Content uploaded by Peter C. Verhoef
Author content
All content in this area was uploaded by Peter C. Verhoef on Aug 15, 2014
Content may be subject to copyright.
46
© 2008, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Journal of Marketing
Vol. 72 (January 2008), 46–64
Ruth N. Bolton, Katherine N. Lemon, & Peter C. Verhoef
Expanding Business-to-Business
Customer Relationships: Modeling
the Customer’s Upgrade Decision
This article develops a model of a business customer’s decision to upgrade service contracts conditional on the
decision to renew the contract. It proposes that the firm’s upgrade decision is influenced by (1) decision-maker
perceptions of the relationship with the supplier, (2) contract-level experiences, and (3) interactions between firm-
and contract-level variables. The authors model the firm’s decision as a binary logit model with random parameters
for the contract-level variables and fixed parameters for the firm-level variables. They estimate the model with data
describing more than 2000 service contracts and find that decision-maker satisfaction, service quality, and price
have a significant effect on the decision to upgrade; price and satisfaction also moderate the effect of service quality
on the decision. Simulations indicate that modest improvements in service quality for a focal contract can have a
relatively large, positive effect on the likelihood that the firm will upgrade. The results suggest that suppliers need
to manage their firm relationships at both the individual contract level and the overall firm level. In addition, the
results suggest specific windows of opportunity for suppliers when firms may be more likely to upgrade to higher-
level service contracts.
Keywords
: customer relationships, upgrades, service quality, satisfaction, retention
Ruth N. Bolton is W.P. Carey Chair and Professor of Marketing, W.P. Carey
School of Business, Arizona State University, and Honorary Professor,
University of Groningen (e-mail: ruth.bolton@asu.edu). Katherine N.
Lemon is Associate Professor of Marketing, Carroll School of Manage-
ment, Boston College (e-mail: katherine.lemon@bc.edu). Peter C. Verhoef
is Professor of Marketing, Faculty of Economics, Department of Market-
ing, University of Groningen (e-mail: p.c.verhoef@rug.nl). The authors are
listed in alphabetical order and contributed equally to the manuscript. The
Center for Customer Relationship Management at Duke University finan-
cially supported this research. The authors thank the anonymous cooper-
ating company for supplying the data. They also thank Stijn van Osselaer
for some theoretical suggestions. P. Rajan Varadarajan served as guest
editor for this article.
This article focuses on the customer’s decision to
upgrade a service contract. A product upgrade is a
form of relationship expansion in which the customer
purchases an expanded offering—a higher-price, aug-
mented good or service (with higher service levels or addi-
tional features)—instead of repurchasing a low-price good
or service (with lower service levels or fewer features) from
the same supplier. Recently, researchers have developed
predictive and normative models that are relevant to con-
sumer service upgrades in industries such as airlines, banks,
and theaters (e.g., Ngobo 2005). For example, Li, Sun, and
Wilcox (2005) model a consumer’s sequential acquisition
of banking services on the basis of “maturity” (e.g., demo-
graphics, prior purchases). We have not been able to dis-
cover any theory-based models of the factors that influence
a firm’s decision to upgrade a service contract over time.
This gap in the literature is rather remarkable because firms
frequently buy fixed-price service contracts from suppliers
in many industries, including computing, telecommunica-
tions, and information services; repair and maintenance ser-
vices for engineering, medical, and/or other equipment; and
services that support financial, health, or energy manage-
ment systems. For example, General Electric, Hewlett-
Packard, ServiceMaster, and Verizon are major suppliers of
service contracts to firms. Service contracts are typically
offered at different levels. For example, a “low-level” con-
tract may promise reactive service (e.g., repairs), whereas a
“high-level” service contract may promise proactive service
(e.g., repairs and preventative maintenance). A large firm
may purchase as many as 20–50 service contracts from a
service supplier. An upgrade may occur on a given service
contract whenever a firm has the opportunity to move (con-
tractually) from a low level of service to a higher level of
service.
Marketing academics and practitioners have long been
interested in the nature of business-to-business (B2B) rela-
tionships (e.g., Dwyer, Schurr, and Oh 1987). An early
study of B2B relationships found that the customer’s overall
assessment of the firm was a key determinant of his or her
decision to continue to conduct business with the firm
(Jackson 1985). A recent study by Cannon and Perreault
(1999) shows how business customers’ evaluations of sup-
plier performance vary across different types of relation-
ships. Research in services marketing has focused on cross-
sectional studies of consumers’ and business customers’
switching behavior (e.g., Ganesh, Arnold, and Reynolds
2000; Keaveney 1995). For example, Heide and Weiss
(1995) find that an organizational buyer’s decision to switch
to a new vendor of a high-technology product depends on
his or her perception of rapid technological change, his or
her prior experiences with vendors, and buying process for-
malization. However, we could find no studies that examine
Expanding B2B Customer Relationships / 47
1Firms may also decide to “downgrade” or purchase a lower
level of service; we find no incidence of downgrading in this con-
text and therefore reserve the issues associated with downgrading
for further research.
the antecedents of B2B customer upgrade behavior (for an
overview of prior research in this area, see Table 1).
We address this gap in the literature by developing a
model of how firms make upgrade decisions for fixed-price
service contracts. We propose that the firm’s decision to
upgrade is influenced by the decision maker’s perceptions
of the supplier (i.e., at the relationship or account level),
contract-level service experiences with the supplier, and
interactions between supplier- and contract-level variables.
Our study context is the repurchase of system support ser-
vices by large firms. The firm’s decision to upgrade a con-
tract is represented by a binary logit model with random
parameters for the contract-level variables and fixed
parameters for the supplier-level variables. We estimate the
model with cross-sectional, longitudinal data from a sup-
plier of system support services with 120 business cus-
tomers in Germany and the United Kingdom, representing a
total of more than 2000 contracts.
Our study contributes to research on business customer–
supplier relationships in several ways. First, it highlights the
importance of understanding the relationship at both the
supplier (i.e., account) level and the individual contract
level because it demonstrates that both dimensions influ-
ence the firm’s upgrade decision. Second, the results sug-
gest a complex role for customer satisfaction, product–
customer fit, and service quality, providing novel opportuni-
ties for the supplier to use sales and service functions to
upgrade customers to higher (more expensive and poten-
tially profitable) levels of services. Third, compared with
prior research, the results show significant differences
between the factors that influence the firm’s decision to
upgrade a contract and those that influence the firm’s deci-
sion to renew a contract. Fourth, from a managerial per-
spective, the results suggest specific actions firms can take
to influence customer upgrades. Finally, the data examined
in this study are of the type available to many B2B firms, so
it is possible for practitioners to apply our model to their
individual business contexts.
Literature Review and Model
Development
In this section, we develop a model of the firm’s decision to
upgrade a service contract conditional on the decision to
renew the contract. In other words, the firm has already
decided to repurchase a service contract from the same sup-
plier; our model describes the firm’s decision about whether
to purchase a contract at the same service level or higher.1
The factors affecting service contract renewal, service
usage, and cross-buying (including upgrading and down-
grading) are not always the same, and when the factors are
the same, the predicted effect (positive, negative, or neutral)
is not always the same (Bolton, Lemon, and Verhoef 2004).
Thus, by framing the firm’s decision problem in this way,
our research isolates the factors that uniquely affect the
upgrade decision. From the firm’s perspective, the upgrade
decision is the process of solving the problem of market
matching and congruence (Alderson 1965). In essence, the
firm and the supplier are engaging in dynamic adaptation
(Dickson 1992); that is, the supplier has distinct levels of
market offerings, and firms are heterogeneous with respect
to their specific market needs. The firm engages in a
problem-solving process to find the level of contract (appro-
priate level of service from the supplier) that best meets its
needs. In this study, we examine specific factors the firm
may use to enhance this market matching process.
All our predictions regarding the upgrade decision are
made under ceteris paribus conditions—that is, after we
control for other factors. In our discussion, we distinguish
between factors associated with the specific contract and
factors associated with the supplier. At the contract level,
we propose that firms are influenced by service quality and
price. At the supplier level, we propose that firms are influ-
enced by decision-maker perceptions of the buyer–seller
relationship—especially assessments of satisfaction and the
criticality of the firm’s needs—which may be considered
account characteristics. We hypothesize interactions
between firm- and contract-level variables. The model also
includes covariates that control for relationship characteris-
tics (share of customer, relationship duration) and contract
type. By incorporating share of customer and price dis-
count, we also account for competitive effects, which tend
to be stronger when share is low or price discount is high.
Service Quality
Firms’ prior service experiences, especially service quality,
at the contract level are likely to influence the decision to
upgrade a service contract (Zeithaml 1988). Unlike an ini-
tial purchase decision, the firm’s upgrade decision is likely
to depend on service quality under the current contract (see
Kalwani and Narayandas 1995). The firm has many oppor-
tunities to assess the service during its experiences with
suppliers, thereby learning to make more effective purchase
decisions (Dodgson 1993). Organizational norms about the
quality of service evolve through ongoing transactions and
are based on prior experiences (Coleman 1990). Thus, when
a focal contract has poor service quality, the firm is likely to
conclude that the service level on the focal contract is inad-
equate for its needs, so it is less likely to renew the contract
(Bolton and Myers 2003). Moreover, when competition
exists and switching costs are low, suppliers that deliver
consistently bad service are forced to exit the marketplace.
At the same time, unlike many other purchase decisions,
the benefits of upgrading a service contract may seem
greater when the firm has experienced poor service quality.
There are three reasons poor service quality can result in
upgrade. First, attribution theory suggests that people are
likely to make causal attributions based on their prior ser-
vice experiences (Folkes 1988). Thus, we believe that if
other experiences are satisfactory and the firm intends to
renew (i.e., ceteris paribus), the decision maker is likely to
attribute poor service quality on the focal contract to an
inadequate level of service contract. Second, prior research
indicates that anticipation of regret may influence repeat
purchase decisions such that anticipation of regret associ-
48 / Journal of Marketing, January 2008
Boulding
et al. (1993)
Kalwani and
Narayandas
(1995)
Rust,
Zahorik, and
Keiningham
(1995)
Rust et al.
