Escolar Documentos
Profissional Documentos
Cultura Documentos
Kumar
easuring, managing, and maximizing customer Mantrala (2002) points out that Blattberg and
Journal of Marketing
Vol. 69 (January 2005), 63–79 Acquisition and Retention Resources / 63
across communication channels, substantively our research how to invest at different points of the customer–firm
is similar to that of Venkatesan and Kumar. However, our relationship;
research takes a more longitudinal perspective and exam- 2. Illustrate the application of the statistical model with an
ines resource allocations more comprehensively because it empirical example; and
begins before the successful acquisition of a customer. We 3. Show by a simulation how varying different inputs to the
depict a conceptual view of our perspective in Figure 1. model (e.g., expenditures, number of communication con-
tacts) affects acquisition rates, retention rates, customer
A natural extension to the contributions made by the profitability, and the magnitude of the firm’s return on
previous studies is a conceptual framework and model that investment.
can be used to balance resource allocations between cus-
tomer acquisition and retention and whose objective is to The context in which we address the issue of balancing
maximize the firm’s long-term profitability. It is important resource allocations is customer contact strategies. How-
that the model be flexible enough to address simultaneously ever, the framework that we provide can extend beyond the
how much and how to invest. This study is that extension. customer contact strategy. With the rise of the Internet and
Specifically, Figure 1 conceptually depicts the key pro- electronic technology, the question of how firms should
cesses in the evolution of the customer–firm relationship. In interact with their customers is gaining in importance, espe-
addition, our study presents a comprehensive system of cially as firms consider the cost differences between tradi-
equations that links the acquisition and retention processes tional communications media, such as television and sales
to customer profitability and can be used for resource allo- forces, and electronic media, such as the Web and e-mail.
cation decisions. Because of the linkage, the system can be
used to assess the trade-offs that occur in resource alloca- Allocating Resources to Customer
tion decisions. It is important to acknowledge that our
approach necessitates that resources can be split between Contact Channels
customer acquisition and retention, which is the case in To answer the questions of how much to spend on cus-
many business-to-business (B-to-B) or direct marketing tomers and how to allocate expenditures, we must under-
contexts. stand the key drivers of customer profitability (for a review,
Based on the existing research, our specific objectives see Berger et al. 2002).1 Given our data, the specific drivers
in this article are to 1We use the terms “customer profitability” and “customer value
1. Present a resource allocation model that addresses the ques- (to the firm)” interchangeably in this research. Both expressions
tions of how much to invest in customer relationships and represent a multiperiod measure of the economic value of a cus-
FIGURE 1
Linking Customer Acquisition, Relationship Duration, and Customer Profitability
Relationship
duration
Acquired Customer
customers profitability
Prospects
Nonacquired
customers
•Firm actions
•Customer actions
•Competitor actions
•Customer characteristics
FIGURE 2
Determinants of Focal Constructs
TABLE 1
Descriptive Statistics
Independent Variables
Acquisition dollars per person 000,581 00,101 ,,508 ,77
Acquisition dollars2 342,118 10,180 ,269,496 5,829
Retention dollars per person ,,1506 ,209
Retention dollars2 2,281,741 43,693
Contact Channel**
Telephone 3.1 1.8 13.4 3.6
Face-to-face .1 .1 2.4 .6
Web .2 .1 1.2 .4
E-mail 2.7 2.1 15.1 4.2
Telephone × e-mail .6 .2 3.1 .4
Face-to-face × e-mail .02 .005 .6 .1
Cross-buying 4.2 1.8
Frequency 8.8 3.1
Frequency2 79.87 72.8
Share-of-wallet 41.6 8.2
Lambda .48–.71 range .2
Firmographics
Industry type (B-to-B) 56% 00.5
Annual sales revenue 40.2 39.6
Size of firm (employees) 212.0 96.0
Dependent Variables
Duration (days) ,1380 ,141
Predicted duration (days) ,1339 ,133
Lifetime values ($) 356,280 28,712
Contact Channel
Firm-initiated:
Telephone .328**
Face-to-face .381**a
E-mail .152*a,b
Telephone × e-mail .093**a,b,c
Face-to-face × e-mail .077**a,b,c
Customer-initiated:
Web .386**
Frequency .417**
Frequency2 –.079*
Cross-buying .288**
Share-of-wallet .335**
Lambda .299**
*p < .10.
