution channels have a higher future pro?t poten-
tial. However, research is just starting to analyze how ?rms can
develop a strategy to communicate with customers effectively in
different channels. Thomas and Sullivan (2005) use data from a
major US retailer to develop a six-step process of how to man-
age marketing communications with multichannel customers.
Additional research by Pauwels and Neslin (2008) uses data
from a major catalog retailer to quantify the impact of opening
a brick-and-mortar retail store when the only channels the ?rm
previously used was catalog and Internet. However, with the
continuing growth of retailers across many different channels,
several questions still remain. These research questions include
how to effectively migrate customers to different channels or
how to measure the impact of channels where no purchases
occur, for example using the Internet for search and the brick-
and-mortar store for purchase. This leaves ample opportunities
for future research to develop metrics that measure the impact
of multi-channel shopping on customer pro?tability.
Product returns
Up until recently many ?rms have seen product returns only
as a necessary hassle of doing business and a drain on pro?ts.
However, recent research has shown that products returns do
play a major role in the exchange process and customers who do
return a moderate amount of products are, ceteris paribus, the
most pro?table in the future. Petersen and Kumar (forthcoming,
2008) used data from a major US catalog retailer to ?rst show
that customers who return from10 to 15%of purchases purchase
more than customer who return too many or too few products.
The authors also showed that product returns are a key driver in
the computation of CLV and ?rms that do not incorporate prod-
uct returns directly into calculations of CLV will overestimate
CLV and improperly allocate marketing resources. In addition,
Anderson et al. (forthcoming) develop a structural model that
shows that it there is an option value of product returns that
102 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
is measurable, suggesting that omitting product returns from
an estimation of demand creates a bias and that it is possi-
ble to ?nd optimal product return policies for different ?rms.
This is mainly due to the fact that customers who have sat-
isfactory product return experiences tend to purchase more in
the future and have a stronger positive relationship with the
?rm. However, for ?rms to link a product returns metric to cus-
tomer value and ?rm performance more research is needed to
continue to identify the drivers and consequences of product
return behavior. Currently, there are two studies which iden-
tify some of the drivers of product return behavior. Anderson et
al. (2008) use data from a mail order catalog ?rm to compare
varying cross-item and within item attributes and their impact
on demand. One main ?nding was that the lower the price of
the item the less likely the item would be returned. Addition-
ally, Petersen and Kumar (forthcoming) empirically determined
several antecedents of customer product returns, including vari-
ables such as cross-buying and multichannel shopping behavior.
However, the sparse research on customer product return behav-
ior still leaves a signi?cant opportunity for future research to
continue to build product return metrics that can be useful for
managing customers for pro?ts.
As a result of all of this research, there exist many different
metrics for retailers to use to manage their customers. It becomes
increasingly important for managers to understand how each of
these metrics can be used strategically for short-term and long-
term business goals, that is how short-term metrics are linked to
long-term metrics and how both short-term and long-term met-
rics are linked to ?nancial performance. To do so, managers need
to also understand which metrics will provide relevant infor-
mation about the past and the current ?nancial position of the
?rm (backward-looking) and which metrics will help managers
lead ?rms into the future (forward-looking). Next, we provide
some discussion on the differences between backward-looking
and forward-looking metrics and which metrics can be linked to
future ?nancial performance.
Backward-looking versus forward-looking metrics
Many marketing metrics used by ?rms currently are
backward-looking, or at best present-looking, in nature
(Zeithaml et al. 2006). Examples of backward-looking met-
rics include measures of customer satisfaction relating to past
purchase experiences, measures of service quality relating to
past service experiences, and measures of perceived loyalty
that re?ect the customer’s perception of their own behavior up
to the current time period. Many of these backward-looking
metrics, along with several other operational and behavioral
measures, are provided for easy viewing on a periodic basis
(e.g., quarterly, monthly, weekly, daily, and even real-time) for
top managers through marketing dashboards, see Reibstein et
al. (2005) for a detailed description of marketing dashboards.
These backward-looking metrics serve the purpose of helping
marketing managers quantify the effectiveness of past marketing
campaigns that provide a clearer picture of current ?rm perfor-
mance. However, while these metrics can show managers why
the ?rm is at its current state, these metrics have been shown to
offer little to no predictive ability to future customer behavior
or ?rm performance.
More recently much of the academic research has focused
on forward-looking metrics that offer some predictive ability
about future customer behavior or ?rm performance (Kumar
2008b). These forward-looking metrics harness the power of
past customer attitudes and behaviors to try and offer some
predictive capabilities about future customer behavior and ?rm
performance. As a result, many of the backward-looking met-
rics have been used as predictors of future customer behavior
and ?rm performance. However, the results tend to be mediocre
at best. For example, when ?rms measure customer satisfaction,
by the time the data is received and analyzed, it re?ects yester-
day’s perceptions of satisfaction. It also does not include any
information about competitors’ actions or potential customer
prospects. All of these factors cause customer satisfaction to be
a less than adequate predictor of future customer behavior or
?rm performance.
This has led to many new forward-looking metrics, such as
customer lifetime value (CLV). Venkatesan and Kumar (2004)
uses behavioral information about past customer interactions
with the ?rm to predict future customer behavior and customer
value. The authors use variables such as average interpur-
chase time and cross-buying behavior and relate these variables
directly with future purchase frequencies and future contribu-
tion margins. In addition, these ?ndings have led to research in
marketing which has been able to account for competitive inter-
actions with customers. Rust et al. (2004b) use a brand switching
matrix and customer survey information to account for switch-
ing behavior among a set of customers. In an alternative method,
Kumar et al. (2008c) impute each customer’s competitive pur-
chase behavior by analyzing deviations in each customer’s
average interpurchase time. With regard to customer acquisi-
tion, Reinartz et al. (2005) account for a prospect’s likelihood
to buy from a ?rm for the ?rst time by comparing demographic
pro?les of prospects with those of current customers who are
pro?table. The end result of these forward-looking metrics, see
Zeithaml et al. (2006) for a more complete list and discussion,
has been to allow manager to more strategically plan effective
marketing actions and better justify the spending of marketing
resources on both current and potential customers.
