v. kumar & werner reinartz creating enduring customer...
TRANSCRIPT
V. Kumar & Werner Reinartz
Creating Enduring Customer ValueOne of the most important tasks in marketing is to create and communicate value to customers to drive theirsatisfaction, loyalty, and profitability. In this study, the authors assume that customer value is a dual concept. First, inorder to be successful, firms (and the marketing function) have to create perceived value for customers. Toward thatend,marketers have tomeasure customer perceived value and have to provide customer perceptions of value throughmarketing-mix elements. Second, customers in return give value through multiple forms of engagement (customerlifetime value, in the widest sense) for the organization. Therefore, marketers need to measure andmanage this valueof the customer(s) to the firm and have to incorporate this aspect into real-time marketing decisions. The authorsintegrate and synthesize existing findings, show the best practices of implementation, and highlight future researchavenues.
Keywords: customer value, perceived value, customer lifetime value, CLV models, customer engagement
A1991 BusinessWeek article describes the notion of“perceived value” as “the new marketing mania” and“the way to sell in the ’90s” (Power 1991). Morris
Holbrook (1994, p. 22) wrote, “despite this obvious importanceof customer value to the study ofmarketing in general and buyerbehavior in particular, consumer researchers have thus fardevoted surprisingly little attention to central questionsconcerning the nature of value.” Moving forward to 2014, theMarketing Science Institute (MSI) specifies in its biannualResearch Priorities (among its top priorities), “One of the mostimportant tasks inmarketing is to create and communicate valueto customers to drive their satisfaction, loyalty, and profitability.Any insights in this area have significant implications for thelong-term financial health of an organization. It truly is at theheart of what marketing is all about.” Out of the 30 Dow Jones(United States) and 30 DAX (Germany) companies, 50% of thecompanies’ vision ormission statements explicitlymention thenotion of value creation for customers and/or stakeholders.For example, American Express’s mission statement includesthe sentence, “We provide outstanding products and un-surpassed service that, together, deliver premium value to our
customers.” Taken together, these high-level observationsserve to underline the importance that the customer valueaspect has had in the past and is continuing to have.
Clearly, business is about creating value. The purpose of asustainable business is, first, to create value for customers1and, second, to extract some of that customer value in theform of profit, thereby creating value for the firm. Thus, thekey underlying premise of this article is that customer valueis a dual concept. First, in order to be successful, firms (andthe marketing function) have to create perceived value forcustomers. In that sense, value is defined as overall assess-ment of the utility of an offering2 according to perceptions ofwhat is received and what is given (Zeithaml 1988). Second,customers provide value (customer lifetime value [CLV], inthe widest sense) for the organization. For the firms/decisionmakers who allocate resources to markets, customers, andproducts, the challenge is to dynamically align resourcesspent on customers and products in order to simultaneouslygenerate value both to and from customers.
We believe that aligning the customer-perceived valuewith customer-generated value vis-a-vis resource allocation is aresearch challenge that needs careful and comprehensive atten-tion. In this article, we focus on this alignment by examining thespecifics of this resource allocation challenge. More specifically,our goal is to answer the following concrete research questions:
1.What is value to the customer? What do we know?2. How should (customer perceptions of) product and servicevalue bemeasured? How can key drivers of customer value beidentified and calibrated?
3. How should the value from the customer be measured andmanaged?
4.What are the drivers of customer value? How can they beincorporated in real-time marketing decisions?
V. Kumar (VK) is Regents Professor, Richard and Susan Lenny Dis-tinguished Chair, Professor in Marketing, Executive Director of the Centerfor Excellence in Brand and Customer Management, and Director of thePhD Program in Marketing, J. Mack Robinson College of Business, GeorgiaState University; V. Kumar is also honored as the Chang Jiang Scholar,Huazhong University of Science and Technology, China; TIAS Fellow,Texas A&M University, College Station, TX; and ISB Senior Fellow, IndianSchool of Business. (e-mail: [email protected]). Werner Reinartz is Professor ofMarketing and Director of the Center of Research in Retailing, Universityof Cologne (e-mail: [email protected]). The authors thank theMarketing Science Institute and Kevin Keller for providing them the opportunityto contribute to this topic. They thank the guest editors of this special issue, aswell as Sarang Sunder, Maren Becker, Manual Berkmann, Mark Elsner,Vanessa Junc, Monika Kauferle, Annette Ptok, Julian Wichmann, andthree anonymous reviewers for their valuable suggestions on an earlierversion of this manuscript. They thank Bharath Rajan for his valuableassistance in this study and Renu for copy editing an earlier version ofthis manuscript.
1We use the term “customer” to denote a consumer as well as abusiness customer.
2We use the notion of “offering” to indicate physical products,services, brands, or a combination thereof.
© 2016, American Marketing Association Journal of Marketing: AMA/MSI Special IssueISSN: 0022-2429 (print) Vol. 80 (November 2016), 36–68
1547-7185 (electronic) DOI: 10.1509/jm.15.041436
What Is Value to Customers?What Do We Know?
Regardless of whether a physical good or a service is demandedby individuals for final consumption (direct demand) or as aninput factor used in providing other goods and services (deriveddemand), the fundamental economic principle of utility max-imization remains the same. Whereas the utility maximizationtheory is used to study productsmeant forfinal consumption, theprofit maximization principle is used to study the inputs used inthe production of other products. The traditional model assumesthat the customer has perfect information about product char-acteristics and associated costs, as well as stable preferences,and thus is able to perfectly construct his/her final objectivefunction—an assumption that has of course been relaxed inmultiple ways (e.g., Hauser and Shugan 1983).
Given the central role of the notion of value to the customerin the marketing literature, it is not surprising that the subjecthas been dealt with repeatedly (Anderson 1998; Anderson, Jainand Chintagunta 1993; Monroe 1971; Wilson 1995; Zeithaml1988). In various definitions of “value,” there is a reasonablevariety in the spirit of describing the trade-off between “give”elements and “get” elements (Anderson, Kumar and Narus2007; Sawyer and Dickson 1984). For the purpose of thisarticle, we define perceived value as customers’ net valuation ofthe perceived benefits accrued from an offering that is basedon the costs they are willing to give up for the needs they areseeking to satisfy.
Perceived customer value of an offering is the aggregationof benefits that the customer is seeking, expecting, or expe-riencing and the undesired consequences (Gutman 1982) thatcome with them. Benefits and undesired consequences are theresults of buying and consuming the offering, and these mayaccrue directly or indirectly and be immediate or delayed. Thecentral aspect of this conceptualization is that customers chooseactions that, ceteris paribus, maximize the desired consequencesand minimize concurrent undesired consequences. Benefits (andundesired consequences) are generated through offering attri-butes. Benefits differ from attributes in that people receive ben-efits, whereas offerings have attributes (Gutman 1982). Thus,attributes are features or properties of an offering. De facto,customers will never perceive all objective attributes (clearly) butwill form a composite perception that recognizes the respectiveattribute salience.
Although much of the literature conceptualizes price asone of many potential attributes, it is of course necessary tosubsume all cost components separately. Besides price, thesetypically include a vast array of different transaction costs, aswell as learning cost and risk. Interestingly, much of themarketing literature typically accounts for price as the onlycost component of an offering, thereby neglecting the manydifferent immediate and delayed types of costs that customersmight incur.
Perceived attributes then are aggregated by customersthrough a categorization process into more abstract benefits inorder to reduce information overload and to facilitate furtherprocessing. Themeans-end chainmodel (Gutman 1982;Howard1977) and the Gray benefit chain (Young and Feigin 1975) areprobably the most widely used conceptualizations behind this
effect chain from concrete offering attributes mapping ontoabstract benefits generated. The “house of quality” concept(Hauser and Clausing 1988) is another very successful attempt,originating from the operations and quality control domain, tomap objective attributes with customer perceived benefits. In-terestingly, this mapping process has been considered virtuallyonly for the offering’s attributes, but an expansive considerationof monetary and nonmonetary cost aspects (price, transactioncosts, risks, privacy) seems absent from the literature.
Once customers construct an aggregate assessment of anoffering’s perceived value, this value acquires meaning in twoways. First, as a necessary condition for customers to perceivean offering to be of positive value, perceived benefits have tooutweigh undesired consequences. Second, once this is the case(and assuming the customer has the required willingness andmeans to transact), the assessment of value of a single offeringamong a set of different offers (e.g., from a consideration set)requires a comparison standard. This comparison standardmay be concurrent competitive offerings, expectations, or pastexperience (similar to Golder, Mitra, and Moorman’s [2012]nomenclature, a “value stock” perspective). Once the customerdetermines an offer to be of highest value, behavioral (choice,loyalty, CLV) and attitudinal (satisfaction, loyalty) outcomesensue.
To summarize, we can observe that customer perceivedvalue is a central mediating construct, separate from quality,perceived benefits, and satisfaction. Moreover, despite therelevance of this construct in practice, the literature has for themost part not been looking at the customer perceived valueconstruct per se and has either omitted it from conceptualmodels (e.g., Golder, Mitra, and Moorman 2012) or hasused proxies, such as customer satisfaction, instead. Finally,the aspects of perceived cost and undesired consequencesin customer value models are arguably underspecified andunderstudied.
How Should (Customer Perceptionsof) Product and Service Value BeMeasured? How Can Key Driversof Customer Value Be Identified
and Calibrated?Measuring customer perceived value is the natural startingpoint before one can even consider giving recommendationswith respect to the value-oriented management.
Customer Value Measurement
The key tasks to be completed are (1) measure overall per-ceived value, (2) measure the associated underlying attributesand benefits, and (3) determine the relative weights that linkattributes/benefits to overall perceived value. Customers areseeking offerings that yield the highest expected value orutility. Because utility cannot be measured or observeddirectly, market researchers, psychologists, and economistshave devised ways to proxy these utilities. There is a longtradition of utility/preference measurement in marketing thatcommonly uses compositional and decompositional methodsto model consumer preferences (see Table 1).
Creating Enduring Customer Value / 37
TABLE1
Ove
rview
ofValueMea
suremen
tApproac
hes
withExe
mplary
Studies
Mea
suremen
tApproac
h
Exe
mplary
Applic
ations/
Sources
Focu
s/Starting
Point
Outcome
Role/Treatmen
tofAttributes
Role/Treatmen
tofBen
efits
Role/Treatmen
tofUndes
ired
Conse
quen
ces
DataInput
Other
Compositional
Multia
ttribute
metho
dGale(199
4)Presp
ecified
quality
andprice
attributes
Ove
rallva
lue
asse
ssmen
tPresp
ecified
Typ
ically
not
cons
idered
Presp
ecified
,typica
llyprice
only
Attributes
are
gene
rated
throug
hfocu
sgrou
psor
man
agerial
intuition
;attribute
ratin
gsan
dim
portan
ceweigh
tsare
collected
directly
bycu
stom
ersu
rvey
Exp
licit
cons
ideration
ofco
mpe
tition
poss
ible
(com
paris
onwith
compe
titive
offerin
g)
Con
sumer
perceive
dva
lue
multiitem
scale
Swee
neyan
dSou
tar(200
1)(PREVALsc
ale)
Presp
ecified
high
er-le
vel
dimen
sion
s(emotiona
lvalue
,so
cial
value,
func
tiona
lvalue
)
Ove
rallva
lue
scoreas
wella
sun
derly
ing
dimen
sion
s
Individu
alpres
pecified
itemsto
mea
sure
high
er-le
velvalue
dimen
sion
s
Sev
eral
pres
pecified
itemsallude
tobe
nefits
ofthe
offerin
g
Individu
alpres
pecified
itemsto
mea
sure
valueformon
ey/
price(broad
erco
stco
ncep
t)
Cus
tomer
survey
Doe
sno
tallow
forhe
teroge
neity
Relations
hip
valueinde
xUlaga
andEgg
ert
(200
6)Presp
ecified
high
er-le
vel
bene
fit
dimen
sion
s
Value
ashigh
er-
orde
rco
nstruc
t,includ
ingweigh
tsthat
determ
ine
theinflue
nceof
bene
fitan
dco
stdimen
sion
s
Individu
alpres
pecified
itemsto
mea
sure
bene
fit
dimen
sion
s
Individu
alpres
pecified
itemsto
mea
sure
vario
usco
stdimen
sion
s
Cus
tomer
survey
Exp
licit
cons
ideration
ofco
mpe
tition
poss
ible
(com
paris
onwith
seco
nd-bes
tsu
pplier)
Dec
ompositional
Con
joint
analys
isTou
bia,
Hau
ser,
andSim
ester
(200
4)(Polyh
edral
adap
tivech
oice
-ba
sedco
njoint
analys
is)
Stated
preferen
cesfor
hypo
thetical
prod
uct
confi
guratio
ns(attributes
and
price)
Part-worth
estim
ates
for
each
attribute
leve
l
Presp
ecified
bymarke
trese
arch
ers
(pric
eis
trea
ted
asan
attribute)
Typ
ically
not
cons
idered
,bu
tco
uldtech
nica
llybe
inco
rporated
Typ
ically
price
only
(tec
hnically,
othe
rco
stdimen
sion
sco
uld
beintegrated
)
Cus
tomer
survey
(relev
ant
attributes
have
tobe
iden
tified
qualita
tively
throug
hprior
stud
ies)
Allowsfor
custom
erhe
teroge
neity
segm
ents
Agg
rega
tepe
rceive
dva
lue
Sinha
and
DeS
arbo
(199
8)Ove
rallpe
rceive
dva
lue
Offe
ringloca
tion
onpe
rceive
dva
luedimen
sion
s
Impliedwith
out
beingsp
ecified
apriori
Impliedwith
out
beingsp
ecified
apriori
Impliedwith
out
beingsp
ecified
apriori;
cons
iderationof
vario
usco
stdimen
sion
s
Cus
tomer
survey
Allowsfor
custom
erhe
teroge
neity/
segm
ents
38 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE1
Continued
Mea
suremen
tApproac
h
Exe
mplary
Applic
ations/
Sources
Focu
s/Starting
Point
Outcome
Role/Treatmen
tofAttributes
Role/Treatmen
tofBen
efits
Role/Treatmen
tofUndes
ired
Conse
quen
ces
DataInput
Other
Auc
tion(W
TP)a
Wan
g,Ven
katesh
,an
dCha
tterje
e(200
7)(in
centive-
compa
tible
elicita
tionof
the
rang
ein
aco
nsum
er’s
rese
rvation
prices
;ICERANGE)
Con
sumers
indica
temax
imum
WTP
andstateatwhich
pricepo
intthey
areindiffe
rent
(ince
ntive-
aligne
d)
Con
sumer’s
rese
rvationprice
rang
e
Presp
ecified
bygive
nprod
uct(no
varia
tion)
Pric
eon
lyAuc
tion/
expe
rimen
t(w
eb-bas
edor
person
alinterview)
Ince
ntive-aligne
d
Pay
wha
tyo
uwan
t(W
TP)a
Kim
,Natter,an
dSpa
nn(200
9)Con
sumers
directlystatetheir
WTP(sellerha
sto
acce
ptan
yam
ount)
Insigh
tinto
individu
alleve
lof
valuepe
rcep
tion
(satisfaction)
Presp
ecified
bygive
nprod
uct(no
varia
tion)
Pric
eon
lyCon
sumer
survey
aftertran
saction
a The
semea
suresap
prox
imateva
lueby
mea
surin
gtheco
nsum
er’s
willingn
essto
pay(W
TP).Strictly
spea
king
,the
yon
lyyieldan
overallv
alue
asse
ssmen
tand
dono
tprovide
thepo
ssibility
ofdise
ntan
glingtheva
lueattach
edto
unde
rlyingch
arac
teris
tics(excep
tin
theca
seof
ave
ryelab
oratestud
yde
sign
).
Creating Enduring Customer Value / 39
Compositional methods begin with a set of explicitlychosen attributes/benefits and use them as the basis fordetermining overall value evaluations. In the compositionalapproach, expected utility is a function of the productattributes or benefits and corresponding costs multiplied bytheir respective importance weights. The key premise here isthat relevant attributes and their respective relevant levels areknown to the decision maker, and customers follow largely arational decision-making approach. This approach has foundreasonable entrance into the managerial domain (e.g., Gale1994; Sweeney and Soutar 2001; Ulaga and Eggert 2006).
In contrast, decompositional techniques attempt to inferunderlying utilities from observed choice, that is, revealedpreferences. They start with measures of preference forofferings or attribute bundles and use them to infer thevalue attached to underlying characteristics. The goal is toapproximate offering value from the customer’s willingnessto pay. A vast literature covers the use of this approach and,therefore, the interested reader should consult those sources(Rao 2014). The key approaches include survey approaches,auctions, conjoint analysis, and field experiments, with someof these methods having been widely used. Moreover, in thepast 30 years, there has been a strong recognition that nor-mative decision models are usually violated (Huber, Payne,and Puto 1982), and an entire stream of behavioral decisiontheory research has unfolded (Simonson 2015) that detailsthe wide range of systematic deviations that can occur whencustomers attempt to assess value.
From the firm’s perspective, the focus of the measurementapproaches pertains mostly to tangible product attributes. Theonly intangible aspects that has seemingly found entry intothese measurement models are the brand name dimension(e.g., Sinha and DeSarbo 1998) and aesthetic design aspects(Kumar 2015). Other intangible aspects seem absent. For ex-ample, in the context of digitization, what seems interesting ishow an individual’s perception of what constitutes value isinfluenced by others (i.e., network effects).
Treatment of Undesired Consequences (Costs)
The vast majority of attention in customer value measurementhas been focused on the “get” side of the offering, namely,documenting the range and depth of attributes and benefits thatare associated with the offering. The situation is starkly differentwith respect to the undesired consequences and cost aspects.Although perceived sacrifice is multidimensional, it is de factooperationalized as a unidimensional aspect in existing research(Teas and Agarwal 2000)—namely, on the price dimension. Forexample, inmuch of the conjoint analyses and logit models, priceis introduced as another (linear) attribute variable (Hauser andUrban 1986, p. 448).
However, besides the mere cost (transaction price) of theoffering to the customer, there is a large set of associatedtransaction costs, learning cost, and maintenance and life cyclecost, which are for the most part overlooked in existing models.In many situations, the monetary value of the sum of thesenonprice costs easily outsizes the transaction price, typical inmany business-to-business (B2B) settings (Anderson, Narus,and Narayandas 2009). Even more importantly, purchasing andconsumption models that are based on usage (i.e., customer
solutions)will becomemuchmore prevalent in the future (Ulagaand Eggert 2006; Ulaga and Reinartz 2011; Vargo and Lusch2004). Therefore, the conceptualization and operationalizationof a broader set of cost attributes is required in future customervalue models.