**p < .05.
aThe coefficient is significantly different (at p < .05) from that of the telephone contact channel.
bThe coefficient is significantly different (at p < .05) from that of the face-to-face contact channel.
cThe coefficient is significantly different (at p < .05) from that of the e-mail contact channel.
expenditure, firm-initiated contacts, and remaining control equations for acquisition, duration, and customer profitabil-
variables, respectively. ity (i.e., Equations 1–3).
TABLE 4
Simulation Results
Scenario 1 Scenario 2 Scenario 3 Scenario 4
Maximize Customer Maximize Relationship
Profitability by Maximize Acquisition Duration by
Optimizing Maximize Customer Profitability by Optimizing Rate by Optimizing Optimizing Retention
Objective Spending Total Contacts Acquisition Contacts Contacts
Spending Spending Spending
Fixed at Fixed at Fixed at
Constraints Means Means Means
Total Total Total Total Acquisition Retention Acquisition Acquisition Retention Retention
Number of Contacts Contacts Effort Contacts Effort Effort Effort Contacts Effort Contacts Effort
Telephone 4.38 11.0% 27.80 43.9% 50.9% 42.1% 4.05 24.9% 21.37 41.5%
Face-to-face .60 1.5% .87 1.4% .3% 1.6% .50 3.1% 1.54 3.0%
Web 2.80 7.1% 2.80 4.4% 3.7% 4.6% .60 3.7% 2.33 4.5%
E-mail 31.90 80.4% 31.90 50.3% 45.1% 51.6% 11.10 68.3% 26.20 50.9%
Telephone × e-mail 1.60 1.60 1.40 1.29
Face-to-face × e-mail .40 .40 .04 .39
Total contacts 39.68 63.37 16.25 51.44
Notes: All numbers are per customer.
cross-buying, frequency, and firmographics) are at their From the cost effectiveness considerations of the e-mail
mean. communication channel, an obvious question that arises is,
To be consistent with the model variables, it is important To what extent should e-mail be used in conjunction with
for this analysis (and all other scenarios) that the expendi- other modes of communication? We address this question in
tures are directly linked to the number and type of contacts. our analysis through the interaction terms. The interaction
Thus, in each simulation, the total expenditure (acquisition terms measure the number of times that any two communi-
plus retention) equals the total cost of contacting the cus- cation modes should occur in the same month but do not
tomer through the various modes of communication. This represent an increase in the total number of communication
necessary condition still allows the allocation across chan- contacts. Thus, in the optimal scenario, we find that when
nels to be different for the same levels of expenditure. telephone communications are used, an e-mail should
As a comparison to Scenario 1, we simulated Scenario 2, accompany it during that same period on 1.6 occasions. In
in which we maximized customer profitability but allowed other words, 37% of the telephone contacts should be
the number of contacts in each channel and the degree of accompanied by an e-mail. With regard to face-to-face com-
simultaneous usage to vary. In Scenario 2, the expenditures munication, the optimal scenario is to accompany the face-
and all the other variables are fixed at their means. We pre- to-face interaction with an e-mail interaction in the same
sent the results for the two scenarios in Table 4. period 67% of the time.
Scenario 1 shows that the profit-maximizing strategy for Although the firm cannot control the degree of
the firm is to invest 78.9% of its budget in retention and customer-initiated interactions, it can create opportunities to
21.1% in acquisition. This allocation is slightly different encourage such communications, and the Web is a tool for
from the firm’s current mean allocation of 74.8% for reten- doing that. The simulation suggests that 7.1% of the firm’s
tion and 25.2% for acquisition (Scenario 2). However, the resources should be directed toward communicating with
dramatic difference between the profit-maximizing solution customers through the Web channel.