(1999)
Ganesh,
Arnold, and
Reynolds
(2000)
Bolton and
Myers (2003)
Li, Sun, and
Wilcox
(2005)
Ngobo
(2005)
Bolton,
Lemon, and
Bramlett
(2006)
Bolton,
Lemon, and
Verhoef
(2006)
Topic Model of
service quality
and repeat
purchase
intentions as
function of “will”
and “should”
expectations
Investigation of
whether
relational
approach to
serving
customers
“pays off” in
sales and
profitability
An investigation
of return on
service quality
Model of
customer
expectation
updating,
focusing on
distributions of
expectations
A comparison
of character-
istics of
consumers who
switched banks
and consumers
who stayed
loyal
Global market
segmentation
for services
Model of how
customer
demand for
multiple
products from a
single provider
evolves over
time
An investigation
of why some
consumers
migrate upward
(“up-buy”) and
some migrate
downward
(“down-buy”) in
a service
relationship
over time
Service
contract
renewal as a
function of
average service
levels,
variability in
service levels
(especially
extreme
outcomes), and
timing of
service delivery
Firm’s service
contract
upgrade deci-
sion modeled as
a function of
perceptions of
its relationship
with a supplier
at the enterprise
level, contract-
level experi-
ences, and
interactions
between firm-
and contract-
level variables
Analysis Longitudinal
experiment and
cross-sectional
survey
Cross-
sectional and
longitudinal
Cross-
sectional
Cross-
sectional
experiment
Cross-
sectional
Cross-
sectional
Longitudinal Longitudinal
and cross-
sectional
Longitudinal Longitudinal
Dependent
Variables
Repeat
purchase
intentions
Changes in net
sales, return on
investment, and
other variables
Market share
and net present
value derived
from consumer
repeat
purchase
intentions
Perceived
quality,
intentions, word
of mouth
No single
dependent
variable
Sales price
elasticity for
low-support
contracts and
sales price
elasticity for
high-support
contracts
Buy/not buy
banking
products (e.g.,
loan, debit
card, money
market
account)
Stay/defect,
migration
intentions
(subscribe or
not)
Renewal
decision for a
single contract
Upgrade
decision for a
single contract,
conditional on
renewal
Sample Business-to-
consumer
(B2C): (1) hotel
business trip,
(2) educational
institution
B2B,
COMPUSTAT
data for
selected
Standard
Industrial
Classification
codes
(n = 114)
Illustrative B2C
example, hotel
chain
(n = 7882)
B2C, 200
undergraduate
students shown
scenarios
describing
restaurant
service
B2C, banks
(n = 200)
B2B: Canada,
Germany,
Japan, Korea,
Singapore,
United King-
dom, United
States (508 low
support, 445
high support)
B2C, banking
panel data
(n = 1201)
B2C, theater-
goers
(n = 200)
B2B: Germany
and United
Kingdom (2442
contracts)
B2B: Germany
and United
Kingdom (2076
contracts)a
Unit of
Analysis
Individual
consumer level
Comparison of
matched
samples over
time
Individual-level
model linked to
aggregate
model
Individual
consumer level
Comparisons
of groups
Enterprise level Individual
household level
Individual
consumer level
Contract level Contract level
TABLE 1
Studies of Repeat Purchases and Service Quality
Expanding B2B Customer Relationships / 49
Boulding
et al. (1993)
Kalwani and
Narayandas
(1995)
Rust,
Zahorik, and
Keiningham
(1995)
Rust et al.
(1999)
Ganesh,
Arnold, and
Reynolds
(2000)
Bolton and
Myers (2003)
Li, Sun, and
Wilcox
(2005)
Ngobo
(2005)
Bolton,
Lemon, and
Bramlett
(2006)
Bolton,
Lemon, and
Verhoef
(2006)
Nature of
Predictor
Variables
(1)
Manipulation or
perception of
service
performance
attributes and
(2) 36 items
from
SERVQUAL
Dichotomous
categorization:
relational
versus
transactional
approach to
service
Perceptual
ratings of
service quality
attributes
Service
variability
(participants
were shown
graphs of up to
20 past
experiences)
Perceptual
ratings of
service quality
attributes
Service
operations
measures,
organizational
characteristics,
national/cultural
variables
budget
Accounts with
competitors,
overall
satisfaction,
switching costs,
demographics
Perceptual
ratings of
overall
satisfaction,
service quality
and pricing
policy,
relationship
variables,
demographics
Service
operations
measures,
price, discount
Decision-maker
perceptions
(satisfaction,
criticality),
service
operations
measures, price,
discount, share
of customer,
relationship
duration
Service Quality
Constructs
(1) Overall
quality of hotel
(unfavorable/fav
orable) and (2)
SERVQUAL
dimensions of
reliability and
empathy
Single
dimensions:
relational or
transactional
approach to
service
Grouped by
business
processes and
their underlying
subprocesses
Overall service
quality
(unfavorable/
favorable)
following
Boulding and
colleagues
(1993)
Three service
factors: people,
convenience,
ease of
transactions
(people factor
is different
across groups)
Responsive-
ness, reliability,
and assurance
(each repre-
sented by two
measures)
None
specifically:
perceptions of
overall
satisfaction with
the bank
Principal
components
analysis
yielded three
factors: quality
of acting and
decor, quality of
performance,
quality of
subscription
system
Design quality
(contract type)
and experience
quality
(engineer work
minutes)
Overall quality
(represented by
mean resolution
time per
contract)
aThis sample is smaller than the sample Bolton, Lemon, and Bramlett (2006) use because (1) satisfaction measures were not available for some firms and (2) high-level contracts are not eligible
for upgrade.
TABLE 1
Continued
50 / Journal of Marketing, January 2008
ated with dropping a service can influence a customer to
retain the service (Inman and Zeelenberg 2002; Lemon,
White, and Winer 2002). Extrapolating to upgrade deci-
sions, we believe that when a decision maker experiences
poor service quality, anticipated regret is likely to cause for-
gone alternatives, such as a higher level of service contract,
to become more attractive. Third, Park, Jun, and MacInnis’s
(2000) experimental research shows that when customers
are committed to buying within a category, “subtractive
framing” (in which customers are asked to delete options
from a fully loaded model) encourages them to choose
more options. In our study context, poor service quality on
the focal contract may encourage subtractive framing and,
thus, upgrading. This is also consistent with the negativity
effect, which suggests that a single negative piece of infor-
mation is often weighed more heavily than a single positive
piece of information in consumer evaluations (Price 1996).
Finally, this is also consistent with learning theory, which
suggests that firms (Dodgson 1993) and consumers
(Janiszewski and Van Osselaer 2000) engage in trial-and-
error behavior over time that results in learning. This
implies that, over time, firms may eventually learn the
appropriate contract match for their needs.
In summary, prior research suggests that a firm’s service
quality has a positive relationship to customer retention;
that is, compared with other contracts the firm holds, a firm
is less likely to renew a service contract when there is poor
service quality on the focal contract. However, the preced-
ing discussion suggests that there are theoretical reasons for
this, as follows:
H1: Conditional on renewal, a firm is more likely to upgrade a
service contract when the quality of service on the focal
contract is poor.
As we discuss subsequently, we expect that certain factors
will moderate the effect of service quality on the firm’s
upgrade decision.
Degree of Product–Customer Fit: Criticality
Studies of B2B purchasing behavior have long distin-
guished between products that have a large impact on the
firm’s financial performance and products that have a small
impact on this performance (Kraljic 1983). For example,
Cisco Systems does not outsource “mission-critical” activi-
ties, defined as “activities that, if performed poorly, would
pose an immediate risk to the company” (Bridge 2004).
Researchers also recognize that business activities that are
“critical” or important to a firm’s profit streams are less
likely to be outsourced (Stremersch et al. 2003). High-level
service contracts are designed for critical service environ-
ments, whereas low-level service contracts are designed for
less critical service environments. Consequently, we predict
that firms with a decision maker who perceives a service as
critical will be more likely to upgrade from a low-level ser-
vice contract. For example, retailers to which copiers may
be critical to their service operations (e.g., FedEx Kinko’s)
are more likely to upgrade their copier maintenance con-
tracts from a low to a higher level of service than other
retailers with less critical needs.
The underlying theoretical rationale for our prediction is
that a poor fit between the product and the customer should
2See also Vancouver and Schmitt (1991) regarding person–
organization fit, as well as the strategy literature, in which
product–market fit is also sometimes called “congruency.”
positively influence the firm’s decision to upgrade. This is
consistent with the idea of market matching (Alderson
1965; Dickson 1992), suggesting that the firm will adapt
and upgrade the contract to create a better fit with its needs.
Following Kristof (1996, pp. 4–5), we define product–
customer fit as the degree of compatibility between cus-
tomer and product that occurs when the product provides
what the customer needs.2Product–customer fit is poor
when a customer perceives service as critical (i.e., a high
level of service is required) but purchases a low-level ser-
vice contract. Mittal and Kamakura (2001) discuss the
nature of the relationship (or fit) between the customer and
the brand, finding that customers with different characteris-
tics have different satisfaction thresholds and, therefore, dif-
ferent probabilities of repurchase. Poor fit will decrease the
likelihood that a contract is renewed because the service
does not appear to deliver on its promise, but at the same
time, it will increase the likelihood that a contract is
upgraded because the firm recognizes the poor fit and
adjusts the contract provisions.
H2: Conditional on renewal, a firm’s likelihood of upgrading a
service contract is positively related to the decision
maker’s perception of the criticality of the service to the
firm’s operations.
Moderating Effect of Criticality
Recent research indicates that when people pay more atten-
tion to a product, service quality has a larger effect (more
positive) on loyalty (Bloemer and Kasper 1995). This find-
ing suggests that decision makers who perceive critical
needs weigh contract-level service experiences more heav-
ily in making their upgrade decision, which results in a
“hyperfocus” on these contract decisions. For this reason,
when the service is perceived as critical, firms are more
likely to upgrade when the focal contract service quality is
low. In contrast, if the service is not perceived as critical,
firms are less likely to upgrade in response to poor service
quality at the contract level (because decision makers are
paying less attention to all contracts).
