Content uploaded by Forrest V. Morgeson
All content in this area was uploaded by Forrest V. Morgeson on Jul 25, 2020
Content may be subject to copyright.
This is a repository copy of Turning Complaining Customers into Loyal Customers:
Moderators of the Complaint Handling–Customer Loyalty Relationship.
White Rose Research Online URL for this paper:
Version: Accepted Version
Morgeson, FV, Hult, GTM, Mithas, S et al. (2 more authors) (2020) Turning Complaining
Customers into Loyal Customers: Moderators of the Complaint Handling–Customer
Loyalty Relationship. Journal of Marketing. ISSN 0022-2429
© American Marketing Association 2020. This is an author produced version of an article
published in Journal of Marketing. Uploaded in accordance with the publisher's
Items deposited in White Rose Research Online are protected by copyright, with all rights reserved unless
indicated otherwise. They may be downloaded and/or printed for private study, or other acts as permitted by
national copyright laws. The publisher or other rights holders may allow further reproduction and re-use of
the full text version. This is indicated by the licence information on the White Rose Research Online record
for the item.
If you consider content in White Rose Research Online to be in breach of UK law, please notify us by
emailing email@example.com including the URL of the record and the reason for the withdrawal request.
Turning Complaining Customers into Loyal Customers:
Moderators of the Complaint Handling – Customer Loyalty Relationship
Forrest V. Morgeson III, PhD
Assistant Professor of Marketing
Michigan State University
Broad College of Business
G. Tomas M. Hult, PhD
Professor and Byington Endowed Chair
Michigan State University
Broad College of Business
Sunil Mithas, PhD
World Class Scholar and Professor
University of South Florida
Muma College of Business
Timothy Keiningham, PhD
J. Donald Kennedy Endowed Chair
St. John’s University
Peter J. Tobin College of Business
Claes Fornell, PhD
Chair and Founder
American Customer Satisfaction Index, LLC
Exponential ETFs, and CFI Group
Turning Complaining Customers into Loyal Customers:
Moderators of the Complaint Handling – Customer Loyalty Relationship
Firms spend substantial resources responding to customer complaints and the marketing
profession has a long history of supporting that enterprise in order to promote customer loyalty.
We question whether this response is always warranted or whether its effectiveness instead
depends on economic, industry, customer-firm, product/service, and customer segment factors
that may alter the firm’s incentives to compete on complaint management. To consider this
question, we integrate economic and marketing theories and investigate factors that influence the
complaint recovery–customer loyalty relationship via a sample of 35,597 complaining customers
spanning a 10-year period across economic sectors, industries, and firms. Overall, we find that
the recovery–loyalty relationship is stronger in faster-growing economies, for industries with
more competition, for luxury products, and for customers with higher satisfaction and higher
expectations of customization. Conversely, the recovery–loyalty relationship is weaker when
customers’ expectations of product/service reliability are higher, for manufactured goods, and for
males compared to females. We discuss implications of these results for managers,
policymakers, and researchers for more effective management of customer complaints.
Keywords: Customer complaint behavior, complaint recovery, customer loyalty, complaint
management incentives, exit-voice-loyalty theory, customer satisfaction
Although customer complaints and the consequences of a firm’s poor complaint handling
are as old as business itself,1 most marketers agree that the financial stakes are higher in today’s
competitive marketing ecosystem. The speed and flexibility with which information and
communications technologies can be used increase the negative risks of customer complaints and
the importance of effective firm recovery of complaints. For example, social media (e.g.,
Facebook, Instagram, LinkedIn, Pinterest, Reddit, Snapchat, Twitter) has created an environment
where a customer’s negative word-of-mouth is often dramatically amplified. A displeased
customer can complain to a firm and simultaneously to potentially millions of other stakeholders.
In severe cases, the amplified complaint environment can create “online firestorms” of
negative publicity with immense financial consequences (Pfeffer, Zorbach, and Carley 2014;
Herhausen et al. 2019). For instance, the negative publicity that was shaped via social media in
regard to service failures by Chipotle (foodborne illness in 2015) and United Airlines (passenger
boarding issues in 2017) illustrate the consequences of poor service and the heightened criticality
of complaint recovery. The result was costs of billions of U.S. dollars to Chipotle and United
Airlines in market value.
On the other hand, there are loyalty payoffs for firms from effective complaint
management. Importantly, studies show that a customer who experiences a failure and lodges a
complaint can still be satisfied and retained if the firm’s recovery is acceptable (e.g., Fornell and
Wernerfelt 1987, 1988; Smith and Bolton 2002). Because the economic benefits of customer
loyalty are sizeable in terms of a firm’s cash flow and market value (e.g., Shah, Kumar, Kim, and
Choi 2017), especially when considering customer acquisition costs, maintaining a complaint
1The oldest known written customer complaint, which was inscribed about 3,800 years ago (c. 1750 BC) on the
ancient Babylonian “Complaint Tablet of Ea-Nasir,” illustrates that customers have long used threats of defection to
express their dissatisfaction and seek recovery (Kilgrove 2018).
management system that helps retain potentially disloyal customers is an economic imperative
for most firms (e.g., Fornell and Wernerfelt 1987). Practically, this means that firms can turn
dissatisfied customers into future loyal customers, albeit the cost of doing so is often high and
requires considerable effort (e.g., Fornell, Morgeson, Hult, and VanAmburg 2020).
Despite important strides made by prior work, significant gaps remain concerning what
we know about the complaint recovery–customer loyalty relationship and what we need to know
in an increasingly dynamic marketing ecosystem. First, given cost and effort implications, the
differing importance of recovery efforts in driving post-complaint satisfaction and loyalty across
diverse consumer industries is largely unclear and needs to be better understood by firms to
optimize their complaint handling. Based on the literature, we do not know much about cross-
industry and cross-sector differences in the importance of complaint recovery to customer
loyalty. Rather, the extant literature has tended to focus on only a small set of consumer
industries (e.g., Mattila 2001), thereby limiting the generalizability of conclusions.
Second, research on complaint recovery has largely failed to account for the potentially
dynamic nature of the recovery–loyalty relationship as it evolves in complex economic
environments. Many studies imply that complaint recovery has a constant effect on customer
loyalty (e.g., Gelbrich and Roschk 2010). Yet, complaint recovery may increase or decrease in
importance to consumers as a determinant of their customer loyalty as macroeconomic and other
exogenous factors change. Given that many consumer perceptions evolve in response to
economic factors – as evidenced by measures like consumer confidence and consumer sentiment
– the relative importance of complaint behavior and a firm’s responses to complaints is likely to
vary over time as well. These interrelated issues (i.e., the differing importance of recovery across
industries and the dynamic exogenous effects impacting customers) illustrate gaps in our
knowledge of the complaint recovery–customer loyalty relationship.
Against this backdrop, we seek to answer the following overarching research question:
How does the relationship between a firm’s customer complaint recovery (i.e., the customer’s
perception of how well the firm handled a complaint) and customer loyalty vary depending on
influences from economic, industry, customer-firm, product/service, and customer segment
factors? We extend theorizing of the complaint recovery–customer loyalty relationship by
integrating two streams: exit-vice-loyalty theory based in economics (e.g., Hirschman 1970) and
the complaint handling literature grounded in expectations-disconfirmation theory (e.g., Fornell
and Wernerfelt 1987, 1988, Fornell and Westbrook 1984). From these literature bases, we derive
a set of factors and mechanisms that influence customers to be more or less responsive to
complaint handling. These factors and mechanisms are, in turn, likely to impact firms’ incentives
to manage complaints, as they alter the expected loyalty pay-offs from recovery efforts. We then
analyze a large and rich sample of consumer data from the American Customer Satisfaction
Index (ACSI), including a sample of 35,597 complaining customers spanning 10 years across
economic sectors, industries, and firms.
The remainder of the paper is organized as follows. First, we review the complaint
management literature. Second, we outline a contingency model of loyalty returns to complaint
management. From this contingency model, we delineate the factors and mechanisms that both
drive and influence customers’ disposition to firms’ complaint management efforts and firms’
incentives to manage complaints. Third, we describe the ACSI data and methods utilized to
analyze the data. Fourth, we present the results from our analyses. Finally, we offer implications
for managers, policymakers, and researchers, and recommend directions for future research.
The Complaint Management Literature
The literature on customer complaints, firms’ complaint management, and customer
loyalty is diverse, emerging nearly a half-century ago (e.g., Etzel and Silverman 1981; Kendall
and Russ 1975). More importantly, the idea of complaint handling as an important strategic
marketing phenomenon with tangible financial impact for firms has gained significant
momentum over the last two decades. Table 1 summarizes findings from studies on customer
complaints, complaint management, and customer loyalty over this period.
Insert Table 1 about here
Previous research has focused in one of three ways on understanding the conditions under
which customers who experience a failure, or are dissatisfied, and complain remain loyal. First,
the literature has observed intervening consumer-psychological variables that moderate or
mediate the failure, complaint, recovery, and/or loyalty perceptions of customers (Dewitt,
Nguyen, and Marshall 2008; Evanschitzky, Brock, and Blut 2011; Hess, Ganesan, and Klein
2003; McCollough, Berry, and Yadav 2000; Simon, Tossan, and Guesquiere 2015; Tax, Brown,
and Chandrashekaran 1998; Umashankar, Ward, and Dahl 2017. Second, studies have examined
complaint management strategies employed by firms (Homburg and Furst 2005; Smith, Bolton,
and Wagner 1999). Third, research has investigated the “service recovery paradox” under unique
circumstances (e.g., across complaints, relative to complaint frequency, longitudinally) (Knox
and van Oest 2014; Maxham and Netemeyer 2002; Michel and Meuter 2008).
Despite progress, significant gaps remain when it comes to understanding the relationship
between complaint handling (recovery) and customer loyalty. The literature has tended to focus
on a small cross-section of consumer industries. Of the studies in Table 1, a plurality focus either
exclusively or partially on failure, complaint, and recovery with restaurants. A handful focus on
hotel and commercial bank customers. A few are “multi-industry” studies of aggregate samples
of consumers spread across contexts. The first two industries (restaurants and hotels) fall into a
single, unique, and service-intensive economic sector (Accommodation and Food Services),
while the multi-industry studies and the studies of bank customers provide a measure of diversity
and exposure to a different kind of service (Finance and Insurance). Nevertheless, research on
complaint management has thus far examined a narrow cohort of industries compared with the
diverse consumer landscape (e.g., Evanschitzky, Brock, and Blut 2011). Given industries differ
and are characterized by variations that may impact both customers’ loyalty and firms’
incentives to manage complaints, this narrow focus on a small cross-section of consumer
experiences results in gaps in our knowledge and potentially faulty complaint-recovery efforts by
Likewise, while the research methods used so far have been somewhat eclectic, most
studies adopt experimental or mixed-design methods with relatively small samples. Of the
studies in Table 1, eight adopt either only experimental methods or a mixed design incorporating
experimental and consumer survey data. Only one is observational (Knox and van Oest 2014),
tracking complaints and actual future purchase behavior with an online retailer. Virtually all of
the remaining studies focus on some type of surveying (of managers or customers), but use
comparatively small, single-point-in-time cross-sectional sampling techniques. In turn, such
studies fail to fully capture the recovery–loyalty relationship as it evolves in complex
environments marked by variations that influence both customers’ loyalty and firms’ strategies.