Marketing metrics and ?rm value
Many of the metrics used by marketing managers can be
split into to main categories: metrics that measure the value of
brands and metrics that measure the value of customers. Many
of the metrics related to the value of brands were the product
of managers who wanted to quantify the intangible value of
?rms for purposes of mergers and acquisitions. Consequently,
this spurred much research on brand value and brand equity
(Aaker 1991; Keller 1993). Recently, there has been a call for
researchthat links brandequitytocustomer equityand?rmvalue
(Madden et al. 2006). Additional research has also showed how
to link an individual’s brand value to that individual’s customer
value. Kumar et al. (2008b) develop a conceptual framework
that ?rst uses the components of an individual’s brand value and
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 103
Fig. 1. Metrics we need to know.
links this value directly to behavioral outcomes such as purchase
frequency and contribution margin. Then the authors link these
behavioral outcomes with individual-level and ?rm-level ?rm
strategies to enhance CLV and ?rm value. These studies have
shown the importance of brand value metrics in their relation
to customer and ?rm value. It also allows managers to justify
marketing spend on increasing individual brand value and brand
equity since they can directly measure the impact of brand value
on customer and ?rm value.
There have been many different types of metrics that measure
customers, from customer satisfaction to customer behavior.
However, what seems to be a common thread amongst much
of the research on customer metrics is that the end goal is to
relate customer metrics to some form of customer value, for
example CLV or customer equity. There is a signi?cant amount
of research that has been providing frameworks for linking mar-
keting spend to customer value to ?rm value (Srivastava et al.
1998). Perhaps one of the biggest challenges to overcome in this
area is the availability of data on customer and ?rm value. Only
recently has research by Kumar and Shah (forthcoming) been
able to empirically link CLV to shareholder value. The authors
?rst calculate CLV and then link those values to a ?rm’s stock
price over time. This research is an important step for managers
to show that marketing efforts are measurable and are directly
related to ?rm value.
What we need to know
In Fig. 1, we organize the relevant metrics into four groups:
at the customer and store levels for the current and the future.
In addition, while the methods with which one estimates these
metrics at the customer or store level (unit of analysis) and for
the current and future (time period) may differ across the three
key retail channels (online, catalog, and brick-and-mortar) giv-
ing rise to 12 cells (2 unit of analysis ×2 time period ×3 retail
channels), we do not expect that the theoretical constructs them-
selves will differ across channels. Thus, we provide the metrics
and their linkages in a single framework in Fig. 1. We break
down the current metrics into three main categories: (1) transac-
tion information, (2) marketing information, and (3) competitive
information. We then provide some discussion on research that
has started to link these current metrics with future metrics –
both at the customer and store level. Finally, we discuss how
these future metrics are linked directly to ?nancial outcomes for
?rms.
Current value measures at the customer level
Pro?tability, of course, is an important metric with which
to evaluate customers. A number of different approaches have
been suggested to understand the drivers of the current level
104 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
of customer pro?tability. A number of these approaches have
already been covered in the “Assessment of the Literature”.
However, research needs to focus on metrics of pro?tability that
are forward-looking. Research should not use metrics such as
past customer value (PCV) which measures the past pro?t of
a customer. Instead, retailers should consider several key mea-
sures that relate to future pro?tability. One is the popular “RFM”
measures that report the time since the customer’s last purchase
(recency), the frequency with which the customer makes pur-
chases (frequency) and the amount a customer typically spends
(monetary value). Each piece of the RFM measure has been
shown to be a key driver in computing future customer prof-
itability and recent research has also shown how managers can
directly predict CLV by using each of the three inputs of RFM
(Fader et al. 2005).
Similarly, understanding the acquisition rate and the reten-
tion rate can give guidance as to the current numbers of active
customers. In addition, many ?rms use measures of “cost of
acquisition” and “cost of retention” as guidance for which cus-
tomers to acquire and retain. However, research has shown that
it is not necessarily ideal to only focus on trying to acquire cus-
tomers who are inexpensive to acquire and inexpensive to retain
(Reinartz et al. 2005). Instead, it is important to understand that
there are also pro?table customers who are expensive to acquire,
expensive toretain, or both. The keyis determiningwhichfactors
make different customers pro?table to the ?rm. Thus, managers
and retailers instead need to focus on metrics related to acquisi-
tion and retention that focus on acquisition pro?ts and retention
pro?ts. Inaddition, these metrics shouldnot be maximizediniso-
lation. There is an inherent dependence between acquisition and
retention requiring managers to determine spending on acquisi-
tion and retention simultaneously (Reinartz et al. 2005; Thomas
2001).
Past research has used past purchase frequencies to predict
future purchase frequencies. However, it is rare for customers,
especially in non-contractual settings to follow the same past
purchase frequencies over time. Instead research needs to focus
on methods to predict future interpurchase times to help describe
expected customer buying behavior. Regularities in interpur-
chase times have been fruitfully modeled with a generalized
gamma distribution. Allenby et al. (1999) use data from a major
investment brokerage to model inter-trading times of investors.
Additionally, Venkatesan and Kumar (2004) use a generalized
gamma distribution to predict the interpurchase times of the cus-
tomers from a high-tech B2B ?rm. Interpurchase times have
also allowed researchers to impute potential competitive pur-
chase behavior from customers. If a customer tends to follow a
general purchasing pattern, it is possible for statistical models
to uncover deviations in that purchasing pattern and attribute
those deviations to purchases from competitors (Kumar et al.
2008a,b,c,d, forthcoming). Thus, research needs to continue to
develop methods that can predict future customer buying pat-
terns.
Product returns are a growing problem with obvious impact
on a customer’s value, with the amount of product returns
exceeding $100 Billion annually (Blanchard 2007). However,
research has shown that there can be some pro?table bene?ts to
customer product returns. For example, customers who return
a moderate amount of purchases, 10–15% in total, have been
shown to on average purchase more into the future than any other
customer. Inaddition, customer product returnbehavior has been
linked directly to future customer buying behavior and CLV
(Petersen and Kumar forthcoming, 2008). This is a key example
where a metric commonly used for a different purpose outside
of marketing, in this case in operations management for sup-
ply chain management, can be used by marketing researchers to
establishlinkages betweencurrent andfuture customer behavior.
Metrics that go beyond the retailer’s own database to shed
light on a customer’s spending with competitor retailers allows
the retailer to understand what “share of wallet” it is attracting.