H3: Conditional on renewal, the strength of the relationship
between a firm’s likelihood of upgrading a service con-
tract and service quality on the focal contract is greater
when the decision maker perceives the service as critical
than when he or she perceives it as noncritical.
Satisfaction with Prior Service
Customer satisfaction with prior service experiences has a
positive effect on the length, breadth, and depth of the
customer–firm relationship (e.g., Oliver 1997). In a field
investigation of B2B relationships in mature industrial mar-
kets, Narayandas and Rangan (2004) report that a decision
maker’s favorable evaluations of performance within the
contract terms (as well as outside them) lead the firm to
increase its commitment to the supplier. Because the
upgrade decision is an increase in the firm’s commitment to
the existing relationship, we predict that a firm with a deci-
Expanding B2B Customer Relationships / 51
sion maker who is highly satisfied with the supplier (i.e., he
or she has high cumulative satisfaction across all aspects of
the relationship) will be more likely to upgrade.
H4: Conditional on renewal, a firm’s likelihood of upgrading a
service contract is positively related to the decision
maker’s overall satisfaction with the existing supplier.
Our hypothesis is consistent with the well-established rela-
tionship between satisfaction and repeat purchase behavior.
However, it is also possible that firms that are satisfied with
the supplier are less likely to upgrade because they believe
that the current service level is exactly meeting their needs.
Moderating Effect of Satisfaction: The Rose-
Tinted-Glasses Effect
Research in psychology and consumer behavior suggests
that people exhibit a confirmatory bias (Hoch and Ha
1986), such that their prior opinions or expectations influ-
ence not only their overall opinions but also their percep-
tions of new information (Oliver and Burke 1999). Bould-
ing, Kalra, and Staelin (1999) find that confirmatory bias
decreases with higher levels of experience and lower levels
of product complexity and ambiguity. Nayakankuppam and
Mishra (2005) find that sellers (people who currently own a
product) overrepresent the positive features of the product
compared with buyers—a “rose-tinted-glasses” effect (Hen-
drick and Hendrick 1988). Because we are studying com-
plex B2B services, these findings suggest that the firm’s
assessment of service quality will be positively biased for
satisfied decision makers and negatively biased for those
who are dissatisfied. Thus, we believe that firms will weigh
contract-level service quality less heavily in their upgrade
decision when decision makers’ satisfaction levels are high.
H5: Conditional on renewal, the strength of the relationship
between a firm’s likelihood of upgrading a service con-
tract and service quality on the focal contract is weaker
when the decision maker’s overall satisfaction with the
supplier is high than when it is low.
Price Effects
When customers consider whether to upgrade a service con-
tract, the price of the service may also influence their deci-
sion making. Prior research suggests that a customer’s ref-
erence price will influence his or her decision to purchase a
good or service. The reference price is the price against
which buyers compare the offered price of a product (Winer
1986). The concept of dual entitlement (Kahneman,
Knetsch, and Thaler 1986) suggests that consumers believe
that firms are entitled to a reference profit, that customers
are entitled to a reference price, and that price increases
commensurate with cost increases will be perceived as fair
(e.g., Bolton and Alba 2006). Similarly, in this research
context, an upgrade represents more benefits to the firm and
higher costs to the supplier. Therefore, firms that upgrade to
a higher level of contract most likely pay a higher price for
that contract than if they had not upgraded. Thus, for com-
pleteness, we predict the following:
H6: Conditional on renewal, a firm’s likelihood of upgrading a
service contract is positively related to the current price of
the service contract.
Bolton and Myers (2003) find that customers who receive
less reliable service over time are more price sensitive than
customers who receive more reliable service. This suggests
that price should moderate the effect of service quality on
the firm’s upgrade decision.
H7: Conditional on renewal, the strength of the relationship
between a firm’s likelihood of upgrading a service contact
and service quality on the focal contract is weaker when
the price of the contract is high than when it is low.
Summary
We hypothesized that service quality on the focal contract
(SQual, H1), perceived criticality (Criticality, H2), satisfac-
tion (Satisfaction, H4), and contract price (LogPrice, H6)
influence the decision to upgrade a service contract. We
also hypothesized that perceived criticality, customer satis-
faction, and contract price moderate the effect of service
quality on the customer’s decision to upgrade (H3, H5, and
H7). Prior research also suggests that models of ongoing
buyer–seller relationships should account for certain covari-
ates. Thus, we include supplier-level variables to capture
relationship characteristics, such as commitment (which can
be represented by measures of share of customer [Share]
and customer relationship duration [Duration]), contract-
level variables for price discount (LogDiscount) to account
for linear effects that have been previously established (Ver-
hoef, Frances, and Hoekstra 2001), and a dummy variable
for contract type (ContractType). Figure 1 depicts the con-
ceptual model of the upgrade decision.
Overall, the firm’s decision to upgrade a contract can be
considered in terms of the overall assessment of value asso-
ciated with the contract—that is, how the firm weighs the
benefits and costs of the upgraded contract. A customer’s
assessment of value from a service has been defined as the
customer perceived trade-off between costs and benefits
(Brady et al. 2005). In a B2B context, Bolton and Drew
(1991) suggest a cost–benefit framework in which a cus-
tomer’s assessment of value depends on benefits received
and sacrifice (i.e., monetary and nonmonetary costs associ-
ated with using the service; e.g., Zeithaml 1988) and the
customer’s frame of reference or context (Holbrook 1999).
At a high level, our model is consistent with the view that
firms weigh the benefits (service quality, satisfaction) and
costs (price) associated with the contract when determining
whether to upgrade and account for their own frame of ref-
erence (criticality, share of requirements, relationship dura-
tion, and contract type).
Model Specification and Estimation
Procedure
Following a long tradition of theoretical and empirical work
modeling firm behavior (e.g., Coughlan 1985; Heide and
Weiss 1995), we represent the firm’s decision to upgrade a
service contract by a binary choice (logit) model. The firm
(i) renews a service contract (k) by choosing the alternative
(upgrade or not) with the highest expected future value
(Vik). That is,
(1) Prob(Upgradeik) = Prob(Vik* > Vik).
52 / Journal of Marketing, January 2008
FIGURE 1
Conceptual Model of Customer Upgrade Decision
aControlling for the renewal decision.
H4
H6
H3
H1
H5
H2
Customer perceptions
(at enterprise level)
H7
Customer
Upgrade
Decisiona
Service Quality
(resolution time:
customer experiences
within contracts)
Price
(price paid for each
contract)
Covariates:
•Customer share
•Relationship duration
•Discount
•Contract type
Satisfaction
Criticality
In this situation, a firm makes repeated decisions (across
contracts), creating dependent observations. The number of
contracts purchased typically differs between companies, so
that we expect the data to be “unbalanced.” To account for
the interdependency within each firm and to allow varying
coefficients across firms, we estimate a random parameters
model (Train 2003). Specifically, we estimate random coef-
ficients for the contract-level variables (Zik) and fixed coef-
ficients for the supplier-level variables (Xi). This specifica-
tion is known as a mixed-effects logit model.
If a firm i makes k contract upgrade decisions, we for-
mulate our model as follows:
(2) Vik = α′Xi+ μ′ikZik + εik,
where α′ represents a vector of fixed parameters for the
supplier-level data Xi, μ′ik is a vector of random coefficients
for the contract-level data Zik, and εik is i.i.d. extreme value.
The terms in Zik are error components that, along with εik,
define the stochastic portion of the utility (Train 2003). As
such, we define the unobserved random portion of utility
(ηik) as
(3) ηik = μ′ikZik + εik.
We assume that the mean and standard deviations of the
random parameters are normally distributed.
In summary, we can write our mixed-effects binary logit
model as follows:
(4) Upgradeik = f(Satisfactioni, Criticalityi, SQualik,
SQualik ×Satisfactioni, SQualik ×Criticalityi,
LogPriceik, SQualik ×LogPriceik,
LogDiscountik, Sharei,
Durationi, ContractTypeik)+ εik.
Recall that we estimate fixed coefficients for the supplier-
level variables (Xi) and random coefficients for the
contract-level variables (Zik). The interactions between per-
ceptions at the supplier level (Satisfactioni, Criticalityi) and
service quality at the contract level (SQualik) and the inter-
action between price (LogPriceik) and service quality create
contract-level variables, so we estimate random parameters
for these interaction terms.
We specify and estimate our model conditional on con-
tract renewal; that is, we estimate the model only for the
2076 contracts that were renewed. However, the (potential)
hierarchical structure of the renewal and upgrade decisions
might suggest the specification of a nested logit model that
allows for a correlation between the error terms of the
unobserved utilities. Because we do not use a nested logit
Expanding B2B Customer Relationships / 53
model, it might be argued that we are biasing the estimates
of our logit model parameters (Ben-Akiva and Lerman
1985). However, this argument ignores the issue of random
parameters arising from unobserved customer (i.e., firm)
heterogeneity. It is cumbersome to account for customer
heterogeneity in a nested logit when both upgrade and
renewal decisions depend on (mostly) the same explanatory
variables. More important, Louviere (2003) argues that if
the researcher uses certain mixed-effects logit models (e.g.,
with a hierarchical structure, random effects), it may not be
possible to use statistical tests to identify the “correct”
model specification. On the basis of these considerations,
we chose a reasonably parsimonious, binary logit model
with mixed effects and estimated it with data describing
renewed contracts only.
Method
Study Context
The study context is computing system support services
purchased by large European firms. Firms spend (approxi-
mately) €650,000 per year on computing system support
services. The service contracts are fixed-price contracts for
one year; customers are not billed on service-usage levels.
Firms purchase system support contracts independently for
systems that have been purchased previously, and an indi-
vidual contract is uniquely associated with a particular sys-
tem. Most firms purchase from multiple suppliers. There
are eight suppliers in the European market, and firms may
also use internal support services. Large firms usually have
many systems, and they buy multiple contracts from multi-
ple suppliers.