From previous research, we draw two conclusions. First, much of the prior complaint
literature focuses on a narrow set of consumer industries, such as restaurants, hotels, and banks.
Already noticing this trend about two decades ago, Mattila (2001, p. 583) suggested that this
focus “…on a single service type … or a specific service industry” has precluded a complete
understanding of the recovery–loyalty relationship, and “consequently, little is known about the
underlying assumptions that cover the entire spectrum.” Second, a large portion of prior studies
use experimental or quasi-experimental methods, and/or analyze small samples of single-point-
in-time cross-sectional data (rather than repeated cross-sectional or longitudinal data). Although
these studies have enriched our understanding, they are not able to effectively inform us about
the influence that the broader, evolving, and dynamic marketing ecosystem has on the
relationship between customer complaint behavior, complaint recovery, and customer loyalty.
Thus, in seeking to answer our research question and to determine if and how the
relationship between a firm’s complaint recovery and a customer’s loyalty vary due to influences
from various factors (i.e., economic, industry, customer-firm, product/service, and customer
segment factors), we seek to close significant knowledge gaps in the complaint recovery
literature. The core focus is on understanding the factors that are stronger/weaker moderators of
the relationship between complaint recovery and customer loyalty, as guided by our contingency
model of loyalty returns to complaint management which follows.
A Contingency Model of Loyalty Returns to Complaint Management
To develop a contingency model of firm-anticipated pay-offs from complaint
management efforts, we synthesize two theories: exit-voice-loyalty theory from economics
(Hirschman 1970) and expectations-disconfirmation theory from marketing (e.g., Fornell et al.
1996; Oliver 1980). These theories illuminate incentives and disincentives that (1) dissatisfied
customers have when making loyalty decisions and (2) firms have to convert complaint recovery
to customer loyalty in their complaint management efforts.
Beginning with exit-voice-loyalty (EVL) theory (Hirschman 1970), a customer who
experiences dissatisfaction with a firm and its products or services has three basic options: (1)
exhibit disloyalty and defect from the firm (i.e., “exit”) to an alternative supplier; (2) complain
and express displeasure to the firm (i.e., “voice”); or (3) do neither, accept the issues causing the
dissatisfaction, and remain “silently loyal” (cf. Dowding, John, Mergoupis, and Van Vugt 2000).
The consumer’s decision about which alternative to pursue is informed by several factors that are
related to the firm (e.g., the firm’s response to “quality deteriorations”) but also external to the
dissatisfying experience. EVL theory focuses primarily on the latter; that is, on industry
conditions and the economic environment surrounding the exchange. These include: the degree
of market competition and the availability of alternatives; the level of investment in or price paid
for the good by the consumer (i.e., the sunk cost); switching costs, the tangible and intangible
costs associated with defecting from one supplier to a competitor; and the individual customer’s
economic situation (and perceived power) at the time of the complaint (Fornell and Davidow
1980; Fornell and Westbrook 1984; Lee and Whitford 2007; Withey and Cooper 1989).
Additionally, much like customers have choices when displeased and making loyalty
decisions, EVL theory specifies that firms have both economic incentives and disincentives to
convert the complaint recovery efforts to customer loyalty outcomes. Take the two extremes of
monopolists and highly competitive markets. Monopolistic firms that market necessity products
during a time of slow economic growth may need to be prepared for greater complaint volume
when quality deterioration results in dissatisfaction. However, because these relatively “weak”
displeased customers are unable to defect and require the good, these monopolistic firms do not
necessarily need to focus on complaint recovery. On the other hand, luxury goods firms in highly
competitive industries with low switching costs during a period of stronger economic growth
have a greater incentive to convert customer complaints to customer loyalty outcomes due to the
reality of relatively frictionless customer defection.
Augmenting EVL theory, we also draw from expectations-disconfirmation theory
(Fornell et al. 1996; McCollough, Berry, and Yadav 2000; Smith and Bolton 2002; Smith,
Bolton, and Wagner 1999; Tax, Brown, and Chandrashekaran 1998). Expectations-
disconfirmation theory and the customer satisfaction perspective focus on the customer-firm
relationship and view loyalty as a function of (1) pre-experience consumer expectations
(positively related to satisfaction and loyalty, unless negatively disconfirmed); (2) the customer’s
expected vs. experienced quality (positively related to satisfaction and loyalty, if a positive gap);
and (3) customers’ overall satisfaction (or fulfillment) with the consumption experience (strongly
and positively related to loyalty). Through the lens of expectations-disconfirmation theory,
confirmed (high) expectations or a positive expectations gap predict stronger customer
perceptions of quality and satisfaction, and thus a stronger customer loyalty likelihood. However,
when the customer has experienced negative expectations disconfirmation, poor quality, and
dissatisfaction, and has chosen to voice this discontent to the firm, customers may be more likely
to defect. A firm’s complaint handling and system to manage the recovery–loyalty relationship is
a reaction to a higher probability of customer disloyalty designed to minimize defection.
Drawing from EVL theory, expectations-disconfirmation theory, and the literature on
customer satisfaction, we next identify factors likely to influence the recovery–loyalty
relationship. Specifically, we argue that this relationship is likely to vary due to a set of
characteristics associated with: (1) economic factors, such as economic growth, surrounding the
complaint and recovery; (2) industry factors, such as industry competitiveness, which impacts
consumers’ switching costs and the availability of alternative suppliers; (3) customer-firm factors
– i.e., customer satisfaction and expectations — both of which frame the complaint and recovery
experience (4) product/service factors, such as whether the good consumed is a lower-priced
necessity good vs. a higher-priced luxury good, or a service vs. a manufactured good; and (5)
customer segment factors (e.g., income, gender, age cohort, and region of residence) related to
the group that is served.
Each of these five factors is observable, but we argue that they effect the recovery–
loyalty relationship through a set of unobserved mechanisms, as shown in Figure 1. The
mechanisms are (1) consumer power; (2) alternatives, switching costs, and barriers; (3) a
negative expectation-disconfirmation gap; (4) a reservoir of consumer goodwill, and (5) latent
segment membership. These mechanisms arise from EVL theory, expectations-disconfirmation
theory, and the literature on customer satisfaction as we highlight in the sections that follow.
Table 2 summarizes the variables that affect the recovery–loyalty relationship and the influential
mechanisms and factors involved.
Insert Figure 1 and Table 2 about here
We predict that the recovery–loyalty relationship is influenced by economic factors (e.g.,
Fornell, Rust, and DeKimpe 2010; Kumar, Umashankar, Kim, and Bhagwat 2014). Specifically,
we expect that a faster-growing economy will positively moderate the link between complaint
recovery and customer loyalty. The reason for this positive moderation is due to the fact that
economic growth is typically accompanied by a variety of features that result in more powerful
consumers (e.g., lower unemployment, stronger income growth, more consumer spending,
stronger consumer confidence). This increased consumer power (e.g., Dubois, Rucker, and
Galinsky 2012; Kim, Park, and Dubois 2018) lead consumers to perceive the market as having
lower switching costs and more viable alternative suppliers, easing defection and disloyalty. As
such, during these faster-growing economic periods firms will be even more determined to
overcome customer complaints effectively and keep customers loyal. Stronger economic growth
will thus positively moderate the link between complaint recovery and customer loyalty, and
firms may have a stronger incentive to convert complaining customers into enduringly loyal
customers via complaint management during these periods.
We predict that the importance of complaint recovery to customer loyalty is not constant
but varies across industries and economic sectors (e.g., Bamiatzi, Bozos, Cavusgil, and Hult
2016; Short, Ketchen, Palmer, and Hult 2007). This variation is due to the diversity of the
competitive economic contexts experienced by consumers. We predict that the most fundamental
factor influencing variance in the recovery–loyalty relationship across industries and sectors is
the degree of competition. We argue that complaint recovery will exhibit a weaker (stronger)
effect on customer loyalty in less (more) competitive industries. This is because in more
competitive industry contexts, customers will recognize their ability to more easily switch to
alternative suppliers and also recognize their greater power vis-à-vis the firm. As such, in more
competitive industries, the expectation is that a stronger relationship will exist between
complaint management efforts by the firm and customers’ future loyalty. Consequently, due to
this competitive industry dynamic, firms will have stronger incentives to manage complaints
given the increased importance customers place on the recovery-loyalty relationship in more
Three customer-firm factors are expected to be important influencers of the recovery-
loyalty relationship (e.g., Fornell, Morgeson, Hult, and VanAmburg 2020). We predict that
customers’ satisfaction and their pre-experience expectations of both the customizability and the
reliability of the products/services consumed will moderate the recovery–loyalty relationship.
Beginning with customer satisfaction, which is defined as the customer’s overall fulfillment
response to a consumption experience (e.g., Fornell 2007; Fornell, Morgeson, Hult, and
VanAmburg 2020; Oliver 2010), we anticipate positive moderation of the recovery-loyalty
relationship. As measured in this study, customer satisfaction is a cumulative phenomenon
reflecting the totality of the consumers’ experiences with the firm. Effectively, this form of
satisfaction can be viewed to represent (a proxy for) the consumer’s reservoir of goodwill toward
the firm and the product/service based in buyer habit and brand identification developed (in
many cases) over a lengthy and deeper customer-firm relationship. Part and parcel to this
relationship, however, is the consumer’s demand that the trusted firm will “go the extra mile” to
resolve a problem when it occurs, as a way to reaffirm the relationship and ensure future loyalty.
Regarding the effect of expectations of customizability, defined as the customer’s pre-
experience perceptions of the product/service’s abilities to meet personal requirements, we
predict a positive moderating effect. Customers with higher customization expectations
anticipate more individualized service from the firm in all areas, including during a failure and
recovery. Higher expectations of customization are likely to lead the customer to demand
personalized service during the recovery and, by design, a heightened positive relationship
between recovery efforts and loyalty. In effect, firms have a greater incentive to manage
complaints to secure loyalty due to higher expectations of customizability.
Regarding expectations of reliability, which we define as the customer’s pre-experience
perceptions of the probability of a lack of failure with the product/service, we predict a negative
moderating effect on the recovery-loyalty relationship. Customers’ stronger expectations of
reliability with a firm are generally created through either multiple problem-free consumption
experiences or through advertising or other marketing communications promising problem-free
experiences. In the event of a failure, however, the result will be a large negative expectations-
disconfirmation gap. Consequently, theoretically we predict that this disconfirmation gap will
negatively frame (and weaken) the consumer’s response to a firm’s complaint recovery efforts
vis-à-vis their loyalty intentions. This is because the unexpected failure resulting in the complaint
and recovery attempt is, from the customer’s perspective, reflective of either a fundamental
disruption of a long problem-free relationship or an indication that the firm’s promises are
Product and Service Factors
We predict that the categorization of the product or service as a necessity good or a
discretionary luxury good is a factor that moderates the recovery–loyalty relationship (e.g.,
Berger and Ward 2010). We define necessity goods as basic products and services customers
often require and therefore must purchase (even when, for example, income is low or declining),
and/or as lower-cost goods for which more expensive substitute goods exist. Discretionary
luxury goods, on the other hand, are defined as superior (and typically more expensive) products
and services sought-out by the customer (often as income is high or rising), even though less
expensive substitute alternative goods are available. Our expectation is that luxury goods
customers will typically have greater financial resources and thus the ability to switch to
alternative luxury providers or less expensive replacement goods more easily. Specifically,
luxury goods customers tend to be financially better-off (e.g., Mandel, Petrova, and Cialdini
2006), and hence they are anticipated to be less impacted by the loyalty-inducing constraints of
sunk costs from earlier purchases as a barrier to switching. On the other hand, if the product or
service is considered a necessity the situation is often reversed, and this – combined with the fact
that this category of goods typically has lower profit margins – decrease demands on firms to
manage customer complaints. Hence, we expect positive moderation of the recovery–loyalty
relationship among luxury goods consumers, and larger loyalty pay-offs via recovery efforts for
firms selling luxury goods.