(Scenario 1) and the mean level of spending (Scenario 2) is As a contrast to the profit-maximizing effort and expen-
in the amount spent. Specifically, the analysis shows that, diture allocation (Scenario 3), we also examine the profit-
assuming an optimal allocation of contacts across channels maximizing communication strategy given a mean level of
in both scenarios, a 68.31% decrease in spending from the expenditures (Scenario 2). At the mean level of expenditure,
mean level increases the cumulative profitability of a cus- which is higher, the profit-maximizing contact strategy is to
tomer by 41.52%. Thus, currently the firm is overspending increase the total number of contacts by nearly 60% (from
on its customers. This overspending apparently increases 40 to 63). The distribution of the contacts is split more
the customer acquisition rate (.2234 in Scenario 1 versus evenly, with 50.3% directed toward e-mail communication
.2579 in Scenario 2) and the expected relationship duration and 43.9% directed toward telephone communication.
(1490 days in Scenario 1 versus 1514 days in Scenario 2). Compared with the e-mail and telephone channels, the
However, the long-term profitability of a customer is emphasis on Web and face-to-face channels is minimal.
declining, which suggests that overspending can result in Comparing the differences in budget and channel
inefficient resource allocation and, more important, a nega- emphasis across Scenarios 1 and 2, we conclude that over-
tive return on investment (ROI). We provide a more detailed spending tends to lead to greater investments in more costly
discussion of ROI subsequently. From a measurement per- and interpersonal communications that do not pay off in
spective, this result supports the claim that firms should not terms of customer profitability. Firms can use this key prac-
use a single measure to assess their performance. Measur- tical insight when faced with the need for cutbacks on
ing acquisition rates, retention rates, and long-term prof- expenditures.
itability is important for understanding customer behavior
and for accurately identifying potential problems in the
Which Is More Critical: Acquisition Expenditures
firm’s customer management practices.
or Retention Expenditures?
In terms of the effort allocation and number of contacts,
the profit-maximizing solution (Scenario 1) suggests that The goal of this analysis is to demonstrate the impact of not
the dominant form of communication should be e-mail optimizing acquisition expenditures versus not optimizing
(80.4% of the total effort). Telephone communication is the retention expenditures. To demonstrate this impact, we
second most impactful contact and should receive 11.0% of focus on customer profitability. We predict profits using
the total effort allocation. From the parameter estimates unstandardized estimates and the model represented by
alone, this finding may appear erroneous because e-mail Equations 1–3. As in the previous part of the simulation,
communications are the least effective in terms of acquisi- before predicting profits, we estimate and use the lambda
tion, duration, and profitability. A rationale for this effect and relationship duration in the profit prediction, and we fix
may be that budget constraints, a reality for almost all firms, all other necessary but nonrelevant variables at their means.
factor into the simulation but are not taken into account To mimic the economic environment in which firms
when the parameters are determined from the estimation. In reevaluate their budgets, we allow the expenditure change
the simulation, we allow expenditures and/or the number of to be either 10% or 25% and be driven by a change in either
contacts to vary within the range of the data used to esti- the acquisition or the retention budgets. In Table 5, we show
mate the model. With budget constraints and in these condi- how the per-customer profitability changes at different
tions, cost effectiveness, not just efficiency, must be a con- expenditure levels. For reference, in Table 6 we report the
sideration and compared across alternatives. total expenditure change as acquisition and retention expen-
TABLE 5
Long-Term Customer Profit Predictions (Percentage Change from Optimal Profits)
Retention Expenditures
Acquisition Expenditures
.75 × optimal $698,676 (–7.35%) $752,459 (–.22%) $752,859 (–.16%) $752,459 (–.22%) $750,359 (–.49%) Diagonal 1
.9 × optimal $727,163 (–3.57%) $753,491 (–.08%) $753,891 (–.03%) $753,491 (–.08%) $751,392 (–.36%)
Optimal $726,273 (–3.69%) $753,688 (–.05%) $754,088 $753,688 (–.05%) $752,588 (–.33%)
1.1 × optimal $736,623 (–2.32%) $753,491 (–.08%) $753,891 (–.03%) $753,491 (–.08%) $751,391 (–.36%)
1.25 × optimal $750,358 (–.49%) $752,458 (–.22%) $752,858 (–.16%) $752,458 (–.22%) $750,358 (–.49%) Diagonal 2
Notes: Predictions assume an optimal communication allocation. All numbers are per customer.