Furthermore, in the context of digitization, a new cost-related aspect has been emerging. For many online services(e.g., Google Maps, Facebook), customers are not expected topay in monetary terms. The core benefit is free of monetarycharge from the end user’s perspective. The monetizationcomes mainly from advertising revenues, with ads targetedat narrow segments or personal individual profiles. Here, wehave a new situation wherein the monetary component withinthe undesired consequences has entirely vanished. Customersnow have to understand the value of the personal informationthat they will give up in this exchange. Thus, customers pay interms of less privacy instead of monetary outlays. In fact, somecustomers value privacy of personal information privacy somuch that they would be willing to pay to preserve privacy—this then creates a market for privacy (Rust, Kannan, and Peng2002). Moreover, Koukova, Kannan, and Kirmani (2012) hintat new option value that digitization may provide. They showhow different formats of media (e.g., offline newspaper sub-scription vs. online formats) may provide complementary andincremental value to customers, depending on usage situation,as opposed to mere substitution. In other words, digitizationwill provide interesting and important new facets to the valuedebate. Table 2 presents a summary of related findingsregarding the measurement of value and value drivers.
How Should Value from CustomersBe Measured and Managed?
Until now we have reviewed (1) how firms can create value tothe customer, (2) the ways to measure customer perceptions ofvalue, and (3) the treatment of undesired consequences whenmanaging customer value. However, creating and communi-cating perceived value to customers is better servedwhen firmsalign perceived value with the resources they spend on cus-tomers, to ensure that the right amount of resources go towardmanaging perceived value. To identify the right amount ofresources, it is important to ascertain the value that customersprovide to the firm. Figure 1 illustrates the approach firms canadopt to derive value from customers.
As seen from Figure 1, the first step in deriving value is torealize the need for a forward-looking metric rather than abackward-looking one. Traditionally, firms have used metricssuch as recency–frequency–monetary value (RFM), past cus-tomer value (PCV), share of wallet (SOW), and tenure/durationto measure the value of customers. The guidance from thesemetrics has driven decisions pertaining to the allocation ofmarketing resources. However, many of the traditional metricsfocus on a backward-looking approach that only takes intoconsideration past activity of a customer, which leads to out-dated information being used for customer selection andresource allocation. In contrast, the CLV metric is a forward-looking metric that takes into account the variable nature ofcustomer behavior and enables firms to treat individual cus-tomers differentially and distinctly from each other depending
40 / Journal of Marketing: AMA/MSI Special Issue, November 2016
on their contributions to the company. Amid the backward-versus-forward-looking distinction, it is also important tonote the role of big data. Using big data that contains pastinformation (e.g., sales, revenue, marketing spending), it ispossible to accurately predict future behavior of customersand, thus, compute their value to the firm. The use of big data,however, does come with the challenges relating to the qualityof data, data visualization, computing capabilities, and ability
to obtain meaningful results. Nevertheless, the point to benoted here is the importance of being able to look into thefuture in determining the future value of customers, as op-posed to relying on insights from past value contributions.
Realizing the need for a forward-looking metric is themost critical step for a firm. Once this has been achieved, thesubsequent steps only strengthen the firm’s position in cultivatinga customer-centric organization. The following sections describe
TABLE 2Summary of Findings Pertaining to the Measurement of Value and Value Drivers
Scholarly Contributions Thus Far Tasks for Future Research
Perceived attributesand benefits
• Deriving underlying choice-relevant attributesusing decompositional approaches
• Understanding of longitudinal, situational, andcross-sectional heterogeneity.
• Understand whether and how attributessimultaneously pay into a given benefit
• Broaden conceptualization and measurementof intangible attributes and benefits
Perceived costs andundesired consequences
• Inclusion of price as key (cost-related) variable • Define conceptualization and operationalizationof a broader set of cost-related attributes
• Include new cost-related aspects due todigitization
Perceived value • Deriving customer perceived value usingcompositional approaches
• Measurement of customer perceived valuethrough willingness to pay
• Accommodating for incentive alignment andheterogeneity across time, customers, andproduct categories
• Recognition of systematic violations of valuemaximization principle
• Examine value of personal information (contextof digitization)
• Examine influence of value perception by socialnetwork
FIGURE 1Approach to Deriving Value from the Customer
Move from backward-looking metrics to a forward-looking metric
Introduce CLV
Understand the drivers of
CLV
Decide on a modeling approach (e.g., individual vs. aggregate)
Collect data for measuring CLV
Choose a type of model (e.g., deterministic vs. probabilistic)
Develop and implement strategies to maximize CLV
Extend the customer value concept to an engagement framework
Rev
ise
and
Ree
stim
ate
Estimate the model (e.g., Bayesian, econometric)
Realize enhanced firm value and shareholder value
Creating Enduring Customer Value / 41
the other steps identified in Figure 1 that put in place a cyclicalprocess involving data collection, deciding on the modelingapproach and the type ofmodel, estimating themodel, identifyingthe drivers of CLV, developing and implementing CLV-basedstrategies, extending CLV into a customer engagement frame-work, and reaping the benefits through higher firm value.
Firms have realized that just as customers derive valuefrom the products/services being offered, firms, too, derivevalue from the customer base. Kumar and Reinartz (2012)define this value from the customer as “the economic value ofthe customer relationship to the firm—expressed on the basisof contribution margin or net profit” (p. 4). When firmsidentify the value provided by customers, they will be able to(1) better manage their costs, (2) post increases in revenuesand profits, (3) realize better return on investment (ROI), (4)acquire and retain profitable customers, and (5) realign mar-keting resources to maximize customer value.
More so than customers’ past and current contributions tothe firm, a crucial factor is their contribution in future periods.It is this future component that is of immense interest toacademicians and practitioners. The concept of future valuecontribution has been conceptualized in the form of CLV. TheCLV metric has been conceptualized as the present value offuture profits generated from a customer over his or her lifetimewith the firm (Venkatesan and Kumar 2004). Estimating CLVhelps the firm to treat each customer differently according tohis or her contribution, rather than treating all customers in asimilar fashion. Furthermore, the sum total of lifetime value ofall customers of the firm represents the customer equity (CE)of the firm. In other words, CLV is a disaggregate measure ofcustomer profitability, and CE is an aggregate measure. Aftercomputing the CLV of its customers, a firm can developstrategies such as optimally allocating its limited resources,identifying the next products that customers are likely topurchase, and balancing acquisition and retention efforts,among others, to achieve maximum return.
Modeling Approaches to CLV
InmeasuringCLV,marketing literature presents a rich variety ofmeasurement approaches. However, the main objective acrossall approaches is clear—identify, maintain, and nurture profit-able customers. The approaches can be categorized into twobroad categories: the aggregate approach and the individualapproach. In the aggregate approach, the average lifetime valueof a customer is derived from the lifetime value of a cohort orsegment, or even a firm. This level of measurement helps firmsin evaluating the overall effectiveness of the marketing plan butnot in customizing strategies for customers. In the individualapproach, the CLV of one customer over his or her entirelifetime with the firm is computed. This level of measurementhelps firms personalize strategies according to customer needsand the future profitability potential of the customer. WebAppendix A lists the aggregate and individual approaches tomeasuring CLV. Table 3 provides an overview of the selectliterature from the extensive field of CLV.
Types of CLV Models
Over the past two decades, research onCLVhas covered awiderange of business conditions through the modeling approaches.
This coverage has expanded the scope and application ofCLV-based models for a multitude of industries and markets.This section discusses some of the popularmodeling approachesfrom the extant literature.
Estimating models independently. Most models, barringfew (Niraj, Gupta, and Narasimhan 2001; Venkatesan, Kumar,and Bohling 2007), focus on predicting future revenue andapply a constant gross margin and retention cost. Venkatesanand Kumar (2004) use a generalized gamma distribution tomodel interpurchase time and employ panel-data regressionmethodologies to model the contribution margin. They con-sider various supplier-specific factors (channel communication)and customer characteristics (involvement, switching costs, andprevious behavior) as the antecedents of purchase frequency andcontribution margin. Web Appendix B provides details on theindependent models. Given the factors identified, purchase fre-quency and contribution margin are modeled separately, andthe equation of CLV is specified as
CLVit = �Ti
t=1
GCit
ð1 + rÞt=fi- �
n
l=1
�mMCi,m,l
ð1 + rÞl ,(1)
where GCi,t is the gross contribution from customer i in pur-chase occasion t;MCi,m,l is the marketing cost for customer i incommunication channel m in time period l; fi, or frequency, is12/expinti (where expinti is the expected interpurchase time forcustomer i); r is the discount rate; n is the number of years toforecast; and Ti is number of purchases made by customer i.
Estimating models simultaneously. Endogeneity is a sta-tistical issue in the CLVmodel that relates directly to causation.When purchase frequency, marketing cost, and gross con-tribution are predicted independently, it becomes unclearwhether it is current MC that leads to future GC or current GCthat leads to future MC. In addition, the issue of heterogeneityrelates to the customer profiles and has been found to influencepurchase quantity and timing significantly (Allenby, Leone,and Jen 1999). When different customers respond differentlyto marketing messages, the contribution margin model mustreflect this variation by allowing the regression weights to bedifferent for each customer. Simultaneously modeling pur-chase frequency, MC, and GC solves both these issues andprovides more accurate results.
Venkatesan, Kumar, and Bohling (2007) use Bayesian de-cision theory to address the uncertainty in customer responseto marketing actions in a B2B setting. In this regard, theymodel the three parameters simultaneously. In the business-to-consumer (B2C) setting, studies have developed a joint modelfor purchase timing and quantity. For instance, Boatwright,Borle, and Kadane (2003) use the Conway–Maxwell–Poissondistribution to jointly model the purchase timing and purchasequantity for an online grocery retailer. Similarly, Chintagunta(1993) models the incidence of purchase at each time intervaland the purchase quantity for grocery purchases by customerswho make regular and frequent visits to a grocery store. Bysimultaneously modeling the parameters, it is possible toobtain an early-warning indication of abrupt changes ininterpurchase times and purchase quantities for a customer.When individual customers/customer segments that exhibit
42 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE3
SummaryofSelec
tCLVLiterature
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
eStudySetting
Insights
Rus
t,Kum
ar,a
ndVen
katesa
n(201
1)
Use
Mon
teCarlo
simulationalgo
rithm
topred
ict
custom
erpu
rcha
seprop
ensity,profi
t,an
dfirm
marke
tingac
tions
.
Empiric
alMon
teCarlo
simulation
High-tech
nology
services
The
prop
osed
mod
elprov
ides
large
improv
emen
tsin
pred
ictio
nov
erthesimpler
mod
elssh
ownin
literature.
Pfeife
r(201
1)Use
compa
ny-rep
ortedda
tato
estim
atethetotal
CLV
ofafirm
’scu
rren
tcu
stom
ers.
Empiric
alAutoreg
ress
ive
(first-orde
r)Netflix
The
stud
yoffers
guidan
ceab
outh
owto
estim
ate
retentionrate
andreve
nuepe
rrene
wal
whe
nthe
repo
rtingpe
riodsp
ansmultip
lerene
wal
perio
ds.
Fad
eran
dHardie
(201
0)Dev
elop
amod
elof
relatio
nshipdu
ratio
nthat
can
beus
edto
pred
ictretentionratesbe
yond
the
obse
rved
retentionrates.
Empiric
alShifte
dbe
tage
ometric
Any
indu
stry
Failingto
acco
untforco
hort-le
velreten
tion-rate
dyna
micslead
sto
bias
edes
timates
ofthe
residu
alva
lueof
acu
stom
er.
Kum
ar(201
0)Study
amultim
edia
multicha
nnel
commun
ication
fram
eworkba
sedon
CLV
.Con
ceptua
l—
Retailing
Max
imizingfirm
’sprofi
tabilityis
critica
lfor
unde
rstand
ingdriversof
CLV
andCRV.
Kum
aran
dSha
h(200
9)Link
custom
ereq
uity
(determined
bytheCLV
metric
)to
marke
tca
pitalization.
Empiric
alMultin
omiallog
itHigh-tech
nology
services
,retailing
Marke
tingstrategies
targeted
atmax
imizingCLV
canincrea
sefirm
valuean
d,thus
,ultim
ately,
stoc
kprice.
Kum
aret
al.
(200
8)Sho
wtheim
plem
entatio
nof
CLV
atIBM
andits
effectson
profi
tabilityan
dreso
urce
alloca
tion.
Empiric
alSee
mingly
unrelated
regres
sion
High-tech
nology
services
CLV
-bas
edrealloca
tionof
marke
tingreso
urce
syielde
da$2
0millionreve
nueincrea
sewith
out
anyad
ditio
nalres
ourceinve
stmen
t.
Ben
oitan
dVan
denPoe
l(200
9)
Ana
lyze
CLV
bymea
nsof
quan
tileregres
sion
and
compa
retheeffectsof
theco
varia
tes.
Empiric
alQua
ntile
regres
sion
Finan
cial
services
The
stud
yprov
ides
insigh
tsinto
theeffectsof
the
cova
riateson
theco
ndition
alCLV
distrib
utionthat
may
bemisse
dby
trad
ition
alleas
tsq
uares.
Don
kers,
Verho
ef,an
dDeJo
ng(200
7)
Predict
CLV
inmultiservice
indu
stry.
Empiric
alProbit,
parametric
duratio
n,no
nparam
etric
duratio
n
Insu
ranc
ese
rvices
Sim
plemod
elspe
rform
well.Foc
usingon
custom
erretentionis
noten
ough
;cros
s-bu
ying
also
need
sto
beac
coun
tedfor.
Fad
er,Hardie,
andJe
rath
(200
7)
Dem
onstrate
theus
eof
Pareto/NBD
mod
elsof
repe
atbu
ying
forco
mpu
tingCLV
.Empiric
alPareto/NBD
Cas
ean
alys
isDifferen
ttype
sof
econ
omic
efficien
cies
for
acqu
isition
strategies
andcu
stom
erse
gmen
tatio
nca
nbe
mad
ethroug
hthis
mod
el.
Fad
er,Hardie,
andLe
e(200
5)Propo
seane
wmod
elthat
links
RFM
metric
with
CLV
usingisov
alue
curves
.Empiric
alPareto/NBD
Onlinemus
icweb
site
Isov
alue
curves
arehigh
lyno
nlinea
rbe
caus
ecu
stom
erswith
low
rece
ncyva
lues
buthigh
freq
uenc
ypres
entlow
erCLV
than
custom
erswho
have
lower
freq
uenc
y.
Rus
t,Le
mon
,an
dZeitham
l(200
4)
Prese
ntafram
eworkthat
enab
lesco
mpe
ting
marke
tingstrategy
optio
nsto
betrad
edoffo
nthe
basisof
projec
tedfina
ncialreturns
.
Empiric
alLo
git
Airline
The
fram
eworken
ablesa“w
hat-if”
evalua
tionof
marke
tingROI,give
napa
rticular
shiftincu
stom
erpe
rcep
tions
.
Creating Enduring Customer Value / 43
TABLE3
Continued
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
eStudySetting
Insights
Ven
katesa
nan
dKum
ar(200
4)Dev
elop
ady
namic
fram
eworkto
maintain/
improv
ecu
stom
errelatio
nships
throug
hop
timal
alloca
tionof
marke
tingreso
urce
san
dmax
imize
CLV
simultane
ously.
Empiric
alGen
eralized
gamma
distrib
ution,
gene
ticalgo
rithm
s
High-tech
nology
services
Marke
tingco
ntac
tsov
erva
rious
chan
nels
influe
nceCLV
nonlinea
rly.CLV
-bas
edcu
stom
erse
lectionprov
ides
high
erprofi
tsin
thefuture
compa
redwith
othe
rmetric
s.
Reina
rtzan
dKum
ar(200
3)Dev
elop
afram
eworkto
compu
teCLV
and
iden
tifyfactorsthat
explaintheva
riatio
nin
profi
tablelifetim
edu
ratio
n.
Empiric
alNBD/Pareto
Catalog
retailer,
high
-tec
hnolog
yse
rvices
The
stud
yde
mon
stratesthesu
perio
rityof
CLV
fram
eworkby
compa
ringitwith
othe
rtrad
ition
alfram
eworks
.
Liba
i,Naray
anda
s,an
dHum
by(200
2)
Introd
ucease
gmen
t-ba
sedap
proa
chfor
custom
erprob
ability
analys
is.
Con
ceptua
l—
Retailer
The
prop
osed
approa
chretainstheac
tiona
ble
inform
ationas
sociated
with
individu
al-le
vel
analys
iswhile
also
maintaining
thesimplicity
ofthemoreag
greg
ate-leve
lmod
els.
Reina
rtzan
dKum
ar(200
0)Propo
seametho
dof
compu
tingCLV
totest
the
prop
osition
sbe
twee
ncu
stom
erloya
ltyan
dprofi
tability.
Empiric
alNBD/Pareto
High-tech
nology
services
Alth
ough
loya
lcon
sumersde
man
dprem
ium
servicean
dbe
lieve
they
dese
rvelower
prices
,they
areno
talway
sprofi
table,
nordo
they
contrib
utemoreto
profi
tsin
thelong
run.
Pfeife
ran
dCarraway
(200
0)
Introd
ucetheus
eof
Marko
vch
ainmod
elsfor
mod
elingcu
stom
errelatio
nships
andCLV
.Empiric
alMarko
vch
ain
Catalog
compa
nyMarko
vch
ainmod
elsca
nha
ndle
complicated
custom
errelatio
nshipsituations
.The
yca
nalso
hand
leso
meco
mplicated
situations
that
alge
braicso
lutio
nsca
nnot.
Berge
ran
dNas
r(199
8)Prese
ntmathe
maticalmod
elsford
eterminationof
CLV
.Con
ceptua
l—
Illus
trative
exam
ples
The
stud
yintrod
uces
thege
neralclass
esof
CLV
mod
els.
44 / Journal of Marketing: AMA/MSI Special Issue, November 2016
such a warning are identified, firms can implement appropriatecustomer relationship management (CRM) initiatives and cus-tomized marketing actions to each individual or to each cus-tomer segment.