This leaves open the possibility that the retailer might focus on
growing customer expenditure from those customers for whom
the retailer currently has a small share of wallet and a large size
of wallet. Du et al. (2007) use data from a major US bank with
information about its customers’ account balances within and
outside of the bank. The authors ?nd little correlation between
the volume of transaction within the bank and with other banks,
along with the fact that a relatively small percentage of cus-
tomers account for a large proportion of external transactions.
This leaves opportunities for the bank to signi?cantly increase
value from customers with low share of wallet and high size of
wallet. In addition, this analysis can also show that it might not
be fruitful to spend signi?cant marketing resources to encourage
cross-selling and up-selling on highly pro?table customers who
are already spending their entire budget (100% share of wallet)
and instead market to customers with potential – those with low
share of wallet and high size of wallet. However, managers need
to be sure not to just base strategic decisions only on a customer’s
current share of wallet. Instead, research needs to focus on link-
ing share of wallet and size of wallet with future metrics such as
CLV by using share of wallet and size of wallet to identify the
optimal set of customers for marketing campaigns.
Current value metrics at the store level
Revenue continues to be an important metric with which
to evaluate stores. Two different models of store revenue were
presented at the Conference on Customer Experience Manage-
ment in Retailing at Babson College, April 24–26, 2008. Fig. 2
presents the model that Len Schlesinger (former COO of The
Fig. 2. The Limited’s model.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 105
Fig. 3. Overstock’s web analytics triangle.
Limited, nowPresident of Babson College) reported having used
to revitalize The Limited’s brick-and-mortar stores. Breaking
store sales down as described in Fig. 2 gave The Limited clear
guidance about how to motivate store personnel. Traf?c to the
store was largely determined by attributes of the shopping cen-
ter and the store’s position in that shopping center. Retail prices
were determined centrally. So incentives were designed to focus
store personnel on the two shaded boxes: conversion percent-
age and units purchased per transaction. The end result was that
these two metrics could be tied directly to future total store sales,
which in turn could be directly linked to future ?rm pro?ts and
shareholder value. Thus, Schlesinger’s experience suggests that
key brick-and-mortar store level metrics should include traf?c,
conversion percentage, units per transaction and average unit
retail price.
Also at that conference, Geoff Atkinson, VP of Tactical Mar-
keting for Overstock.com(an online lowprice retailer) presented
the model in Fig. 3. Atkinson suggested that many websites
make the mistake of looking primarily at conversion percentage
(as The Limited brick-and-mortar store did), which misses the
importance of average order size. Overstock chooses to focus on
revenue per visit as the primary metric of website performance
since that metric takes both conversion and average order size
into account. Overstock uses these metrics when comparing two
or more variables and tracks the results using Omniture and sev-
eral internal systems. Overstock measures site performance on
these metrics daily and weekly. Trends in these metrics are used
to evaluate website design improvements. Overstock’s expe-
rience suggests that online store level metrics should include
orders, revenue, visits, conversions, average order size and rev-
enue per visit – similar to those in brick-and-mortar and other
of?ine channels (e.g., catalogs).
In addition to the revenue focused metrics presented by The
Limited and Overstock.com, we also need metrics linking store
revenue to store pro?t. Such measures include marketing spend-
ing, pro?t margins, contribution dollars, and awareness of retail
stores and speci?c retail brands. In general if we follow the
hierarchy of effects model (awareness, liking, trial, repeat, loy-
alty), it is thought that raising awareness of the retail store or
a given brand will eventually lead to an initial and potentially
repeat purchase. In addition, if the store has a grasp on its pro?t
from the sales of goods (pro?t margin and contribution dollars)
and its ability to acquire and retain customers (effectiveness of
marketing spend), with the general awareness of the store and
its products/services, then a direct link can be drawn to ?rm
pro?tability and shareholder value.
Future value measures at the customer level
Virtually all measures of the important metric “Customer
Lifetime Value” are built with historical data and hence re?ect
the customer’s expected future value if nothing changes as the
customer moves into the future. But, of course, things will
change and some of those changes are under the control of
marketers. There is need for forward-looking measures of a
customer’s expected lifetime value and those measures should
include the impact of different marketing programs on that life-
time value.
Questioning Reichheld’s (2003) work on a “net promoter
score” (the percentage of surveyed customers who report that
they are willing to recommend this company to a friend) led to
exploring that a customer’s value can be more than the value
of the purchases that customer makes from the company as
observed by Kumar et al. (2007). Speci?cally, Kumar et al.
(2007) de?ne a customer’s referral value or CRV as the value
of the business that a customer brings in minus the marketing
costs that prompted that customer to make a referral. However,
this research is just the beginning for metrics related to word of
mouth and referrals. Research needs to continue to focus on the
drivers of the value of word of mouth and its implications on
customer and ?rm value.
Future value metrics at the store level
Positive “word of mouth”, particularly as spread on the Inter-
net, has been shown to be an important indicator of future
success. Godes and Mayzlin (2004b) linked chat room com-
ments to TV show ratings, Chevalier and Mayzlin (2006) linked
online reviews and ratings to online book sales, and Dhar
and Chang (2007) linked CD sales to blog posts and reviews.
Measures of retailer-related online chat, reviews and blogs are
relevant future value metrics at the store level. Research needs
to continue to work on linking general word of mouth or ‘buzz’
about products to future sales. This way managers can predict
the impact of a word of mouth marketing campaign is likely to
have on future sales to determine the optimal level of investment.
A retailer’s own brand equity, as measured by consumer sur-
veys, should be related to the future value of the ?rm. As noted
by Leone et al. (2006), one could also measure a retailer’s brand
equity as the sum of the customer equity (closely related to an
aggregate measure of customer lifetime value) associated with
each of the retailer’s customers. On a related point, Leone et
al. (2006) point out as well that the value of a manufacturer’s
brand to a retailer is not the same as the value of that brand to
the manufacturer. In particular, the value of a manufacturer’s
brand to a retailer is related to the value, to the retailer, of the
customers who buy that brand fromthe retailer. Thus, going for-
ward research should focus on trying to link the increase of a
?rm or a retailer’s brand value to a corresponding increase in
total sales. Further, it is also important for some ?rms to iden-
tify an individual’s brand value (IBV) and relate that IBV to
106 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
that customer’s lifetime value (Kumar et al. 2008b). Then the
?rm can also make strategic decisions to build brand value and
in turn customer value at the individual level through targeted
marketing campaigns.