TABLE 6
Percentage Change in Total Budget from Optimal Total Budget
Retention Expenditures
Acquisition Expenditures
.75 × optimal –25.00% –13.16% –5.27% 2.62% 14.46% Diagonal 1
.9 × optimal –21.84% –10.00% –2.11% 5.78% 17.62%
Optimal –19.73% –7.89% .00% 7.89% 19.73%
1.1 × optimal –17.62% –5.78% 2.11% 10.00% 21.84%
1.25 × optimal –14.46% –2.62% 5.27% 13.16% 25.00% Diagonal 2
diture deviations from their optimal levels.6 When interpret- retention results in an ROI of –55.29, whereas overspending
ing Table 5, it is important to recall that acquisition and by 25% results in an ROI of only –4.27.
retention expenditures represent different proportions of the The results on the diagonal of Table 7 also provide
total budget (Table 4). Thus, a 10% deviation in the acquisi- insight into the acquisition/retention trade-off. Specifically,
tion budget (which is a significantly smaller portion of the the diagonals help answer the following question: When
total budget) from its optimal level represents only a 2.11% faced with the need to make a budget change, how should
deviation from the optimal total budget. firms make that change? In most cases, firms consider this
The profit projection in Table 5 shows that the percent- question not as “either/or” but rather as how much to pull
age changes in profit are small, particularly at the level of a from or add to each budget. The diagonals provide insight
10% change in the total budget. This projection suggests into this question.
that relatively substantial deviations from optimal expendi- Diagonal 1. Diagonal 1 describes the situation in which
tures are associated with relatively modest changes in the neither the acquisition nor the retention budget is at its
dependent variable (i.e., a flat maximum). However, despite respective optima, and instead, one type of expenditure is
the modest percentage changes in profits for a single cus- increased while the other is decreased. At comparable
tomer, on an absolute basis and for multiple customers, the absolute total budget changes (see Table 6), the results in
loss in profitability is significant. For example, the smallest Table 7 show that increases in acquisition expenditures and
difference occurs when the acquisition budget is .90 of the decreases in retention expenditures (from the optimal)
optimal and the retention budget is optimal.7 At this expen- result in lower ROIs than do decreases in acquisition and
diture level, a firm with a portfolio of 200,000 customers8 increases in retention. This important outcome is not obvi-
will save approximately $2.69 million from reduced mar- ous from the parameter estimates because the ROI metric
keting expenditures but lose approximately $39.3 million in we used in this analysis taps further into a firm’s perfor-
long-term customer profitability because of suboptimal mance than just profitability. Here, the ROI measure
budget allocations. These figures only become more magni- focuses on the incremental change in customer value and
fied as the budgets deviate further from the optimal level. compares the size of that change for different levels of
There are several important insights regarding the ROI investment. Thus, consistent with our prior assertion, multi-
when budgets deviate from the optimal level. To compute ple metrics are useful for assessing firm performance.
the ROI, we use the approach described by Rust, Lemon, Diagonal 2. The highlighted Diagonal 2 provides
and Zeithaml (2004), which focuses on the size of the insight into the impact of jointly increasing or decreasing
investment and the change in customer equity that results the acquisition and retention budgets. At comparable
from the investment. In this context, we measure the change absolute changes in the total budget, the results indicate that
in customer equity as the deviation in the long-term prof- underspending decreases the ROI more than does over-
itability from the optimal level. Thus, the change in cus- spending. This effect becomes significantly more magnified
tomer equity and the resultant ROI calculations are nega- at greater deviations from the optimal expenditure levels.
tive. We report the exact numbers in Table 7. The first This result has important implications for customer portfo-
insight from the ROI analysis pertains to acquisition lio management because it suggests that firms must invest
spending. sufficiently to acquire and maintain relationships. Although
Acquisition Spending. Under the assumption that reten- the investment in the customer may become less efficient
tion spending is optimal and contacts are optimally allo- beyond the optimal expenditure level, the decreased effi-
cated, the misallocation (i.e., deviation from the optimal ciency does not outweigh the lost value that the firm would
acquisition expenditure) is asymmetric; underspending on incur if the customer’s potential was not fully realized.
acquisition is worse than overspending on acquisition by
the same amount. For example, overspending the optimal Does the Optimal Contact Strategy Maximize
acquisition budget by 25% results in an ROI of –2.83, Customer Profits, Acquisition Likelihood, and
whereas underspending by 25% results in an ROI of –3.03. Relationship Duration?