Likewise, we anticipate that the importance of complaint recovery to customer loyalty
varies between customers of services and manufactured goods. In particular, complaint recovery
will have a weaker effect on loyalty for customers of manufactured goods relative to customers
of services (i.e., negative moderation). For a significant proportion of manufactured goods, such
as frequently purchased and inexpensive nondurable goods, customer complaint behavior is itself
far less likely following a dissatisfying experience. That is, customers are less likely to seek
recovery when displeased with this class of nondurable goods, choosing to either remain silently
loyal or to defect without complaint (Fornell, Morgeson, Hult, and VanAmburg 2020). This
suggests that, in the aggregate, complaint recovery is relatively less important to loyalty
decisions for these necessity goods (often price-based commodities), and possibly also among
the smaller group of customers who do complain. Moreover, prior research has confirmed that
complaint recovery after a failure, as a type of interactional justice, has a stronger effect on
loyalty in personal services contexts relative to less personal non-services goods (Gelbrich and
Roschk 2010; Malshe and Agarwal 2015), supporting the negative moderation of the recovery-
loyalty relationship for manufactured goods.
Customer Segment Factors
We examine four customer segment factors (i.e., customer age, gender, income, and
region of residence) that will potentially influence the complaint recovery–customer loyalty
relationship. Given the context and focus of our study, these customer segment factors are
important inclusions in the analyses to holistically understand the recovery–loyalty link.
However, limited theoretical and empirical evidence exists regarding the nature of the potential
moderation for these factors within the complaint management literature. Consequently, we draw
on research and seek guidance from the related literature regarding influencers of the customer
satisfaction and customer loyalty relationship (e.g., Cronin, Brady, and Hult 2000). We also draw
broadly on the consumer behavior literature related to age, gender, income, and region.
Beginning with income (Kapferer and Bastien 2009), research indicates that customer
satisfaction is less influential as a determinant of loyalty for wealthier consumers (Mandel,
Petrova, and Cialdini 2006), possibly due to a more expansive choice set and lower barriers to
switching, and thus so too might dissatisfaction and ratings of complaint recovery matter less to
loyalty. This suggests a negative moderating effect for income (Walsh, Evanschitzky, and
Wunderlich 2008) on the recovery-loyalty relationship.
On the other hand, research has shown that generally a stronger customer satisfaction-
customer loyalty relationship exist among women than among men, in particular as it relates to
individual providers, brands, and exchanges (e.g., Fornell 1997; Melnyk, Osselaer, and Bijmolt
2009). As a result, we predict a stronger complaint recovery–customer loyalty relationship
among females (Homburg and Giering 2001).
Considering age and generational cohort, research has shown that the impact of
satisfaction on loyalty increases with age, possibly due to these customers’ stronger reliance on
their own evaluative abilities developed through lengthy personal experience. For this reason,
complaint recovery may likewise more strongly impact customer loyalty for the older
generational cohorts (Homburg and Giering 2001; Walsh, Evanschitzky, and Wunderlich 2008).
Finally, while it is reasonable to anticipate an effect on the recovery-loyalty relationship
across regions within the United States (cf. Kim, Park, and Dubois 2018), given the prevalence
of geography-specific marketing strategies (“geo-marketing”) deployed by national firms such as
mobile-service providers (“geo-fencing”), no theory or research offers strong predictions for
moderation of the recovery-loyalty relationship based on customers’ regions of residence.
Sample and Data
To test how the factors in Figure 1 impact the complaint recovery–customer loyalty
relationship, we analyze a 10-year period of data drawn from the large-scale samples included in
the American Customer Satisfaction Index (ACSI). Since 1994, the ACSI has annually
interviewed customers of the largest firms in the U.S. economy. ACSI measures customer
satisfaction as its central focus but includes additional variables on customer complaint behavior,
complaint recovery, and post-complaint repurchase intention, among others (e.g., Fornell et al.
1996; Fornell, Morgeson, and Hult 2016; Hult et al. 2017; Johnson and Fornell 1991;
Keiningham, Morgeson, Aksoy, and Williams 2014; Morgeson, Mithas, Keiningham, and Aksoy
2011). Only the most economically significant firms with the largest market shares in an industry
are included in the ACSI sample each year, resulting in a dataset that primarily include
customers of Fortune 1000 consumer products and services companies.
The ACSI sample analyzed covers a recent 10-year period (2005 to 2014). We began
with a sample that includes 41 distinct industry categories which span seven of the ten NAICS
economic sectors (Manufacturing, Retail Trade, Transportation and Warehousing, Information,
Finance and Insurance, Health Care and Social Assistance, and Accommodation and Food
Service – see Appendix 1 for more detail on the sectors and industries). After excluding non-
complaining respondents and ensuring availability of at least 25 non-missing firm-year
observations for firms/brands, we have a sample of n = 35,597 complaining customers across
firms, industries, and economic sectors with data available on all relevant variables. The volume
of responses in our dataset is significantly larger than what has been studied in prior customer
complaint studies (see Table 1) and provides an opportunity to more deeply understand the roles
of the factors and mechanisms in Figure 1 as they pertain to the recovery–loyalty relationship.
Specifically, this rich ACSI sample enables us to rigorously assess how the relationship between
recovery and loyalty varies across the factors (i.e., economic, industry, customer-firm,
product/service, and customer segment factors).2
2Data from two economic sectors were removed prior to analysis – Energy Utilities (gas and electric power) and
Public Administration. Data for these sectors differ from the remaining sectors in the ACSI. Energy Utilities
includes a far larger number of companies (nearly 30) than the average ACSI industry – due to regional monopolies
in the industry – and thus includes far more completed interviews, a fact that could bias our aggregate model results.
Regarding public administration, the study parameters for this sector were changed by ACSI in 2007, and samples
before and after that date have only limited comparability. Pre-testing confirmed suspicions, and thus the data from
the two sectors were eliminated to not confound our findings.
The variables used to operationalize the core factors (customer loyalty and customer
complaint handling) and moderating factors (i.e., economic, industry, customer-firm,
product/service, and customer segment factors), obtained from the ACSI dataset as well as
several secondary data sources, are detailed in Table 3. The core variables of loyalty and
complaint handling were measured via survey variables as a part of the data collection efforts by
the American Customer Satisfaction Index. Customer loyalty is operationalized via a variable
measuring the customer’s stated likelihood to repurchase from the same firm in the future
(REPUR). Complaint handling (recovery) is measured as a variable that assesses how well, or
poorly, a customer’s most recent complaint was handled (HANDLE).
The moderators were assessed via a combination of survey data from ACSI and objective
data from the U.S. Bureau of Economic Analysis, Compustat (obtained via the Wharton
Research Data Services), NAICS codes, and the U.S. Census’ Regions and Divisions of the
United States (see Table 3). To represent the economic factors, we use quarterly changes in
annualized U.S. gross domestic product growth (GDPGR). Industry factors are represented by
the degree of competition in an industry, operationalized with the Herfindahl-Hirschman Index
(HHI). The customer-firm factors are operationalized as the respondents’ overall, cumulative
customer satisfaction with the purchase and consumption experience (SATIS), and the customers’
pre-experience expectations regarding both the customizability (CUSTOMX) and reliability
(RELYX) of the good. Product and service factors – necessity vs. luxury goods and services vs.
manufactured goods – are measured via the LUXURY and MFG variables described in Table 3.
The customer segment factors are operationalized through latent membership in various
demographic groups. These include income (INCDUM), measured categorically as the
respondent’s total annual household income and transformed (based on the sample median) to a
low-high dummy variable; the respondent’s gender, self-identified as male or female (FEMALE);
customer age, measured as membership in one of four generational cohorts (Silent Generation,
Baby Boomers, Generation X, and Millennials) and operationalized as three dummy variables
(BOOMDUM, GENXDUM, and MILLDUM); and region of residence in the United States,
measured as the West, Northeast, Midwest, or Southeast regions and operationalized via three
dummy variables (NEDUM, MIDWDUM, and SOUTHDUM).
Insert Table 3 about here
Table 4 reports descriptive statistics and correlations, including summary statistics, for all
of the variables included in the model. For the core variables (REPUR and HANDLE), the mean
score for repurchase intention across all ACSI respondents (n = 319,330) during the study period,
including complainant and non-complainant customers, is 8.05 (on a 1–10 scale, “very unlikely”
to “very likely”). The score drops significantly (p<0.01) to 6.19 among complaining customers.
The mean complaint rate across all sectors and years in the full sample of customers is 11.1
percent, meaning that over the 10-year study period roughly one in nine respondents had a
product or service failure or other source of dissatisfaction about which they complained. The
average complaint recovery (i.e., complaint handling) score is 6.31 (on a 1–10 scale ranging
from “very poor” to “very well”), slightly higher than the customer loyalty score. None of the
correlations in Table 4 are unusually high. Regarding potential concerns about multicollinearity,
the final model has average variance inflation factors of less than 10.3
3The model contains a few variables for which the maximum VIF is greater than 10. However, with the exception of
the interaction involving Generation X and the HANDLE variable, all other variables are statistically significant
despite the high VIF. As Disatnik and Sivan (2016) note, multicollinearity should be of less concern when high VIFs
are due to product terms in interactions. Nonetheless, we further verified that results for the moderators are stable
when we enter them sequentially in blocks.
Insert Table 4 about here
We examine our research question and the theoretically developed contingency model of
loyalty returns to complaint management by examining the effects of customer complaint
handling (HANDLE) on customer repurchase intention (REPUR) while simultaneously
examining how this relationship is moderated by economic (GDPGR), industry (HHI), customer-
firm (SATIS, CUSTOMX, and RELYX), product/service (LUXURY, MFG), and customer segment
factors (INCDUM, FEMALE, BOOMDUM, GENXDUM, MILLDUM, NEDUM, MIDWDUM,
and SOUTHDUM). Given that the nesting of customers in the same firm/brand within an
industry/sector across multiple years creates a multilevel structure, we use hierarchical linear
modeling (HLM) to analyze the complaint handling/recovery–repurchase intention/loyalty
relationships (e.g., Hofmann 1997; Raudenbush and Bryk 2002).4
Analysis of multilevel data poses three types of potential estimation difficulties relevant
to our study: aggregation bias, misestimated errors, and heterogeneity of regression. First,
aggregation bias occurs when a variable takes different meanings at different levels of analysis.
For example, by aggregating individual customer ratings for complaint recovery data across
firms, we can conceptualize how firms vary in their ability to handle customer complaints.
Recovery can be assessed at both the customer and firm levels by aggregating customer-level
data.5 HLM addresses these potential confounding effects on variable interpretation by
decomposing the effects of variables at separate levels. Second, misestimated standard errors
4The authors gratefully acknowledge the input of Stephen Raudenbush on the HLM modeling.