Just as we noted that share of wallet is an important cus-
tomer level, current value metric that requires data beyond that
found in a retailer’s CRM database, “growth of the retailer’s
customer base” is a store level, future value metric that requires
data beyond that in a retailer’s CRMdatabase. In order to growa
store’s customer base, the retailer needs information about con-
sumers who are not currently shopping at the store. With this
information a retailer can compute its share of customers across
the industry or in the marketplace and also identify the best
prospects to target in order to grow the customer base. There
are signi?cant challenges to gathering data across all ?rms in an
industry or even to gather panel data from a large sample of cus-
tomers. However, it is still necessary for managers to understand
what factors impact a growth in their customer base – whether it
includes customers that switch from one company to another or
whether it includes new customers who have never purchased in
a given product category. Either way, research has to continue to
develop methods for predicting a ?rm’s growth in its customer
base.
Finally, the ultimate store level, future value metric is a ?nan-
cial outcome: shareholder value or stock price. The market’s
expectations about the future of the ?rm are impounded in a
?rm’s shareholder value or stock price. It would be particularly
useful to develop an understanding of the factors that in?uence a
retailer’s stock price. In other words, what are the current/future
and customer/store level metrics that can be linked to a retailer’s
stock price and in turn how can these metrics be strategically
used to increase a retailer’s stock price? To this point few stud-
ies have been able to build links between marketing metrics
and shareholder value or stock price, with few exceptions (e.g.,
Kumar and Shah forthcoming). However, there are still ample
opportunities to continue to develop these linkages, especially
since it is often necessary for marketers to continue to justify
their resource allocations with empirical evidence showing the
impact of their decisions.
How do we get there?
As noted earlier in this paper, it is critical to develop metrics
in twelve “cells” of activity: at both the customer and store level
and each with implications for both the current and the future
time period. Then, each of those cells must be measured for
the three key channels – online, catalog, and brick-and-mortar.
An immediate question is how do we obtain the data neces-
sary for these measurements? There are immediate challenges
that are faced by both researchers and practitioners when devel-
oping and implementing marketing strategies based on metrics
that are directly linked to customer value and ?rm performance.
First, ?rms need to be able to capture the appropriate customer
and marketing data to even develop measures for different met-
rics. Then, managers and academics alike need to determine
how to operationalize these variables so that they can be used
effectively. This can only be done when the ?rm captures data
at relevant intervals of time. Managers then need to properly
track the incoming data and the effectiveness of marketing pro-
grams to continue to develop new programs. Next, it is crucial
to use this data to build ?eld experiments to test the effective-
ness of these predictions. Finally, it is necessary for these ?rms
to disseminate this knowledge within ?rms to gain buy-in from
not only marketing and sales, but also from top management.
Only then can marketing justify its position within the ?rm. We
discuss each of these points in detail below.
Data capture
Let us assume that we can use the basic data obtained for
both current metrics and for forecasting, extrapolation into the
future, or both. This implies that we need measures from the
three channels at two levels of aggregation. To take measures at
the customer level implies that the ?rmmust have an information
systemin place to systematically track and capture that informa-
tion. Consider the online case. There is a considerable amount
of data generated from online behavior but only purchase data
can be captured longitudinally.
1
Brick-and-mortar channels only
permit longitudinal data collection through devices like super-
market panels which address a very small sample of consumers.
The best data are obtained from catalogs where it is relatively
easy and customary to capture customer-level data. Ultimately,
these data can then be aggregated to the store level. However,
take the instance of a ?rm that desires to implement a CRM
system to capture data about current and potential customers.
Research suggests that about 70% of CRM system implementa-
tions fail in improving the bottom-line. How might a ?rm create
a successful approach to capturing data? Much of the success of
technology adoption in a ?rm comes from a ?rm’s willingness
to develop an incentive structure based on CRM usage and per-
formance and the willingness to initiate, maintain, and terminate
relationships with customers based on the recommendations of
the CRM system (Krasnikov et al. 2008; Reinartz et al. 2004).
Thus, it is important for ?rms to not just begin to collect data
about customers, but to adopt data capture technologies that are
aligned with the incentive structure of the ?rm.
Operationalization
Then once the data is collected, the next task is to identify
howeach of the metrics will be operationalized. In simpler cases,
metrics such as pro?t at the customer level can directly mean
how much pro?t a customer has provided the retailer in a given
time frame. However, many metrics are much more dif?cult to
operationalize. We only need to look at metrics such as customer
satisfaction and customer loyalty to see the issues of metric oper-
ationalization. How should a ?rm operationalize a metric such
as loyalty? We could de?ne loyalty any number of ways. First, in
terms of behavioral loyalty, it could refer to customer tenure, or
the length of time since the customer made a purchase. It could
1
Considerable progress is being made on “behavioral targeting” which per-
mits search behavior across web site visits to be captured, but these data are not
typically added to a customer’s data ?le.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 107
also be de?ned as the number of times a customer has purchased
from a given ?rm. Finally, it could also be a combination of sev-
eral metrics, such as an RFM score. In addition, loyalty could
also be attitudinal. Firms could measure loyalty through sur-
veys as the perception of the ?rm in the eyes of the customer.
Also, other metrics which may seemstraightforward to measure,
such as product returns, can give different implications to man-
agers based on their operationalization. For example, a recent
study showed that three different operationalizations of product
returns – (1) number of product returns, (2) value of product
returns, and (3) ratio of products returned to total products pur-
chased – gave three different conclusions to managers. This is
a major problem when marketing managers try to link market-
ing metrics to ?nancial performance. In different cases, different
variable operationalizations provide vastly different results. This
makes it extremely important for marketers to begin to standard-
ize a single method of measuring each metric and to always use
that method when linking a metric to a ?nancial outcome. This
can increase the validity of the measure and also allow for com-
parisons of metrics across ?rms, across research studies, and
across time.
Measurement interval
Another relevant issue is the measurement or data interval.
The appropriate interval should be matched to the product cat-
egory. For frequently purchased products, weekly measures are
appropriate. While not all customers purchase every week, it is
critical particularly for the store-level measures to have short
measurement intervals. This also permits the analyst to develop
the best measures for the metrics describing future behavior.