A similar insight can be drawn from retention spending. Intuitively, an optimal strategy should maximize the cus-
Retention Spending. Under the assumption that acquisi- tomer’s profitability, the acquisition likelihood, and the
tion spending is optimal and contacts are optimally allo- relationship duration. However, the relative magnitude of
cated, overspending on retention is better than underspend- the parameter estimates across the three equations suggests
ing on retention by the same amount. This result is most that this is not the case. To investigate this proposition, we
significant at the 25% level. At this level, underspending on vary the number of contacts and the interaction level
between the contacts but fix the other variables (including
6In Tables 5 and 6, we round all dollar amounts to the nearest expenditures) at their means. Then, we use the estimates
whole number and all percentages to two decimal points. from each equation of the model to determine the number
7As a result of rounding, increasing and decreasing the acquisi- and distribution of contacts that would maximize customer
tion budget by 10% appears to give the same result. However, profitability (Scenario 2), the probability of acquisition
through our analysis, we find that reducing the budget results in (Scenario 3), and the length of the customer–firm relation-
profits that are slightly lower than those that result from increasing
the acquisition budget by the same amount. The absolute differ-
ship (Scenario 4). We show this comparison in Table 4.
ence between a 10% increase and a 10% decrease is $.42. In Table 4, note that the contact strategy, which maxi-
8This number is representative of the size of this firm’s cus- mizes customer profitability, results in a .26 probability of
tomer base. acquisition and a relationship duration of 1514 days. In con-
TABLE 7
Effect of Marketing Expenditures on ROI
Retention Expenditures
Acquisition Expenditures
.75 × optimal –116.74 –3.94 –3.03 –3.49 –6.10 Diagonal 1
.9 × optimal –54.97 –2.04 –1.31 –1.88 –4.59
Optimal –55.29 –1.68 n/a –1.58 –4.27
1.1 × optimal –34.21 –1.99 –1.30 –1.85 –4.47
1.25 × optimal –7.83 –3.62 –2.83 –3.26 –5.67 Diagonal 2
Notes: n/a = not applicable.
trast, the contact strategy that maximizes the acquisition which allocation scenario is the ROI more attractive (see
likelihood results in a probability of acquisition of .30. The Table 7)?
contact strategy that maximizes relationship duration results
in a duration of 1587 days. Thus, the profit-maximizing
contact strategy maximizes neither acquisition likelihood Conclusion
nor relationship duration. Because the communication bud- In this study, we investigated several drivers of customer
get is fixed in this scenario, it is appropriate to consider it in profitability and derived some implications for resource
terms of maximization, not in terms of optimization. The allocations to customers and prospects. A key assertion
maximization of acquisition and retention rates results from demonstrated in this analysis is that both the amount of
various channel usage scenarios within the budget investment and how it is invested in a customer relate
restriction. directly to the acquisition, retention, and profitability of that
A reason for this outcome could be that the profit equa- customer. Expanding on this basic principle, we arrive at
tion parameters reflect the cost effectiveness (i.e., cost– several substantive conclusions for our empirical setting.
benefit) of each contact mode, whereas the acquisition and
How Much to Invest
duration equation parameters reflect only the effectiveness
of each contact mode. This distinction leads to a difference We find that underspending is more detrimental and results
in the relative magnitude of the parameters when profit is in smaller ROIs than does overspending. When firms trade
the response objective versus when acquisition likelihood or off between expenditures for acquisition and those for
relationship duration is the response objective. Therefore, retention, a suboptimal allocation of retention expenditures
the optimal solution for acquisition or duration will not nec- will have a greater impact on long-term customer profitabil-
essarily be the optimal solution for a profit objective. ity than will suboptimal acquisition expenditures. Consis-
Another interesting finding from this analysis pertains tent with prior research (Chintagunta 1993; Tull et al.
to the strategy for allocating resources across the contact 1986), there appears to be a flat maximum with respect to
channels. Note that in Scenarios 3 and 4, it is not the total acquisition and retention expenditures. Specifically, we find
number of contacts that is determined but rather the acquisi- that a 10% deviation in either acquisition or retention
tion or retention contacts. Thus, when we compare Scenario expenditures from their respective optimal levels results in
2 with Scenarios 3 or 4, we consider the percentage of less than a 1% change in the long-term profitability of a
either the acquisition or retention effort, not the percentage customer.
of total effort.