Suppliers in many industries give the different levels of
their service contracts descriptive names, such as “mission-
sensitive support,” “proactive support,” or “managed ser-
vices,” that are intended to help the buyer achieve a good
product–customer fit. System support is provided out of
centralized facilities, so that it is standardized across cus-
tomers and contracts. In our study, the supplier offers con-
tracts that are available at three levels of system support,
which we call low, medium, and high support. We do not
study high support (which cannot be upgraded); we study
upgrades from low to medium levels or from medium to
high levels. For each level of support, the supplier has spe-
cific, contractually defined service performance levels, such
as “24 hours, 7 days a week availability, with a guaranteed
response within 2 hours.” Each level of support provides a
subset of the services of the next (higher) support level. A
higher level of support contract is more expensive than a
lower level of support contract, but it provides higher levels
of support that can reduce resolution times, thus increasing
service quality. Importantly, over time, a customer may
decide to upgrade the support level of a contract. No cus-
tomers downgraded their contracts in our database.
The Database
For our study, a cooperating supplier provided a probability
sample of customer records that described purchases of sys-
tem support service contracts by 120 large firms (i.e., busi-
ness customers) in Germany and the United Kingdom. The
3The cooperating company drew the sample with the restriction
that the equipment (covered by the service contract) was still in
use and would continue in use, thus removing any cases in which
the customer no longer owns or uses the equipment and thus does
not renew because the equipment is no longer in service.
4Service operations data (i.e., resolution time measures) were
collected from January 1997 to February 1999; billing records
(i.e., purchase data, including upgrade decision [dependent
variable], contract type, and price information) were collected in
February 1998 and January 1999; the survey data (i.e., satisfac-
tion, criticality, duration, and customer share measures) were col-
lected in July 1998.
sampling procedure ensured that the sampled firms oper-
ated the same system, thus reducing the effects of variables
extraneous to our study (e.g., factors unique to after-sales
service for a particular system).3The data described the
firm–supplier relationship over a three-year period (1997–
1999). We combined information from four primary
sources: (1) a master file describing the characteristics of
each customer account, including number and type of con-
tracts; (2) annual billing data for all contracts of each cus-
tomer, including price and any discounts; (3) internal
monthly service operations data (i.e., service quality met-
rics) for all contracts of each customer; and (4) self-report
data regarding satisfaction and criticality from a survey of
the decision maker at each firm. (The survey was adminis-
tered to the person identified by the supplier as the decision
maker, after he or she was rigorously screened by telephone
to ensure that he or she either recommended or made the
final decision on system support contracts.4) We merged
these sources to create a comprehensive database that incor-
porates both cross-sectional and longitudinal observations
at the customer account and contract level. Note that we
measure the predictor variables (Satisfactioni, Criticalityi,
SQualik, Sharei, Durationi) in the period before the upgrade
decision, so it is possible to make causal inferences from
our data. The service contract’s price and any discount are
from the time of purchase.
All firms purchased five or more contracts from the
same supplier; the average number of contracts was 19.7,
and the standard deviation was 23.7. Of the 120 firms,
43.3% upgraded at least one contract—upgrading occurs
only if firms renew contracts (there were no firms in our
database that downgraded contracts); of the contracts, 88%
(2076) were subsequently renewed, and 8% (165) of the
renewed contracts were upgraded. Thus, our analysis data-
base contains the 2076 renewed contracts for the 120 firms.
The relatively low percentage of upgraded contracts is
likely because most firms purchase a service level that is
appropriate for their needs. We conducted preliminary sta-
tistical analyses and determined that the upgrade process
was the same for both contract levels (low to medium or
medium to high); overall, more medium-level contracts
were upgraded (18%) than low-level contracts (5%). There
was no evidence of a need for interaction effects between
contract type and any of the other predictor variables. Thus,
we capture this feature of the data by incorporating a
dummy variable that represents medium-level contracts. We
discuss this issue further in the “Results” section.
54 / Journal of Marketing, January 2008
Measures
Table 2 reports the predictor variables, their measures, the
data source and level (i.e., entire customer account or con-
tract), and descriptive statistics (average, standard devia-
tion, and skewness). The average firm had been purchasing
from the supplier for six years and spent 30% of its total
system support budget on the supplier’s service contracts.
Because the majority of the predictor variables were mea-
sured in a straightforward way, we discuss only measures
that require additional explanation.
Satisfaction. A survey elicited self-report measures from
chief information officers or management information sys-
tem managers who were identified from company records
and screened by telephone to ensure that they either recom-
mended or made the final decision on system support con-
tracts. We calculated the average of three survey items to
measure the decision maker’s satisfaction. The items are
similar to items used in other studies (e.g., Bolton 1998;
Oliver 1997). They elicit perceptions of (1) the decision
maker’s overall satisfaction with his or her relationship with
the supplier, (2) the value that the supplier adds to the firm’s
business activities, and (3) the overall satisfaction with cus-
tomer service. We assessed each question on a five-point
scale (1 = “extremely unsatisfied,” and 5 = “extremely
satisfied”).
Criticality. We measure the criticality of the service to
the firm by asking the decision maker, “When your system
fails, what effect does that have on your business (“not at all
serious/very serious”)?” Because the distribution of
responses was heavily skewed, we collapsed the multiple-
response variable to a dichotomous variable based (approx-
imately) on a median split. Thus, the variable is coded as 1
if the service is highly critical and as 0 if otherwise. In total,
71% of the firms consider their services critical for their
operations.
Contract-level service quality. Before this study, the
supplier had hired a market research company to conduct
research with its European business customers. The com-
pany conducted face-to-face interviews with 40 decision
makers and administered large-scale telephone surveys to
decision makers with the dual goals of understanding per-
ceptions of service quality and areas for improvement. The
supplier also participated in an industrywide benchmarking
study that showed that its satisfaction ratings were roughly
comparable to the other two major suppliers in Europe.
However, timeliness, completeness, and effectiveness of
system support consistently emerged as frequent customer
concerns. Specifically, respondents identified speedy reso-
lution time as a key aspect of service quality that they con-
sidered in repeat purchase decisions. The following are ver-
batim comments from four decision makers:
[The supplier needs] to have a more timely response; it
would be better to get rid of the callback system—replac-
ing it with a direct connect [to an engineer].
I have a highly technical staff and [the supplier] should
recognize that ... we go through our own diagnosis and
exhaust our resources. We get a junior helper [at the cus-
tomer support center] and have to go through the same
process with [the supplier] and waste our time.
5Note that resolution time is not the same as response time. Res-
olution time is the time elapsed until the customer request has
been resolved, not the time until first response. However, not sur-
prisingly, quantitative analyses of the operations data (not shown
herein) indicated that response time, engineer work time, number
of repeat requests, and other service quality/business process met-
rics were correlated with resolution time (and they might be con-
sidered its antecedents).
6Our approach is a different from cross-sectional studies that
rely on perceptual measures of service quality (which can be stan-
dardized across companies and industries). Because we study
firms that purchase from a single supplier, we can measure service
quality using business process metrics from the service operations
database. Our approach is consistent with prior within-company
research that uses business process metrics to represent service
quality, thus producing actionable results for managers (e.g.,
Bolton and Myers 2003). In models estimated with perceptual
measures, it is frequently possible to capture multiple dimensions
of service quality, such as reliability, responsiveness, assurance,
empathy, and tangibles. However, our statistical tests indicated
that a single measure of service quality was sufficient to explain
the upgrade decision. The inclusion of additional variables did not
improve the explanatory power of the model.
[The supplier] should shorten the resolution time.
[The supplier] needs to keep up with all the information
that we provide them on our problems and resolve them in
a more timely manner.
These decision makers’ comments reflect the notion that
additional supplier resources (e.g., faster connect time,
more customer support staff, access to knowledgeable engi-
neers) were necessary to reduce resolution time, thus reduc-
ing system downtime and its negative effect on business
performance.5Preliminary statistical analyses also sup-
ported the notion of a single variable—resolution time—as
an indicator of service quality. Respondents’ intense dissat-
isfaction with a single service quality attribute is consistent
with recent research. Slotegraaf and Inman (2004) report
that as customers approach the end of a service contract
period, satisfaction with attributes that can be remedied
declines at a faster rate. In our study context, concerns
about slow resolution time can be remedied by upgrading
the service contract so that resolution time becomes a
prominent dissatisfier and a key input into the upgrade
decision.
For these reasons, we measured service quality at the
contract level by calculating the average resolution time per
service request for the focal contract (expressed as minutes/
request) from the monthly service operations data during
the year before the purchase decision.6A large value of this
variable indicates poor service quality (rather than good ser-
vice quality). We calculated the log of this measure to cre-
ate a variable with a less skewed distribution. Thus, our
model incorporates service quality measured as the log of
average resolution time per request for the focal contract.
This transformation implies that as resolution time becomes
larger, there are diminishing marginal effects of service
quality at the contract level.
Price. We obtained the price of each contract from the
billing data and measured it at the contract level. We
included two aspects of price in the model: contract price
(in dollars) and price discount (expressed as a percentage)
Expanding B2B Customer Relationships / 55
TABLE 2
Measures and Descriptive Statistics (N = 2076)
Variable
(Name) Description and Measurementa
Data
Source
Data
Level AveragebSkewness
Upgrading
(Upgradeik)
Contract upgraded or not.
A dichotomous variable determined by comparing billing records for January
1999 and January 1998
Billing Contract .079
(.27)
3.11
Satisfactionc
(Satisfactioni)
Average of the following questions (Coefficient α= .70)
How satisfied … (1 = “extremely unsatisfied,” 5 = “extremely satisfied”)
•as a whole with the customer service which you buy from [the supplier]?
•with the value that [the supplier] adds to the business activities?
•are you overall with your relationship to [the supplier]?
Survey Supplier 4.04
(.55)
–.86
Service quality
(SQualik)
Log of the average resolution time (in minutes) for the focal contract:
1998 average resolution time per request for focal contract
Operations Contract .38
(1.09)
3.11
Criticalityc
(Criticalityi)
“When your system fails, what effect does that have on your business?”
Dichotomous variable derived from self-report data (1 = “very serious”)
Survey Supplier .71
(.46)
–.90
Covariates
Share of customer
(Sharei)
Percentage of service contracts purchased by customer from the supplier.