Another setting in which simultaneousmodeling has beenoffered is in winning back lost customers. Losing customersfor good can be a significant setback for a firm’s customermanagement efforts. While studies have demonstrated themerits of preventing customers from defecting (Bolton 1998,Bolton and Lemon 1999, Lemon, White, and Winer 2002),it is also important to look into ways of winning back lostcustomers. Kumar, Bhagwat, and Zhang (2015) studywhether firms should chase a lost customer by investigatingthe impact of the reason for defection on reacquisition andsecond-lifetime duration and profitability. They achieve thisby jointly estimating customer reacquisition, second-lifetimeduration, and second-lifetime profitability per month. WebAppendix B provides details on these simultaneous models.
Brand-switching approach. Using information about thefocal brand and the competing brands, Rust, Lemon, andZeithaml (2004) model acquisition and retention of customers inthe context of brand switching. This approach requires collectinginformation on the brand purchased in the previous purchaseoccasion, the probability of purchasing different brands, andindividual-specific CE driver ratings from the customers.Through a Markov switching matrix, the individual customers’probability of switching from one brand to another based onindividual-level utilities is modeled. The probability thus cal-culated ismultiplied by the contribution per purchase to arrive atthe customer’s expected contribution to each brand for eachfuture purchase. The summation of expected contribution over afixed time period, after adjustments are made for the time valueof money, produces the CLV for the customer. Web AppendixB provides more details on the CLV model specification.
Monte Carlo simulation algorithm. Another approachadopted by researchers to investigate the value created bycustomers is the simulation method. While managerial heu-ristics and empirical models have uncovered important in-sights on customer value, studies have also investigated theuse of simulation models. The reason for exploring simulationmethods stems from relatively unsuccessful attempts at pre-dicting future customer profitability, with very simple modelsoften performing just as well as more sophisticated ones. Forinstance, Campbell and Frei (2004) find that it is easier topredict future profitability for some customers than for others,even for customers within the same profit tier. Similarly,Malthouse and Blattberg (2005) find that their best modelsmisclassify most customers who are predicted to have highprofitability. Furthermore, Donkers, Verhoef, and De Jong(2007) show that the simplest model they test performs thebest and provides better predictions than other models. Thesestudies make the case for alternative approaches, such as asimulation method.
Rust, Kumar, andVenkatesan (2011) present aMonteCarlosimulation method that can accurately predict future customerprofitability and performs better than existing methods. Theyaccomplish this in an “always-a-share” setting wherein there isno dormancy in a customer–firm relationship and customers
never completely terminate their relationship with a firm. Thus,in this approach, firmsmeasure future profitability of a customerby predicting his/her purchase pattern over a prediction period,but they do not predict when a customer will terminate therelationship with the firm. In this model, the profitability of acustomer is measured in terms of total profits and net presentvalue of profit. The predictions regarding customer behaviorinclude (1) the propensity for customer i to purchase in eachfuture time period t and (2) the profit provided by customer igiven purchase in future time period t. While the predictions ofcustomer behavior capture the revenue aspect, the marketingactions by the firm (e.g., number of sales calls, direct mail sentto a customer) capture the cost aspects; these aspects need to bepredicted for accurate CLV measurement. Toward this end, thestudy proposes a singlemodel framework that predicts customerpurchase incidence (Purit), customer gross profit (Pit) condi-tional on purchase, and firm marketing contacts (Xit) and alsomodels the potential correlations among these factors. WebAppendix B provides more details on the joint probabilitymodel.
Customer migration model. Dwyer (1997) presents a cus-tomermigrationmodel for CLVanalysis that is applicable for the“always-a-share” typology. Accordingly, customer behavior canbe predicted on the basis of historical probabilities of purchase,depending on recency and the current recency state in which thecustomer is located. Further generalizations include segmenta-tion variables such as the RFM metric and other demographicvariables. In several situations, RFM is used as a segmentationtool by classifying segments as “low RFM” and “high RFM.”Other segmentation approaches include SOW and customer lifecycle.WebAppendixB provides the approach to expressingCE.
Deterministic model. Deterministic models preciselymodel outcomes as determined by parameter values, rela-tionship states, and initial conditions. These models focus oninputs and outputs and leave little room for variations. Typ-ically, these models are used to study firm actions such ascustomer acquisition, customer retention, customer profit-ability, cross-buying behavior, and product return behavior(Reinartz and Kumar 2000, 2002, 2003). The basic form of thedeterministic approach used tomodel CLV is described by Jainand Singh (2002) (see Web Appendix B for the basic model).While this model does not account for acquisition costs, othermodels assume a constant gross contribution margin andmarketing costs (Berger and Nasr 1998). Several variationsof this basic model have been proposed (e.g., Dwyer 1997).Blattberg, Getz, and Thomas (2001) provide a comprehensiveway to calculate customer equity by accounting for the numberof prospects, acquisition spending, and consumer segments.
Probabilistic model. In a probabilistic model, the observedbehavior is viewed as the realization of an underlying stochasticprocess governed by latent (unobserved) behavioral charac-teristics, which, in turn, vary across individuals. The focus ofthis type ofmodel is on describing (and predicting) the observedbehavior instead of trying to explain differences in observedbehavior as a function of covariates (as is the case with anyregression model). In other words, this type of model assumesthat consumers’ behavior varies across the population accord-ing to some probability distribution (Gupta et al. 2006). Web
Creating Enduring Customer Value / 45
Appendix B provides details on the basic probabilistic model.Literature provides many models to estimate the CLV of acustomer using the internal data of a company. One such ap-proach defines CLV as a function of the time interval betweene-mail contacts sent to a customer (Dreze and Bonfrer 2009).Using the data from the entertainment industry, this modelestimates the relationship between the time interval and CLV.The objective is to find the optimal time interval for permission-based e-mails to a customer base. Apart from internal data,survey data can also be used to estimate the CE of a firm (seeRust, Lemon, and Zeithaml 2004; Rust, Zeithaml, and Lemon2001).
Structural model. The issue of multiple discreteness(wherein consumers may purchase more than one brand inone purchase occasion) in studies of CLV has received at-tention in the literature (e.g., Kim, Allenby, and Rossi 2002;Manchanda, Ansari, and Gupta 1999). However, the resultshave been suboptimal because conclusions regarding thequantity decision could not be made effectively. Sunder,Kumar, and Zhao (2016) adopt a direct utility approach tostructurally model multiple discreteness while accounting forvariety-seeking behavior in the demand model in assessingCLV of consumers in the consumer packaged goods (CPG)setting. Furthermore, the study is conducted on a longitudinaltransaction database and allows the budget to vary deter-ministically with time. In addition, this is the first study tounify choice, timing, and quantity decisions in a singleequation, thereby providing a direct approach for assessingCLV. Web Appendix B details this approach.
As a summary, Table 4 provides a comparison of theapproaches discussed in the previous sections, according totheir merits and shortcomings.
What Are the Drivers of CustomerValue? How Can Customer Value BeIncorporated in Making Real-Time
Marketing Decisions?Until now, we have discussed various models proposed forunderstanding and measuring value from customers. Thechoice of model is ultimately decided by the availability ofdata (volume and variety), time, and technical resources, aswell as the intended use of the CLV measure. Furthermore,information on competitive actions brings marketplacerealities into the decision-making process through game-theoretic approaches. The final choice of model notwith-standing, the results have to lead to managerial decisionmaking that is also conducive to real-time modifications. Thisinvolves understanding the drivers of customer value. Indiscussing the real-time applications, we adopt a relativeapproach wherein we focus more on time intervals (e.g.,frequency of buying, pattern of buying cycles, the need torevise CLV scores periodically) rather than immediate actions.In this regard, we survey and present research that hasexamined this aspect of real-time applications and providerelated insights for implementation. This, we believe, willpresent a new perspective on the real-time nature of decisionmaking.
The drivers of CLV determine the nature of the rela-tionship between the firm and the customer, and they helpestimate the level of profitability and the CLV of eachcustomer. They have been classified into two types: exchangecharacteristics and customer heterogeneity (Reinartz andKumar 2003). Exchange characteristics encompass the setof variables that define and describe relationship activitiesin the broadest sense. Customer heterogeneity refers to thedemographic and psychographic indicators that help a firmin segmenting customers and managing customer–firm rela-tionships. As expected, the nature of a business (whether B2Bor B2C) determines the exchange characteristics and customerheterogeneity factors. Figure 2 lists the classification of thesedrivers in B2B and B2C settings.
Similarly, Palmatier et al. (2006) provide a synthesis of theseveral empirical studies that have investigated the driversthat lead to stronger customer–firm relationships, as observedthrough WOM, customer loyalty, sales-related perfor-mance, customer likelihood to repurchase, and customer–firm cooperation. These studies clearly identify the driversthat have the greatest impact on customer–seller relation-ships. However, as more investigations on the nature ofcustomer–firm relationships are uncovered, we can expect thelist of drivers to change.
Strategies for Maximizing CLV
Once the computation of CLV is completed, firms proceed tomaximize this metric in order to reap its full benefits. TheCLV metric assists marketers to increase future profitabilityof not just current customers but also prospects. Furthermore,the CLV metric is not just about the dollar value of futurecustomer profitability. It extends beyond that and aids instrategy development on one or more of the following: cus-tomer acquisition, customer retention, balancing customeracquisition and retention, customer churn, and customer win-back.3 While the importance of these five tasks in ensuringprofitability is duly noted, this does not mean that “max-imizing” each individual metric is the correct recipe forsuccess. Firms can look into maximizing CLV from anoptimization perspective wherein the elasticities of each ofthese factors can be studied. Such an endeavor might be apromising avenue for future research. This section highlightsthe importance of understanding these five tasks in devel-oping the CRM playbook of an organization.
Customer acquisition. The expansive literature on cus-tomer acquisition has probed several important questions, suchas the following:
• How likely is it that prospects will respond to our acquisitionpromotion?
• How many new customers can we acquire in this campaign?• How many orders will each of our newly acquired customers
place?• How do the marketing variables, such as shipping fee, WOM
referral, and promotion depth, influence prospects’ responsebehavior?
3For more details, see the “Wheel-of-Fortune” strategies in Kumar(2008).
46 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE4
ComparisonofSelec
tCLVModels
CLVModel
Typ
eMerits
Shortco
mings
Typ
ical
DataReq
uirem
ents
Exe
mplarStudySetting
Agg
rega
teap
proa
ch•Aidsin
evalua
tingov
erall
effectiven
essof
marke
tingac
tions
•Not
able
tocu
stom
izemarke
ting
strategies
•Pub
liclyav
ailablefirm
-leve
ldatathat
includ
esinform
ationon
margin,
disc
ount
rate,an
dretentionrate
•Bothbrick-an
d-mortarfirm
san
don
linefirm
sac
ross
allind
ustriesthat
arepu
blicly
trad
ed
Inde
pend
ent
estim
ation
•Eas
yto
use
•Aidsincu
stom
er-le
veland
firm
-leve
lstrategy
deve
lopm
ent
•Req
uiresen
d-us
ertran
sactionda
ta•Creates
endo
gene
ityan
dhe
teroge
neity
issu
es•Doe
sno
tinclud
eco
mpe
tition
•Cus
tomer-le
veldataon
tran
sactions
andcu
stom
er-firm
interactions
from
firm
’sinternal
reco
rds
•Hightech
nology
Sim
ultane
ous
estim
ation
•Acc
ountsforen
doge
neity
and
heteroge
neity
•Moreac
curate
resu
ltsthan
inde
pend
entes
timation
•Mod
elde
velopm
entan
des
timation
ismoreco
mplex
•Cus
tomer-le
veld
atafrom
firm
’sinternal
reco
rdsthat
includ
esinform
ationon
purcha
sequ
antity,
prod
uctup
grad
es,cros
s-bu
ying
,marke
tingco
mmun
ication,
prod
uct
returns,
andfreq
uenc
yof
contac
ts
•Telec
ommun
ications
•Hightech
nology
•Groce
rystores
Brand
-sw
itching
approa
ch
•Can
beus
edwhe
nthefirm
has
cros
s-se
ctiona
land
long
itudina
lda
taba
se•Acc
ountsforalltyp
esof
marke
ting
expe
nditu
res
•Can
acco
mmod
ateco
mpe
tition
•Sam
plese
lectionca
nplay
anim
portan
trolein
theac
curacy
ofthe
metric
•Ofte
nrelieshe
avily
onsu
rvey
-bas
edda
ta,thus
lead
ingto
anincrea
sein
samplingco
stan
dsu
rvey
bias
es
•Surve
yda
tafrom
asa
mpleof
custom
ersthat
includ
esda
taon
purcha
sefreq
uenc
y,co
ntrib
ution
margin,
andbran
dspu
rcha
sed
rece
ntly
•Airlines
•Ren
talc
ars
•Electronicstores
•Groce
rystores
Mon
teCarlo
simulation
algo
rithm
•Betterp
redictivepo
wer
over
simpler
compe
tingmod
els
•Betterun
derstand
ingof
custom
erprofi
tabilityan
dfirm
value
•Can
notbe
used
inalost-for-goo
dse
tting
•Hea
vyrelianc
eon
long
purcha
sehistories
•Individu
al-le
veld
atafrom
firm
’sinternal
reco
rdsthat
includ
esinform
ationon
purcha
seprop
ensity,
marke
tingco
ntac
ts,a
ndgros
sprofi
t
•Hightech
nology
(can
also
beus
edforac
tual
andsimulated
data)
Cus
tomer
migratio
nmod
el
•Con
side
rsprob
abilistic
nature
ofcu
stom
erpu
rcha
ses
•Can
beus
edon
lyinlim
itedbu
sine
ssse
tting
s•Datafrom
firm
’sinternalreco
rdsthat
includ
espu
rcha
seprop
ensitie
san
drece
ncyof
purcha
se
•Catalog
mailing
•Dire
ctmarke
ting
Creating Enduring Customer Value / 47
TABLE4
Continued
CLVModel
Typ
eMerits
Shortco
mings
Typ
ical
DataReq
uirem
ents
Exe
mplarStudySetting
Deterministic
mod
el•Highe
rpred
ictiveac
curacy
•Aidsin
firm
-leve
lstrateg
yde
velopm
ent
•Req
uireshu
geam
ountsof
individu
alcu
stom
erda
ta•Doe
sno
tco
nsider
therelatio
nship
betwee
nmod
elpa
rameters
•Des
criptive,
butn
otpres
criptive,
and
thereforeless
helpfulinman
agerial
decision
mak
ing
•Doe
sno
tac
coun
tforco
mpe
tition
•Seg
men
t-leve
ldataon
marke
ting
costs,
acqu
isition
rate,retention
rate,an
dco
ntrib
utionmarginfrom
firm
’sinternal
reco
rds
•Finan
cial
services
•Hightech
nology
•Insu
ranc
e•App
arel
•Catalog
retailing
Proba
bilistic
mod
el•Can
beus
edwhe
nthefirm
does
not
have
along
itudina
ldatab
ase
•Iden
tifica
tionof
subd
riversaids
inbe
tterreso
urce
alloca
tion
•Ass
umes
purcha
sevo
lumean
dinterpurch
asetim
eto
beex
ogen
ous
•Calls
forfreq
uent
upda
tingof
the
mod
el•Hea
vyrelianc
eon
data
andless
errelianc
eon
man
agerialins
ight
•Individu
al-le
veld
atafrom
firm
’sinternal
reco
rds,
such
asrece
ncyof
purcha
ses,
freq
uenc
yof
purcha
ses,
andva
lueof
tran
sactions
•Mag
azines
/catalog
s•Entertainmen
t•Internet
•Hightech
nology
•Airlines
Struc
tural
mod
el•Mod
elba
sedon
theo
retical
unde
rpinning
sof
cons
umer
beha
vior
•Can
acco
untforva
rious
salient
aspe
ctsof
cons
umer
beha
vior
(e.g.,
multip
ledisc
retene
ss,bu
dgeting)
that
cann
otbe
addres
sedby
othe
rmetho
ds•Aidsin
accu
rate
out-of-sam
ple
pred
ictio
nan
dman
agerialp
olicy
simulations
•Mod
elde
velopm
entan
des
timation
isve
ryco
mplex
•Relieshe
avily
onac
ross
andwith
inva
riatio
nin
custom
erpu
rcha
ses
•Sca
nner
pane
ldataon
bran
dspu
rcha
sedin
aninstan
ce,im
puted
priceinform
ation,
hous
ehold-leve
lmarke
tsh
areof
compe
tingbran
ds,
andbu
dgetaryco
nstraints
•Con
sumer
pack
aged
good
s
48 / Journal of Marketing: AMA/MSI Special Issue, November 2016
• How long will the newly acquired customers stay with ourcompanies?
• How much profit or value will this acquisition campaignbring to our companies?
To understand customer acquisition, it is important to payattention to the business setting: noncontractual or contractual.In a noncontractual setting, customers can and do split theirspending across several firms. As a result, observing when acustomer ceases to be a customer becomes difficult for thefirm.However, situations wherein customers develop strong rela-tionships with firms do exist (e.g., strong preference towarda particular brand of coffee). In a contractual setting, firmsenjoy a relatively continuous future cash flow for a period oftime, and they know when customers terminate the relation-ship. Even in such settings, it is possible for customers to defectwithout notifying the firm (e.g., failure to renew a magazinesubscription). Studies have developed different models tostudy these two business settings regarding factors such as theexpected duration or time of the relationship with customers,the likelihood of a customer continuing the relationship, andindicators of defection at the end of a service period, amongothers. Table 5 lists a representative set of studies that haveconsidered these issues and accounted for many of theproblems that might occur in the model-building process.
A fundamental research interest in understanding cus-tomer acquisition is in identifying the probability of a cus-tomer being acquired. Several studies have explored thisquestion and have uncovered valuable insights (Lix et al.1995; Reinartz, Thomas, and Kumar 2005). For instance,Von Wangenheim and Bayon (2007) find that customersatisfaction influenced the number of WOM referrals, whichhad an impact on customer acquisition, and that the recep-tion of a WOM referral had an increased marginal effect onthe likelihood of a prospect to purchase. This is achievedby studying (1) whether prospects’ purchase likelihood is afunction ofWOM referrals, and (2) whether the characteristics
of the WOM referral source influence the probability that re-ceived WOM induces a purchase behavior. In a B2B setting,studies have investigated the role of referrals in customeracquisition. For instance, Hada, Grewal, and Lilien (2010)recognize the types of referrals (customer–to–potential cus-tomer referrals, horizontal referrals, and supplier-initiatedreferrals) and have developed the concept of referral equity tocapture the net effect of all referrals for a supplier firm inthe market. These authors further propose a framework formanaging supplier-initiated referrals that incorporates thesupplier and the supplier’s management of the communi-cation between the referrer and the potential customer. Godes(2012) explores the conditions and subsequent impact offirms launching referral programs. The study demonstratesthat launching such programs increases the willingness to payof the early adopters in the high-technology B2B market. Inaddition, Kumar, George, and Leone (2013) develop and testan approach to compute business reference value (BRV),identify the behavioral drivers of BRV, and offer strategies totarget and manage the most promising customers on the basisof their CLV and BRV scores.