For durable goods or other products/services purchased less
frequently, weekly data are probably unnecessary. In any case,
the measurement interval should always be less than the aver-
age interpurchase time in order to account for heterogeneity in
purchasing across consumers. A main issue that arises out of
different measurement intervals occurs when different depart-
ments of a ?rm capture and measure data at different intervals.
For example, take a large multi-national CPG company. It may
be the case that this company measures sales across all stores on
a weekly basis. This company may also measure advertising and
other marketing spend on a monthly basis. Finally, this company
may measure changes in distribution of products on a quarterly
basis. Take another example of a high-tech B2B?rm. This high-
tech B2B ?rm collects information about sales to customers on
a monthly level, but keeps information about marketing expen-
ditures to customers (e.g., sales calls) as they occur. Thus the
question is: Which level of time aggregation should be used for
a longitudinal study? Should the ?rm aggregate the data to the
monthly quarterly level and lose the variability of the weekly or
daily transaction data? The problem here is a loss of variance
in the more frequently measured data due to aggregation. Or,
should the ?rmbring the quarterly or monthly data to the weekly
or daily level by using average weekly or daily values for mar-
keting and distribution data? The problem here is that the ?rm
maymake not gainthe full picture of the purchasingor marketing
cycle by tracking it too frequently. In addition, the correct answer
to both questions may depend on the research question that we
are trying to answer. To avoid this problem in the ?rst place,
?rms need to consider appropriate data measurement intervals
andcapture data across all facets of the business at those intervals
most appropriate for the products and services being sold.
Tracking
A basic level of analysis is tracking the metrics. While this
does not show relationships between metrics, two-dimensional
visual representations (metric over time) can stimulate discus-
sion about the causes of a particular up or down trend. In
addition, tracking is easy to communicate to senior management
and throughout the organization. However, absolutely crucial is
going beyond tracking to develop analyses that link marketing
activities and the metrics. This gives the company the ability
to allocate resources over marketing programs that will pro-
vide the greatest leverage in expanding the particular metric. For
example, if the metric is customer lifetime value in the catalog
channel, linking number of catalogs sent to a customer’s CLV
aids in the determination of the optimal number of catalogs to
mail to each customer. Venkatesan and Kumar (2004) used data
from a high-tech B2B ?rm to track the frequency and timing of
different types of marketingcommunications andtheir impact on
a customer’s decision to purchase. This gives the ?rm an under-
standing of how decisions to allocate resources to customers are
likely to affect a customer’s decision to purchase in the future.
Similarly, if the metric is at the ?rmlevel such as market share or
excess stock price returns, companies can evaluate the impact of
programs such as R&D, branding, and so forth on these metrics.
Measuring alone is insuf?cient – developing models and estab-
lishing relationships between “inputs” and “outputs” is essential.
How can this be done? A recent study which won the Market-
ing Science Practice Prize competition in 2003, exhibits exactly
how a German catalog company successfully tracked its mar-
keting communications over time to optimize catalog sending
to customers (Elsner et al. 2004). The authors used a dynamic
multi-level modeling approach with the elasticities of catalog
mailing along with segmenting customers using RFM scores to
determine optimal catalog mailing practices. This led the catalog
company to go from the 5th to the 2nd market position.
Similarly, it is important to develop relationships between the
metrics themselves to see if there are “suf?cient statistics,” that
is, measures that are highly correlated with others implying that
the dashboard can be smaller. For example, ?rm-level metrics
such as pro?tability and share price should be highly correlated.
In addition, it would be useful to know if a particular metric is
a leading indicator for another. For example, changes in brand
equity should lead changes in sales, market share, or both.
Experimentation
The tracking and linking analyses just described are essential
parts of a metrics program. However, a very useful step beyond
these is to run controlled ?eld experiments where the manager
manipulates some aspect of the marketing program and moni-
tors how it affects a selected metric. Particularly, amenable to
108 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
experimentation are the online and catalog channels. The latter
has been the subject of experimentation for many years, the for-
mer since the inception of the Web-based channel. For example,
a clothing retailer that uses both a catalog and an Internet site
might be interested in seeing the impact of changing the mer-
chandise mix or pricing on sales. The company would randomly
selected customers who either visit the Web site or receive a
catalog and offer the changed merchandise or price and keep
another group as a control. Given the availability of customer-
level data, however, the impact of the policy change on CLV
could also be determined. While this can be done in a brick-and-
mortar channel, it is much more dif?cult to implement given the
need to coordinate over stores and measure the impact at the
customer level (at least for the CLV metric). However, there is
limited research in marketing that deals with direct experimenta-
tion. We ?rst see experimentation in marketing research Eastlack
and Rao (1986) with the Campbell Soup company. However,
since that time, there have been few studies that have conducted
actual experiments with ?rms. An example of a more current
?eld experiment includes Kumar et al. (2006b, 2008d) where the
authors used data from a high-tech B2B ?rm and B2C telecom-
munications and ?nancial services ?rms to guide salesperson
decision-making. While it is more dif?cult to publish studies
that deal with experiments, it is much needed and can take care
of many confounding factors.
Dissemination
From an organizational perspective, it is important that key
metrics be disseminated throughout the company. Employees
should be aware that senior management is very interested in
both company- and customer-level metrics as it not only gives
direction for data collection, but also demonstrates that man-
agement is seriously interested in its customers. There is also
a rationale for “common ground” in that all managers need to
know how their brands are being evaluated. Thus, there is a
serious signaling effect that communicating key metrics has in
the organization. Employees who own stock in the company are
used to tracking the share price daily; they should also get used
to tracking the key metrics at their desks.
Most companies are trying to do the communication through-
out the organization through the use of a dashboard. This allows
senior management the ability to indicate what are the important
measures that all within the organization should be looking at
in common (Pauwels et al. 2008). It has been reported that the
McDonald’s CEO used to have the corporate dashboard strate-
gically placed behind his desk. As such, anyone that came in
to meet with him was looking at the CEO AND the dashboard.