5Indeed, we make use of this property in an exploratory analysis when we add the mean of the complaint handling
variable in the model for the intercept at Level 2. Our key findings for the moderating effects of HHI, GDPGR, and
LUXURY remain unchanged when we do so, and we find that the mean complaint recovery variable in Level 2 for
the intercept is positive and statistically significant. This suggests that firms with better complaint management have
higher customer loyalty, even after controlling for an individual customer’s assessment of complaint handling.
may arise as a result of failure to account for the dependence of observations, in this case within
a firm in an economic sector or for a particular year. HLM avoids this problem by incorporating
a unique random effect for each firm-year. Third, heterogeneity of regression could arise when
relationships between complaint recovery and loyalty vary across sectors or years. HLM permits
the modeling of variation in the intercepts and slopes of loyalty across firm-years by utilizing
industry or economic characteristics, such as HHI or GDPGR, as Level 2 variables.
The HLM analyses are conducted incrementally in three steps. In Step 1, we partition the
total variance in customer loyalty into levels (“within” variance at the customer level and
“between” variance across firm-years) through a fully unconditional model. This model specifies
no predictors at the customer (Level 1) or firm/year (Level 2) levels. In Step 2 of the HLM
analysis, we fit a random coefficients regression model by allowing predictors at the customer
level only (Level 1). The random coefficients regression model provides Level 1 coefficients that
can subsequently be modeled with Level 2 variables. In Step 3 of the HLM analysis, we model
the randomly varying intercepts and slope coefficients (obtained in Step 2) through Level 2
predictors. Thus, we estimate the following equations at the customer and firm/year levels.
The Level 1 model is:
Yijt = 0jt + 1jt*(HANDLEijt)
+ 2jt*(SATISijt) + 3jt*(CUSTOMXijt) + 4jt*(RELYXijt)
+ 5jt*(FEMALEijt) + 6jt*(INCDUMijt) + 7jt*(MILLDUMijt) + 8jt*(GENXDUMijt) + 9jt*(BOOMDUMijt)
+ 10jt*(NEDUMijt) + 11jt*(MIDWDUMijt) + 12jt*(SOTHDUMijt) + 13jt*(IMRijt)
+ 14jt*(SATIS * HANDLEijt) + 15jt*(CUSTOMX * HANDLEijt) + 16jt*(RELYX * HANDLEijt)
+ 17jt*(FEMALE *HANDLEijt) + 18jt*(INCDUM * HANDLEijt)
+ 19jt*(NEDUM * HANDLEijt) + 20jt*(MIDWDUM * HANDLEijt) + 21jt* (SOTHDUM * HANDLEijt)
+ 22jt* (MILLDUM* HANDLE) + 23jt*(GENXDUM * HANDLEijt) + 24jt*(BOOMDUM * HANDLEijt)
where Y represents the individual customer’s repurchase intention (customer loyalty) rating
(REPUR) as an outcome variable, and subscript i indexes customers, subscript j indexes firms
(nested in sectors), and subscript t indexes years. Explanatory variables at Level 1 include the
customer complaint handling rating (HANDLE); customer satisfaction (SATIS); expectations of
customization (CUSTOMX); expectations of reliability (RELYX); a gender dummy variable
(FEMALE); an income dummy variable (INCDUM); the respondent age cohort represented by
the dummy variables MILLDUM, GENXDUM, and BOOMDUM; and geographical regions
represented by the dummy variables NEDUM, MIDWDUM, and SOTHDUM. The Level 1 model
also includes interaction terms involving HANDLE and the individual-level variables SATIS,
CUSTOMX, RELYX, FEMALE, INCDUM, region dummies (e.g., NEDUM, MIDWDUM,
SOTHDUM), and age/generational cohort dummies (MILLDUM, GENXDUM, BOOMDUM).
We centered all variables at Level 1 before creating the interaction terms, as explained below.
Finally, we include an Inverse Mills Ratio (IMR) in the Level 1 model to account for any
potentially non-random selection in that the sample of complaining customers may be different
from those who did not complain. We used a Probit model for calculating the IMR and in that
model we included a variable representing the fraction of complaints in a particular year for a
particular firm as an instrumental variable. We verified the relevance of this variable and it was
positive and significant in the first-stage equation. This instrumental variable also satisfies the
exclusion restriction conceptually because a particular customer’s loyalty to a firm is unlikely to
be related to what fraction of customers of that firm choose to voice their complaints.
At Level 2, we model the intercept and slope of the recovery–loyalty relationship by the
three economic, industry, and product/service moderators: necessity vs. luxury goods
(LUXURY), services vs. manufactured goods (MFG), GDP Growth (GDPGR), and the
Herfindahl-Hirschman Index (HHI). We fixed all other slopes. Thus, the Level 2 models are:
0jt = 00 + 01*(LUXURYjt) + 02*(GDPGRjt) + 03*(HHIjt) + 04*(MFGjt) + u0jt
1jt = 10 + 41*(LUXURYjt) + 12*(GDPGRjt) + 13*(HHIjt) + 14*(MFGjt) + u4jt
xjt = x0 if x =2-24
To estimate the coefficients, we account for differential precision of the information
provided by each firm-year using the generalized least squares (GLS) procedure. Additionally,
because the customers and Level 1 parameters vary across firm-years, we employ an iterative
technique using an expectation maximization algorithm and Fisher scoring to obtain maximum
likelihood estimates of the Level 1 and Level 2 variance components (Raudenbush et al. 2016).
We centered the variables as suggested by Raudenbusch and Bryk (2002). In the Level 1
model, because our primary interest is in modeling the recoveryloyalty relationship, and while
we use LUXURY, MFG, GDPGR and HHI at Level 2, we do not expect these variables to explain
the entire variance in the slope. Hence, we allow the slope of HANDLE to vary across firm-years,
and we group-mean center the HANDLE variable across firm-years (Raudenbush and Bryk
2002). For the remaining predictors at Level 1 (i.e., SATIS, CUSTOMX, RELYX, FEMALE,
INCDUM, MILLDUM, GENXDUM, BOOMDUM, NEDUM, MIDWDUM, and SOTHDUM), we
constrain the variances of their slope to be zero at Level 2 across firm-years and we grand-mean
centered these variables. For the interaction terms involving HANDLE at Level 1, we used
group-mean centered HANDLE, grand-mean centered satisfaction and expectations variables
(SATIS, CUSTOMX, and RELYX), and uncentered INCDUM and FEMALE variables. Use of
such centering decisions at Level 1 implies that the intercept at Level 1 represents loyalty for a
customer with an average rating of HANDLE within a firm-year and at the average values of all
other variables in our sample. At Level 2, one can either grand-mean center variables or leave
them uncentered (see Raudenbush and Bryk 2002, pp. 32-35); we use uncentered variables for
Level 2 (LUXURY, MFG, HHI, and GDPGR) for easier interpretation of results.6
6Our results for the key moderators of the recovery–loyalty relationship are qualitatively similar and robust even if
we use different centering choices such as group-mean centering of customer expectations and customer satisfaction
variables at Level 1, and interaction of group-mean centered HANDLE with such group-mean centered variables.
From the model specifications, we first assess the model fit improvement by comparing
the fully unconditional model (FUM), which specifies no predictors at either the customer (Level
1) or firm-year (Level 2) levels, the random coefficients (RC) model which allows predictors at
the customer level (Level 1) only, and the “full” model with randomly varying intercepts and
slope coefficients. Based on the AIC, BIC and deviance values, we find the “full” model reported
in Table 5 to be significantly better than the FUM and RC models. We next discuss the results
for the complaint recovery–customer loyalty relationship (slope) and the moderating effects of
the various factors we examine on this relationship (economic, industry, customer-firm,
product/service, and customer segment factors), followed by the intercept results. To
complement Table 5, Figure 2 summarizes the economic significance of the various moderators
of the recovery–loyalty relationship.
Insert Table 5 and Figure 2 about here
Main Effect Results
Before discussing the moderating effects, we report on the main effects of key study
variables on customer loyalty (repurchase intention) at the group-mean value of complaint
recovery in Table 5 (see Panel A). Among the customer-firm factors, customer satisfaction
(Coefficient = 0.664, p <.001) and expectations of customization (Coefficient = 0.049, p <.001)
positively and significantly influence customer loyalty, while customer expectations of
product/service reliability negatively influence loyalty (Coefficient = -0.011, p <.05). These
findings suggest that customers who are more demanding of customizability are also more loyal,
while those who anticipate stronger reliability are relatively more fickle. The results are
consistent with theoretical reasoning and the related literature that highlights the importance of
competing based on differentiation and customization rather than on cost or reliability (e.g.,
Rust, Moorman, and Dickson 2002).
Among the results for the customer segment factors, at the mean value of complaint
handling, higher income households tend to have higher customer loyalty (Coefficient = 0.062,
p<.01) compared to those with lower income. However, at the mean value of complaint handling,
there are no differences in loyalty across men and women (Coefficient = -0.018, ns). We find
that Millennials and Generation X customers have lower loyalty when compared to those from
the reference group of the Silent Generation, and that those residing in the Midwest and
Southeast have lower loyalty than those from the West. Because there is no strong theory for
predicting differences in loyalty across regions (cf. Kim, Park, and Dubois 2018), we avoid
overinterpretation of these results but document them here for further research and theorizing.
Finally, the instrumental Inverse Mills Ratio, added to our models to control for potential non-
selection bias between complaining and non-complaining customers, shows a positive and
significant effect on loyalty, as expected (Coefficient = 0.510, p<.01).
Results for intercept modeling. Although our principal interest in this study is to
understand the factors that moderate the recovery–loyalty relationship, we also provide
complementary results from modeling of the intercept in Table 5 (see Panel B). First, there is no
statistically significant difference in mean repurchase intention (customer loyalty) related to
GDP growth (Coefficient = 0.006, ns) and being a provider of luxury goods (Coefficient = -
0.004, ns). Second, we find that mean repurchase intention is higher (Coefficient = 8.680,
p<.001) for industries with higher market concentration (HHI), as one would expect when
customers have few or no viable product or service alternatives and higher barriers to switching.
Third, manufacturing firms have on average lower customer repurchase intention than service
firms (Coefficient = -0.989, p<.001).
Predicted Moderating Effect Results
Economic factors. Table 5 (Panel D) indicates that Gross Domestic Product growth
(Coefficient = 0.006, p < .05) has a positive and statistically significant moderating effect on the
recovery–loyalty relationship, meaning that complaint recovery is positively enhanced under
these conditions. In terms of economic significance, changing the score on the GDPGR variable
by 3.6 (from one standard deviation below the mean to one standard deviation above the mean)
is associated with a change in slope of the recovery–loyalty relationship by a 9.4 percent increase
in the mean HANDLE slope to 0.022 (i.e., 40 in Table 5). This finding suggests that loyalty
payoffs from customer complaint handling are stronger when the economy is doing relatively
better, and that managers should not underinvest in complaint handling when market conditions
are otherwise favorable.