In addition to customer acquisition, the number of newlyacquired customers and their initial order quantity are alsoimportant. While Lewis (2006b) identifies that shipping feesinfluence the customer acquisition process and the initialorder size, Villanueva, Yoo, and Hanssens (2008) show thatmarket channels also influence the customer acquisitionprocess and the value that newly acquired customers willbring to the company. The latter study develops two func-tions: (1) the value-generating function, which links newlyacquired customers’ contributions to the firm’s equitygrowth, and (2) the acquisition response function, whichexpresses the interactions between marketing spending andthe number of acquisition. Using a three-variable vectorautoregression modeling technique for data from an onlineretailer, the study finds that marketing-induced customers add
FIGURE 2Drivers of CLV in B2B and B2C Settings
• Past customer spending level• Cross-buying behavior• Purchase frequency• Recency of purchase• Past purchase activity• Marketing contacts by the firm
• Past customer spending level• Cross-buying behavior• Focused buying behavior• Average interpurchase time• Participation in loyalty programs• Customer returns• Customer-initiated contacts• Frequency of marketing contacts• Type of marketing contacts• Multichannel shopping• Consumer deal usage intensity• Coupon usage intensity
Includes variables such as age, gender, spatial income, and
physical location of the customers
Includes variables such as industry, annual revenue, and
location of the business
B2B Firm B2C Firm
Exchange
Characteristics
Customer
Heterogeneity
Creating Enduring Customer Value / 49
TABLE5
SummaryofSelec
tCustomer
Acq
uisitionStudies
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
eStudySetting
Insights
Had
a,Grewal,an
dLilien(201
4)Propo
seafram
eworkthat
inco
rporates
the
supp
lieran
dtheirman
agem
entof
commun
icationbe
twee
nthereferrer
andthe
potentialc
ustomer.
Empiric
alTria
dic
commun
ication
Bus
ines
san
alytic
solutio
nsInflue
nceof
thesu
pplier-se
lected
referral
onpo
tentialc
ustomersde
pend
son
supp
lier
unce
rtainty,
andpe
rceive
dbias
sign
ifica
ntly
redu
cespo
tentialc
ustomers’
supp
lier
evalua
tion.
Kum
ar,Geo
rge,
and
Leon
e(201
3)Mea
sure,un
derstand
,an
dman
ageBRV.
Empiric
alBinarych
oice
mod
elTelec
ommun
ications
,fina
ncials
ervice
sThe
stud
yde
fine
stheco
ncep
tof
BRV,
determ
ines
thedriversof
BRV,a
ndillum
inates
therole
ofmea
suresin
drivingBRV.
God
es(201
2)Und
erstan
dwhe
nan
dwhy
abu
sine
sssh
ould
anno
unce
areferral
prog
ram.
Empiric
alGam
etheo
ryHigh-tech
nology
B2B
Referen
ceprog
ram
canse
rveas
apa
rtial
subs
tituteforan
exclus
ive-us
eco
ntract.
Villan
ueva
,Yoo
,an
dHan
ssen
s(200
8)Propo
sean
dtest
anem
piric
almod
elthat
captures
long
-term
effectsof
custom
erac
quisition
onCEgrow
th.
Empiric
alVec
tor
autoregres
sion
mod
elus
ing
threeva
riables
Web
hostingco
mpa
nyMarke
ting-indu
cedcu
stom
ersad
dmoresh
ort-
term
value,
butWOM
custom
ersad
dne
arly
twiceas
muc
hlong
-term
valueto
thefirm
.
Von
Wan
genh
eim
andBay
on(200
7)Exa
minethelinks
betwee
ncu
stom
ersa
tisfaction,
WOM,an
dcu
stom
erac
quisition
.Empiric
alLo
gistic
regres
sion
Ene
rgymarke
tThe
satisfaction–
WOM
linkis
nonlinea
ran
dis
mod
erated
byse
veralc
ustomer
invo
lvem
ent
dimen
sion
s.
Lewis
(200
6a)
Exa
minetherelatio
nshipbe
twee
nac
quisition
disc
ount
depthan
dtheva
lueof
custom
eras
sets.
Empiric
alSurviva
lan
alys
isNew
spap
eran
don
line
groc
erAcq
uisitio
ndisc
ount
depthisne
gativelyrelated
torepe
at-buy
ingratesan
dcu
stom
eras
set
value.
Lewis
(200
6b)
Study
how
shipping
feesc
hedu
lesaffect
custom
erac
quisition
,cu
stom
erretention,
and
orde
rsize
.
Empiric
alSys
tem
oflinea
rregres
sion
Onlinegroc
erHighe
rsh
ipping
fees
areas
sociated
with
redu
cedorde
ringrates,
andpe
nalties
forlarge
rorde
rslead
toredu
cedorde
rsize
.
Tho
mas
,Reina
rtz,
andKum
ar(200
4)Determinewhe
ther
max
imizingcu
stom
erac
quisition
andcu
stom
erretentionse
parately
max
imizes
profi
ts.
Empiric
alStand
ardrig
ht-
cens
ored
Tob
itB2B
high
-tec
hnolog
yman
ufac
turer,
pharmac
euticals,
and
catalogretailer
Dec
reas
ingmarke
tingsp
ending
fortheB2B
firm
andca
talogretailer,an
dincrea
sing
spen
ding
tocu
stom
ersof
theph
armac
eutical
firm
,wou
ldincrea
setotalcus
tomer
pro fi
tability.
Han
sotia
andWan
g(199
7)Determinewhich
pros
pectssh
ould
beco
ntac
ted,
andpres
entmod
elsto
estim
ate
resp
onse
andcu
stom
erva
lueat
theindividu
alleve
l.
Empiric
alProbitan
dTob
itmod
els
Motor
club
mem
bership
The
authorsreco
mmen
dthat
firm
sad
opta
long
-term
view
bylook
ingbe
yond
resp
onse
totheac
tual
beha
vior
ofcu
stom
erson
cethey
are
acqu
ired.
Lixet
al.(199
5)Link
purcha
se-in
tentioninform
ationfrom
survey
data.
Empiric
alLine
arregres
sion
and
log-linea
r
Retailer
The
stud
yprov
ides
ametho
dology
toac
hiev
eefficien
cyin
othe
rdirect
maile
fforts.
50 / Journal of Marketing: AMA/MSI Special Issue, November 2016
more short-term value, but WOM customers add nearly twiceas much long-term value to the firm. The study also dem-onstrates the long-term impact of different resource alloca-tions for acquisition marketing using dynamic simulations.
Of course, making marketing decisions using the responseprobability, initial order quantity, and duration is not enough.Companies should select prospects to acquire according totheir lifetime contribution, which can be termed lifetime value,CLV, or CE. For instance, Reinartz, Thomas, and Kumar (2005)estimate customer profitability with a standard right-censoredTobit model using a set of exogenous variables and including thepredicted duration and response probability. The authors findthat decreasing marketing spending for a B2B firm and catalogretailer and increasing spending to customers for a pharma-ceutical firm leads to increases in total customer profitability.
Customer retention. Aside from customer acquisition,firms devote large amounts of resources toward customerretention practices. To gain a better understanding of cus-tomer retention and to advise firms in a better manner, studieshave typically focused on answering the following questions:
• Will the recently acquired customer repurchase or not?• What will be the lifetime duration of the customer (i.e., when
will the customer churn)?• Given that the customer is going to repurchase, (1) How
many items is that customer going to purchase? (2) Howmuch is that customer likely to spend? (3) Will the customerpurchase in multiple product categories?
• What will firms have to do in order to retain a customer worthretaining?
• What is the long-term impact of this customer’s purchasebehavior on firm value?
Various studies have been conducted and many models havebeen developed to answer these questions. Table 6 shows arepresentative set of studies that have considered these issuesand accounted for many of the problems that might occur inthe model building process.
When to engage in the activity of retaining a customer canbe a highly misunderstood and undervalued component incustomer retention. Monitoring a customer’s purchasing andattitudinal behavior is vital in understanding when a firmshould aggressively and actively pursue his or her retention.This is important for three reasons. First, firms can often losesight of a customer’s loyalty and lose their profitable cus-tomers, thereby creating undue financial stress. Second,monitoring customer behavior allows the firm to identify theattitudinal changes in a customer. This is important becauseunderstanding the attitudinal changes of a customer withregard to the firm’s brand advises the firm on how and whento be aggressive in its retention strategies for that particularcustomer. Finally, a defecting customer can cause harm tothe firm’s brand through negative WOM if the customer’sdefection is due to unmet needs. If such unmet needs prevailover a large set of customers, and if this possibility is notaddressed, the negative WOM may quickly snowball into aserious issue, thereby damaging both the brand and firm value.
Determining how much to spend on a customer is an im-portant assessment involved in identifying who and when toretain. Innovations in statistical modeling now allow firms to
measure a customer’s future value and profitability to the firm,whichmakes it easier tomake decisions on howmuch to spendas compared with the future value. This logical approach tocustomer retention calls for data pertaining to several aspects ofcustomer transactions over a period of time. With respect tomodeling techniques required to understand customer reten-tion, simulated maximum likelihood estimation is the mostcommonly used procedure. However, more advanced esti-mation techniques, such as Markov chain Monte Carlo,Bayesian estimation, and generalized method of moments,have been used. Understanding customer retention also callsfor accounting for customers’ responsiveness to retention effortsbecause this determines the method of communicating theintention to retain and the costs related to it.
Researchers and managers alike are placing higher im-portance on the study of customer retention and its impact oncompany profits. In both B2B and B2C firms, model-basedapproaches are becoming increasingly available, and thusnecessary, in both contractual and noncontractual relation-ships. In developing strategies for firms, research studies havedrawn insights from (1) explaining customer retention ordefection (Bolton and Lemon 1999; Borle, Singh, and Jain2008); (2) predicting the continued use of the service rela-tionship through the customer’s expected future use andoverall satisfactionwith the service (Lemon,White, andWiner2002; Lewis 2006a); (3) predicting renewal of contracts usingdynamic modeling (Bolton, Kannan, and Bramlett 2000), (4)modeling the probability of a member lapsing at a specific timeusing survival analysis (Bhattacharya 1998); (5) modelingthe duration of relationship using the negative binomial dis-tribution (NBD)/Pareto model and the proportional hazardmodel (Fader, Hardie, and Lee 2005; Reinartz and Kumar2000; Schmittlein, Morrison, and Colombo 1987); (6) use ofloyalty and rewards programs for retention (Leenheer et al.2007;Meyer-Waarden 2007); and (7) assessing the impact of areward program and other elements of the marketing mix(Anderson and Simester 2004; Schweidel, Fader, and Bradlow2008).
An important strategy for retaining customers is tonurture their cross-buying behavior. It has been shown thatwhen customers purchase more products or services fromthe same firm, they extend the duration of their relationshipwith the firm (Reinartz and Kumar 2003) and increasepurchase frequency (Reinartz, Thomas, and Bascoul 2008).Cross-buying results not only in an increase in revenue con-tribution for the firm but also in more engagement with thefirm, higher profit contribution, and higher switching costs(Kumar, George, and Pancras 2008). Although cross-buyinghas the potential to increase profitability, some firms (e.g.,financial services firms) have encountered unprofitable cus-tomers despite taking efforts to promote cross-buying (Brown2003). Greater customer cross-buying does not alwaysresult in higher customer profitability. Shah et al. (2012)study this phenomenon of unprofitable cross-buying andfind that (1) customer cross-buying is not necessarily prof-itable for all customers of the firm and can in fact adverselyaffect a firm’s bottom line, (2) persistent adverse customerbehavior drives unprofitable customer cross-buying overtime, and (3) a company’s marketing policies and practices
Creating Enduring Customer Value / 51
TABLE6
SummaryofSelec
tCustomer
Reten
tionStudies
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
eStudySetting
Insights
Borle,S
ingh
,and
Jain
(200
8)Estim
atepu
rcha
se-le
velC
LVby
jointly
mod
eling
purcha
setim
ing,
purcha
seam
ount,an
dris
kof
custom
erde
fection
Empiric
alDiscrete-ha
zard,
log-no
rmal
Mem
bership-ba
sed
direct
marke
ting
compa
ny
Long
erinterpurch
asetim
esareas
sociated
with
larger
purcha
seam
ountsan
dgrea
terris
kof
leav
ingthefirm
.
Sch
weide
l,Fad
er,an
dBradlow
(200
8)
Mod
elcu
stom
erretentionthat
acco
unts
for
duratio
nde
pend
ence
,prom
otiona
leffe
cts,
subs
criber
heteroge
neity,cros
s-co
horteffects,
andse
ason
ality.
Empiric
alPropo
rtiona
lha
zard
Telec
ommun
ications
service
Inclus
ionof
prom
otiona
leffe
ctsim
prov
esthe
foreca
stac
curacy
ofretentionbe
havior,w
hereas
includ
ingcros
s-co
horteffectsdo
esno
tsign
ifica
ntly
improv
eit.
Fad
eran
dHardie(200
7)Provide
analternativeap
proa
chto
survivor
analys
isfores
timatingcu
stom
ertenu
re.
Empiric
alShifte
dbe
tage
ometric
Con
trac
tual
subs
criptio
n-ba
sed
busine
ss
Propo
sedmod
eloffers
useful
diag
nostic
insigh
tsan
dis
very
easy
toim
plem
entus
ingMicroso
ftExc
el.
Leen
heer
etal.
(200
7)Dev
elop
amod
elon
therelatio
nbe
twee
nloya
ltyprog
ramsan
dpu
rcha
sebe
havior.
Empiric
alTyp
eIITob
itGroce
ryretailing
Mod
elac
coun
tsforen
doge
neity
ofloya
ltyprog
rams.
Predictiveva
lidity
oftheprop
osed
mod
elismuc
hbe
ttertha
nthat
ofthena
ıvemod
el.
Mey
er-W
aarden
(200
7)Exa
minetheim
pact
ofloya
ltyprog
ramson
custom
erlifetim
edu
ratio
n.Empiric
alPropo
rtiona
lha
zard
Sup
ermarke
tThe
high
erthesh
areof
cons
umer
expe
nditu
resin
astore,
thelong
erthelifetim
edu
ratio
nwillbe
.
Bolton,
Lemon
,an
dBramlett
(200
6)
Mod
elafirm
’srepa
tron
agebe
havior
forse
rvice
contracts.
Empiric
alRan
dom
intercep
tsEnterprise-leve
lsy
stem
sMod
elsof
custom
erretentionsh
ould
inco
rporate
theex
tent,va
riability,
andtim
ingof
asu
pplier’s
servicede
liveryov
ertim
e.
Verho
efan
dDon
kers
(200
5)
Inve
stigatetheim
pact
ofch
anne
lsfirm
sfreq
uently
useon
custom
erloya
ltyan
dcros
s-bu
ying
.
Empiric
alProbit
Finan
cial
services
prov
ider
Dire
ct-m
aila
cquisitio
nch
anne
lperform
spo
orly
onretentionan
dcros
s-se
lling
;radioan
dTV
perform
poorly
forretentionon
ly;an
dweb
site
performswellfor
retention.
And
erso
nan
dSim
ester
(200
4)
Inve
stigateho
wthede
pthof
acu
rren
tprice
prom
otionaffectsfuture
purcha
sing
offirst-tim
ean
des
tablishe
dcu
stom
ers.
Empiric
alPoiss
on,linea
rregres
sion
Mail-o
rder
durable
good
sDee
perpricedisc
ountsin
thecu
rren
tpe
riod
increa
sedfuture
purcha
sesby
first-tim
ecu
stom
ersbu
tredu
cedfuture
purcha
sesby
establishe
dcu
stom
ers.
Verho
ef(200
3)Inve
stigatethediffe
rentiale
ffectsof
custom
errelatio
nshippe
rcep
tions
andrelatio
nship
marke
tinginstrumen
tson
custom
erretentionan
dsh
areov
ertim
e.
Empiric
alTob
it,prob
itFinan
cial
services
prov
ider
Affe
ctiveco
mmitm
entan
dloya
ltyprog
ramsthat
prov
ideec
onom
icince
ntives
positivelyaffectbo
thcu
stom
erretentionan
dsh
arede
velopm
ent.
52 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE6
Continued
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
eStudySetting
Insights
Lemon
,White,
andWiner
(200
2)
Exa
minetheinflue
nceof
custom
erfuture-foc
used
cons
iderations
,ov
eran
dab
ovetheeffectsof
satisfaction,
onthecu
stom
er’s
decision
todisc
ontin
uease
rvicerelatio
nship.
Empiric
alLo
git
Interactivetelevision
entertainm
ent
service
Con
sumersaresign
ifica
ntly
forw
ard-look
ing
whe
nthey
mak
ethede
cision
toco
ntinue
(or
discon
tinue
)ase
rvicerelatio
nship.
Bolton,
Kan
nan,
andBramlett
(200
0)
Inve
stigatetheco
ndition
sin
which
aloya
ltyprog
ram
willha
veapo
sitiveeffect
oncu
stom
erev
alua
tions
,beh
avior,an
drepu
rcha
seintentions
.
Empiric
alLo
git
Finan
cial
services
Loya
ltyprog
ram
mem
bers
overlook
nega
tive
evalua
tions
oftheco
mpa
nyvis-a-visco
mpe
tition.
Reina
rtzan
dKum
ar(200
0)Tes
twhe
ther
long
-life
custom
ersarealway
sprofi
table.
Empiric
alNBD/Pareto
Catalog
retailer
Long
-life
custom
erskn
owtheirva
lueto
the
compa
nyan
dde
man
dprem
ium
service;
they
believe
they
dese
rvelower
prices
,an
dthey
spread
positivewordof
mou
thon
lyifthey
feelan
dac
tloya
l.
Boltonan
dLe
mon
(199
9)Qua
ntify
therelatio
nshipbe
twee
ncu
stom
ersa
tisfactionan
dsu
bseq
uent
serviceus
age.
Empiric
alTyp
eITob
itInteractivetelevision
entertainm
ent
servicean
dce
llular
service
Cus
tomers’
usag
eleve
lsca
nbe
man
aged
throug
hpricingstrategies
,co
mmun
ications
,an
ddy
namic
custom
ersa
tisfactionman
agem
ent.