It was a strong message of what was important and what was
being viewed by the top of the organization. Adashboard further
keeps everyone up to date on the status of the brand/?rm. One
could view the dashboard in the form of a tree structure – there
are the key output measures, as discussed previously, and then
the drivers of each of these measures. As one gets further down
the organization it is possible to “drill down” on each of these
measures and understand the causal factors. For example, share-
holder value is a key output measure and CLV being a driver.
Then, the question is what drives CLV? That would be acqui-
sition, purchase amount, purchase frequency, and retention. We
could then look at what marketing actions tend to drive each of
these components. As can be quickly seen, with multiple out-
put measures, and most actions affecting multiple drivers, the
complexity starts to proliferate.
Linkages
Ideally, the linkages between drivers and output variables
need to be well understood, and hopefully, empirically deter-
mined. Unfortunately, these relationships are not all well known.
In part, this is because of the lack of data, and in part, because
they are not all in the same place. In the absence of data driven
relationships, judgmental parameters are necessary, often via
decision calculus. Ironically, it is often the case that managers do
not feel comfortable specifying their own judgment, yet have to
make decisions based on this judgment all the time. If we can get
the judgmental or empirically based relationships established,
then it would be possible to continually experiment and update
the relationships of these linkages. Some companies are very
good at constantly experimenting, while others are not. How-
ever, without continual experimentation, these linkages cannot
be well established or well understood.
The best place to start is by specifying the marketing objec-
tives. Each retailer will have their own objectives. Getting
alignment on the objectives is always a critical, and not nec-
essarily, an easy step. The next step is to establish hypotheses of
what are the drivers. Only then does it make sense to estimate
the relationships. Reibstein et al. (2005), discuss the ?ve steps
in developing such a dashboard which include:
1. Selecting the key metrics
2. Populating the dashboard with data
3. Establishing the relationships between dashboard items
4. Forecasting and “what if” analysis
5. Connecting to ?nancial consequences
In this article we have provided much of the material nec-
essary to help managers develop a marketing dashboard for
their ?rm. We have provided an assessment of the metrics that
exist and those metrics which are necessary to monitor. We have
provided guidance as to how ?rms can overcome the common
challenges of migrating to these new metrics and capturing and
tracking this data over time. Finally, we have shown which met-
rics have been shown to provide linkages to ?nancial outcomes
such as CLV and shareholder value. However, this research still
needs to continue to re?ne the selection and measurement of
marketing metrics and continue to develop linkages between
marketing metrics and ?nancial performance. Only then can
marketing justify its place in the board room.
Guidelines for future research
Ongoing research on marketing metrics will continue to
provide insights to marketing managers as they establish opti-
mal marketing strategies which are directly linked to ?nancial
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 109
outcomes. Thus, we provide a roadmap of much needed research
in the area of marketing metrics. First and foremost, research
needs to continue to establish the linkages between marketing
metrics and ?rm performance. Next, research needs to continue
to focus on the two tenets of marketing – customer acquisi-
tion and customer retention. In addition, marketing is constantly
evolving and managers need to understand how “new” mar-
keting, such as social content, can impact current marketing
practice. Most importantly, the emphasis on these studies should
continue to focus on metrics that can guide future decision-
making and not focus only on the past. Finally, there needs to
be a focus on research that relates directly to retailers, both in
the measurement of retailer brand value and retailer market cap-
italization. Too often research is focused at the ?rm-level giving
retailers few insights about strategic marketing decisions. We
discuss each of these calls for new research below.
Establishing linkages
While we have outline the taxonomy cells for retailer met-
rics, there is much work which remains ahead. The American
Marketing Association ran a conference in Atlanta in July 2008
for the Knowledge Coalition. This ?rst endeavor of its sort was
to establish what we know, in research and in practice, about
the relationships between marketing spending and marketing
results. The result was that much more work still needs to be
done in this area. Too often when a company comes to cut a bud-
get within an organization, marketing is cut off ?rst. Often this
is because marketing is unable to establish empirical linkages
between marketing metrics, such as awareness, and ?nancial
outcomes, such as stock price. Thus, it is crucial that marketing
managers begin to not only use their intuition to drive marketing
decisions, but to also empirically link marketing metrics to ?nan-
cial outcomes. Then it will become easier to justify the need for
marketing resources and also quantify both the negative short-
termand long-termimpact of a reduction in marketing resources.
Customer acquisition and retention
We also know that customer lifetime value is an important
metric for retailers as well as others and that the sum of the
customer lifetime values of all customers is customer equity.
All ?rms strive to grow customer equity over time. This effort
can only happen when resources are spent on trying to retain
current pro?table customers and to acquire new pro?table cus-
tomers. Some research has empirically shown that it is necessary
to balance this acquisition and retention budget (Reinartz et al.
2005; Thomas 2001), however continual work is needed to bet-
ter understand what drives customer acquisition and retention.
Only then can marketing managers leverage the drivers of acqui-
sition and retention to continue to groweach customer’s lifetime
value and overall customer equity.
Social content
A new emerging ?eld, and still not well known, is the impact
of social content, such as product and store reviews, and blogs
(Godes and Mayzlin 2004a,b). Bizrate.com, Shopzilla.com, and
Shopping.com all were established to provide user feedback to
future shoppers about retail shopping sites. There is now the
continual growth of such websites such as Bazaarvoice.comthat
empowers retail customers to help informfuture shoppers. Many
?rms have turned to these forms of “new marketing” to try and
grab a ?rm hold on the emerging trends. However, it is still
unclear how ?rms can learn anything from these new media or
even use these new media in predictive customer behavior and
in turn ?rm pro?ts. The impact of such social networking is
needs to be better understood and provides many opportunities
for future research.
Relating current actions to future actions
The big question facing marketers today is the impact of their
current behavior on future performance (Zeithaml et al. 2006).
It is often hard to use any current metrics to predict future per-
formance. Most measures are short-term, and certainly drawing
the link is easier when actions and results are contemporaneous.
The question we are still in need of better exploration is the long-
termimpact of marketing actions. Thus, future research needs to
begin to identify potential metrics that have the ability to predict
future events (beyond just CLV and customer equity) to some
degree of accuracy so that managers can track these leading indi-
cators to get a better grasp on where the ?rmis heading – or even
any interventions that are necessary to continue to increase ?rm
value.