Industry factors. Turning to the industry factors and cross-sectoral differences, HHI
negatively and significantly moderates the recovery–loyalty relationship (Coefficient = -1.413, p
< .01). This means that firms in more concentrated industries derive fewer benefits from
complaint handling to drive future customer loyalty than those in more competitive industries. In
terms of economic significance, changing the score on the HHI variable by 0.02 (from one
standard deviation below the mean to one standard deviation above the mean) is associated with
a change in slope of the recovery–loyalty relationship of 0.028, which is a 12.3 percent decrease
in the mean HANDLE slope of 0.229 in Table 5 (Panel D). Put differently, this finding indicates
that complaint handling is more important for loyalty in more competitive sectors where
consumers have a larger number of viable alternative suppliers to choose from (and potentially
defect to) and thus have lower switching barriers than in the opposite. These results are
consistent with theory and the mechanisms impacting the recovery–loyalty relationship. That is,
while customers of monopolists or firms in more concentrated industries may indeed care about
effective complaint recovery, they also understand that defection due to poor complaint handling
may not be an option (e.g., Fornell and Wernerfelt 1987, 1988; Hirschman 1970). Although
dissatisfied enough to complain, the customers’ narrow (or non-existent) alternative choice-set
(i.e., fewer/no alternative supplier options) delimits the importance of complaint recovery to their
customer loyalty intentions. As such, firms in more competitive industries should anticipate
higher payoffs from complaint recovery.
Customer-firm factors. Among the examined customer-firm relationship variables,
findings indicate that while expectations of customization positively and significantly moderate
the recovery–loyalty relationship (Coefficient = 0.009, p <.01), customer expectations of
product/service reliability negatively moderate the relationship (Coefficient = -0.003, p <.05). In
terms of economic significance, changing the score on the CUSTOMX variable by 4.14 (from
one standard deviation below the mean to one standard deviation above the mean) is associated
with a change in slope of the recovery–loyalty relationship of 0.037, which is a 16.7 percent
increase in the mean HANDLE slope of 0.229 in Table 5. In contrast, changing the score on the
RELYX variable by 5.10 (from one standard deviation below the mean to one standard deviation
above the mean) is associated with a change in slope of the recovery–loyalty relationship of
0.015, which is approximately a 6.7 percent decrease in the mean HANDLE slope of 0.229 in
Table 5 (Panel C). These findings indicate a stronger effect of complaint handling on customer
loyalty for firms with customers who are, on average, more demanding of goods customizable to
their personal use, though the opposite is true of firms whose customers have higher expectations
Our findings also indicate a positive moderating effect of customer satisfaction on the
recovery–loyalty relationship (Coefficient = 0.015, p <.001). In terms of economic significance,
changing the score on the SATIS variable by 5.22 (from one standard deviation below the mean
to one standard deviation above the mean) is associated with a change in slope of the recovery–
loyalty relationship by 0.078, which is about a 34.2 percent increase in the mean HANDLE slope
of 0.229 in Table 5 (Panel C). This finding extends prior work in that it shows the relatively high
returns on customer satisfaction (cf. Otto, Szymanski, and Varadarajan 2020) through its
moderating effect on the recovery–loyalty relationship. These results are interesting given that
customer satisfaction theory also suggests that firms with higher customer satisfaction are
relatively insulated from occasionally less-exceptional complaint handling to secure future
customer loyalty (Fornell, Mithas, Morgeson, and Krishnan 2006). Taken together, these
findings indicate that the customer-firm relationship provides important information about
variance in the recovery-loyalty relationship, and thus in firms’ expected payoffs as increased
loyalty through complaint management.
Product and service factors. Regarding the moderating effect of the product/service
factors on the recovery–loyalty relationship, we find that the variable for necessity vs. luxury
goods positively moderates the relationship (Coefficient = 0.010, p < .05). In terms of economic
significance, changing the score on the LUXURY variable by 3.2 (from one standard deviation
below the mean to one standard deviation above the mean) is associated with a change in slope
of the recovery–loyalty relationship of 0.032, which is a 14 percent increase in the mean
HANDLE slope of 0.229 in Table 5 (Panel D). This finding is consistent with our theory and
mechanisms and suggests that firms providing goods tending towards luxuries get a bigger return
in customer loyalty from strong complaint handling, and vice versa for basic, necessity goods
providers. For firms predominantly marketing to consumers of necessity goods, on the other
hand, the incentive to manage complaints were anticipated to be weaker and the results confirm
In addition, we find that the manufacturing vs. services variable (MFG) negatively
moderates the recovery–loyalty relationship (Coefficient = - 0.037, p < .01). In terms of
economic significance, changing the score on the MFG variable by 1 (from zero for service firms
to 1 for manufacturing firms) is associated with a change in slope of the recovery–loyalty
relationship of 0.037, which is a 16.2 percent decrease in the mean HANDLE slope of 0.229 in
Table 5 (Panel D). This finding confirms that the loyalty of consumers of more personal services
is more strongly impacted by complaint recovery than is the case with consumers of
manufactured goods, indicating that manufacturing firms have a lower payoff in customer
loyalty from strong complaint handling when compared to service-delivering firms.
Customer segment factors. Turning to the customer segment results, we find a steeper
recovery–loyalty slope among females (Coefficient = 0.019, p <.01) when compared to males,
consistent with earlier research that has observed positive moderation in the satisfaction-loyalty
relationship among women (e.g., Homburg and Giering 2001). In terms of economic
significance, changing the score on the FEMALE variable from 0 to 1 (male to female) is
associated with a change in slope of the recovery–loyalty relationship by 0.019, which is about
an 8.3 percent increase in the mean HANDLE slope of 0.229 in Table 5 (Panel C). For the other
customer segment factors, we fail to find a statistically significant moderating effect of income,
age, or region on the recovery–loyalty relationship. While difficult to interpret within theory (or
the scant record of empirical studies), we find these null results interesting in their own right.
Discussion and Implications
Turning complaining customers into loyal customers was the central focus in this
research. We captured the dynamics of this focus via an overarching research question: How
does the relationship between a firm’s customer complaint recovery (i.e., the customer’s
perception of how well the firm handled a complaint) and customer loyalty vary depending on
influences from economic, industry, customer-firm, product/service, and customer segment
factors? To address the nuances in this question, we developed a contingency model of loyalty
returns to complaint management based on exit-voice-loyalty theory (Hirschman 1970),
expectations-disconfirmation theory (e.g., Oliver 1980), and conceptualizations related to the
mechanisms connecting complaint handling to customer loyalty (e.g., Fornell et al. 1996). We
conducted a moderated multilevel analysis of the complaint handling (recovery)–customer
loyalty relationship by utilizing an ASCI dataset of 35,597 complaining customers over a 10-year
period across firms, industries, and economic sectors. Within the contingency modeling, we set
out to better understand the implications of the recovery–loyalty relationship as moderated by the
economic, industry, customer-firm, product/service, and customer segment factors. The
implications of these influences (moderating effects) are next discussed for managers,
policymakers, and researchers. We conclude with directions for future research.
Implications for Practice
Without a deeper understanding of the boundaries of the complaint handling–customer
loyalty relationship – via the practical incorporation of the implications stemming from the
moderators of economic, industry, customer-firm, product/service, and customer segment factors
– firms will likely allocate cost estimates to complaint management that are too low for the
required recovery actions or customer loyalty estimates that are too high, or both, instead of
achieving an optimal point of recovery-loyalty yield. First, managers should recognize not only
that industries vary widely in terms of the percentage of customers who complain, but also that
the characteristics of the economic sectors and industries can dictate the importance of complaint
recovery to their customers. In an industry (i.e., market research) where “best practices” from
“leading service providers” are often recommended for adoption without regard to industry
distinctiveness (e.g., Goodman 2006; Johnston and Mehra 2002), our findings indicate that
merely transposing a firm’s complaint management from one industry to another is ill-advised
and can be detrimental to a firm’s performance. While this may sound self-evident, many
managers are obsessed with seeking out cross-industry leaders to emulate towards improving
their own customers’ experiences (e.g., Berry and Seltman 2017; Michelli 2006, 2008). There are
clear differences across sectors and industries in customer experience management, customers as
strategic assets, and the accompanying complaint management that should be undertaken.
Our findings also indicate that the financial ramifications of firms’ complaint
management efforts will likely differ significantly. Since complaint management matters more or
less to customer loyalty across sectors in variant economic contexts and firms offering different
classes of goods, the expected economic benefit to the firm seeking to reaffirm customer loyalty
via aggressive complaint recovery will vary as well. Efforts that would produce a positive return-
on-investment (ROI) for firms in some industries offering certain goods may, at times, produce a
negligible or even negative ROI for firms in other industries. For example, we find that the
recovery-loyalty relationship is stronger for customers with higher expectations of customization
but weaker when the customers’ expectations of product/service reliability are higher. Combine
these findings with the sector and industry differences, and it is relatively easy to grasp that
developing complaint management systems cannot be undertaken solely through cross-industry,
best-practice benchmarking, but instead must incorporate a more refined approach (i.e., based on
the relevance of the economic, industry, customer-firm, product/service, and customer segment
factors). Succinctly, sensitivity to economic sectors and industry contexts can save a firm from
focusing too much or too little on complaint management. To be clear, this is not a call for
industries with weaker recovery–loyalty relationships to ignore customer complaints. Rather, it is
a call for managers to assess the most cost-effective means of soliciting and responding to
customer complaints and having the dexterity to adjust those efforts when conditions warrant it.
Without context, the implications suggest that a profit-maximizing strategy simply
requires that managers understand the impact of complaint recovery on customer loyalty in their
industry. Added to this complexity, however, is the reality that profitability is not evenly
distributed throughout the customer base. Profitability is driven by a small percentage of
customers, with most customers not producing an adequate level of return (Kaplan and
Narayanan 2001). Consequently, complaint management systems designed to maximize financial
performance are complex. They are likely too complex for frontline customer service
representatives to handle unaided, particularly as they relate to the level of remuneration used to
compensate complaining customers. Decision support systems need to be implemented that
consider economic factors (and impact on the expectations of customers), industry factors, and
the relative profitability of customers. This will make it easier for customer service personnel to
respond to complaining customers in ways that optimize customer satisfaction, customer loyalty,
and the economic contribution of customers (while, importantly, also being mindful of
customers’ potential social media amplification of their dissatisfaction).
As possible solutions, some complaint management channels are less expensive to
operate for firms than others. Often these channels vary in terms of personal contact.
Interestingly, contemporary alternatives to the traditional channels of direct in-person or
personalized telephone support may enhance customers’ perceptions of complaint handing. For
example, online customer service options, such as self-service and agent-assisted digital
communication channels are on the rise and preferred by many consumers to more personal
channels because of their speed of response. Customized and personalized web-based systems
are clearly on the rise, and these systems appear to offer a preferred balance of customization and
attentiveness and a less personalized approach. Of course, the effectiveness of different recovery
strategies will be impacted by the environment in which the business operates.
For policymakers, our findings reposition previous thinking about customer complaints,
complaint handling, and customer loyalty and, by extension, the health of the overall economy.
Although variations in intensity across political administrations should be considered, many
governments take active roles in monitoring both customers’ complaints against firms and firms’
responses to these complaints (e.g., in the United States via the Federal Trade Commission and
the Better Business Bureau). Our findings suggest not only that varying complaint levels should
be expected across industries and firms but also that customers’ perceptions of how well firms
have resolved their complaint issues should be expected to vary. These variations are due at least
in part to the industry context within which the complaint was filed. Thus, striking a balance
between over-reaching in regulations (i.e., too much/constraining regulations) and under-
reaching in regulations (i.e., too few/flexible regulations) needs to be considered in policy.