Calculated as: the total dollar value of contracts purchased from the supplier
(from the billing data) divided by the dollar value of the total support budget
(as reported by the decision maker)
Survey,
master file
Supplier .30
(.22)
1.07
Relationship
duration
(Durationi)
Duration of relationship in years.
Calculated as the years between January 1999 and start of relationship
Master file Supplier 6.41
(3.19)
.15
Price
(LogPriceik)
Log of the price level per contract.
Dollars per contract (scaled for reasons of confidentiality)
Billing Contract .74
(1.55)
–.56
Discount
(LogDiscountik)
Log of the price discount per contract.
Discount is expressed as a percentage of the dollar price per contract
Billing Contract –4.46
(4.04)
–.32
ContractTypeik Contract level (medium = 1, otherwise = 0) Master file Contract .16
(.36)
1.88
a“Supplier” should be interpreted to mean the business customer’s perception of the supplier or service.
bStandard deviation is in parentheses.
cFrom a survey administered on July 1998.
56 / Journal of Marketing, January 2008
for each contract. To allow for nonlinear effects of price and
discount while minimizing the potential for multicollinear-
ity, we used a log form for the price variables. This formu-
lation is more parsimonious than including both linear and
quadratic price variables.
Covariates. We measured two covariates at the cus-
tomer account level: share of customer and relationship
duration. We calculated share of customer by dividing the
dollar value of the number of contracts purchased from the
supplier by the dollar value of the firm’s total system sup-
port service budget (as reported in the survey). The firm–
supplier relationship duration is a self-report measure from
the survey; it was measured in years. (Note that it is not the
same as the length of the decision maker’s relationship with
the supplier.) Finally, we incorporated a dummy variable to
represent contract type (1 = medium, and 0 = otherwise).
Table 3 displays the correlation matrix of the predictor
variables. The majority of the coefficients are low. An
exception is the correlation (.30) between list price and per-
centage price discount, suggesting that contracts with
higher list prices have higher percentage discounts. This
correlation is not surprising given that the European market
for system support contracts is highly competitive. Overall,
we concluded that multicollinearity among the predictor
variables is reasonably low.
Model Estimation
We estimated our model using LIMDEP 8.0, which pro-
vides accessible estimation procedures for the random
parameter versions of the binary logit model (Greene 2003).
We performed simulation in the random parameter model
using 100 Halton draws (Greene 2003). We estimated three
models: (1) a covariates-only model, (2) a main effects–
only model, and (3) a full model including the interaction
effects. The fixed coefficients, random coefficients, scale
parameters of the random coefficients, and fit statistics of
these three models appear in Table 4. This table shows sig-
nificant scale parameters for the random coefficients for
multiple variables. The significant scale parameters imply
that there is heterogeneity in the estimated parameters, so
we conclude that accounting for heterogeneity with random
coefficients is a useful modeling strategy. This feature is
important to remember when interpreting the estimated
parameters. For example, the scale parameter for LogPrice
is rather large, suggesting significant heterogeneity across
the sampled firms. Thus, for some customers, the price
effect may be smaller, whereas for other customers, the
price effect may be larger.
Results
Model Fit
Of the three estimated models, Models 2 and 3 have a much
better fit than Model 1. Model 3 has a higher likelihood and
R-square than Model 2 and an equal Akaike information
criterion (AIC). However, note that AIC imposes a heavy
penalty in a random coefficients model because the addition
of each predictor variable adds more than one parameter
estimate.
7We tested models with other functional forms, such as a non-
linear effect of satisfaction (i.e., quadratic). The fit of this model
was worse than our model (AIC = 972 versus 970). We also con-
sidered whether firms make comparisons across contracts by
including the difference between resolution time of the focal con-
tract and the average resolution of all other contracts. The model
fit of our reported model was much better than the model fit of the
model accounting for a comparison across contracts (AIC = 996
versus 970).
We believe that the full model (Model 3) is preferable
for three reasons. First, most of the hypothesized predictor
variables in the upgrade model are significant (p< .05).
Second, according to log-likelihood ratio test, the estimated
model is highly significant (p< .0001). Third, for compari-
son purposes, we estimated a model that excluded the
contract-level variables with their random coefficients. That
is,
(5) Prob(Upgradeik) = g(Satisfactioni, Criticalityi,
Sharei, Durationi)+ εik.
We compared our proposed model (Equation 4) with the
“firm-only” model (Equation 5) and found that the pro-
posed model produces an improved fit (AIC = 1130.28 ver-
sus 970).7Thus, our model comparisons suggest that a
binary logit model with random coefficients, plus both firm-
and contract-level variables, is an appropriate modeling
strategy.
We assessed the stability of our estimated model (Equa-
tion 4) as follows: We estimated our model using 75% of
our sample. We repeated this procedure three times (i.e.,
three random samples). Of the estimated parameters, 97.2%
have the same sign as in our reported model. The signifi-
cance levels are the same for 86% of the estimated parame-
ters. Consequently, we conclude that our model results are
stable.
Hypothesis Tests
We find strong support for the model. The results for the
three hypotheses regarding main effects are as follows:
Firms experiencing poor service quality at the contract level
are indeed more likely to upgrade (H1, p= .013). However,
H2is not supported; the decision maker’s perception of the
criticality of the firm’s service does not influence upgrading
(H2, p= .24). Firms with decision makers who are more sat-
isfied with the overall relationship with the firm are more
likely to upgrade contracts, in support of H4(H4, p< .001).
Current contract price has a positive effect (with diminish-
ing marginal returns) on upgrading (H6, p< .001). We pre-
dicted an interaction effect between service quality and crit-
icality (H3). When decision makers perceive the firm’s
service needs as critical, we expected that poor service
quality at the contract level would lead to a higher probabil-
ity of upgrading. This hypothesis was not supported (H3,
p= .49).
We predicted that there would be an interaction effect
between satisfaction and service quality. This hypothesis
was supported (H5, p< .001). We conducted simulations for
different satisfaction/service quality levels and discovered
that this interaction effect captures the following feature:
For poor service quality, business customers who have low
Expanding B2B Customer Relationships / 57
TABLE 3
Correlation Matrix (N = 2361)
Variable Upgradeik SatisfactioniCriticalityiSQualik LogPriceik LogDiscountik DurationiShareiContractTypeik
Upgradeik 1.00
Satisfactioni.04 1.00
Criticalityi.05 .22 1.00
SQualik –.01 –.03 –.02 1.00
LogPriceik .09 .16 .16 .01 1.00
LogDiscountik –.15 .06 –.14 –.13 .21 1.00
Durationi–.04 .28 .15 –.03 .08 –.03 1.00
Sharei.07 .01 –.04 .09 .30 .09 .14 1.00
ContractTypeik .20 .03 .02 .34 .08 –.12 .02 .04 1.00
58 / Journal of Marketing, January 2008
TABLE 4
Estimation Results (N = 2076)
Variable Expected Covariates- Main Effects– Full
(Hypothesis) Hypothesis Sign Only Model Only Model Model
Constant –2.259** –5.069** –6.141**
Supplier Level
Satisfactioni4 + .498** .712**
Criticalityi2 + –.061 –.159
Contract Level
SQualik (resolution time) 1 + –.490** 1.179*
LogPriceik 6 + .297** .429**
LogDiscountik –.131** –.138**
Criticalityi×SQualik 3 + –.103
Satisfactioni×SQualik 5 – –.592**
LogPriceik ×SQualik 7 – –.241**
Covariates
Sharei.979** .869** .999**
Durationi–.062** –.070** –.159**
ContractTypeik .799** 1.084** 1.011**
Scale Parameters
Constant 1.097** .305** .126*
ContractTypeik .943** .631** .964**
SQualik (resolution time) .507** .429**
LogPriceik .420** .469**
LogDiscountik .082** .140**
Criticalityi×SQualik .720**
Satisfactioni×SQualik .437**
LogPriceik ×SQualik .217**
Model Fit
Log-likelihood –502 –471 –465
AIC 1016 970 970
McFadden R2.13 .18 .19
*
p
< .05.
**
p
< .01.
Notes: Two-sided
p
-values for hypothesis tests. Number of observations = 2076.
satisfaction are more likely to upgrade than those who have
high satisfaction. However, for good service quality, cus-
tomers who have high satisfaction are much more likely to
upgrade than those who have low satisfaction. This finding
is consistent with our hypotheses. Overall, high satisfaction
leads to upgrading, except when there is poor service qual-
ity on the focal contract. We also predicted an interaction
between current contract price and service quality (H7). We
find strong support for this effect, suggesting that poor ser-
vice quality reduces the effect of price on the upgrade deci-
sion (negative interaction effect, H7, p< .001).
Covariates
Firms that devote a larger portion of their total support bud-
get to the supplier (i.e., higher “share of customer”) are
more likely to upgrade a contract (p< .001), and firms that
have lengthy relationships with the supplier are less likely
to upgrade (p< .001), perhaps because they have already
achieved an optimal product–customer fit. In addition to the
price effects we reported, percentage price discount has a
8Another approach would be to specify one model (Model A) to
represent the decision to upgrade contracts from low to medium
and another model (Model B) to represent the decision to upgrade
from medium to high. However, there are some statistical reasons
we did not choose to use this approach. First, the number of low-
level contracts is relatively small compared with the number of
medium-level contracts (325 versus 1751). Because only a small
percentage of contracts are upgraded (8%), the number of low-
negative influence on the upgrade decision (p< .001), with
diminishing marginal returns. The significant effects of
price in this study are rather surprising because price is
often not a statistically significant predictor in models of
customer retention (e.g., Bolton 1998; Bolton, Lemon, and
Bramlett 2006). Contracts at the low service level are more
likely to be upgraded (p< .001) than service contracts at the
medium service level, ceteris paribus. This result raises the
following question: Are the model results different when we
omit the low-service-level contracts from the analysis?
However, we conducted this analysis and still found that the
effects are much the same as we reported previously.8
Expanding B2B Customer Relationships / 59
level contracts that are upgraded becomes small in absolute terms.