Bha
ttach
arya
(199
8)Und
erstan
dho
wmem
bers’ch
arac
teris
ticsrelate
tolaps
ingbe
havior
inpa
idmem
bershipco
ntex
ts.
Empiric
alHaz
ard
Artmus
eum
Haz
ardof
laps
ingislowered
with
long
erdu
ratio
n,en
rollm
entin
spec
ialinteres
tgrou
ps,gift
freq
uenc
y,an
dhigh
erinterren
ewal
times
.
Bolton(199
8)Dev
elop
amod
elof
thedu
ratio
nof
prov
ider–cu
stom
errelatio
nshipan
dtherole
ofcu
stom
ersa
tisfaction.
Empiric
alPropo
rtiona
lha
zard
Cellularse
rvice
Relations
hipbe
twee
ndu
ratio
ntim
esan
dsa
tisfactionis
strong
erforcu
stom
erswho
have
moreex
perie
ncewith
these
rviceorga
niza
tion.
Creating Enduring Customer Value / 53
could facilitate (or deter) the persistence of adverse customerbehavioral traits associated with unprofitable cross-buying.Using data from five firms, the study finds that the firm withthe most liberal product-return policy had the maximumlevel of unprofitable cross-buying due to excessive productreturns.
Even though several studies have highlighted theimportance of customer retention as a single link in the chainof CRM, many firms have not yet understood the largercomprehensive view of the CRM process. For instance, somemanagers still view customer acquisition and retention asseparate processes. Although most studies assess acquisitionand retention separately, the literature provides an abundanceof direction through which firms can link the two together inorder to improve their CRM process.
Balancing acquisition and retention. Apart from iden-tifying the balance in marketing effort between acquisitionand retention in order to maximize profitability, studies haveaddressed questions such as the following:
• What are the drivers of customer acquisition?• After being acquired, how long can the customer be expected
to stay with the firm?• How much investment is required in order to keep the
firm–customer relationship alive?• Given the resource constraints, how much should be spent on
acquisition efforts versus retention efforts to maximize long-term profitability?
In investigating such topics, researchers have modeledacquisition and retention both independently and jointly.
Table 7 lists a select set of studies that have consideredbalancing customer acquisition and retention.
In modeling independently, Lewis (2006b) investigateswhether shipping fees differentially influence customeracquisition and retention. In a system of simultaneousequations, the study examines the effects of shipping feesand other marketing variables on the number of newcustomers acquired, the average order size for new cus-tomers, and the number of daily orders and the averageorder size for established customers. Furthermore, toaccount for the possible correlation between various equationsand the possibility of endogenously determined explanatoryvariables, the author estimates the equations using three-stageleast squares.
In another study, Lewis (2006a) investigates the influ-ence of customer acquisition promotion depth on customerretention, including repeat purchasing and duration. In anonline retailing setting, the study adopts a logistic regressionto model whether the customer makes a subsequent pur-chase within the next three quarters, using acquisitiondiscount as an explanatory variable. Using data from thenewspaper industry, the study adopts accelerated failuretime models to model the time as a subscriber, usingacquisition discount and its quadratic form as an explan-atory variable. The study also examines the effects ofacquisition discount on customer asset value.
Berger and Bechwati (2001) optimize the allocation ofpromotion budget between acquisition spending and reten-tion spending. When companies are considering one marketsegment and using one promotion method, such as direct
TABLE 7Summary of Select Balancing Customer Acquisition and Retention Studies
RepresentativeStudies Study Focus
StudyType Model Type
StudySetting Insights
Reinartz, Thomas,and Kumar (2005)
Develop a modelingframework for balancingresources betweencustomer acquisition effortsand customer retentionefforts.
Empirical Probit two-stageleast squares
B2B high-technology
manufacturer
Firms can manage theircustomer bases profitablythrough resource allocationdecisions that involve trade-offs between acquisition andretention initiatives.
Thomas (2001) Establish that the customeracquisition process affectsthe customer retentionprocess.
Empirical Tobit Airplane pilotmembership
The proposed methodologycorrects for biases in thecustomer retention analysisthat result from assumingthat customer acquisitionand retention areindependent processes.
Berger and Bechwati(2001)
Create a framework for theoptimal allocation ofpromotion budget forcustomer acquisition andretention.
Conceptual Decision calculus — The study discussesdecisions about expenditureallocation betweenacquisition and retention indifferent market conditions.
Blattberg andDeighton (1996)
Find the optimal balancebetween acquisition andretention that maximizescustomer equity.
Conceptual Decision calculus — The authors recommendthat once managers havedetermined the balancebetween acquisition andretention, they plan for eachtask separately.
54 / Journal of Marketing: AMA/MSI Special Issue, November 2016
mailing, managers decide the allocation of existing budgetbetween acquisition and retention to maximize CE.
In modeling acquisition and retention jointly, Thomas(2001) proposes a method known as a Tobit model withselection to account for the impact of the customer acquisitionprocess on the retention process. The proposed model suc-cessfully links customer acquisition to retention because thelength of a customer’s lifetime is observed conditional onthe customer being acquired. The model also establishes thatthe error term in the acquisition equation is possibly corre-lated with that in the retention equation. Finally, Reinartz,Thomas, and Kumar (2005) model customer acquisition,retention, and profitability together as a system of equations,using a probit two-stage least squares model and estimate itas a system of equations.
In summary, customer acquisition modeling is actuallya probability prediction, and customer retention modelingessentially concerns the duration of customer lifetime.Acquisition probability can be estimated by a probit or logitmodel; hazard models can also be used if the timing ofincidence is concerned. Duration data are usually right-censored, so Tobit or hazard models are by nature suit-able estimation techniques. Researchers link acquisition andretention modeling either by specifying the correlationof the error terms in probability and duration models or byspecifying the joint distribution of acquisition and reten-tion for estimation.
Customer churn. Determining who is likely to churn isan essential step. This is possible by monitoring customerpurchase behavior, attitudinal response, and other metricsthat help identify customers who feel underappreciated orunderserved. Customers who are likely to churn demonstrate“symptoms” of their dissatisfaction, such as fewer purchases,lower response to marketing communications, longer timebetween purchases, and so on. It is important to note thatwhile customer retention and customer churn are similarconcepts, the need to study customer churn separately isrooted in the business setting. Whereas customer retentionapplies to both noncontractual and contractual businesssettings, customer churn is relevant only in a contractualsetting. Therefore, understanding customer churn becomescrucial in certain customer–firm relationships. In this regard,the CLV-based churn models provide directions on severaltopics, including the following:
• When are the customers likely to defect?• Can we predict the time of churn for each customer?• When should we intervene and prevent the customers from
churning?• Howmuch dowe spend on churn prevention with respect to a
particular customer?
Table 8 shows a representative set of studies that investigatethe customer churn process.
Researchers have strong interest in the causes of customerchurn and switching behaviors. They have developed variousmodels that include (1) modeling churn with time-varyingcovariates (Jamal and Bucklin 2006; Van den Poel andLariviere 2004), (2) analyzing themediation effects of customerstatus and partial defection on customer churn (Buckinx
and Van den Poel 2005), (3) modeling churn using twocost-sensitive classifiers (Glady, Baesens, and Croux 2009;Lemmens and Croux 2006), (4) determining the factors thatinduce service switching (Capraro, Broniarczyk, and Srivastava2003), and (5) analyzing the impact of price reductions onswitching behavior (Danaher 2002).
Customer win-back. It is no surprise that it is not pos-sible for firms to retain all acquired customers. When cus-tomers churn, managers usually consider this indicative of theend of the customers’ life cycle with the firm. However, itdoes not have to be: firms can still win the lost customers backand give them a second life. Unlike new customers, lostcustomers have certain knowledge about the products andservices of the company and have their own judgment onthe attributes and functions of the products and services.While it is easier to approach lost customers because they havefamiliarity with the firm, lost customers often switch firmsbecause they are not satisfied with the product, which makesit difficult for the reacquiring firm to change customers’attitudes and persuade them to come back. Furthermore, firmsalso have to consider whether it is worth trying to bringcustomers back to the firm—not all customers are worthchasing after they leave the firm.
Once a customer is likely to churn, the firm response vis-a-vis letting a customer go or winning the customer back iscritical. Researchers have studied this phenomenon fromthree aspects:
• Should the firm intervene in customer churn?• If so, how should the firm approach customers to win them
back?• What elements would help firms in reacquiring lost
customers?
The answers to these questions will drive the optimal cus-tomer win-back strategy. Table 9 lists select literature that hasexplored the topic of customer win-back.
Stauss and Friege (1999) provide a conceptual frameworkfor regaining lost customers that consists of analysis, actions,and controlling. The study contends that to determine cus-tomer value in the context of regain management, the lifetimevalue of the terminated relationship is not an appropriatemeasurement. Furthermore, Griffin and Lowenstein (2002)highlight the importance of highly trained win-back teamsand provide a general outline for winning back lost cus-tomers. This study proposes that not all churned customersare contenders to be won back. Instead, firms should calculatesecond-lifetime value (SLTV), segment customers on thebasis of SLTV, and evaluate customers in each segment todetermine why they defected. In addition, Thomas, Blattberg,and Fox (2004) investigate the best price strategy for reac-quisition of lapsed customers, using a split hazard model (orcensored duration model). Using data from the newspaperpublishing industry, this study finds that lowering reac-quisition prices to increase the likelihood of reacquisition isan optimal strategy. Finally, Kumar, Bhagwat, and Zhang(2015) investigate whether lost customers are worth theinvestment in reacquisition and whether they will remainprofitable if reacquired. The study finds that (1) the lostcustomers’ first-lifetime experiences and behaviors, (2) the
Creating Enduring Customer Value / 55
TABLE 8Summary of Select Customer Churn Studies
RepresentativeStudies Study Focus
StudyType Model Type Study Setting Insights
Glady, Baesens, andCroux (2009)
Present a frameworkusing a profit-sensitiveloss function for theselection of the bestclassificationtechniques withrespect to theestimated profit.
Empirical Logistic regression,neural network,
decision tree, cost-sensitive classifier
Retail financialservices
Cost-sensitiveapproaches achievegood results in terms ofthe defined profitmeasure and overallclassification.
Jamal and Bucklin(2006)
Study the empirical linkbetween customerchurn and factors suchas customer serviceexperience, failurerecovery, and paymentequity.
Empirical Weibull hazard Satellite television Prediction of customerchurn is significantlyimproved whenheterogeneity is addedto the churn rates andto the responseparameters.
Lemmens and Croux(2006)
Investigate whetherbagging andstochastic gradientboosting can improvechurn predictionaccuracy.
Empirical Bagging and boostingclassification trees
Wirelesstelecommunications
Bagging and boostingtechniquessignificantly improvethe classificationperformance, andbalanced calibrationsample reduces theclassification errorrate.
Buckinx and Van denPoel (2005)
Identify which of thecurrently behaviorallyloyal customers arelikely to (partially)churn in the future.
Empirical Logistic regression,automatic relevancedetermination, neuralnetwork, random
forests
CPG retailer RFM variables are thebest predictors ofpartial customerdefection; variablessuch as customerrelationship duration,mode of payment,cross-buying behavior,usage of promotions,and brand purchasebehavior aremoderately useful inattrition models.
Van den Poel andLariviere (2004)
Develop acomprehensiveretention modelincluding time-varyingcovariates related tocustomer behavior.
Empirical Proportional hazard Financial services Demographiccharacteristics,environmentalchanges, andinteractive andcontinuous customerrelationships affectretention.
Capraro,Broniarczyk, andSrivastava (2003)
Study repurchasedecisions that involvean information-basedevaluation ofalternatives.
Empirical Hierarchical logisticregression
Health insurance After satisfaction levelis accounted for,objective andsubjective knowledgeabout alternativesdirectly affectsdefection likelihood.
Danaher (2002) Derive a revenue-maximizing strategyfor subscriptionservices.
Fieldexperiment
Time-seriesregression
Cellular service Access and usageprices have differentrelative effects ondemand and retention.
56 / Journal of Marketing: AMA/MSI Special Issue, November 2016
reason for defection, and (3) the nature of the win-back offermade to lost customers are all related to the likelihood of thecustomers’ reacquisition, their second-lifetime duration, andtheir second-lifetime profitability per month.
Customerwin-back is a critical componentwithin the overallCRM strategy, but it has received relatively less researchattention than customer acquisition, retention, and churn. Firmsmust first be able to understand the drivers of customer
TABLE 9Summary of Select Customer Win-Back Studies
RepresentativeStudies Study Focus Study Type Model Type Study Setting Insights
Kumar, Bhagwat,and Zhang(2015)
Demonstrate how lostcustomers’ first-lifetimeexperiences andbehaviors, the reasonfor defection, and thenature of the win-backoffer made to lostcustomers are relatedto reacquisitionlikelihood, their second-lifetime duration, andsecond-lifetimeprofitability per month.
Empirical Probit, right-censoredTobit, regression
Telecommunicationsservice
The stronger acustomer’s first-lifetime relationshipwith the firm, the morelikely the customer isto accept the win-backoffer.
Homburg, Hoyer,and Stock(2007)
Test the relationshipbetween customers’first-lifetime satisfactionand their reacquisition.
Empirical Logistic regression Telecommunicationsservice
Health of the first-lifetime relationship ispositively related toreacquisitionlikelihood; customerperceptions offairness regardingwin-back offer arepositively related toreacquisition.
Tokman et al.(2007)
Identify the factorsdriving win-back offereffectiveness.
Quasi-experimental
design
Analysis of variance Auto repair andmaintenance service
Price and servicebenefits provided inthe win-back offer,social capital, andservice importanceare important inshaping customerswitch-backintentions, regardlessof previoussatisfaction, regret, ordelight with the newservice provider.
Thomas,Blattberg, andFox (2004)
Study the probability ofcustomer reacquisitionand the duration of thesecond lifetime, with afocus on the impact ofthe depth of the pricediscount.
Empirical Split-hazard,Bayesian Markovchain Monte Carlo
Newspaper industry For the win-back offerto be effective, itshould consist of lowpromotional prices;successfulreacquisition shouldbe followed by priceincreases.
Stauss andFriege (1999)
Conceptual basis for“regain management”aimed at winning backcustomers who eithergive notice to terminatethe businessrelationship or whoserelationship has alreadyended.
Conceptual — — Retention policy inservice companiesneeds to becomplemented byregain management.
Creating Enduring Customer Value / 57
reacquisition, customer duration in a second lifetime, andsecond-lifetime customer profitability before they can makeaccurate strategic decisions regarding which customers to try towin back, and what offers to provide to maximize customerprofitability.
Extending CLV to Engage with Customers
Typically, customer contribution to firm profitability oc-curs (1) directly, through their purchases, and (2) indirectly,through their nonpurchase reactions, which include referringpotential customers, influencing current and potential cus-tomers in their social network, and offering review/feedbackfor improvements. In this regard, Kumar et al. (2010)propose the customer engagement value framework, whichcan be used to identify and evaluate the right customer,who is successfully engaged with the firm and who generatesvalue and positively contributes to the profits of the firm. Thefollowing discussion elaborates on customers’ indirect con-tribution to value.
Role of referrals. Studies in this area have found thatcustomers acquired through referrals are not only lessexpensive to acquire but also more valuable for the firm, ascompared with customers acquired through traditional meth-ods (Schmitt, Skiera, and Van den Bulte 2011; Villanueva,Yoo, and Hanssens 2008). For instance, Schmitt, Skiera, andVan den Bulte (2011) examine the difference in value derivedfrom each customer depending on acquisition method, usingcustomer data from a financial services firm. The study findsthat referred customers are more profitable than nonreferredcustomers. However, Villanueva, Yoo, and Hanssens (2008)find that customers acquired through traditional techniques aremore valuable in the short term to some extent, but referredcustomers are twice as valuable in the long term. In addition,the value differential becomes even largerwhen the disparity inacquisition costs is considered, due to the considerably lowercosts of referrals. Table 10 summarizes the other key contri-butions of past research in this area.
Kumar, George, and Leone (2007) introduce the customerreferral value (CRV) metric, which captures the net presentvalue of the future profits of new customers who purchased thefirm offerings as a result of the referral behavior of the currentcustomer. Simply put, it represents the value of how firm-initiated and firm-incentivized customer referral programs canimprove the profitability of the customer base by acquiringcost-effective prospects (Kumar, George, and Leone 2010).Most notably, this metric makes a distinction between thecustomers that initiated a relationship with the firm solelybecause of the referral and those that would have becomecustomers even in the absence of a referral. Therefore, CRVincludes both the acquisition savings and the profits of thefuture transactions of customers who join as a result of thereferral. However, it includes only the acquisition savings ofcustomers who would join anyway, because their futuretransactions would happen even in absence of the referral, sothey cannot be attributed to the referrer.
Research in this area has focused on (1) developing modelsthat use only a customer’s actual past referral behavior tocompute an individual customer’s referral value (Kumar,
George, and Leone 2007), (2) the identification of thebehavioral drivers of CRV and the methods of solicitingreferrals according to each customer’s CLV and CRV scores(Kumar, George, and Leone 2010), and (3) studying referralbehavior in a business setting, that is, BRV (Kumar, George,and Leone 2013). The BRV is defined as the ability of anddegree to which a client’s reference influences prospects topurchase. The BRV of each referencing client and the CLV ofeach referencing client and newly acquired client from theprospect pool are measured using a three-step methodcomprising (1) determining whether client referencesinfluenced the prospect’s decision to adopt, (2) determiningthe influence each client reference had on the prospect’sdecision to adopt, and (3) computing the CLV of theconverted prospect. Furthermore, Kumar, George, andLeone (2013) empirically determine the key drivers of BRVand recommend that a firm using a client-referencingstrategy needs to leverage a portfolio of client referencesthat includes clients with varied characteristics in order tomatch the client to the prospect successfully.