Retailer brand value
Marketers for years have been focused on brand establish-
ment. Retailers have brands of their own, as well. There has
been very little research that has focused speci?cally on the
establishment of retail brands, whether these are speci?c prod-
ucts with a retailer’s brand name (Kumar and Steenkamp 2007)
or the brand equity of the name of the retail store. One question
is the impact of the brands that a retailer carries and its impact on
the retailer’s brand itself. Future research needs to help quantify
retail and private label brand values so that managers can make
more strategic decisions to increase ?rm value.
Retailer market capitalization
Lastly, the ultimate is drawing the link between marketing
spending and market capitalization. While some work has begun
for manufacturing ?rms, little has been done for speci?cally for
retailers. Given the increased complexity of real estate, and its
?uctuating value, do the same principles hold for retailers as
what has been found to date for manufacturers? This is another
great opportunity for future research to investigate.
So, while progress has been made, there is considerable work
still to be done to better understand the role of retailer metrics.
In summary, this article has identi?ed the directions retailers
and manufacturers can take to not only build a pro?table cus-
tomer database, but also focus on building the brand value/equity
through establishment of superior marketing metrics.
110 J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111
References
Aaker, David A. (1991), “Managing Brand Equity,” New York: Free Press.
Allenby, Greg M., Robert P. Leone and Lichung Jen (1999), “A Dynamic Model
of Purchase Timing with Application to Direct Marketing,” Journal of Amer-
ican Statistical Association, 94 (June), 365–74.
Ambler, Tim (2003), “Marketing Metrics,” second ed. Prentice-Hall.
Anderson, Eric T., Karsten Hansen, and Duncan Simester (forthcoming), “The
Option Value of Returns: Theory and Empirical Evidence,” Marketing Sci-
ence.
and Lei K. Wang (2008), “Howare Demand and Returns
Related?,” Working Paper, Kellogg School of Management.
Blanchard, David (2007), “Supply Chains Also Work in Reverse,” in Industry
Week.
Chevalier, Judith A. and Dina Mayzlin (2006), “The Effect of Word of Mouth on
Sales: Online Book Reviews,” Journal of Marketing Research, 43 (August),
345–54.
Dhar, Vasant and Elaine Chang (2007), “Does Chatter Matter? The Impact of
User-Generated Content on Music Sales,” Working Paper, New York Uni-
versity.
Du, Rex Yuxing, Wagner A. Kamakura and Carl F. Mela (2007), “Size and Share
of Customer Wallet,” Journal of Marketing, 71 (April), 94–113.
Eastlack, Joseph O. Jr. and Ambar G. Rao (1986), “Advertising Experiments at
the Campbell Soup Company,” Marketing Science, 8 (1), 57–71.
Elsner, Ralf, Manfred Krafft and Arnd Huchzermeier (2004), “Optimizing
Rhenania’s Direct Marketing Business Through Dynamic Multilevel Mod-
eling (DMLM) in a Multicatalog-Brand Environment,” Marketing Science,
23 (2), 192–206.
Fader, Peter S., Bruce G.S. Hardie and Ka Lok Lee (2005), “RFM and CLV:
Using Iso-Value Curves for Customer Base Analysis,” Journal of Marketing
Research, 42 (November), 415–30.
Farris, Paul W., Neil T. Bendle, Phillip E. Pfeifer and David J. Reibstein (2006),
“Marketing Metrics: 50+ Metrics Every Executive Should Master,” 3rd ed.
Upper Saddle River, NJ: Wharton School Publishing.
Godes, David and Dina Mayzlin (2004a), “Firm-Created Word-of-Mouth Com-
munication: AField-Based Quasi-Experiment,” Working Paper, Yale School
of Management.
(2004b), “Using Online Conversations to Study Word of
Mouth Communication,” Marketing Science, 23 (4), 545–60.
Gupta, Sunil, Donald R. Lehmann and Jennifer Ames Stuart (2004), “Valuing
Customers,” Journal of Marketing Research, 41 (1), 7–18.
and Valarie A. Zeithaml (2006), “Customer Metrics
and Their Impact on Financial Performance,” Marketing Science, 25 (6),
718–39.
Hogan, John E., Katherine N. Lemon and Barak Libai (2003), “What is the true
value of a lost customer?,” Journal of Service Research, 5 (3), 196–208.
Keller, Kevin Lane (1993), “Conceptualizing, Measuring, and Managing
Customer-Based Brand Equity,” Journal of Marketing, 57 (1), 1–22.
Kerin, Roger A. and Raj Sethuraman (1998), “Exploring the Brand Value-
Shareholder Value Nexus for Consumer Goods Companies,” Journal of the
Academy of Marketing Science, 26 (4), 260–73.
Knott, Aaron I., Andrew Hayes and Scott A. Neslin (2002), “Next-Product-
to-Buy Models for Cross-Selling Applications,” Journal of Interactive
Marketing, 16 (3), 59–75.
Krasnikov, Alexander, Satish Jayachandran, and V. Kumar (2008), “The Impact
of CRM Technology on Cost and Pro?t Ef?ciencies: Evidence from US
Commercial Banking Industry,” Working Paper, University of South Car-
olina.
Kumar, Nirmalya and Jan-Benedict E.M. Steenkamp (2007), “Brands Under
Attack fromPrivate Labels,” in Private Label Strategy Boston, MA: Harvard
Business School Press.
Kumar, V. (2008a), “Managing Customers for Pro?t,” Upper Saddle River, NJ:
Wharton School Publishing.
(2008b), “Customer Lifetime Value: The Path to Prof-
itability,” The Netherlands: NOW Publishers, Inc.
, Morris George and Joseph Pancras (2008a), “Crossbuy-
ing in Retailing: Drivers and Consequences,” Journal of Retailing, 84 (1),
15–27.
, Anita Man Luo, and Vithala R. Rao (2008b), “Linking
an Individual’s Brand Value to the Customer Lifetime Value: An Integrated
Framework,” Working Paper, Georgia State University.
Kumar, V. and Andrew Petersen (2005), “Using a Customer-Level Market-
ing Strategy to Enhance Firm Performance: A Review of Theoretical and
Empirical Evidence,” Journal of the Academy of Marketing Science, 33 (4),
504–19.
and Robert P. Leone (2007), “How Valuable is Word of
Mouth?,” Harvard Business Review, 85 (10), 139–46.