Second, the small sample also leads to a relatively small number
of contracts per enterprise. For a random coefficients model, this is
problematic because a sufficient number of observations per enter-
prise (roughly five contracts per customer) is required. Finally, two
separate models are less parsimonious than one model with a
dummy variable for contract type and (if appropriate) interaction
terms.
Theoretical Implications
Our results suggest that the customer upgrade decision is
significantly different from the new product/service pur-
chase decision, because prior experience with the customer
is critical, and different from the customer renewal decision,
because the firm’s experience with the supplier appears to
influence these decisions in different ways. Overall, we find
that the results are consistent with the basic ideas of market
matching and congruence. Firms use their experiences with
the service contract to determine whether the current ser-
vice contract fits their needs, and they upgrade when neces-
sary to find the appropriate level of service they require
from the supplier. The ways firms use this information to
obtain the best match with their needs provide new insights
into the firm’s upgrade decision and our understanding of
B2B customer behavior.
Accounting for Multiple Inputs into the Upgrade
Decision
Traditional approaches to choice modeling focus on con-
necting information on the characteristics of an offering and
prior customer experience to behavior (e.g., Guadagni and
Little 1983). More recent attention has shifted to under-
standing how customers’ overall attitudes toward and
evaluations of an offering explain behavior (Bolton 1998;
Gustafsson, Johnson, and Roos 2005; Mittal and Kamakura
2001; Verhoef 2003). Our study underscores the need to
combine both approaches to understand the complexities of
buyer behavior.
In particular, the results highlight the circumstances
under which buyers may pay more attention to objective
information (e.g., service quality, price) or summary evalua-
tions (e.g., satisfaction) and the extent to which summary
evaluations may actually interact with contract-level experi-
ences to influence buyer behavior. Thus, at the macro level,
the results of this study suggest that (1) both contract-level
experiences and firm-level decision-maker perceptions
influence contract-level decisions directly and (2) decision-
maker perceptions of the supplier interact with contract-
level experiences to influence these decisions. A model of
firm decision making estimated solely at the supplier level
or solely at the contract level omits interesting antecedents
of the upgrade decision. This finding is particularly novel
and important because the majority of prior B2B research
has been conducted solely at the supplier or customer
account level (i.e., eliciting measures of perceptions of the
entire relationship, across contracts) and is based on key-
informant data. The significant contract-level effects sug-
gest that it is important to examine these decisions at both
the firm level (e.g., decision-maker perceptions, relation-
ship characteristics) and the contract level (e.g., service
quality, price) to gain a complete understanding of the fac-
tors that influence firm upgrade decisions.
Specific insights regarding key factors that influence the
customer upgrade decision also emerge from this research.
First, we find that a firm’s experience of poor service qual-
ity can increase the likelihood of upgrading an individual
contract. Second, high satisfaction (at the overall firm level)
leads to a higher probability of upgrade. Third, customers
who pay greater prices for a contract are more likely to
upgrade. Finally, we find two key interactions that also
influence the upgrade decision: service quality ×satisfac-
tion and service quality ×price. Next, we discuss each of
these findings and elaborate on how service quality influ-
ences the upgrade decision through (combined) main and
interaction effects.
Poor Service Quality: How Can It Be Good for the
Supplier?
At the contract level, poor service quality can increase the
likelihood that the firm will upgrade to a higher level of ser-
vice contract. On the surface, it may seem counterintuitive
that by providing poor service, the supplier can gain addi-
tional business from the firm. However, it is important to
remember that the firm is implicitly making comparisons
across contracts; the decision maker evaluates satisfaction
with its relationship with the supplier as well as service
quality on the focal contract. We speculate that the multi-
dimensional nature of the customer–supplier relationship
allows three mechanisms to operate that create a positive
main effect of service quality. First, if service quality is
poor for the focal contract in an otherwise satisfactory rela-
tionship, the decision maker may attribute poor service
quality to having purchased a contract level that is insuffi-
cient to meet the firm’s needs rather than to intrinsically
poor service by the supplier. Second, the evaluation of the
focal contract quality may lead the decision maker to antici-
pate regret if he or she repurchases the same level of service
contract rather than a higher level of service contract. Third,
because customers are committed to buying the service,
poor service quality on the focal contract may lead the deci-
sion maker to use subtractive framing, whereby he or she
realizes that the lower-level service contract is a subset of
the higher-level (i.e., augmented) service contract.
Our finding also demonstrates that “snapshots” of
contract-level experiences are remembered and that rela-
tively poor service stands out in the mind of the customer in
an otherwise satisfactory relationship. It is also consistent
with substantial qualitative research showing that critical
incidents or moments of truth are important influences on
service purchase decisions (Bitner, Booms, and Tetreault
1990) and with prior research on the negativity effect,
which suggests that a single negative piece of information
carries more weight than a single positive piece of informa-
tion (Price 1996).
Does this result imply that poor service quality is good
for firms? No, certainly not. For this specific firm, the deci-
sion to upgrade a contract ultimately depends on the inter-
action effects. Because most customers are satisfied with
the firm and prices are high, the net effect is as follows: The
60 / Journal of Marketing, January 2008
business customer is less likely to upgrade when quality is
poor on the focal contract and the interaction effects are
taken into account. We subsequently illustrate this outcome
using simulation results.
Higher Prices—More Upgrades?
Firms that currently pay higher prices on a contract (or have
lower discounts) are more likely to upgrade. Prior research
examining antecedents of the service contract renewal deci-
sion (Bolton, Lemon, and Bramlett 2006) has found that
price has a negative effect on the renewal decision; the
higher the price, the less likely firms are to renew the con-
tract. Thus, it appears that price operates differently on the
upgrade decision, perhaps acting as a current reference
price against which the increase in price (to move up to a
higher-level contract) is considered. This is consistent with
recent research on price fairness (Bolton and Alba 2006),
which finds that when increases in price are perceived as
aligning with increases in vendor costs, such prices are con-
sidered fair.
Decision-Maker Perceptions Also Influence the
Decision
In addition to effects of contract-level experiences, such as
service quality and price, decision makers’ perceptions of
the supplier, which are firm-level constructs, directly influ-
ence firms’ contract-level decisions. Examining main
effects, we observe that firms with decision makers who are
satisfied are more likely to upgrade service contracts. This
result is consistent with research in the customer satisfac-
tion and loyalty literature. Notably, we do not find support
for a main effect of criticality on the upgrade decision. We
believe that this may be a ceiling-effect phenomenon; firms
that perceive the equipment as critical to their needs may
already have chosen a high level of contract, thereby mask-
ing this effect in our data.
Satisfaction Moderates the Effect of Service
Quality
Decision-maker perceptions of the supplier moderate the
effects of service quality on customer upgrade decisions.
Customer satisfaction (the decision maker’s evaluation of
the supplier’s management of the entire account relation-
ship) moderates the effect of service quality on the cus-
tomer upgrade decision—a rose-tinted-glasses effect. The
results suggest that a highly satisfied decision maker can
reduce the effects of poor service quality at the contract
level, so the firm is less likely to upgrade. A way to interpret
this interaction is to consider a firm with low satisfaction
(which we believe is driving this result). In this case, it
appears that decision makers pay more attention to objec-
tive quality information, such as service quality. This
implies that they may be in a more active problem-solving
mode and is consistent with research by Olsen and Johnson
(2003), who find that less satisfied customers focus more
immediately on quality perceptions than more satisfied cus-
tomers. The interplay between supplier perceptions (satis-
faction) and contract-level objective experiences of service
quality suggests that the firm’s decision to upgrade a service
contract entails a consideration of information and assess-
ments at multiple levels and that it is more complex than
prior research has indicated. Note that from the firm’s per-
spective, our results suggest that decision makers’ percep-
tions can “bias” contract-level service experiences, leading
customers to choose contract levels that may not necessarily
be optimal.
Price
×
Service Quality Interaction
In addition to main effects of the contract-level measures of
service quality and price, we find that these two factors
interact to influence the firm’s upgrade decision. This nega-
tive interaction effect suggests that, consistent with prior
research, firms that experience higher service quality are
less price sensitive or, conversely, firms that experience
lower service quality are more price sensitive (see Bolton
and Myers 2003). This result also implies that there may be
a ceiling effect for the positive effects of two somewhat
negative aspects of a product (poor quality and high price),
such that when the price is too high or the quality is too
poor, these positive effects are lessened.
Illustrative Simulation Results
It is important for suppliers to assess the “net effect” of
changes in service quality (i.e., main and interaction effects
combined) on the buying firm’s likelihood of upgrading a
service contract. For this supplier, we simulated the effects
of a 5% change in satisfaction, resolution time, and price.
For each of these three scenarios, we calculated the relative
effect on the business customer’s probability of upgrading
the focal service contract (i.e., not the absolute change in
probability). First, a 5% increase in satisfaction leads to a
44% increase in the probability that the business customer
will upgrade a contract (ceteris paribus). Second, for a 5%
increase in resolution time (i.e., service quality is worse for
the focal contract), we find a 25% decrease in upgrade
probability. However, if the firm decreases the resolution
time by 5% (i.e., service quality is better for the focal con-
tract), the upgrade probability increases by 123%. (This
asymmetry arises because there are diminishing marginal
returns to service quality.) Finally, for comparison purposes,
we simulated the effects of a 5% increase in price. The
result is a 6% increase in the upgrade probability (because
customers that pay higher prices are less price sensitive as a
result of reference price effects). Note that modest service
improvements (i.e., average resolution time for the focal
contract decreases by approximately two minutes) have a
much greater impact on business customer behavior than
price changes.
Methodological Contributions
This study’s approach to firm decision making is novel in
several respects. First, it incorporates perceptual measures
(from surveys) and operations measures (from internal
process metrics) into a single model of customer behavior.
Second, the model is estimated with a cross-sectional, lon-
gitudinal database that enables us to show the causal nature
of our findings (i.e., antecedents in time t – 1 affect the
upgrade decision in time t). Third, the inclusion of random
effects in the model enables us to account for individual
Expanding B2B Customer Relationships / 61
firm effects (i.e., customer heterogeneity) and to distinguish
them from the hypothesized effects. Thus, our study shows
that comprehensive customer relationship databases can
help researchers construct richer models of customer
behavior.