Valuing customer influences. The spread of informationbyWOM is a critical component in converting prospects intocustomers. To better understand the impact of WOM, severalstudies have called for its inclusion in customer value models(Hogan, Lemon, and Libai 2003; Libai et al. 2010). Researchhas investigated WOM propagation in the context of onlinereferrals (Trusov, Bucklin, and Pauwels 2009) and the evo-lution of the network as a whole with every consecutiveinstance of WOM (Robins et al. 2007). Kumar et al. (2013)develop a framework to capture the value of an individual’sWOM in terms of both its viral impact and the net sales thatit facilitates through two key metrics—customer influenceeffect (CIE) and customer influence value (CIV).Whereas theCIE measures the net spread and influence of a messagefrom a particular individual; the CIV calculates the monetarygain or loss realized by a firm that is attributable to a cus-tomer, through that customer’s spread of positive or negativeinfluence. The study implements the framework by creatingand deploying a social media strategy at an ice creamretailer. By tracking these two metrics for the retailer, thisstudy is able to demonstrate a 49% increase in brandawareness, 83% increase in ROI, and 40% increase in salesrevenue growth rate. While most firms are still strugglingwith social media accountability, this study effectivelyshows the importance of CIE and CIV. Although theseearly studies illustrate the importance of valuing customerinfluences, more research is required to get a deeperunderstanding of this phenomenon.
Valuing customer knowledge contributions. Customerscan add value to the company by helping elucidate customerpreferences and participating in the knowledge developmentprocess. In this regard, beta site customers that use/reviewproducts and provide valuable feedback to firms are a greatexample of customer knowledge contribution. Studies haveinvestigated the amount of value added by customers throughtheir knowledge contributions. For instance, Fuller, Matzler,and Hoppe (2008) find that brand community members whohave a strong interest in the product and in the brand usually
58 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE10
SummaryofSelec
tCustomer
ReferralBeh
aviorStudies
Rep
rese
ntative
Studies
StudyFocu
sStudy
Typ
eModel
Typ
e
B2B vs.
B2C
StudySetting
Insights
Kum
ar,Geo
rge,
andLe
one
(201
3)Und
erstan
dtherole
andva
lueof
client
referenc
esin
abu
sine
ssco
ntex
t.
Empiric
alLo
git
B2B
Telec
ommun
ications
service,
fina
ncial
services
Firm
sca
neffectivelyus
ebu
sine
ssreferenc
esto
pullin
newcu
stom
ers.
Richmed
iaan
dsimilarco
mpa
nies
aretheke
ydriversof
succ
ess.
Sch
mitt,Skiera,
andVan
den
Bulte
(201
1)Determinetheex
tent
ofprofi
tability
andloya
ltyof
referred
custom
ers.
Empiric
alReg
ress
ion,
prop
ortio
nal
haza
rd
B2C
Ban
king
services
Referredcu
stom
ers(1)ha
vea
high
erco
ntrib
utionmargininitially,
(2)h
aveasu
staine
dhigh
erretention
rate,a
nd(3)a
remoreva
luab
leinthe
shortan
dlong
run.
Kum
ar,Geo
rge,
andLe
one
(201
0)Determinetheop
timal
custom
ertargetingforreferral
marke
ting
campa
igns
.
Empiric
al,field
stud
yBay
esianTob
itB2C
Finan
cial
services
,retailing
Firm
sca
neffectivelyse
lect
custom
erswith
inse
gmen
tsof
high
andlow
CLV
/CRVforreferral
marke
tingca
mpa
igns
usingthe
driversof
CRV(dyn
amic
targeting).
Villan
ueva
,Yoo
,an
dHan
ssen
s(200
8)Estab
lishava
lue-ge
neratin
gproc
essby
mod
elingtheinteractions
betwee
nne
wcu
stom
erac
quisition
andthegrow
thof
firm
value.
Empiric
alVec
tor
autoregres
sion
B2B
Web
-hos
ting
compa
nyMarke
ting-indu
cedcu
stom
ersad
dmoresh
ort-term
value,
butWOM
custom
ersad
dne
arlytwiceas
muc
hlong
-term
valueto
thefirm
.
Kum
ar,Geo
rge,
andLe
one
(200
7)Com
pare
custom
erse
gmen
tatio
nac
cordingto
CLV
andCRV.
Con
ceptua
l—
B2C
Telec
ommun
ications
service,
fina
ncial
services
Cus
tomerswith
high
CLV
areno
tthe
sameas
custom
erswith
high
CRV.
Ryu
andFeick
(200
7)Determinetheeffectiven
essof
rewards
forreferral
marke
ting.
Field
stud
yAna
lysisof
cova
rianc
eB2C
MP3play
ers
Offe
ringarewarddo
esincrea
sereferral
likelihoo
d,bu
tthe
size
ofthe
rewardis
notsign
ifica
nt.
Hog
an,Le
mon
,an
dLiba
i(200
4)Qua
ntify
thead
vertisingrip
pleeffect.
Con
ceptua
l—
—Hairstylingse
rvices
The
WOM
gene
ratedafteran
adve
rtising-motivated
purcha
seca
nbe
sign
ifica
nt.
Hog
an,Le
mon
,an
dLiba
i(200
3)Dev
iseametho
dto
acco
untfor
social
effectsin
theman
agem
entof
custom
ers.
Empiric
alNon
linea
rleas
tsq
uares
regres
sion
B2C
Internet
bank
ing
The
valueof
alost
custom
erch
ange
sas
afunc
tionof
thetim
ein
theprod
uctlifecy
cle.
Creating Enduring Customer Value / 59
have extensive product knowledge and engage in product-related discussions and support each other in solving prob-lems and generating new product ideas. With respect to theroles played by customers in this process, studies havecategorized them into two categories: information providersand codevelopers (Fang 2008). In capturing the value con-tribution through knowledge sharing, Kumar et al. (2010)introduce the customer knowledge value (CKV) metric asthe value a customer adds to the firm through his or herfeedback. The need for computing the knowledge valuecontributed by customers has been recognized by studiesthat have posited that (1) tracking a customer’s product orservice expertise (which is correlated with the customer’sCLV) can be valuable in assessing CKV (Von Hippel 1986),and (2) defected customers might contribute to CKV bysharing their reasons for leaving, allowing the firm toidentify service improvement opportunities and increase itscapability to detect at-risk customers (Stauss and Friege1999; Tokman, Davis, and Lemon 2007). These studiesconstitute a nascent stream of research that has muchpromise.
Realizing Enhanced Firm and Shareholder Value
Customer equity represents the sum of the lifetime valuesof all customers of the firm. This topic has been a subjectof constant attention; it has been shown to maximize thereturn on marketing investments and guide the allocationof the marketing budget (Blattberg, Getz, and Thomas2001; Reinartz, Thomas, and Kumar 2005; Rust, Lemon,and Zeithaml 2004). For instance, Gupta, Hanssens, andStuart (2004) show across multiple firms that customervalue can be a method of firm valuation that is as good as abetter than traditional accounting. They also quantify theimpact of firm value resulting from improvements inretention, margin, and acquisition costs. Customer sat-isfaction has also been linked to higher firm value in that itincreases immediate cash flows and creates future growthoptions (Fornell et al. 2006; Gruca and Rego 2005; Malsheand Agarwal 2015).
Kumar and Shah (2009) propose a framework to linkthe outcome of marketing initiatives (as measured by CE)to the firm’s market capitalization (MC) (as determined bythe stock price of the firm). The intuition behind thisframework is that the stock price of the firm is based on theexpected future cash flows of the firm. If cash flow isprimarily generated from customers, an increase in CE (orcash flow from customers) should relate to an increase inMC (or the stock price of the firm). Such an approachwould answer three important questions: (1) Can marketingstrategies that increase CE also increase the stock price of thefirm? (2) If so, can such increases in stock price be predictedon the basis of changes in CE? (3) Can the increases in thestock prices of the two firms be attributed to shareholder valuecreation?
To answer these questions, Kumar and Shah (2009) im-plement CLV-based strategies that are designed to strengthenthe CRM practices of a firm. The implementation was doneover a nine-month period in a B2B firm and a B2C firm, bothFortune 1,000 companies. At the end of the implementation, the
study finds that the increases in stock price for the B2B firm andthe B2C firm were approximately 32.8% and 57.6%, respec-tively. Furthermore, when the average monthly values of thestock price are plotted over time and compared with the actualaverage values of the stock prices of the two firms, the resultsindicate that it is possible to track the actual movement of stockprices within a maximum deviation range of 12%–13%. It isalso found that the stock price of the B2B firm outperformedthe S&P 500 by 200%, and the B2C firm outperformed theS&P 500 by 360%. Finally, the study also finds that theB2B firm’s stock increases by 32.8% during the observationperiod, while its three closest competitors’ stocks go up byan average of 12.1% over the same period. Similarly, theB2C firm’s stock increases by 57.6%, while its three closestcompetitors’ stocks go up by an average of 15.3% over thesame period. These findings clearly demonstrate the link-age between CE and MC.
The studies discussed here sufficiently demonstrate thepower of real-time decision making that firms would gainwhen the drivers of customer value and timely marketinginterventions are carefully understood and coordinated. Thisimplementation would not only refine marketing actions butalso enhance value to the firm.
Key Insights and Future ResearchDirections
Extant marketing literature has investigated the concept ofcustomers providing value to firms and have sought tounderstand the process of deriving value from customers,ways of measuring it, and strategies for maximizing it. Thecurrent study has discussed the knowledge created byresearchers thus far in understanding customer value. Thisresearch has shown that value is created to and from thecustomers. While this creation happens fairly regularly, thisprocess is beneficial only if it is long-lasting, which is pos-sible only when the customer perceived value (i.e., value tocustomers) and firm value (i.e., value from customers) arealigned. This alignment, however, only happens over time.Furthermore, the alignment process is constantly shaped bywhat we know about value and the knowledge we still seek.When these two value sources are aligned, it leads to thecreation of enduring customer value, as opposed to just value.The alignment is materialized through the appropriate settingof prices in a manner that reflects customer and competitivefactors. In effect, there is a constant reevaluation from thefirm’s side vis-a-vis prices to ensure the value alignment toand from the customers. The setting of prices is subsequentlyreflected in the development of products and services, as wellas in the designing of marketing campaigns. Figure 3 presentsan organizing framework that describes the process of valuealignment.
This study has uncovered some valuable insights that willbe helpful in understanding the creation and communicationof value. With regards to value to customers, this study hasdefined customer perceived value that incorporates both the“give versus get” perspective and the value-communicatedaspect. Such an approach to defining and understanding per-ceived value highlights the importance of thinking beyond
60 / Journal of Marketing: AMA/MSI Special Issue, November 2016
the value component, to include the costs given up for thebenefits that are being sought. Following this definition, thisstudy has identified three tasks for measuring customerperceptions of value: measuring overall perceived value,measuring the associated underlying attributes and benefits,and determining the relative weights of the attributes/benefitslinked to overall perceived value. With respect to value fromthe customers, this study has identified that a forward-lookingmetric such as CLV is ideal in metricizing customer valuecontributions. Furthermore, this study has observed that tocreate net value for the firm, the perceived value that cus-tomers receive must be aligned with the resources spent onthe customers, through the adoption of forward-lookingmetrics.To this end, CLV is identified as the metric (amongthe popularly used metrics) that most accurately com-putes the net present value of a customer according to his orher future transactions with the firm, after accounting forrevenue, expense, and customer behavior. Various CLV-based strategies have been proposed that firms can use tomaximize value from customers after computing CLV.Table 11 lists the key insights from this study.
Looking ahead, we would like to offer three key areas thatcould spur future research. First, we believe the Internet willattain even more dynamism in terms of the uses, abilities, andopportunities it presents. In this regard, the Internet of Things(IoT) is bound to offer several avenues for researchers andpractitioners to identify and maximize ways to create value forfirms and customers. With its origins in the supply chaincontext (Ashton 2009), the IoT now encompasses uses invirtually all areas, including health care, urban planningand management, emergency services, construction, envi-ronment monitoring, lifestyle, and sports management. Theready applicability to a wide range of industries is largely dueto the connectivity to media and transportation channels theIoT provides to firms.
For instance, consider the case of a professional sportsteam. Sports fans would like to get to the sporting ven-ues in time for the game and the other attractions offered.However, a game day brings with it the problems of trafficand changing weather conditions, which create challengesin finding the fastest route to the venue. With a key goal ofensuring fan satisfaction, a sports team could considerproviding a wearable device to fans that would inform themof traffic conditions, road closures, and weather updates,in addition to team-related and game-related information.Such a device from the team management, apart from being apiece of fan merchandise, would also keep fans engaged withthe team. The technology to create such a device is availablethrough IoT, but how can firms create value from such anoffering? To begin with, what value-based metric can be usedto identify the fans who would receive such a device? Cansuch a device capture fan satisfaction effectively? Howwouldthe team view fan satisfaction at the venue and fan sat-isfaction getting to and coming from the venue? In otherwords, would an unpleasant experience away from the venueadversely influence a fan’s satisfaction at the game andsubsequently lead to a decline in loyalty? Furthermore, whatinsights could the team gain regarding fan experience ongame day and the ways to mitigate any loss in following?How could the team capture these fan sentiments and use thedevice to offer a valuable proposition that both engages thefan with the team and ensures value to the team throughconsistent ticket sales?
Second, health and fitness consciousness has been on therise, with fitness studios offering varied programs to suitevery customer need. These paid options provide severalcustomization options for users to design their visit andstructure their fitness regimen. These options includescheduling visit times (some studios are even open 24 hoursa day), personal training, designing meal plans, and studio
FIGURE 3Ensuring Value Alignment: An Organizing Framework
What We Know
Alignment
What We Know
What We Would Like to Know
• Impact of products’ intangible aspects on perceived value• Value of privacy• Source of perceived value
Productsand
ServicesPerceived
CostUndesired
Consequences
BenefitsPerceivedAttributes
FirmValue
CustomerPerceived
Value
Marketing Programs
CustomerChurn
Customer Win-Back
Customer Retention
Customer AcquisitionProfitable
Customer Engagement
What We Would Like to Know
• CLV for a basket of goods• Effect of customer attitudes on CLV• Impact of macroeconomic trends on CLV• Nature of relationship on value creation• Customer engagement and privacy• Level of engagement and successive generation of products
Pricing Decisions
Creating Enduring Customer Value / 61
amenities, among others. In addition, fitness studios nowhave a strong online presence through blogs and socialnetworking sites. They now interact with customers andfitness enthusiasts by exchanging fitness-related information.Consumers also have access to unpaid options such as fitnessvideos and predesigned fitness routines available on theInternet. While this option lacks customization, beingavailable for free compensates for this dearth. Perhaps themajor attraction of the unpaid option is the availability evenduring travel. However, some fitness studios also now havededicated members’ websites that offer health tips, alongwith tools to manage workout schedules through videos andinstruction manuals, in addition to regular studio admission.From the perspective of a fitness studio, the challenge ofsuch a site is twofold. First, how does the firm present itsvalue proposition vis-a-vis other competing studios (e.g.,www.my360gym.com)? Second, how does it fight the com-petition from the unpaid fitness options? In such cases,what other novel and innovative programs can be offered byfitness studios to attract customers? Furthermore, how canstudios design, price, and offer online fitness material (text,audio, and video) that can effectively competewith the existingunpaid options? In addition, can studios exist only in the onlineformat (e.g., www.booyafitness.com) without any physicalpresence, and if so, how would the value then be determinedfor a fitness customer?
Finally, household purchase decisions have receivedsubstantial research attention (e.g., Epp and Price 2008;Gupta, Hagerty, andMyers 1983). The decision-making rolestypically include influencers, gatekeepers, deciders, buyers,users, and disposers. In situations in which the head of thehousehold is the decider and buys a good, the product isintended either for personal use or common use. Householdconsumption is also known to be affected by scale econo-mies, wherein the utilization rate of a good can be raised byincreases in family size (Lazear and Michael 1980). In suchconditions, for the purposes of identifying value and designingcustomer strategies, firms would benefit if they had answers toquestions such as the following: (1) Is the good being pur-chased for personal use or common use? (2) Who will bepaying for the good—the individual or the family? (3) Will atrade-off be necessary to buy the good, either by the individualor the family? (4) In the case of a common good, will thegood hold the same value to all members of the household,and if not, how can the disparity be alleviated? (5) In the caseof a common good, will there be any design changes necessaryto satisfy all the users? (6) How should the media mix be de-signed such that the influencers of the household are reachedwith the right message, in the right format, and at the righttime? (7) Given these questions regarding the decision-makingroles, the usage of the good, and the awareness of the good,should firms compute the value as a comprehensive amount
TABLE 11Key Insights from This Study
Value to Customers Value from Customers
• Customer perceived value is different from quality, perceivedbenefits, and satisfaction. This article defines perceived valueas the customer’s net valuation of the perceived benefitsaccrued from an offering that is based on the costs they arewilling to give up for the needs they are seeking to satisfy.
• Benefits and undesired consequences are the results of buyingand consuming the offering (i.e., the attributes of the offering),and these may accrue directly or indirectly and may beimmediate or delayed.
• Perceived value is measured according to attributes, whichmay be objective attributes (i.e., produced attributes) orperceived attributes (i.e., experienced attributes).
• Approaches to modeling consumer preferences have adoptedtwo types of methods: compositional and decompositional.While the former is a set of explicitly chosen attributes/benefitsthat are used as the basis for determining overall valueevaluations, the latter attempts to infer underlying utilities fromobserved choice.
• Firms can create perceived value for customers through (1)leveraging their own capabilities, (2) aligning with customers’perception of what is valuable for them, and (3) claiming adifferential advantage (e.g., premium or margin) overcompetitive offerings.
• Customers form a judgment of value as a function of perceived,not actual, benefits and costs.
• Measuring customer perceptions of value involves three keytasks: (1) measuring overall perceived value, (2) measuring theassociated underlying attributes and benefits, and (3)determining the relative weights of the attributes/benefits linkedto overall perceived value.
• Firms need to align the perceived value customers receive withthe resources spent on them through the adoption of forward-looking metrics, to create net value for the firm.
• Unlike backward-looking metrics, forward-looking metricsfocus on the customer, rather than the product, and can beused to understand current clients as well as prospects.
• CLV calculates the net present value of a customer accordingto his or her future transactions with the firm, after accountingfor revenue, expense, and customer behavior.
• CLV can be modeled at the aggregate level or the individualcustomer level.
• CLV-based strategies can be used not only to maximizerevenue, minimize costs, or both, but also help with customeracquisition and retention, balancing acquisition and retention,customer churn, and customer win-back.
• The customer contribution to firm profitability occurs directly,through customer purchases, and also indirectly, throughactions that might include referrals or influencing others viasocial networks, customer reviews, and feedback to the firm.
• The concept of customer engagement value helps in theidentification and evaluation of the right customer, who issuccessfully engaged with the firm, who generates value, andwho positively contributes to the profits of the firm.