(2008c), “Exploring the Effectiveness of Marketing
Communications on Customer Buying and Referral Behavior,” Working
Paper, Georgia State University.
Kumar, V. and Denish Shah (forthcoming), “Expanding the Role of Marketing:
From Customer Equity to Market Capitalization,” Journal of Marketing.
and Rajkumar Venkatesan (2006a), “Managing Retailer
Pro?tability—One Customer at a Time!,” Journal of Retailing, 82 (4),
277–94.
and Rajkumar Venkatesan (2005), “Who are Multi-
channel Shoppers and how do they Perform?: Correlates of Multichannel
Shopping Behavior,” Journal of Interactive Marketing, 19 (2), 44–62.
, Timothy R. Bohling, and Denise Beckman (forthcom-
ing), “The Power of CLV: Managing Customer Value at IBM,” Marketing
Science.
, Rajkumar Venkatesan and Werner Reinartz (2006b),
“Knowing What to Sell, When, and to Whom,” Harvard Business Review,
84 (3), 131–7.
(2008d), “Performance Implications of Adopting a
Customer-Focused Sales Campaign,” Journal of Marketing, 72 (September),
50–68.
Leone, Robert P., Vithala R. Rao, Kevin Lane Keller, Anita Man Luo, Leigh
McAlister and Rajendra K. Srivastava (2006), “Linking Brand Equity to
Customer Equity,” Journal of Service Research, 9 (2), 125–38.
Madden, Thomas J., Frank Fehle and Susan Fournier (2006), “Brands Matter:
An Empirical Demonstration of the Creation of Shareholder Value Through
Branding,” Journal of the Academy of Marketing Science, 34 (2), 224–35.
Pauwels, Koen, Tim Ambler, Bruce Clark, Pat LaPointe, David J. Reibstein,
Bernd Skiera, Berend Wierenga and Thorsten Wiesel (2008), “Dashboards &
Marketing: Why, What, How and Which Research is Needed?,” Cambridge,
MA: Marketing Science Institute.
and Scott A. Neslin (2008), “Building with Bricks and
Mortar: The Revenue Impact of Opening Physical Stores in a Multichannel
Environment,” Report 08-101, MarketingScience Institute, Cambridge, MA.
Petersen, J. Andrewand V. Kumar (forthcoming), “Are Product Returns a Neces-
sary Evil? The Antecedents and Consequences of Product Returns,” Journal
of Marketing.
(2008), “CLV and Optimal Resource Allocation: The
In?uence of Marketing, Buying and Product Returns,” Working Paper, Uni-
versity of North Carolina at Chapel Hill.
Reibstein, DavidJ., YogeshJoshi, DavidNortonandPaul W. Farris (2005), “Mar-
keting Dashboards: A Decision Support System for Assessing Marketing
Productivity,” in Presentation at the Marketing Science Conference.
Reichheld, Frederick F. (2003), “The One Number You Need to Grow,” Harvard
Business Review, 81 (12), 46–54.
Reinartz, Werner, Manfred Krafft and Wayne D. Hoyer (2004), “The Customer
Relationship Management Process: Its Measurement and Impact on Perfor-
mance,” Journal of Marketing Research, 41 (August), 293–305.
, Jacquelyn S. Thomas and V. Kumar (2005), “Balancing
Acquisition and Retention Resources to Maximize Customer Pro?tability,”
Journal of Marketing, 69 (1), 63–79.
Roberts, John and Tim Ambler (2006), “Beware the Silver Metric: Marketing
Performance Measurement Has to Be Multidimensional,” Report 06-113,
Marketing Science Institute, Cambridge, MA.
Rust, Roland T., Tim Ambler, Gregory S. Carpenter, V. Kumar and Rajendra K.
Srivastava (2004a), “MeasuringMarketingProductivity: Current Knowledge
and Future Directions,” Journal of Marketing, 68 (4), 76–89.
, Katherine N. Lemon and Valarie A. Zeithaml (2004b),
“ReturnonMarketing: UsingCustomer EquitytoFocus MarketingStrategy,”
Journal of Marketing, 68 (1), 109–27.
J.A. Petersen et al. / Journal of Retailing 85 (1, 2009) 95–111 111
Simon, Carol J. and Mary W. Sullivan (1993), “The Measurement and Determi-
nants of Brand Equity: A Financial Approach,” Marketing Science, 12 (1),
28–52.
Srivastava, Rajendra K., TasadduqA. Shervani andLiamFahey(1998), “Market-
based assets and shareholder value: A framework for analysis,” Journal of
Marketing, 62 (1), 2–18.
Thomas, Jacquelyn S. (2001), “A methodology for linking customer acquisition
to customer retention,” JMR, Journal of Marketing Research, 38 (2), 262–8.
and Ursula Y. Sullivan (2005), “Managing Marketing
Communications with Multichannel Customers,” Journal of Marketing, 69
(October), 239–51.
Venkatesan D Rajkumar and V. Kumar (2004), “A Customer Lifetime Value
Framework for Customer Selection and Resource Allocation Strategy,” Jour-
nal of Marketing, 68 (4), 106–25.
Venkatesan, Rajkumar, V. Kumar and Nalini Ravishanker (2007), “Multichannel
Shopping: Causes and Consequences,” Journal of Marketing, 71 (April),
114–32.
Verhoef, Peter C. (2003), “Understanding the Effect of Customer Relationship
Management Efforts on Customer Retention and Customer Share Develop-
ment,” Journal of Marketing, 67 (4), 30–45.
, Philip Hans Franses and Janny C. Hoekstra (2001),
“The Impact of Satisfaction and Payment Equity on Cross-buying: A
Dynamic Model for a Multi-service Provider,” Journal of Retailing, 77 (3),
359–78.
Villanueva, Julian, Shijin Yoo and Dominique M. Hanssens (2008), “The
Impact of Marketing-Induced vs. Word-of-Mouth Customer Acquisition
on Customer Equity Growth,” Journal of Marketing Research, 45 (1),
48–59.
Zeithaml, Valarie A., Ruth N. Bolton, John Deighton, Timothy L. Keiningham,
N. Katherine, J. Lemon and Andrew Petersen (2006), “Forward Looking
Focus: Can Firms Have Adaptive Foresight?,” Journal of Service Research,
9 (2), 168–83.
doc_582821112.pdf