How to Invest
Comparison of Scenario 2 (maximizing profits) and
Scenario 3 (maximizing acquisition rates) reveals a similar If the firm initiates the contact, the relative effectiveness of
but reverse emphasis. In both scenarios, telephone and e- highly interpersonal and interactive communication chan-
mail are the dominant forms of communications. Specifi- nels is greater than that of less interpersonal and interactive
cally, the results suggest that more than 93% of the acquisi- communication channels. However, this generalization does
tion effort should be directed toward these two channels. not hold for all response variables if the customer initiates
However, when cumulative profits are the objective, tele- the interaction. The customer communications strategy that
phone is more dominant than e-mail, whereas the reverse is maximizes long-term customer profitability maximizes nei-
true when acquisition is the objective. In contrast to the ther the acquisition rate nor the relationship duration.
acquisition scenario, the comparison between Scenario 2 Instead, developing a communications strategy to manage
(maximizing profits) and Scenario 4 (maximizing duration) long-term customer profitability generally requires a long-
suggests a similar optimal distribution of communications term and holistic perspective toward the relationship. This
across the channels. Specifically, the relative ordering of perspective tends to give more emphasis to more interper-
retention communication efforts across the channels is the sonal and interactive communications than does a limited
same for cumulative profit maximization and duration max- focus on acquisition.
imization. In combination, Scenarios 2, 3, and 4 suggest Each of these conclusions can be used for the strategic
that when the goal is short term, such as customer acquisi- advantage of a firm and can have a potentially large impact
tion, interpersonal interactions are not as critical. However, on the cost and/or revenue aspect of customer profitability.
for objectives managed over a longer time horizon (e.g., We acknowledge that generalizability of some of the results
cumulative profitability), interpersonal interactions aid in can be achieved only with further testing on multiple data
the achievement of the objective. sets.9 However, many of the conclusions would not have
In summary, this simulation addresses three general been revealed if our model did not integrate acquisition,
questions that a practicing manager may confront: retention, and customer profitability into a single frame-
work that accounts for the natural linkages and endogeneity
1. If marketing budgets deviate from their optimal level, what
is the impact on customer profitability (see Table 5)? 9We also estimated the model presented in this research on a
2. If managers are mandated to cut their budgets by, say, a data set for a B-to-C firm that used similar customer contact
fixed percentage, from which “bucket” should they make the modes. Although this empirical context was not as rich, the effects
cuts—acquisition, retention, or both (see Tables 5 and 6)? that are of focal interest (decreasing returns to acquisition and
3. Increasingly, marketing managers are assessed on the basis retention, flat maximum, asymmetric response to acquisition, and
of not only their profits but also their ROI. Thus, under the retention spending changes) were consistent with our original
assumption of a suboptimal allocation of the budget, in findings.
REFERENCES
Aaker, David (1975), “ADMOD: An Advertising Decision Chintagunta, Pradeep (1993), “Investigating the Sensitivity of
Model,” Journal of Marketing Research, 8 (February), 37–45. Equilibrium Profits to Advertising Dynamics and Competitive
Allenby, Greg M., Robert P. Leone, and Lichung Jen (1999), “A Effects,” Management Science, 38 (3), 287–306.
Dynamic Model of Purchase Timing with Application to Direct Dwyer, R. Robert, Paul H Schurr, and Sejo Oh (1987), “Develop-
Marketing,” Journal of American Statistical Association, 94 ing Buyer–Seller Relationships,” Journal of Marketing, 51
(June), 365–74. (April), 11–27.
Amemiya, Takeshi (1974), “Multivariate Regression and Simulta- Farris, Paul (2003), “Getting the Biggest Bang for Your Marketing
neous Equation Models When the Dependent Variables Are Buck: A Point of View,” in Measuring and Allocating Marcom
Truncated Normal,” Econometrica, 42 (6), 999–1012. Budgets: Seven Expert Points of View. Cambridge, MA: Mar-
Anderson, Erin and Bart Weitz (1992), “The Use of Pledges to keting Science Institute, 7–14.