Implications for Managerial
Practice
A recent study suggests that in most industries, more cus-
tomers change their spending behavior than defect, imply-
ing that gains due to increases in the breadth of the relation-
ship may account for up to 25% of increases in revenue,
whereas losses due to defections may account for only 3%
of losses in revenue (Coyles and Gokey 2005). This obser-
vation is consistent with our findings. Small increases in the
likelihood of upgrading are potentially highly lucrative.
They yield substantial changes in revenues because there
are many customers purchasing relatively high-priced ser-
vice contracts. Our study offers several insights that enable
marketers to manage their relationships with customers.
Given the enhanced focus on customer relationships and
organic growth, this research is relevant to most industries
that provide goods and services in B2B environments,
including service firms and traditional manufacturers that
now differentiate their offerings by augmenting their tangi-
ble goods with service (e.g., delivery guarantees, contracts,
consulting). In the following subsections, we offer some
guidelines for managers.
Manage at Both the Account and the Contract
Level
Suppliers must consider (and manage) business customer
experiences simultaneously at the account level (i.e., per-
ceptions that encompass the entire relationship) and at the
contract level (e.g., service quality, price, discount) over
time. This observation is consistent with prior qualitative
research that conceptualizes the evolution of B2B relation-
ships in terms of social exchange processes that arise from
discrete episodes or transactions (Narayandas and Rangan
2004).
Understand the Role of Goods and Services
At the account level, suppliers should strive for high cus-
tomer satisfaction and should also monitor information
regarding the objective aspects of the service. The inter-
action between satisfaction and service quality suggests that
account management teams—both sales and service profes-
sionals—within the supplier’s organization must gain a
deep understanding of each business customer’s needs so
that they can justify to the decision maker the need to
upgrade a service contract when satisfaction is low and the
service quality is poor (Piercy and Lane 2006). As one deci-
sion maker remarked, “[The supplier] needs to be more
knowledgeable about our configuration so as to help us
more with purchases.” In other words, customer dissatisfac-
tion can be an opportunity for account management teams
to “migrate” contracts to higher service levels because it
may represent a situation in which the decision maker is
paying particular attention to the service contracts and is in
problem-solving mode. The persuasion process requires an
intimate understanding of the environment within the cus-
tomer organization and how the supplier’s product can serve
it.
Measure and Manage Service Delivery Metrics
Suppliers must also manage the relationship at the individ-
ual contract level. In a B2B context, customers hold many
contracts and the contracts are fairly complex, so contract
management is a significant challenge. Consequently, sup-
pliers should measure and manage service operations met-
rics and price levels at the contract level so that account
managers are aware of the actual (not promised) service
quality levels for every contract the customer purchases and
the current price paid on each contract. This knowledge will
ensure that appropriate levels of service are being delivered
across all contracts. More important, this knowledge will
enable the supplier to identify proactively contracts with
poor service quality delivery because these contracts repre-
sent both a risk and an opportunity. Situations in which the
decision maker perceives the supplier’s management of the
account relationship as satisfactory but objective service
quality varies noticeably across contracts simultaneously
represent an opportunity for upgrading to a higher-level ser-
vice contract and a risk of losing the contract. Account
management teams should view this situation not only as a
critical moment but also as a “teachable moment,” when
they can discuss an upgrade to a higher-level contract and
demonstrate its value (encourage subtractive framing).
Similarly, firms that hold contracts that were negotiated at a
higher price may be key targets for upgrades because they
constitute a market segment that may be more willing to
(and interested in) an upgrade.
Manage a Multiple Service Contract Environment
Popular services management heuristics, such as “under-
promise, overdeliver” or “a rising tide raises all boats” (i.e.,
always attempt to exceed customers’ quality expectations),
are appropriate slogans for encouraging customer retention
but not for encouraging upgrading of service contracts.
Indeed, these approaches are overly simplistic and difficult
to execute in an environment with multiple service con-
tracts. Extremely high service quality on all contracts elim-
inates the critical moments that suppliers can use to demon-
strate the value of a service upgrade for a particular
contract. Moreover, if the supplier delivers extremely high
service quality for some (but not all) service contracts, the
buying firm’s expectation or norm is likely to be higher, so
contracts with good, but not outstanding, service quality
will seem poor in comparison. There may be a tipping point
at which rather than upgrading, the buying firm infers that
service is unsatisfactory and chooses not to renew the
contract.
Improve Customer Decision Making
We must sound a note of caution for managers, such as
chief information officers and telecommunication decision
makers, who are responsible for managing service con-
tracts. From the buying firm’s perspective, our results indi-
62 / Journal of Marketing, January 2008
cate that a decision maker’s perception may color the
contract-level experiences, leading him or her to choose
contract levels that may not be optimal. It behooves the firm
to ensure that the decision maker does not rely solely on his
or her impression of how the account is managed. The deci-
sion maker must be familiar with individual service con-
tracts, collecting information from the end users of the ser-
vice so that he or she knows which contracts should be
upgraded and which should not.
Overall
In summary, our research suggests that the supplier firm
needs to understand the nature and history of customers’
experiences with the firm over time, at both the account and
the contract levels. Both superior account management and
service delivery (through people and technology) are neces-
sary if a supplier firm intends to grow its relationships with
customers by means of service contract upgrades. Achiev-
ing these synergies may be particularly difficult in firms
that isolate marketing, sales, and operations in separate
functional silos. Our research is also related to recent
research in the area of key account management (KAM).
For example, Homburg, Workman, and Jensen (2002) find
that activity proactiveness and activity intensity on the part
of the supplier firm increase KAM effectiveness. Under-
standing and monitoring service quality of each contract is
a specific example of such proactive activity. In addition,
Piercy and Lane (2006) suggest that understanding cus-
tomer requirements is a key element of KAM effectiveness.
The findings of this study suggest specific aspects of cus-
tomer requirements on which supplier firms should focus.
The findings also suggest that suppliers should be asking
the following questions about their relationships with their
business customers: What is happening in the buyer–seller
relationship (i.e., at the account level)? What is happening
(operationally) on a contract-by-contract basis? and What is
happening when we consider the decision maker’s frame of
reference (prior satisfaction, perceptions of the environ-
ment) and place the focal contract in that context? Answers
to these three questions will help suppliers proactively man-
age their relationships with business customers and enable
them to determine which customers may be interested in
upgrading and which service contracts they may be willing
to upgrade.
Concluding Remarks
There are many avenues for further research regarding
upgrading in B2B contexts. First, we conducted our study
in only one industry context. However, we believe that this
limitation is reduced by the richness of our data, which
allows us to eliminate competing explanations for our find-
ings. In a cross-sectional, longitudinal database with multi-
ple contracts at each customer account, we can gain many
insights into the factors that influence the upgrade decision.
Nevertheless, further research should investigate this deci-
sion in different study contexts, as well as experimental set-
tings in which the constructs (e.g., satisfaction, service
quality, price) can be manipulated rather than measured. In
addition, we incorporated heterogeneity across firms
through a random coefficients model. Alternatively, a latent
class model could be used to capture unobserved
heterogeneity.
Second, we were able to link service operations mea-
sures to the customer upgrade decision. However, managers
are interested in understanding which improvements in ser-
vice will “pay off.” Thus, further research should link ser-
vice operations to other business performance metrics, such
as customer lifetime value or customer equity. Moreover, it
should investigate the effect of service operations improve-
ments on new customer acquisition, retention, upgrading,
and cross-buying, perhaps by specifying and estimating sys-
tems of equations. For example, understanding the relative
influence of service quality variables versus other
marketing-mix variables (e.g., price) on these three distinct
customer behaviors would also be helpful.
Third, our analyses focus on short-term (i.e., year-to-
year decisions) rather than long-term decisions, whereby a
customer might choose to stop purchasing any contracts
from a given supplier (i.e., completely terminate the rela-
tionship). Our model provides insight into the customer
upgrade decision over time, but it cannot explain disconti-
nuities, such as when a firm creates a committee to formally
evaluate its relationships with suppliers). A worthwhile
extension of our work would be to examine switching and
upgrade behavior and incorporate transaction costs. It
would be helpful to understand the role of competition and
brand equity in such decisions, an issue that was beyond the
scope of this study.
REFERENCES
Alderson, Wroe (1965), Dynamic Marketing Behavior: A Func-
tionalist Theory of Marketing. Homewood, IL: Richard D.
Irwin.
Ben-Akiva, M. and S.R. Lerman (1985), Discrete Choice Analysis.
Cambridge, MA: MIT Press.
Bitner, Mary Jo, Bernard H. Booms, and Mary Stanfield Tetreault
(1990), “The Service Encounter: Diagnosing Favorable and
Unfavorable Incidents,” Journal of Marketing, 54 (January),
71–84.
Bloemer, José M.M. and Hans D.P. Kasper (1995), “The Complex
Relationship Between Consumer Satisfaction and Brand Loy-
alty,” Journal of Economic Psychology, 16 (2), 311–29.
Bolton, Lisa and Joseph Alba (2006), “Price Fairness: Good and
Service Differences and the Role of Vendor Costs,” Journal of
Consumer Research, 33 (2), 258–65.
Bolton, Ruth N. (1998), “A Dynamic Model of the Duration of the
Customer’s Relationship with a Continuous Service Provider:
The Role of Satisfaction,” Marketing Science, 17 (1), 45–65.
——— and James H. Drew (1991), “A Multistage Model of Cus-
tomers’ Assessments of Service Quality and Value,” Journal of
Consumer Research, 17 (4), 375–84.
———, Katherine N. Lemon, and Matthew D. Bramlett (2006),
“The Effect of Service Experiences Over Time on a Supplier’s
Retention of Business Customers,” Management Science, 52
(12), 1811–23.
Expanding B2B Customer Relationships / 63
———, ———, and Peter C. Verhoef (2004), “The Theoretical
Underpinnings of Customer Asset Management: A Framework
and Propositions for Future Research,” Journal of the Academy
of Marketing Science, 32 (3), 271–92.
——— and Matthew B. Myers (2003), “Price-Based Global Mar-
ket Segmentation for Services,” Journal of Marketing, 67
(July), 108–128.