Implications for Research
Our study offers important implications for customer relationship researchers, in
particular those focused on firm and brand-related strategic issues and customer asset
management. First, while complaint recovery is positively linked with customer loyalty across all
economic sectors studied (which included 7 of 10 economic sectors in the marketplace), the
strength of the relationship varies. We find that the recovery-loyalty relationship is stronger in
faster-growing economies, for industries with more competition, for luxury products, and for
customers with higher satisfaction and higher expectations of customization. Equally important,
the recovery-loyalty relationship is weaker when the customers’ expectations of product/service
reliability are higher, for manufactured goods, and for males compared to females. Given the
richness of the data, these findings raise important questions about the limitations of existing
theory and empirical research to adequately explain the effectiveness of complaint recovery in
securing customer loyalty. Consequently, we advance both theory and empirical understanding
of the link between complaint handling and customer loyalty, including the theoretical and
empirical boundaries captured by the economic, industry, customer-firm, product/service, and
customer segment factors. Our contribution, as guidance for future research, is critical in that
virtually all of the previous findings in this literature are derived from either lab experiments
(largely with student subjects) and/or single-point-in-time cross-sectional survey research,
neither of which are designed to capture variance in these factors.
Second, consider existing meta-analyses from within the recovery–loyalty literature
(Gelbrich and Roschk 2010; Matos, Henrique, and Rossi 2007; Orsingher, Valentini, and
Deangelis 2010). These meta-analyses examining and aggregating recovery–loyalty effects
across studies have often mentioned, but almost universally failed to test, the possibly
confounding effects of industry and economic contexts (e.g., Matos, Henrique, and Rossi 2007;
Orsingher, Valentini, and Deangelis 2010). The few studies that have included such
examinations have focused on and tested macro effects at only the highest, aggregate levels (e.g.,
“service” industries vs. “non-service” industries) (Gelbrich and Roschk 2010). Such limited tests
are understandable given the nature of the data that are aggregated for the meta-analyses, yet our
results suggest the need for taking these effects into account at a disaggregated level for richer
insights. We captured 7 of the 10 NAICS economic sectors and 41 industries within these
sectors, and modeled a set of comprehensive moderators involving economic, industry,
customer-firm, product/service, and customer segment factors. This modeling helped create a
better understanding of the boundaries of the recovery-loyalty link. At the very least, our
findings should spur further research to developing theories of customer complaint management
and interpreting and comparing the effects observed across prior studies.
Third, we sought to expand the theoretical foundations of the recovery–loyalty literature
by blending theories from economics (exit-voice-loyalty theory – Hirschman 1970) with
traditional marketing theory (expectancy-disconfirmation theory and the customer satisfaction
literature). While most marketing studies that have examined the recovery–loyalty relationship
have focused almost exclusively on marketing theory, with some complementary underpinnings
in consumer psychology (e.g., justice theory), our blended economics-marketing approach
provides a different and advantageous theoretical lens to expand knowledge. Through this
broader and deeper lens, micro- and macroeconomic factors moderating the recovery–loyalty
relationship are clearer and will contribute to the continued development and refinement of the
contingency theory of loyalty returns to complaint management we offered in this research.
Directions for Future Research
While we have sought to close some of the enduring gaps in the marketing literature on
customer complaints, complaint recovery, and customer loyalty, additional work remains. First,
and as suggested earlier, future research should work to systematically reassess existing findings
from the marketing literature on the complaint handling–customer loyalty relationship based on
the results of this study. For instance, a meta-analysis that more systematically integrates
economic, industry, customer-firm, product/service, and customer segment factors as influencers
of the recovery-loyalty relationship across the many studies produced over the last two decades
could both reinforce our findings and substantially alter accepted conclusions. It is clear that
previous findings have significant limitations and continually having a state-of-the-art
understanding of the recovery-loyalty relationship is critically important to well-functioning
firms’ operational performance (Katsikeas, Morgan, Leonidou, and Hult 2016).
Second, future research should integrate our findings into models for determining the
value of customer retention initiatives and customer loyalty, such as customer lifetime value.
Customer lifetime value (CLV) models aim to illustrate the economic value of long-term
customer loyalty and the financial benefits of customer retention for firms. CLV results are
generally referenced to show that efforts that reduce churn often produce more valuable long-
term customer relationships that increase profitable firm growth. Customer complaint recovery
is, of course, just one of many customer retention tools. Like virtually any customer loyalty
initiative that is examined through the lens of a CLV model, complaint management comes at a
cost that can influence the profitability and margins of the customer segment being targeted. For
example, firms may need to invest in and deploy relevant information technology and CRM tools
to handle complaints (Mithas, Krishnan, and Fornell 2016), including deciding on appropriate
levels of human touch versus technology in dealing with complaint recovery interactions.
Increasingly, advances in IT tools such as Artificial Intelligence and Machine Learning can both
facilitate managing many aspects of customer relationships in a cost-effective manner. However,
the key for success will be to align deployment of IT tools with a firm’s strategy (e.g., Mithas
and Rust 2016; Rust, Moorman, and van Beauningen 2016), and to not lose sight of the revenue
impacts of marketing and technology decisions.
Ultimately, the value of a loyalty-building initiative can be deemed advisable or
inadvisable based on its impact on a customer segment’s CLV. Some customer retention
strategies can be predicted to pay-off in the long run via higher CLV while others will not.
Determining the difference between profitable and unprofitable loyalty efforts is important as it
relates to customer complaints and complaint handling. As our results show, an essential element
in gauging the true effect of customer complaint management on customer loyalty is
understanding the moderators of this relationship. In particular, having a clear understanding of
how the macro and micro moderators impact the strength of the relationship between complaint
recovery and customer loyalty is vital to achieving superior firm performance. We examined a
set of critical moderating factors on this relationship, but the dynamics of the marketplace keep
evolving and so will the influencers of the recovery-loyalty relationship. For example, some
indications exist that political ideology and partisanship may influence customers’ complaining
behavior, consumption experience, and loyalty (e.g., Jung, Garbarino, Briley, and Wynhausen
2017; Kim, Park, and DuBois 2018).
Finally, we recommend two avenues for theory development in the complaint recovery-
customer loyalty relationship literature. While our study has made strides in providing theoretical
support for the relationship between economic, industry, customer-firm, product/service, and
customer segment factors as moderators of this relationship, additional theorizing that more fully
clarifies the varied and complex connections between these factors and the mechanisms driving
consumer behavior could help inspire future empirical research and valuable practical insights.
Relatedly, theorizing that helps explain the moderating effects of customer segment factors and
demographic characteristics – such as gender and region of residence, where little theoretical or
empirical work now exists – is needed. While significant in their own right, our findings
regarding customer segment factors would have more robust practical implications if founded in
a guiding theory. For example, most national firms tailor marketing and product offerings to men
and women and variably across geographic regions, but would also likely do so for complaint
management if given a compelling explanation for the moderating effects of these and other
customer segment factors on the recovery-loyalty relationship.
Bamiatzi, Vassiliki, Konstantinos Bozos, S. Tamer Cavusgil, and G. Tomas M. Hult (2016),
“Revisiting the Firm, Industry and Country Effects on Profitability under Recessionary
and Expansion Periods: A Multi-level Analysis,” Strategic Management Journal, 37 (7),
Berger, Jonah and Morgan Ward (2010), “Subtle Signals of Inconspicuous Consumption,”
Journal of Consumer Research, 37 (4), 555-569.
Berry, Leonard L. and Kent D. Seltman (2017), Management Lessons from Mayo Clinic: Inside
One of the World's Most Admired Service Organizations, New York, NY: McGraw-Hill.
Chen, Ming-jer, Jing-lih Fahr, and Ian C. MacMillan (1993), “An Exploration of the Expertness
of Outside Informants,” Academy of Management Journal, 36 (6), 1614-1632.
Combs, James G. and David J. Ketchen, Jr. (1999), “Explaining Interfirm Cooperation and
Performance: Toward a Reconciliation of Predictions from the Resource-Based View and
Organizational Economics,” Strategic Management Journal, 20 (9), 867-888.
Cronin, J. Joseph, Jr., Michael K. Brady, and G. Tomas M. Hult (2000), “Assessing the Effects
of Quality, Value, and Customer Satisfaction on Consumer Behavioral Intentions in
Service Environments,” Journal of Retailing, 76 (2), 193-218.
DeWitt, Tom, Doan T. Nguyen, and Roger Marshall (2008), “Exploring Customer Loyalty
Following Service Recovery: The Mediating Effects of Trust and Emotions,” Journal of
Service Research, 10 (3), 269-281.
Disatnik, David, and Liron Sivan (2016), “The Multicollinearity Illusion in Moderated
Regression Analysis,” Marketing Letters, 27 (2), 403-408.
Dowding, Keith, Peter John, Thanos Mergoupis, and Mark Van Vugt (2000), “Exit, Voice and
Loyalty: Analytic and Empirical Developments,” European Journal of Political
Research, 37 (4), 469-495.
Dubois, David, Derek D. Rucker, and Adam D. Galinsky (2012), “Super Size Me: Product Size
as a Signal of Status,” Journal of Consumer Research, 38 (56), 1047-1062.
Etzel, Michael J. and Bernard I. Silverman (1981), “A Managerial Perspective on Directions for
Retail Customer Dissatisfaction Research,” Journal of Retailing, 57 (3), 124-136.
Evanschitzky, Heiner, Christian Brock, and Markus Blut (2011), “Will You Tolerate This? The
Impact of Affective Commitment on Complaint Intention and Post Recovery Behavior,”
Journal of Service Research, 14 (4), 410-425.
Fornell, Claes (2007), The Satisfied Customer: Winners and Losers in the Battle for Buyer
Preference, New York, NY: Palgrave Macmillan.
Fornell, Claes and N. Moshe Davidow (1980), “Economic Constraints on Consumer
Complaining Behavior,” Advances in Consumer Research, Vol. VII, Ann Arbor, MI:
Association for Consumer Research, 318-323.
Fornell, Claes, Michael D. Johnson, Eugene W. Anderson, Jaesung Cha, and Barbara Everitt
Bryant (1996), “The American Customer Satisfaction Index: Nature, Purpose, and
Findings,” Journal of Marketing, 60 (4), 7-18.
Fornell, Claes, Sunil Mithas, Forrest V. Morgeson III, and M. S. Krishnan (2006), “Customer
Satisfaction and Stock Prices: High Returns, Low Risk,” Journal of Marketing, 70 (1), 3-
Fornell, Claes, Forrest V. Morgeson III, and G. Tomas M. Hult (2016), “Stock Returns on
Customer Satisfaction Do Beat the Market: Gauging the Effect of a Marketing
Intangible” Journal of Marketing, 80 (5), 92-107.
Fornell, Claes, Forrest V. Morgeson III, G. Tomas M. Hult, and David VanAmburg (2020). The
Reign of the Customer: Customer-Centric Approaches to Improving Satisfaction, New
York, NY: Palgrave Macmillan.
Fornell, Claes, Roland Rust, and Marnik Dekimpe (2010), “The Effect of Customer Satisfaction
on Consumer Spending Growth,” Journal of Marketing Research, 47 (1), 28-35.
Fornell, Claes and Birger Wernerfelt (1987), “Defensive Marketing Strategy by Customer
Complaint Management: A Theoretical Analysis,” Journal of Marketing Research, 24
Fornell, Claes and Birger Wernerfelt (1988), “A Model for Customer Complaint Management.”