Build and Sustain Commitment in Distribution Channels,” Hanssens, Dominique (2003), “Allocating Marketing Communica-
Journal of Marketing Research, 24 (February), 18–34. tion Expenditures: A Long Run View,” in Measuring and Allo-
Anderson, James and James A. Narus (1990), “A Model of Dis- cating Marcom Budgets: Seven Expert Points of View. Cam-
tributor Firm and Manufacturer Firm Working Relationships,” bridge, MA: Marketing Science Institute, 15–20.
Journal of Marketing, 54 (January), 42–58. Heckman, James (1976), “The Common Structure of Statistical
——— and ——— (1999), Business Market Management. Upper Models of Truncation, Sample Selection and Limited Depen-
Saddle River, NJ: Prentice Hall. dent Variables, and a Simple Estimator for Such Models,”
——— and ——— (2003), “Selectively Pursuing More of Your Annals of Economic and Social Measurement, 5, 475–92.
Customer’s Business,” Sloan Management Review, 44 (Spring), ——— (1979), “Sample Selection Bias as a Specification Error,”
42–49. Econometrica, 47 (January), 153–61.
Berger, Paul D., Ruth N. Bolton, Douglas Bowman, Elten Briggs, Jagpal, Harsharanjeet (1981), “Measuring Joint Advertising
V. Kumar, A. Parasuraman, and Terry Creed (2002), “Market- Effects in Multiproduct Firms,” Journal of Advertising
ing Actions and the Value of Customer Assets,” Journal of Ser- Research, 21 (1), 65–69.
vice Research, 5 (August), 39–54. Kalwani, Manohar and Narakesari Narayandas (1995), “Long-
——— and Nada Nasr-Bechwati (2001), “The Allocation of Pro- Term Manufacturer–Supplier Relationships: Do They Pay Off
motion Budget to Maximize Customer Equity,” The Interna- for Supplier Firms?” Journal of Marketing, 59 (January), 1–16.
tional Journal of Management Science, 29 (1), 49–61. Kamakura, Wagner and Gary Russell (1989), “A Probabilistic
Blattberg, Robert C. and John Deighton (1996), “Manage Market- Choice Model for Market Segmentation and Elasticity Struc-
ing by the Customer Equity Test,” Harvard Business Review, ture,” Journal of Marketing Research, 26 (November), 379–90.
74 (July-August), 136–44. ———, Michel Wedel, Fernando de Rosa, and Jose A. Mazzon
———, Gary Getz, and Jacquelyn S. Thomas (2001), Customer (2003), “Cross-Selling Through Database Marketing: A Mixed
Equity: Building and Managing Relationships as Valuable Data Factor Analyzer for Data Augmentation and Prediction,”
Assets. Boston: Harvard Business School Press. International Journal of Research in Marketing, 20 (March),
Bolton, Ruth (1998), “A Dynamic Model of the Duration of the 45–65.
Customer’s Relationship with a Continuous Service Provider: Krishnamurthi, Lakshman and S.P. Raj (1988), “A Model of Brand
The Role of Satisfaction,” Marketing Science, 17 (1), 45–65. Choice and Purchase Quantity Price Sensitivities,” Marketing
——— and Katherine N. Lemon (1999), “A Dynamic Model of Science, 7 (1), 1–20.
Customers’ Usage of Services: Usage as an Antecedent and Lee, Lung-Fei, G.S. Maddala, and R.P. Trost (1980), “Asymptotic
Consequence of Satisfaction,” Journal of Marketing Research, Covariance Matrices of Two-Stage Tobit Methods for Simulta-
36 (May), 171–86. neous Equations Models with Selectivity,” Econometrica, 48
———, ———, and Peter C. Verhoef (2004), “The Theoretical (2), 491–503.
Underpinnings of Customer Asset Management: A Framework Mantrala, Murali K. (2002), “Allocating Marketing Resources,” in
and Propositions for Future Research,” Journal of the Academy Handbook of Marketing, Barton Weitz and Robin Wensley, eds.
of Marketing Science, 32 (3), 271–92. London: Sage Publications, 409–35.