62 / Journal of Marketing: AMA/MSI Special Issue, November 2016
TABLE 12Future Research Directions
Value to the Customers Value from the Customers
• The effect of customer costs and consumer needs onperceived value has been studied largely from the viewpoint ofa product’s tangible features. Intangible aspects, such as, forinstance, the network effect on perceived value, have not beenexplored. Network effects have been addressed in the “valuefrom customers” perspective (e.g., Kumar et al. 2013; Robinset al. 2007) to strengthen profitable customer–firmrelationships. Future research could explore individuals’perception of value and its constituents in a network settingthat exhibits influences across customers.
• Studies of undesired consequences have explored the realmof privacy concerns primarily in an online business setting.However, a more comprehensive treatment of the value ofprivacy is required to better our understanding in terms of (1)quantifying the overall value of privacy, (2) determining thevalue of personal information that customers will be willing togive up for products with lower prices, and (3) establishing theoffline product categories that privacy costs apply tocustomers.
• In identifying the source of perceived value, while usage, costs,and profits involved have been considered, the nature ofproduct has not been considered. Future studies could focuson the source of perceived value—simple product design orsophisticated product design.
• While Sunder, Kumar, and Zhao (2016) have bridged the gap inliterature by proposing a structural approach tomeasuring CLVthat incorporates the choice, timing, and quantity decisions ofconsumers to assess CLV in the CPG setting, future studies inthis area can look into expanding the analysis for a basket ofgoods, and including stochastic shocks to the system thatmight influence the consumption.
• Literature has shown that customer behavior influences CLV(Reinartz and Kumar 2003) and that customer attitudesinfluence customer behavior (e.g., Anderson 1998; Hogan,Lemon, and Libai 2003), which in turn influences CLV. But docustomer attitudes directly influence CLV? Answering thisquestion is bound to provide insights into the reasoning behindcustomer behavior and how they affect customer profitability.
• While most CLV implementation studies account for marketingand financial variables, the macroeconomic trends remainunaccounted for. What approaches can we adopt to accountfor macroeconomic factors such as GDP growth rate, rate ofunemployment, and so on, in the CLV implementation to makeit more accurate and reflective of market realities?
• Although the benefits of implementing customer engagementhave been demonstrated, we need to know the nature of therelationship between level of engagement and value creation(e.g., linear or inverted U shaped).
• Furthermore, what is the role of customer engagement incustomer privacy issues? That is, if customers are engagedmore, will they be less sensitive to sharing private information?
• In realizing value through customer engagement, is it possibleto identify which form of engagement will work with a certaintype of customer? That is, can a customer provide value in allforms of engagement?
• Research has identified other forms of engagement, such asgifting behavior. Bhagwat and Kumar (2015) propose thatencouraging customers to take part in gifting behavior is oneway to effectively engage them with the firm and toconsequently see profitable outcomes. In this regard, are thereany other forms of engagement that can lead to profits for thefirm?
• When firms launch products, the performance of the previousgeneration of products and the overall firm performance arechallenged. For instance, whenever Apple launches a product(e.g., iPhones), the prices of the previous-generation productsare lowered, prompting many consumers to wait for such aprice drop to buy the earlier version. In this regard, does thelevel of engagement facilitate the adoption of successivegeneration of products, and if so, how?
• CLV maximization can also be viewed from an optimizationstandpoint. Firms often plan and change their resource inputs.When the elasticities of factors such as customer acquisition,retention, and win-back are considered, the resultingoptimization may lead to better-informed resource planning.
• Despite the prevalence of coalition loyalty programs, they havereceived little attention in academic research. Future studiescould investigate the profitability drivers of such loyaltyprograms.
• The identification of the lifetime value for dealers (e.g., dealerlifetime value) would help firms such as car dealerships to finda reliable method of (re)allocating scarce marketing resourcesto the “best”-performing dealerships.
Creating Enduring Customer Value / 63
that accounts value derived by all the users of the household,or as a summation of individual values of all the users?
This study has also presented specific research areasregarding value to and value from the customers.With respectto value to the customers, for instance, this study recom-mends looking into the network effects because the effect ofcustomer costs and consumer needs on perceived value hasbeen studied largely from the viewpoint of a product’stangible features, while the intangible aspects, such as thenetwork effects, have been explored by only a few studies(Kumar et al. 2013; Robins et al. 2007). Therefore, a moredetailed investigation into this is required to broaden ourunderstanding on delivering customer value.
Another issue that requires research attention is the caseof privacy. In order to use online services, customers arerequired to share their personal information instead of beingcharged a fee. In effect, while customers get to use theservices for free, the cost incurred is sharing personalinformation. Amid this growing digitization climate, firmsnow will have to understand (1) the value of shared privatedata, (2) the business settings in which such an option will(and will not) work, and (3) how the value proposition can becommunicated to customers. Future research along theselines will be helpful in understanding the drivers of value in adigitized format and in customizing offerings to maximizevalue to and from customers.
With respect to value from customers, this study hasidentified several areas for future research. For instance,while a structural approach to measuring CLV that incor-porates the choice, timing, and quantity decisions of con-sumers to assess CLV in the CPG setting has been proposed(Sunder, Kumar, and Zhao 2016), more studies in this areaare needed to formalize this learning. Specifically, studiesthat can expand the specific product-level categorization toinclude a basket of goods will facilitate the study of CLVfrom a retailer’s perspective. Such studies will likely result in(1) further refinements to the drivers of customer and retailerprofitability, (2) better customization of product offerings,and (3) targeted store-level product promotions.
Another area that could be fruitful for research on CLVinvolves accounting for macroeconomic trends and newproduct introductions by competitors. Specifically, futurestudies should look into identifying approaches that can beadopted to account for macroeconomic factors such as GDPgrowth rate, rate of unemployment, inter- and intracountryvariations, consumption and investment spending, and inter-national trade balance in the CLV implementation to make itmore accurate and reflective of market realities. In this regard,Kumar and Pansari (2016) have shown that national culturaldimensions affect the drivers of purchase frequency andcontribution margin and that economic factors influence thecomponents of CLV directly. In effect, the relative effects ofthe drivers of CLV are influenced by the differences in themacroeconomic trends of the countries as well as in the cul-tures of the countries. However, more studies across countrieswith different levels of economic stability/performance areneeded to more thoroughly understand this result. Theseinsights will be of use to firms that have international oper-ations to plan their resource allocation well ahead of time.
Future studies could also focus on identifying the natureof the relationship between level of customer engagement andvalue creation (e.g., linear or inverted U shaped). This willhelp in refining the engagement strategies aimed at enhanc-ing customer value creation. Furthermore, identifying newerforms of engagement would expand the number of ave-nues through which a firm could establish engagement. Forinstance, research has determined that encouraging cus-tomers to take part in gifting behavior effectively engagesthem with the firm and consequently leads to higher profits(Bhagwat and Kumar 2016). Future studies could investigateother forms of engagement that, when encouraged amongcustomers, can lead to profits for the firm.
An area of loyalty that is gaining significant traction iscoalition loyalty. Such a program brings together two or morecompanies to offer loyalty incentives across a wide spectrumof retail and service offerings (e.g., Star Alliance in the air-line industry). These programs provide significant advan-tages such as shared operational costs, higher cross-sellingpotential, and a symbiotic relationship among participatingcompanies (Lee, Lee, and Sohn 2013). Limited research inthis area has found that service failure by one partner has aspillover effect on other partners (Schumann, Wunderlich,and Evanschitzky 2014), and positive spillover effects fromone partner are reflected in cross-buying of other partners(Lemon and Von Wangenheim 2009). Future research couldlook into the profitability drivers of these programs to gain abetter understanding of the outcomes.
Another area of research opportunity lies in determiningthe optimal balance of resource allocation between tradi-tional and new media, and the role of each in maximizingROI. The challenge here for researchers lies in determiningthe optimal mix of media options and how they should beprioritized among the customer deciles in order to max-imize ROI. When the rebalancing of resources is linkedback to customer acquisition, retention, and win-back ini-tiatives, the results would lead to better informed resourceplanning.
The identification of lifetime value of distributors/dealers(e.g., car dealerships) is another area with opportunities forfuture research. In the case of U.S. car dealerships, automanufacturers sell their cars through exclusive dealershipsand/or dealerships that carry multiple brands. Here, the automanufacturers need to find a reliable method of (re)allocatingscarcemarketing resources to the “best-”performing dealerships.In this regard, creating a “dealer lifetime value,” classifyingdealers as “high-value” or “low-value” and placing them inappropriate deciles, would help manufacturers solve thereallocation problem. Table 12 provides a compilation oftopics that might be considered for future research.
This study has discussed the major developments in theCLV area of research and highlighted potential future re-search directions. Specific refinements and improvementscan be expected in (1) approaching measurement of CLV, (2)understanding the drivers of CLV, and (c) gathering moreempirical evidence regarding the various business applica-tion of CLV. When these developments feed into real-timedecision making for marketers, marketing’s accountability tothe corporate boardroom will have been enhanced.
64 / Journal of Marketing: AMA/MSI Special Issue, November 2016
REFERENCESAllenby, Greg, Robert Leone, and Lichung Jen (1999), “ADynamic
Model of Purchase Timing with Application to Direct Market-ing,” Journal of the American Statistical Association, 94 (446),365–74.
Anderson, Eric T. and Duncan I. Simester (2004), “Long-RunEffects of Promotion Depth on New Versus EstablishedCustomers: Three Field Studies,”Marketing Science, 23 (1),4–20.
Anderson, Eugene W. (1998), “Customer Satisfaction and Word ofMouth,” Journal of Service Research, 1 (1), 5–17.
Anderson, James C., Dipak Jain, and Pradeep K. Chintagunta(1993), “Understanding Customer Value in Business Markets:Methods of Customer Value Assessment,” Journal of Business-to-Business Marketing, 1 (1), 3–30.
———, Nirmalya Kumar, and James A. Narus (2007), ValueMerchants: Demonstrating and Documenting SuperiorValue in Business Markets. Boston: Harvard BusinessPublishing.
———, James A. Narus, and Das Narayandas (2009), BusinessMarket Management: Understanding, Creating, and DeliveringValue, 3rd ed. Upper Saddle River, NJ: Prentice Hall.
Ashton, Kevin (2009), “That ‘Internet of Things’ Thing,” RFiDJournal, 22 (7), 97–114.
Benoit, Dries F. andDirk Van den Poel (2009), “Benefits of QuantileRegression for the Analysis of Customer Lifetime Value in aContractual Setting: An Application in Financial Services,”Expert Systems with Applications, 36 (7), 10475–84.
Berger, Paul D. and Nada I. Nasr (1998), “Customer Lifetime Value:Marketing Models and Applications,” Journal of InteractiveMarketing, 12 (1), 17–30.
——— and Nada Nasr Bechwati (2001), “The Allocation of Pro-motion Budget to Maximize Customer Equity,” Omega, 29 (1),49–61.
Bhagwat, Yashoda and V. Kumar (2016), “The Impact of GiftingBehavior on Customer Profitability,” working paper, GeorgiaState University.
Bhattacharya, C.B. (1998), “When Customers are Members: Cus-tomer Retention in Paid Membership Contexts,” Journal of theAcademy of Marketing Science, 26 (1), 31–44.
Blattberg, Robert C. and John Deighton (1996), “ManageMarketingby the Customer Equity Test,”Harvard Business Review, 74 (4),136–44.
———, Gary Getz, and Jacquelyn S. Thomas (2001), CustomerEquity: Building andManaging Relationships as Valuable Assets.Boston: Harvard Business Publishing.
Boatwright, Peter, Sharad Borle, and Joseph Kadane (2003), “AModel of the Joint Distribution of Purchase Quantity andTiming,” Journal of the American Statistical Association, 98 (463),564–72.
Bolton, Ruth N. (1998), “A Dynamic Model of the Duration of theCustomer’s Relationship with a Continuous Service Provider:The Role of Satisfaction,” Marketing Science, 17 (1), 45–65.
———, P.K. Kannan, and Matthew D. Bramlett (2000), “Impli-cations of Loyalty Program Membership and Service Experi-ences for Customer Retention andValue,” Journal of the Academyof Marketing Science, 28 (1), 95–108.
——— and Katherine N. Lemon (1999), “A Dynamic Model ofCustomers’ Usage of Services: Usage as an Antecedent andConsequence of Satisfaction,” Journal of Marketing Research,36 (May), 171–86.
———, ———, and Matthew D. Bramlett (2006), “The Effect ofService Experiences over Time on a Supplier’s Retention ofBusiness Customers,” Management Science, 52 (12), 1811–23.
Borle, Sharad, Siddharth S. Singh, and Dipak C. Jain (2008),“Customer Lifetime Value Measurement,”Management Science,54 (1), 100–12.
Brown, Thomas K. (2003), “The Cross-Sell Cross-Up,” BankDirector, (1st Quarter), [available at http://www.bankdirector.com/magazine/archives/1st-quarter-2003/the-cross-sell-cross-up].
Buckinx, Wouter and Dirk Van den Poel (2005), “Customer BaseAnalysis: Partial Defection of Behaviourally Loyal Clients in aNon-Contractual FMCG Retail Setting,” European Journal ofOperational Research, 164 (1), 252–68.
Campbell, Dennis and Frances Frei (2004), “The Persistence ofCustomer Profitability: Empirical Evidence and Implicationsfrom a Financial Services Firm,” Journal of Service Research,7 (2), 107–23.
Capraro, Anthony J., Susan Broniarczyk, and Rajendra K. Srivastava(2003), “Factors Influencing the Likelihood of CustomerDefection: The Role of Consumer Knowledge,” Journal of theAcademy of Marketing Science, 31 (2), 164–75.
Chintagunta, Pradeep K. (1993), “Investigating Purchase Inci-dence, Brand Choice, and Purchase Quantity Decisions ofHouseholds,” Marketing Science, 12 (2), 184–208.
Danaher, Peter J. (2002), “Optimal Pricing of New SubscriptionServices: Analysis of a Market Experiment,”Marketing Science,21 (2), 119–38.
Donkers, Bas, Peter Verhoef, and Martijn de Jong (2007), “ModelingCLV: A Test of Competing Models in the Insurance Industry,”Quantitative Marketing and Economics, 5 (2), 163–90.
Dreze, Xavier and Andre Bonfrer (2009), “Moving from CustomerLifetime Value to Customer Equity,” Quantitative Marketingand Economics, 7 (3), 289–320.
Dwyer, Robert F. (1997), “Customer Lifetime Valuation to SupportMarketing Decision Making,” Journal of Interactive Marketing,11 (4), 6–13.
Epp, Amber M. and Linda L. Price (2008), “Family Identity: AFramework of Identity Interplay in Consumption Practices,”Journal of Consumer Research, 35 (1), 50–70.
Fader, Peter and Bruce Hardie (2007), “How to Project CustomerRetention,” Journal of Interactive Marketing, 21 (1), 76–90.
——— and ——— (2010), “Customer-Base Valuation in a Con-tractual Setting: The Perils of Ignoring Heterogeneity,” Mar-keting Science, 29 (1), 85–93.
———, ———, and Kinshuk Jerath (2007), “Estimating CLVUsing Aggregated Data: The Tuscan Lifestyles Case Revisited,”Journal of Interactive Marketing, 21 (3), 55–71.
———, ———, and Ka Lok Lee (2005), “‘Counting Your Cus-tomers’ the Easy Way: An Alternative to the Pareto/NBDModel,” Marketing Science, 24 (2), 275–84.
Fang, Eric (2008), “Customer Participation and the Trade-OffBetween New Product Innovativeness and Speed to Market,”Journal of Marketing, 72 (July), 90–104.
Fornell, Claes, Sunil Mithas, Forrest Morgeson, and M.S. Krishnan(2006), “Customer Satisfaction and Stock Prices: High Returns,Low Risk,” Journal of Marketing, 70 (January), 3–14.
Fuller, Johann, Kurt Matzler, and Melanie Hoppe (2008), “BrandCommunity Members as a Source of Innovation,” Journal ofProduct Innovation Management, 25 (6), 608–19.
Creating Enduring Customer Value / 65
Gale, Bradley T. (1994),Managing Customer Value. NewYork: theFree Press.
Glady, Nicolas, Bart Baesens, and Christophe Croux (2009),“Modeling Churn Using Customer Lifetime Value,” EuropeanJournal of Operational Research, 197 (1), 402–11.
Godes, David (2012), “The Strategic Impact of References inBusiness Markets,” Marketing Science, 31 (2), 257–76.
Golder, Peter N., Debanjan Mitra, and Christine Moorman (2012),“What Is Quality? An Integrative Framework of Processes andStates,” Journal of Marketing, 76 (July), 1–23.
Griffin, Jill andMichaelW. Lowenstein (2002),Customer Winback:How to Recapture Lost Customers—and Keep Them Loyal. NewYork: John Wiley & Sons.
Gruca, Thomas S. and Lopo L. Rego (2005), “Customer Sat-isfaction, Cash Flow, and Shareholder Value,” Journal of Mar-keting, 69 (July), 115–30.
Gupta, Sunil, Michael R. Hagerty, and John G. Myers (1983), “NewDirections in Family Decision Making Research,” in Advancesin Consumer Research, Vol. 10, Richard P. Bagozzi and AliceM. Tybout, eds. Ann Arbor, MI: Association for ConsumerResearch, 445–50.
———, Dominique Hanssens, Bruce Hardie, William Kahn, V.Kumar, Nathaniel Lin, et al. (2006), “Modeling CustomerLifetime Value,” Journal of Service Research, 9 (2), 139–55.
———, ———, and Jennifer Ames Stuart (2004), “Valuing Cus-tomers,” Journal of Marketing Research, 41 (February), 7–18.
Gutman, Jonathan (1982), “A Means-End Chain Model Based onConsumer Categorization Processes,” Journal of Marketing,46 (April), 60–72.
Hada, Mahima, Rajdeep Grewal, and Gary L. Lilien (2010), “Re-ferral Equity and Referral Management: The Supplier Firm’sPerspective,” Review of Marketing Research, 7, 93–144.
———,———, and——— (2014), “Supplier-Selected Referrals,”Journal of Marketing, 78 (March), 34–51.
Hansotia, Behram J. and Paul Wang (1997), “Analytical ChallengesinCustomerAcquisition,” Journal of InteractiveMarketing, 11 (2),7–19.
Hauser, John R. and Don Clausing (1988), “The House of Quality,”Harvard Business Review, 66 (3), 63–73.
——— and Steven M. Shugan (1983), “Defensive MarketingStrategies,” Marketing Science, 2 (4), 319–60.