Boulding, William, Ajay Kalra, and Richard Staelin (1999), “The
Quality Double Whammy,” Marketing Science, 18 (4), 463–84.
———, ———, ———, and Valarie Zeithaml (1993), “A
Dynamic Process Model of Service Quality: From Expecta-
tions to Behavioral Intentions,” Journal of Marketing Research,
30 (February), 7–27.
Brady, M.K., G.A. Knight, J.J. Cronin, G.T.M. Hult, and B.D.
Keillor (2005), “Removing the Contextual Lens: A Multi-
National, Multi-Setting Comparison of Service Evaluation
Models,” Journal of Retailing, 81 (3), 215–30.
Bridge, R. Gary (2004), “New and Better Managed Services—
Maybe,” paper presented at Competing Through Service Sym-
posium, Arizona State University (November).
Cannon, Joseph P. and William D. Perreault Jr. (1999),
“Buyer–Seller Relationships in Business Markets,” Journal of
Marketing Research, 36 (November), 439–60.
Coleman, J.S. (1990), Foundations of Social Theory. Cambridge,
MA: Harvard University Press.
Coughlan, Anne T. (1985), “Competition and Cooperation in Mar-
keting Channel Choice: Theory and Application,” Marketing
Science, 4 (2), 110–39.
Coyles, Stephanie and Timothy C. Gokey (2005), “Customer
Retention Is Not Enough,” Journal of Consumer Marketing, 22
(2), 101–105.
Dickson, Peter (1992), “Toward a General Theory of Competitive
Rationality,” Journal of Marketing, 56 (January), 69–83.
Dodgson, Mark (1993), “Organizational Learning: A Review of
Some Literatures,” Organization Studies, 14 (3), 375–94.
Dwyer, F. Robert, Paul H. Schurr, and Sejo Oh (1987), “Develop-
ing Buyer–Seller Relationships,” Journal of Marketing, 51
(April), 11–27.
Folkes, Valerie S. (1988), “Recent Attribution Research in Con-
sumer Behavior: A Review and New Directions,” Journal of
Consumer Research, 14 (March), 548–65.
Ganesh, Jaishankar, Mark J. Arnold, and Kristy E. Reynolds
(2000), “Understanding the Customer Base of Service
Providers: An Examination of the Differences Between
Switchers and Stayers,” Journal of Marketing, 64 (July),
65–87.
Greene, William H. (2003), LIMDEP VERSION 8.0: Econometric
Modeling Guide, Vol. 1. Plainview, NY: Econometric Software.
Guadagni, Peter and John D.C. Little (1983), “A Logit Model of
Brand Choice Calibrated on Scanner Data,” Marketing Science,
2 (3), 203–238.
Gustafsson, Anders, Micheal D. Johson, and Inger Roos (2005),
“The Effects of Customer Satisfaction, Relationship Commit-
ment Dimensions, and Triggers on Customer Retention,” Jour-
nal of Marketing, 69 (October), 210–18.
Heide, Jan B. and Allen M. Weiss (1995), “Vendor Consideration
and Switching Behavior for Buyers in High-Technology Mar-
kets,” Journal of Marketing, 59 (July), 30–43.
Hendrick, Clyde and Susan S. Hendrick (1988), “Lovers Wear
Rose Colored Glasses,” Journal of Social and Personal Rela-
tionships, 5 (2), 161–83.
Hoch, Stephen J. and Youn-Won Ha (1986), “Consumer Learning:
Advertising and the Ambiguity of Product Experience,” Jour-
nal of Consumer Research, 13 (September), 221–33.
Holbrook, Morris B. (1999), “Introduction to Consumer Value,” in
Consumer Value: A Framework for Analysis and Research,
Morris B. Holbrook, ed. New York: Routledge, 1–28.
Homburg, Christian, John Workman Jr., and Ove Jensen (2002),
“A Configurational Perspective on Key Account Management,”
Journal of Marketing, 66 (April), 38–60.
Inman, J. Jeffrey and Marcel Zeelenberg (2002), “Regret in Repeat
Purchase Versus Switching Decisions: The Attenuating Role of
Decision Justifiability,” Journal of Consumer Research, 29
(June), 116–28.
Jackson, Barbara Bund (1985), “Build Customer Relationships
That Last,” Harvard Business Review, 63 (6), 120–28.
Janiszewski, Chris and Stijn Van Osselaer (2000), “A Connection-
ist Model of Brand–Quality Associations,” Journal of Market-
ing Research, 37 (August), 331–50.
Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler
(1986), “Fairness as a Constraint on Profit Seeking: Entitle-
ments in the Market,” American Economic Review, 76 (4),
728–41.
Kalwani, Manohar U. and Narakesari Narayandas (1995), “Long-
Term Manufacturer–Supplier Relationships: Do They Pay Off
for the Supplier Firm?” Journal of Marketing, 59 (January),
1–16.
Keaveney, Susan M. (1995), “Customer Switching Behavior in
Service Industries: An Exploratory Study,” Journal of Market-
ing, 59 (April), 71–82.
Kraljic, Peter (1983), “Purchasing Must Become Supply Manage-
ment,” Harvard Business Review, (September–October),
109–117.
Kristof, Amy (1996), “Person–Organization Fit: An Integrative
Review of Its Conceptualizations, Measurement and Implica-
tions,” Personnel Psychology, 49 (Spring), 1–49.
Lemon, Katherine N., Tiffany Barnett White, and Russell S. Winer
(2002), “Dynamic Customer Relationship Management: Incor-
porating Future Considerations into the Service Retention
Decision,” Journal of Marketing, 66 (January), 1–14.
Li, Shibo, Baohong Sun, and Ronald T. Wilcox (2005), “Cross-
Selling Sequentially Ordered Products: An Application to Con-
sumer Banking Services,” Journal of Marketing Research, 42
(May), 233–39.
Louviere, Jordan J. (2003), “Complex Statistical Choice Models:
Are the Assumptions True, and If Not, What Are the Conse-
quences?” keynote address, Discrete Choice Workshop in
Health Economics, University of Oxford (September 22).
Mittal, Vikas and Wagner A. Kamakura (2001), “Satisfaction,
Repurchase Intent, and Repurchase Behavior: Investigating the
Moderating Effect of Customer Characteristics,” Journal of
Marketing Research, 38 (February), 131–42.
Narayandas, Das and V. Kasturi Rangan (2004), “Building and
Sustaining Buyer–Seller Relationships in Mature Industrial
Markets,” Journal of Marketing, 68 (July), 63–77.
Nayakankuppam, Dhananjay and Himanshu Mishra (2005), “The
Endowment Effect: Rose-Tinted and Dark-Tinted Glasses,”
Journal of Consumer Research, 32 (December), 390–95.
Ngobo, Paul Valentin (2005), “Drivers of Upward and Downward
Migration: An Empirical Investigation Among Theatergoers,”
International Journal of Research in Marketing, 22 (2),
183–201.
Oliver, Richard L. (1997), Satisfaction: A Behavioral Perspective
on the Consumer. Boston: Irwin/McGraw-Hill.
——— and Raymond R. Burke (1999), “Expectation Processes in
Satisfaction Formation: A Field Study,” Journal of Service
Research, 1 (3), 196–214.
Olsen, Line Lervik and Michael D. Johnson (2003), “Service
Equity, Satisfaction and Loyalty: From Transaction-Specific to
Cumulative Evaluations,” Journal of Service Research, 5 (3),
184–95.
Park, Whan C., Sung Youl Jun, and Deborah J. MacInnis (2000),
“Choosing What I Want Versus Rejecting What I Do Not Want:
An Application of Decision Framing to Product Option Choice
Decisions,” Journal of Marketing Research, 37 (May),
187–202.
64 / Journal of Marketing, January 2008
Piercy, N. and N. Lane (2006), “The Underlying Vulnerabilities in
Key Account Management Strategies,” European Management
Journal, 24 (2–3), 151–62.
Price, Lydia (1996), “Understanding the Negativity Effect: The
Role of Processing Focus,” Marketing Letters, 7 (1), 53–62.
Rust, Roland T., Jeffrey Inman, Jianmin Jia, and Anthony Zahorik
(1999), “What You Don’t Know About Customer-Perceived
Quality: The Role of Customer Expectation Distributions,”
Marketing Science, 18 (1), 77–92.
———, Anthony J. Zahorik, and Timothy L. Keiningham (1995),
“Return on Quality (ROQ): Making Service Quality Finan-
cially Accountable,” Journal of Marketing, 59 (April), 58–70.
Slotegraaf, Rebecca J. and J. Jeffrey Inman (2004), “Longitudinal
Shifts in the Drivers of Satisfaction with Product Quality: The
Role of Attribute Resolvability,” Journal of Marketing
Research, 41 (August), 260–80.
Stremersch, Stefan, Alan Weiss, Benedict Dellaert, and Ruud
Frambach (2003), “Buying Modular Systems in Technology
Intensive Markets,” Journal of Marketing Research, 40
(August), 335–50.
Train, Kenneth E. (2003), Discrete Choice Methods with Simula-
tion. Cambridge, MA: Cambridge University Press.
Vancouver J.B. and N.W. Schmitt (1991), “An Exploratory Exam-
ination of Person-Organization Fit: Organizational Goal Con-
gruence,” Personnel Psychology, 44 (2), 333–52.
Verhoef, Peter C. (2003), “Understanding the Effect of Customer
Relationship Management Efforts on Customer Retention and
Customer Share Development,” Journal of Marketing, 67
(October), 30–45.
———, Philip Hans Franses, and Janny C. Hoekstra (2001), “The
Impact of Satisfaction and Payment Equity on Cross-Buying: A
Dynamic Model for a Multi-Service Provider,” Journal of
Retailing, 77 (Fall), 359–78.
Winer, Russell (1986), “A Reference Price Model of Brand Choice
for Frequently Purchased Products,” Journal of Consumer
Research, 13 (September), 250–56.
Zeithaml, Valarie A. (1988), “Consumer Perceptions of Price,
Quality, and Value: A Means-End Model and Synthesis of Evi-
dence,” Journal of Marketing, 52 (July), 2–22.