Marketing Science, 7 (3), 287-298.
Fornell, Claes and Robert A. Westbrook (1984), “The Vicious Circle of Consumer Complaints,”
Journal of Marketing, 48 (3), 68-78.
Gelbrich, Katja and Holger Roschk (2010), “A Meta-Analysis of Organizational Complaint
Handling and Customer Responses,” Journal of Service Research, 14 (1), 24-43.
Goodman, John (2006), “Manage Complaints to Enhance Loyalty,” Quality Progress, 39 (2), 28-
Herhausen, Dennis, Stephen Ludwig, Dhruv Grewal, Jochen Wulf, and Marcus Schoegel (2019),
“Detecting, Preventing, and Mitigating Online Firestorms in Brand Communities,”
Journal of Marketing, 83 (3), 1-21.
Hess, Ronald L., Shankar Ganesan, and Noreen M. Klein (2003), “Service Failure and Recovery:
The Impact of Relationship Factors on Customer Satisfaction,” Journal of the Academy
of Marketing Science, 31 (2), 127-145.
Hirschman, Albert O. (1970), Exit, Voice, and Loyalty: Responses to Decline in Firms,
Organizations, and States, Cambridge, MA: Harvard University Press.
Hofmann, David A. (1997), “An Overview of the Logic and Rationale of Hierarchical Linear
Models,” Journal of Management, 23 (6), 723-744.
Hoffman, K. Douglas, Scott W. Kelley, and Holly M. Rotalsky (1995), “Tracking Service
Failures and Employee Recovery Efforts,” Journal of Services Marketing, 9 (2), 49-61.
Homburg, Christian and Andreas Fürst (2005), “How Organizational Complaint Handling Drives
Customer Loyalty: An Analysis of the Mechanistic and the Organic Approach,” Journal
of Marketing, 69 (3), 95-114.
Homburg, Christian and Annette Giering (2001), “Personal Characteristics as Moderators of the
Relationship between Customer Satisfaction and Loyalty – An Empirical Analysis,”
Psychology & Marketing, 18 (1), 43-66.
Hult, G. Tomas M., Forrest V. Morgeson III, Neil A. Morgan, Sunil Mithas, and Claes Fornell
(2017), “Do Managers Know What Their Customers Think and Why?” Journal of the
Academy of Marketing Science, 45 (1), 37-54.
Johnson, Michael D. and Claes Fornell (1991), “A Framework for Comparing Customer
Satisfaction across Individuals and Product Categories,” Journal of Economic
Psychology, 12 (2), 267-286.
Johnston, Robert, and Sandy Mehra (2002), “Best-Practice Complaint Management,” Academy
of Management Executive, 16 (4), 145-154.
Jung, Kiju, Ellen Garbarino, Donnel A. Briley, and Jesse Wynhausen (2017), “Blue and Red
Voices: Effects of Political Ideology on Consumers’ Complaining and Disputing
Behavior,” Journal of Marketing, 44 (3), 477-499.
Kapferer, Jean-Noel and Vincent Bastien (2009), The Luxury Strategy Break the Rules of
Marketing to Build Luxury Brands, London, UK: Kogan Page.
Kaplan, Robert S., and V. G. Narayanan (2001), “Measuring and Managing Customer
Profitability,” Journal of Cost Management, 15 (5), 5-15.
Katsikeas, Constantine S., Neil A. Morgan, Leonidas C. Leonidou, and G. Tomas M. Hult
(2016), “Assessing Performance Outcomes in Marketing,” Journal of Marketing, 80 (2),
Kau, Ah-Keng and Elizabeth Wan-Yiun Loh (2006), “The Effects of Service Recovery on
Consumer Satisfaction: A Comparison Between Complainants and Non-Complainants,”
Journal of Services Marketing, 20 (2), 101-111.
Keiningham, Timothy L., Forrest V. Morgeson III, Lerzan Aksoy, and Luke Williams (2014),
“Service Failure Severity, Customer Satisfaction, and Market Share: An Examination of
the Airline Industry,” Journal of Service Research, 17 (4), 415-431.
Kendall, C. L. and Frederick A. Russ (1975), “Warranty and Complaint Policies: An Opportunity
for Marketing Management,” Journal of Marketing, 39 (2), 36-43.
Kilgrove, Kristina, “Meet the Worst Businessman of the 18th Century BC,” Forbes.com, May
Kim, Jeehye Christine, Brian Park, and David Dubois (2018), “How Consumers’ Political
Ideology and Status-Maintenance Goals Interact to Shape Their Desire for Luxury
Goods,” Journal of Marketing, 82 (6), 132-149.
Knox, George and Rutger van Oest (2014), “Customer Complaints and Recovery Effectiveness:
A Customer Base Approach,” Journal of Marketing, 78 (5), 42-57.
Kumar, V., Nita Umashankar, Kihyun Hannah Kim, and Yashoda Bhagwat (2014), “Assessing
the Influence of Economic and Customer Experience Factors on Service Purchase
Behaviors,” Marketing Science, 33 (5), 673-692.
Lee, Soo-Young and Andrew B. Whitford (2007), “Exit, Voice, Loyalty, and Pay: Evidence
from the Public Workforce,” Journal of Public Administration Research and Theory, 18
Malshe, Ashwin and Manoj K. Agarwal (2015), “From Finance to Marketing: The Impact of
Financial Leverage on Customer Satisfaction,” Journal of Marketing, 79 (September),
Mandel, Naomi, Petia K. Petrova, and Robert B. Cialdini (2006), “Images of Success and the
Preferences for Luxury Brands,” Journal of Consumer Psychology, 16 (1), 57-69.
Matos, Celso Augusto, Jorge Luiz Henrique, and Carlos Alberto Vargas Rossi (2007), “Service
Recovery Paradox: A Meta-Analysis,” Journal of Service Research, 10 (1), 60-77.
Mattila, Anna S. (2001), “The Effectiveness of Service Recovery in a MultiIndustry Setting,”
Journal of Services Marketing, 15 (7), 583-596.
Maxham, James G. III and Richard G. Netemeyer (2002), “A Longitudinal Study of
Complaining Customers’ Evaluations of Multiple Service Failures and Recovery
Efforts,” Journal of Marketing, 66 (4), 57-71.
McCollough, Michael A., Leonard L. Berry, and Manjit S. Yadav (2000), “An Empirical
Investigation of Customer Satisfaction after Service Failure and Recovery,” Journal of
Service Research, 3 (2), 121-137.
Melnyk, Valentyna, Stijn M. J. van Osselaer, and Tammo H. A. Bijmolt (2009), “Are Women
More Loyal Customers Than Men? Gender Differences in Loyalty to Firms and
Individual Service Providers,” Journal of Marketing, 73 (4), 82-96.
Michel, Stefan and Matthew L. Meuter (2008), “The Service Recovery Paradox: True but
Overrated?” International Journal of Service Industry Management, 19 (4), 441-457.
Michelli, Joseph (2006), The Starbucks Experience: 5 Principles for Turning Ordinary Into
Extraordinary, New York, NY: McGraw-Hill.
Michelli, Joseph (2008), The New Gold Standard: 5 Leadership Principles for Creating a
Legendary Customer Experience Courtesy of the Ritz-Carlton Hotel Company, New
York, NY: McGraw-Hill.
Mithas, Sunil, M. S. Krishnan, and Claes Fornell (2016), “Information Technology, Customer
Satisfaction, and Profit: Theory and Evidence,” Information Systems Research, 27 (1),
Mithas, Sunil and Roland T. Rust (2016), “How Information Technology Strategy and
Investments Influence Firm Performance: Conjecture and Empirical Evidence,” MIS
Quarterly 40 (1), 223-245.
Morgeson, Forrest V. III, Sunil Mithas, Timothy L. Keiningham, and Lerzan Aksoy (2011), “An
Investigation of the Cross-National Determinants of Customer Satisfaction,” Journal of
the Academy of Marketing Science, 39 (2), 198-215.
Oliver, Richard L. (1980), “A Cognitive Model of the Antecedents and Consequences of
Satisfaction Decisions,” Journal of Marketing Research, 17 (4), 460-469.
Oliver, Richard L. (2010), Satisfaction: A Behavioral Perspective on the Consumer. Milton Park,
Abingdon, Oxfordshire, United Kingdom: Routledge.
Orsingher, Chiara, Sara Valentini, Matteo Deangelis (2010), “A Meta-Analysis of Satisfaction
with Complaint Handling in Services,” Journal of the Academy of Management Science,
38 (2), 169-186.
Otto, Ashley S., David M. Szymanski, and Rajan Varadarajan (2020), “Customer Satisfaction
and Firm Performance: Insights from Over a Quarter Century of Empirical Research,”
Journal of the Academy of Marketing Science, In Press (https://doi-
Pfeffer, Jürgen, Thomas Zorbach, and Kathleen M. Carley (2014), “Understanding Online
Firestorms: Negative Word-of-Mouth Dynamics in Social Media Networks,” Journal of
Marketing Communications, 20 (1-2), 117-128.
Raudenbush, Stephen W. and Anthony S. Bryk (2002), Hierarchical Linear Models:
Applications and Data Analysis Methods, Thousand Oaks, CA: Sage Publications.
Raudenbush, Stephen W., Anthony S. Bryk, Yuk Fai Cheong, Richard T. Congdon, and
Mathilda du Toit (2016), HLM7: Hierarchical Linear and Nonlinear Modeling. Chicago,
IL: Scientific Software International.
Rust, Roland T., Christine Moorman, and J van Beuningen (2016), “Quality Mental Model
Convergence and Business Performance,” International Journal of Research in
Marketing, 33 (1), 155-71.
Rust, Roland T., Christine C. Moorman, and Peter R. Dickson (2002), “Getting Return on
Quality: Revenue Expansion, Cost Reduction, or Both?,” Journal of Marketing 66 (4), 7-
Shah, Denish, V. Kumar, Kihyun H. Kim, and Jeewon B. Choi (2017), “Linking Customer
Behaviors to Cash Flow Level and Volatility: Implications for Marketing Practices,”
Journal of Marketing Research, 54 (February), 27-43.
Short, Jeremy C., David J. Ketchen, Jr., Timothy B. Palmer, and G. Tomas M. Hult (2007),
“Firm, Strategic Group, and Industry Influences on Performance,” Strategic Management
Journal, 28 (2), 147-167.
Simon, Françoise, Vesselina Tossan, and Chantal Connan Guesquière (2015), “The Relative
Impact of Gratitude and Transactional Satisfaction On Post-Complaint Consumer
Response,” Marketing Letters, 26 (2), 153-164.
Smith, Amy K. and Ruth N. Bolton (2002), “The Effect of Customers’ Emotional Responses to
Service Failures on Their Recovery Effort Evaluations and Satisfaction Judgments,”
Journal of the Academy of Marketing Science, 30 (1), 5-23.
Smith, Amy K., Ruth N. Bolton, and Janet Wagner (1999), “A Model of Customer Satisfaction
with Service Encounters Involving Failure and Recovery,” Journal of Marketing
Research, 36 (3), 356-372.
Spreng, Richard A., Gilbert D. Harrell and Robert D. Mackoy (1995), “Service Recovery:
Impact on Satisfaction and Intentions,” Journal of Services Marketing, 9 (1), 15-23.