——— and Glen L. Urban (1986), “The Value Priority Hypothesesfor Consumer Budget Plans,” Journal of Consumer Research,12 (4), 446–62.
Hogan, John E., Katherine N. Lemon, and Barak Libai (2003),“What Is the True Value of a Lost Customer?” Journal of ServiceResearch, 5 (3), 196–208.
———, ———, and ——— (2004), “Quantifying the Ripple:Word-of-Mouth and Advertising Effectiveness,” Journal of Ad-vertising Research, 44 (3), 271–80.
Holbrook, Morris B. (1994), “The Nature of Customer Value: AnAxiology of Services in the Consumption Experience,” in ServiceQuality: New Directions in Theory and Practice, Roland T. Rustand Richard L. Oliver, eds. Thousand Oaks, CA: Sage Pub-lications, 21–71.
Homburg, Christian, Wayne D. Hoyer, and Ruth Maria Stock(2007), “How to Get Lost Customers Back?” Journal of theAcademy of Marketing Science, 35 (4), 461–74.
Howard, John A. (1977), Consumer Behavior: Application ofTheory. New York: McGraw-Hill.
Huber, Joel, John W. Payne, and Christopher Puto (1982), “AddingAsymmetrically Dominated Alternatives: Violations of Regu-larity and the Similarity Hypothesis,” Journal of ConsumerResearch, 9 (1), 90–98.
Jain, Dipak and Siddhartha S. Singh (2002), “Customer LifetimeValue Research in Marketing: A Review and Future Directions,”Journal of Interactive Marketing, 16 (2), 34–46.
Jamal, Zainab and Randolph E. Bucklin (2006), “Improving theDiagnosis and Prediction of Customer Churn: A HeterogeneousHazard Modeling Approach,” Journal of Interactive Marketing,20 (3/4), 16–29.
Kim, Jaehwan, Greg M. Allenby, and Peter E. Rossi (2002),“Modeling Consumer Demand for Variety,”Marketing Science,21 (3), 229–50.
Kim, Ju-Young, Martin Natter, and Martin Spann (2009), “PayWhat You Want: A New Participative Pricing Mechanism,”Journal of Marketing, 73 (January), 44–58.
Koukova, Nevena, P.K. Kannan, and Amna Kirmani (2012),“Multiformat Digital Products: How Design Attributes Interactwith Usage Situations to Determine Choice,” Journal of Mar-keting Research, 49 (February), 100–15.
Kumar, Minu (2015), “Aesthetic Principles of Product Form andCognitive Appraisals,” in The Psychology of Design: CreatingConsumer Appeal, R. Batra, C. Seifert, and D. Brei, eds. NewYork: Routledge, 234–51.
Kumar, V. (2008), Managing Customers for Profit: Strategies toIncrease Profits and Build Loyalty. Upper Saddle River, NJ:Wharton School Publishing.
——— (2010), “A Customer Lifetime Value-Based Approach toMarketing in the Multichannel, Multimedia Retailing Environ-ment,” Journal of Interactive Marketing, 24 (2), 71–85.
———, Lerzan Aksoy, Bas Donkers, Rajkumar Venkatesan,Thorsten Wiesel, and Sebastian Tillmanns (2010), “Under-valued or Overvalued Customers: Capturing Total CustomerEngagement Value,” Journal of Service Research, 13 (3),297–310.
———, Yashoda Bhagwat, and Xi Zhang (2015), “Regaining‘Lost’ Customers: The Predictive Power of First-LifetimeBehavior, the Reason for Defection, and the Nature of theWin-Back Offer,” Journal of Marketing, 79 (July), 34–55.
———, Vikram Bhaskaran, Rohan Mirchandani, and Milap Shah(2013), “Creating a Measurable Social Media Marketing Strat-egy: Increasing the Value and ROI of Intangibles and Tangiblesfor Hokey Pokey,” Marketing Science, 32 (2), 194–212.
———, Morris George, and Robert P. Leone (2007), “How Valu-able Is Word of Mouth?” Harvard Business Review, 85 (10),139–46.
———, ———, and ——— (2010), “Driving Profitability byEncouraging Customer Referrals: Who, When, and How,”Journal of Marketing, 74 (September), 1–17.
———, ———, and ——— (2013), “Defining, Measuring, andManaging Business Reference Value,” Journal of Marketing,77 (January), 68–86.
———, ———, and Joseph Pancras (2008), “Cross-Buying inRetailing: Drivers and Consequences,” Journal of Retailing,84 (1), 15–27.
——— and Anita Pansari (2016), “National Culture, Economy, andCustomer Lifetime Value: Assessing the Relative Impact of theDrivers of Customer Lifetime Value for a Global Retailer,”Journal of International Marketing, 24 (1), 1–21.
——— and Werner Reinartz (2012), Customer Relationship Man-agement: Concept, Strategy, and Tools. Heidelberg, Germany:Springer.
——— and Denish Shah (2009), “Expanding the Role of Mar-keting: From Customer Equity to Market Capitalization,”Journal of Marketing, 73 (November), 119–36.
66 / Journal of Marketing: AMA/MSI Special Issue, November 2016
———, Rajkumar Venkatesan, Tim Bohling, and DeniseBeckmann (2008), “The Power of CLV: Managing CustomerLifetime Value at IBM,” Marketing Science, 27 (4), 585–99.
Lazear, E.P. and R.T. Michael (1980), “Family Size and the Dis-tribution of Real Per Capita Income,” American Economic Re-view, 70, 91–107.
Lee, Moo Yeob, Ann Sung Lee, and So Young Sohn (2013),“Behavior Scoring Model for Coalition Loyalty Programs byUsing Summary Variables of Transaction Data,” Expert Systemswith Applications, 40 (5), 1564–70.
Leenheer, Jorna, Harald J. Van Heerde, Tammo H.A. Bijmolt, andAle Smidts (2007), “Do Loyalty Programs Really EnhanceBehavioral Loyalty? An Empirical Analysis Accounting forSelf-Selecting Members,” International Journal of Research inMarketing, 24 (1), 31–47.
Lemmens, Aurelie and Christophe Croux (2006), “Bagging andBoosting Classification Trees to Predict Churn,” Journal ofMarketing Research, 43 (May), 276–86.
Lemon, Katherine N. and Florian von Wangenheim (2009), “TheReinforcing Effects of Loyalty Program Partnerships and CoreService Usage: A Longitudinal Analysis,” Journal of ServiceResearch, 11 (4), 357–70.
———, Tiffany White, and Russell Winer (2002), “DynamicCustomer Relationship Management: Incorporating FutureConsiderations into the Service Retention Decision,” Journal ofMarketing, 66 (January), 1–14.
Lewis, Michael (2006a), “Customer Acquisition Promotions andCustomer Asset Value,” Journal of Marketing Research,43 (May), 195–203.
——— (2006b), “The Effect of Shipping Fees on CustomerAcquisition, Customer Retention, and Purchase Quantities,”Journal of Retailing, 82 (1), 13–23.
Libai, Barak, Ruth Bolton, Marnix S. Bugel, Ko De Ruyter,Oliver Gotz, Hans Risselada, et al. (2010), “Customer-to-Customer Interactions: Broadening the Scope of Word ofMouth Research,” Journal of Service Research, 13 (3),267–82.
———, Das Narayandas, and Clive Humby (2002), “Toward anIndividual Customer Profitability Model,” Journal of ServiceResearch, 5 (1), 69–76.
Lix, Thomas S., Paul D. Berger, and Thomas L. Magliozzi (1995),“New Customer Acquisition: Prospecting Models and the Useof Commercially Available External Data,” Journal of DirectMarketing, 9 (4), 8–18.
Manchanda, Puneet, Asim Ansari, and Sunil Gupta (1999), “The‘Shopping Basket’: A Model for Multicategory Purchase Inci-dence Decisions,” Marketing Science, 18 (2), 95–114.
Malshe, Ashwin and Manoj K. Agarwal (2015), “From Finance toMarketing: The Impact of Financial Leverage on CustomerSatisfaction,” Journal of Marketing, 79 (September), 21–38.
Malthouse, Edward C. and Robert C. Blattberg (2005), “Can WePredict Customer Lifetime Value?” Journal of Interactive Mar-keting, 19 (1), 2–16.
Meyer-Waarden, Lars (2007), “The Effects of Loyalty Programs onCustomer Lifetime Duration and Share of Wallet,” Journal ofRetailing, 83 (2), 223–36.
Monroe, Kent B. (1971), “Psychophysics of Prices: A Reappraisal,”Journal of Marketing Research, 8 (May), 248–50.
Niraj, Rakesh, Mahendra Gupta, and Chakravarthi Narasimhan(2001), “Customer Profitability in a Supply Chain,” Journal ofMarketing, 65 (July), 1–16.
Palmatier, Robert W., Rajiv P. Dant, Dhruv Grewal, and Kenneth R.Evans (2006), “Factors Influencing the Effectiveness of
Relationship Marketing: A Meta-Analysis,” Journal of Mar-keting, 70 (October), 136–53.
Pfeifer, Phillip E. (2011), “On Estimating Current-Customer EquityUsing Company Summary Data,” Journal of Interactive Mar-keting, 25 (1), 1–14.
——— and Robert L. Carraway (2000), “Modeling CustomerRelationships as Markov Chains,” Journal of Interactive Mar-keting, 14 (2), 43–55.
Power, Christopher (1991), “Value Marketing,” Bloomberg.com,(November 10), (accessed June 8, 2016), [available at http://www.bloomberg.com/news/articles/1991-11-10/value-marketing].
Rao, Vithala R. (2014), “Conjoint Analysis,” in The History ofMarketing Science, Russell S. Winer and Scott A. Neslin, eds.Singapore: World Scientific Publishing, 47–76.
Reinartz, Werner and V. Kumar (2000), “On the Profitability ofLong-Life Customers in a Noncontractual Setting: An Empiri-cal Investigation and Implications for Marketing,” Journal ofMarketing, 64 (October), 17–35.
——— and ——— (2002), “The Mismanagement of CustomerLoyalty,” Harvard Business Review, 80 (7), 86–94.
——— and——— (2003), “The Impact of Customer RelationshipCharacteristics on Profitable Lifetime Duration,” Journal of Mar-keting, 67 (January), 77–99.
———, Jacquelyn S. Thomas, and Ganael Bascoul (2008), “In-vestigating Cross-Buying and Customer Loyalty,” Journal ofInteractive Marketing, 22 (1), 5–20.
———,———, and V. Kumar (2005), “Balancing Acquisition andRetention Resources to Maximize Customer Profitability,”Journal of Marketing, 69 (January), 63–79.
Robins, Garry, Tom Snijders, Peng Wang, Mark Handcock, andPhilippa Pattison (2007), “Recent Developments in ExponentialRandom Graph (p*) Models For Social Networks,” SocialNetworks, 29 (2), 192–215.
Rust, Roland T., P.K. Kannan, and Na Peng (2002), “The CustomerEconomics of Internet Privacy,” Journal of the Academy ofMarketing Science, 30 (4), 451–60.
———, V. Kumar, and Rajkumar Venkatesan (2011), “Will theFrog Change into a Prince? Predicting Future Customer Prof-itability,” International Journal of Research in Marketing,28 (4), 281–94.
———, Katherine N. Lemon, and Valarie A. Zeithaml (2004),“Return on Marketing: Using Customer Equity to Focus Mar-keting Strategy,” Journal of Marketing, 68 (January), 109–27.
———, Valarie A. Zeithaml, and Katherine N. Lemon (2001),Driving Customer Equity: How Customer Lifetime Value IsReshaping Corporate Strategy. New York: Simon & Schuster.
Ryu, Gangseog and Lawrence Feick (2007), “A Penny for YourThoughts: Referral Reward Programs and Referral Likelihood,”Journal of Marketing, 71 (January), 84–94.
Sawyer, Alan G. and Peter R. Dickson (1984), “Psychologicalperspectives on Consumer Response to sales Promotion,” inResearch on Sales Promotion: Collected Papers, Katherine Jocz,ed. Cambridge, MA: Marketing Science Institute, 1–21.
Schmitt, Philipp, Bernd Skiera, and Christophe Van den Bulte(2011), “Referral programs and Customer Value,” Journal ofMarketing, 75 (January), 46–59.
Schmittlein, David C., Donald G. Morrison, and Richard Colombo(1987), “Counting Your Customers: Who Are They and WhatWill They Do Next?” Management Science, 33 (1), 1–24.
Schumann, Jan, Nancy Wunderlich, and Heiner Evanschitzky(2014), “Spillover Effects of Service Failures in CoalitionLoyalty Programs: The Buffering Effect of Special TreatmentBenefits,” Journal of Retailing, 90 (1), 111–18.
Creating Enduring Customer Value / 67
Schweidel, David A., Peter Fader, and Eric Bradlow (2008),“Understanding Service Retention Within and Across CohortsUsing Limited Information,” Journal of Marketing, 72 (January),82–94.
Shah, Denish, V. Kumar, Yingge Qu, and Sylia Chen (2012),“Unprofitable Cross-Buying: Evidence from Consumer andBusiness Markets,” Journal of Marketing, 76 (May), 78–95.
Simonson, Itamar (2015), “Mission (Largely) Accomplished: What’sNext for Consumer BDT-JDM Researchers?” Journal of Market-ing Behavior, 1 (1), 9–35.
Sinha, Indrajit and Wayne S. DeSarbo (1998), “An IntegratedApproach Toward the Spatial Modeling of Perceived CustomerValue,” Journal of Marketing Research, 35 (May), 236–49.
Stauss, Bernd and Christian Friege (1999), “Regaining ServiceCustomers: Costs and Benefits of RegainManagement,” Journalof Service Research, 1 (4), 347–61.
Sunder, Sarang, V. Kumar, and Yi Zhao (2016), “Measuring theLifetime Value of a Customer in the Consumer Packaged Goods(CPG) Industry,” Journal of Marketing Research, (publishedelectronically February 25), [DOI: 10.1509/jmr.14.0641].
Sweeney, Jillian C. and Geoffrey N. Soutar (2001), “ConsumerPerceived Value: The Development of a Multiple Item Scale,”Journal of Retailing, 77 (2), 203–20.
Teas, R. Kenneth and Sanjeev Agarwal (2000), “The Effects ofExtrinsic Product Cues on Consumers’ Perceptions of Quality,Sacrifice, and Value,” Journal of the Academy of MarketingScience, 28 (2), 278–90.
Thomas, Jacquelyn S. (2001), “A Methodology for Linking Cus-tomer Acquisition to Customer Retention,” Journal of MarketingResearch, 38 (May), 262–68.
———, Robert C. Blattberg, and Edward J. Fox (2004), “Recap-turing Lost Customers,” Journal of Marketing Research, 41(February), 31–45.
———, Werner Reinartz, and V. Kumar (2004), “Getting the MostOut of All Your Customers,”Harvard Business Review, 82 (7/8),116–23.
Tokman, Mert, Lenita M. Davis, and Katherine N. Lemon (2007),“TheWOWFactor: Creating Value ThroughWin-BackOffers toReacquire Lost Customers,” Journal of Retailing, 83 (1), 47–64.
Toubia, Olivier, John R. Hauser, and Duncan I. Simester (2004),“Polyhedral Methods for Adaptive Choice-Based ConjointAnalysis,” Journal of Marketing Research, 41 (February), 116–31.
Trusov, Michael, Randolph E. Bucklin, and Koen Pauwels (2009),“Effects of Word-of-Mouth Versus Traditional Marketing:Findings from an Internet Social Networking Site,” Journalof Marketing, 73 (September), 90–102.
Ulaga, Wolfgang and Andreas Eggert (2006), “Value-Based Differ-entiation in Business Relationships: Gaining and Sustaining KeySupplier Status,” Journal of Marketing, 70 (January), 119–36.
——— and Werner J. Reinartz (2011), “Hybrid Offerings: HowManufacturing Firms Combine Goods and Services Success-fully,” Journal of Marketing, 75 (November), 5–23.
Van den Poel, Dirk and Bart Lariviere (2004), “Customer AttritionAnalysis for Financial Services Using Proportional Hazard Mod-els,”European Journal ofOperational Research, 157 (1), 196–217.
Vargo, Stephen L. and Robert F. Lusch (2004), “Evolving to aNew Dominant Logic for Marketing,” Journal of Marketing,68 (January), 1–17.
Venkatesan, Rajkumar and V. Kumar (2004), “ACustomer LifetimeValue Framework for Customer Selection and Resource Allo-cation Strategy,” Journal of Marketing, 68 (October), 106.
———,———, and Timothy Bohling (2007), “Optimal CustomerRelationship Management Using Bayesian Decision Theory:An Application for Customer Selection,” Journal of MarketingResearch, 44 (November), 579–94.
Verhoef, Peter (2003), “Understanding the Effect of CustomerRelationship Management Efforts on Customer Retentionand Customer Share Development,” Journal of Marketing,67 (October), 30–45.
——— and Bas Donkers (2005), “The Effect of AcquisitionChannels on Customer Loyalty and Cross-Buying,” Journal ofInteractive Marketing, 19 (2), 31–43.
Villanueva, Julian, Shijin Yoo, andDominiqueM. Hanssens (2008),“The Impact of Marketing-Induced Versus Word-of-MouthCustomer Acquisition on Customer Equity Growth,” Journalof Marketing Research, 45 (February), 48–59.
Von Hippel, Eric (1986), “Lead Users: A Source of Novel ProductConcepts,” Management Science, 32 (7), 791–805.
Von Wangenheim, Florian and Tomas Bayon (2007), “The Chainfrom Customer Satisfaction via Word-of-Mouth Referrals toNew Customer Acquisition,” Journal of the Academy of Mar-keting Science, 35 (2), 233–49.
Wang, Tuo, Ramaswamy Venkatesh, and Rabikar Chatterjee(2007), “Reservation Price as a Range: An Incentive-CompatibleMeasurement Approach,” Journal of Marketing Research,44 (May), 200–13.
Wilson, D.T. (1995), “An Integrated Model of Buyer-Seller Rela-tionships,” Journal of the Academy of Marketing Science, 23 (4),335–45.
Young, Shirley and Barbara Feigin (1975), “Using the BenefitChain for Improved Strategy Formulation,” Journal of Mar-keting, 39 (July), 72–74.
Zeithaml, Valarie A. (1988), “Consumer Perceptions of Price,Quality, and Value: A Means-End Model and Synthesis ofEvidence,” Journal of Marketing, 52 (July), 2–22.
68 / Journal of Marketing: AMA/MSI Special Issue, November 2016