Tax, Stephen S., Stephen W. Brown, and Murali Chandrashekaran (1998), “Customer
Evaluations of Service Complaint Experiences: Implications for Relationship
Marketing,” Journal of Marketing, 62 (2), 60-76.
Umashankar, Nita, Morgan K. Ward, and Darren W. Dahl (2017), “The Benefit of Becoming
Friends: Complaining After Service Failures Leads Customers with Strong Ties to
Increase Loyalty,” Journal of Marketing, 81 (6), 79-98.
Walsh, G., Heiner Evanschitzky and Maren Wunderlich (2008). “Identification and Analysis of
Moderator Variables Investigating the Customer Satisfaction-Loyalty Link,” European
Journal of Marketing, 42(9/10), 977-1004.
Weekley, Jeff A. and Joseph A. Gier (1989), “Ceilings in the Reliability and Validity of
Performance Ratings: The Case of Expert Raters,” Academy of Management Journal, 32
Withey, Michael J. and William H. Cooper (1989). “Predicting Exit, Voice, Loyalty, and
Neglect,” Administrative Science Quarterly, 34 (4), 521-539.
Wirtz, Jochen and Anna S. Mattila (2004), “Consumer Responses to Compensation, Speed of
Recovery and Apology After a Service Failure,” International Journal of Service
Industry Management, 15 (2), 150-166.
Factors and Mechanisms for the Study of the Complaint Recovery–Customer Loyalty Relationship
Significant Moderating Effects on the Complaint Recovery–Customer Loyalty Relationship
(as a Percent of Mean HANDLE Slope in Table 5)
Notes: Figure 2 shows the effects as percentage increases of the continuous variables (all
variables except FEMALE and MFG) when the value of that variable increases from one
standard deviation below the mean to one standard deviation above the mean. For the FEMALE
and MFG variables, the effect is when the value of the variable changes from 0 to 1.
Sample Research on Customer Complaints, Complaint Management,
and Customer Loyalty Relationships
n = 373
n = 410
Tax, Brown, and
n = 239
n = 375 and
n = 602
n = 615
n = 441
n = 355 and
n = 549
n = 1,356
and Klein (2003)
n = 346
n = 187
n = 550
B2B vs. B2C;
Kau and Loh
n = 153
n = 1,189
n = 459
Brock, and Blut
n = 146 and
n = 233
Knox and van
al; n = 922
n = 144
Ward and Dahl
“Strong tie” vs.
n = 35,597
Note: For the sake of parsimony, we only summarize the contents of these studies for the different factors included
or excluded. For example, we include only primary customer-firm factors examined within each article, e.g.,
“justice” rather than distributive justice (and its sub-factors), procedural justice (and its sub-factors), and
interactional justice (and its sub-factors). The sample sizes for each study relate to complaining customers.
How Economic, Industry, Customer-Firm, Product/Service, and Customer Segment
Factors Moderate the Complaint Recovery–Customer Loyalty Relationship
(Factors Often Adhere to Multiple Mechanisms)
Consumer power: Economic growth is typically accompanied by
a variety of features that result in more powerful consumers (e.g.,
lower unemployment, stronger income growth, more consumer
spending, stronger consumer confidence).
Alternatives, switching costs and barriers: In competitive
industries, customers recognize their ability to easily switch to
alternative suppliers and also recognize their greater power vis-à-
vis the firm.
Reservoir of consumer goodwill: Cumulative customer
satisfaction represents the customer’s reservoir of goodwill
toward the firm and product/service based in buyer habit and
brand but mandates additional firm attention after failures.
Negative expectation-disconfirmation gap: Customers with
higher customization expectations anticipate more individualized
service from the firm in all areas, including during a failure and
Negative expectation-disconfirmation gap: The unexpected
failure resulting in the complaint and recovery attempt is, from
the customer’s perspective, reflective of either a fundamental
disruption of a long problem-free relationship or an indication
that the firm’s promises are hollow.
Necessity vs. Luxury
Alternatives, switching costs and barriers: Luxury goods
customers will typically have greater financial resources and thus
the ability to switch to alternative luxury providers or less
expensive replacement goods more easily.
Consumer power: For a significant proportion of manufactured
goods, such as frequently purchased and inexpensive nondurable
goods, complaints are less likely, with customers choosing to
either remain silently loyal or defect without complaint.
Customer Segment Factors
Latent segment membership: Satisfaction is less influential as a
determinant of loyalty for wealthier consumers due to a more
expansive choice set, and so too might dissatisfaction and
complaint recovery matter less to loyalty.
Latent segment membership: Research has shown a stronger
satisfaction-loyalty relationship among women, which suggests a
stronger recovery–customer loyalty relationship as well.
Latent segment membership: Research has shown that the impact
of satisfaction on loyalty increases with age, and complaint
recovery may likewise more strongly impact customer loyalty for
older generational cohorts.
Latent segment membership: Customer region is anticipated to
have a moderating effect given the prevalence of geography-
specific marketing strategies (“geo-marketing,” “geo-fencing”).
Summary of Variables and Operationalization
ACSI Survey Question:1 “The next time you are going to purchase the
same product or service, how likely is it that it will be with (COMPANY)
again? Using a 10-point scale on which "1" means "very unlikely" and
"10" means "very likely," how likely is it that it will be with (COMPANY)
again?” on a scale of 1-10.
ACSI Survey Question: “How well, or poorly, was your most recent
complaint handled? Using a 10-point scale on which “1” means “handled
very poorly” and “10” means “handled very well”, how would you rate the
handling of your complaint?” on a scale of 1-10.
Annual Gross Domestic Product (GDP) growth data obtained via the U.S.
Bureau of Economic Analysis website at: www.bea.gov.
Annual Herfindahl-Hirschman Index at the sub-sector (industry) level,
calculated as: sum of the squared company-level market share percentages
of the largest firms measured in the industry. The data are from Compustat,
obtained via the Wharton Research Data Services.
ACSI Survey Question: “Please consider all your experiences to date with
(COMPANY). Using a 10-point scale on which "1" means "very
dissatisfied" and "10" means "very satisfied," how satisfied are you with
(COMPANY)?” on a scale of 1-10.
Expectations of Customization
ACSI Survey Question: “At the same time, you probably thought about
things you personally require from (COMPANY). Using a 10-point scale
on which "1" now means "not very well" and "10" means "very well," how
well did you expect (COMPANY) to meet your personal requirements?” on
a scale of 1-10.
Expectations of Reliability
ACSI Survey Question: “Thinking about your expectations before you
purchased from (COMPANY), you probably thought about how often
things could go wrong. Using a 10-point scale, on which "1" now means
"very often" and "10" means "not very often," how often did you expect
that things could go wrong with (COMPANY)?” on a scale of 1-10.
Necessity vs. Luxury
ACSI Survey Question: “Thinking about (COMPANY), do you think of it
more as a supplier of basic necessity goods and services or a supplier of
exclusive luxury goods and services? On a scale from 1 to 10, where 1 =
"necessity goods and services provider" and 10 = "luxury goods and
services provider," how would you rate (COMPANY)?” on a scale of 1-10.
Manufacturing vs. Service
Manufacturing (Services = 0, Manufacturing = 1) based on NAICS codes.
The data are from the U.S. Census Bureau website at:
Annual household income from the prior year (0 = $60K or below; 1 =
Above $60K). Data on income came from the ACSI database.
Female (0 = Male; 1 = Female). Data on gender came from the ACSI
Indicator variables for whether consumer-respondent is part of the
Millennial, Generation X, Baby Boomer, or Silent Generations (reference
category). Generational cohorts were determined uniquely for each sample
year based on accepted categorizations (Millennials 1980-2000
(MILLDUM); Generation X 1965-1979 (GENXDUM); Baby Boomers
1946-1964 (BOOMDUM); and Silent Generation pre-1946). Data on
customer cohort (age) came from the ACSI database.
Indicator variables for residence of the respondent in the Northeast
(NEDUM), Midwest (MIDWDUM), Southeast (SOTHDUM), or West of
the United States, with the West as the reference category in our models.
Regions were defined following the U.S. Census’ “Regions and Divisions
of the United States” (www.census.gov/prod/1/gen/95statab/preface.pdf).
1 Most questions in the ACSI survey are asked on 1–10 scales and then transformed to 0–100 index scores for
official reporting purposes. In this study, we analyze the variables on their original 1–10 scales.
2The luxury variable data were collected for each firm using “expert raters” (e.g., Chen, Farh, and MacMillan 1993;
Combs and Ketchen 1999; Weekley and Gier 1989) affiliated with the American Customer Satisfaction Index (n =
15). Each expert rater was asked to assess each ACSI-measured brand/company on a 10-point scale, as a firm which
supplies basic necessity goods (1) to a supplier of high-end luxury goods (10). The average rating for each firm
among the expert raters was then associated with each respondent for that firm in the sample.
Descriptive Statistics and Correlations
All correlations equal to or above 0.02 are statistically significant at p< .05.
HLM Estimation of Fixed Effects with Robust Standard Errors
(Dependent Variable is Customer Loyalty)
Panel A: Level 1 Main Effects
Customer Satisfaction (SATIS), 2
Expectations of Customizability (CUSTOMX), 3
Expectations of Reliability (RELYX), 4
Gender (FEMALE), 5
Income (INCDUM), 6
Millennial (MILLDUM), 7
Generation X (GENXDUM), 8
Baby Boomers (BOOMDUM), 9
Northeast (NEDUM), 10
Midwest (MIDWDUM), 11
South (SOTHDUM), 12
Inverse Mills Ratio (IMR), 13
Panel B: Level 2 Modeling of Intercept 0
GDP GROWTH, 02
Panel C: Level 1 Interaction Effects
SATIS * HANDLE, 14
CUSTOMX * HANDLE, 15
RELYX * HANDLE, 16
FEMALE * HANDLE, 17
INCDUM * HANDLE, 18
NEDUM * HANDLE 19
MIDWDUM * HANDLE, 20
SOTHDUM * HANDLE, 21
MILLDUM* HANDLE, 22
GENXDUM * HANDLE, 23
BOOMDUM * HANDLE, 24
Panel D: Level 2 Modeling of Complaint Handling (HANDLE), 1
GDP GROWTH, 12
Panel E: Variance Explained
Proportion of Variance Explained by Level 1 model
Proportion of Variance Explained by Level 2 model for HANDLE
Deviance (-2 Log Likelihood)
Note: Moderating effects on the complaint handling - loyalty relationships are bolded. ** p<0.01, * p<0.05
Economic Sectors and Consumer Industries in the Sample
Beverages – Beer
Beverages – Soft Drinks
Tobacco – Cigarettes
Personal Care Products
Automobiles and Light Trucks
Cellular Phone Manufacturers
Retail Trade (RETAIL)
Department and Discount Stores
Specialty Retail Stores
Gasoline Service Stations
Health and Personal Care Stores
E-Commerce Retail Websites
E-Commerce Travel Websites
Transportation and Warehousing (TRSPRT)
Parcel Delivery – Express Mail
U.S. Postal Service
Telecommunications – Local and Long-Distance Telephone
Publishing – Newspapers
Telecommunications – Cable Television
Cellular Telephone Service Providers
Telecommunications – Internet Service Providers
Finance and Insurance (FIN)
Personal Property Insurance
E-Commerce Financial Services Websites
Health Care and Social Assistance (HLTH)
Accommodation and Food Services (ACCO)
Restaurants – Limited Service
Restaurants – Full Service