Journal of Business Research 92 (2018) 25–35

Contents lists available at ScienceDirect

Journal of Business Research

journal homepage: www.elsevier.com/locate/jbusres

The effect of relationship and transactional characteristics on customer T

retention in emerging online markets
⁎
Anand K. Jaiswala, Rakesh Nirajb, , Chang Hee Parkc, Manoj K. Agarwalc
a
Indian Institute of Management Ahmedabad, Vastrapur, Ahmedabad, 380015, India
b
Weatherhead School of Management, Case Western Reserve University, Cleveland, OH 44106, United States of America
c
School of Management, Binghamton University, SUNY, Binghamton, NY 13902, United States of America

A R T I C LE I N FO A B S T R A C T

Keywords: Trust is important for maintaining customer relationships in online retailing, as customers have only a virtual
Trust connection with sellers. This is especially true in online markets of emerging economies, given their lack of trust-
Retention enhancing infrastructure and well-functioning regulatory institutions. We investigate the effect of trust and a set
Customer relationship management of other relationship and transactional characteristics—mode of customer acquisition, length of relationship,
Online retailing
service communication, product return activity, and type of products purchased—on retention in the context of
Latent attrition model
emerging online markets. We obtain data from an online retailer in India that include both survey and trans-
action information. Using a latent attrition model, we find that trust positively affects customer retention be-
havior. We also find that relationship length, service communication, product return experience, and the type of
products purchased affect retention. Furthermore, we conduct split-sample analysis and suggest some man-
agerial actions on spending efforts to enhance retention.

1. Introduction
Online retailing has experienced rapid growth globally in recent
decades. While Internet penetration is over 85% in developed countries,
the penetration is much lower in developing countries: for example,
52% in China and 35% in India (Internet Live Stats, 2016). The number
of regular online buyers in these countries is even smaller, as payment
and other e-commerce-enabling infrastructure (e.g., logistics, tele-
communications, effective regulatory institutions) is not as well de-
veloped (Martinsons, 2008). Emerging markets are also characterized
by the widespread use of unbranded products by consumers (Sheth,
2011). Given that brands enhance consumer trust and confidence in a
purchase (Aaker, 1992), trust deficiency becomes even more salient in
these markets due to a lack of branded goods. These factors influence
customers' trust for online stores and consequently affect loyalty and
retention (Ashraf, Thongpapanl, & Auh, 2014; Lee & Turban, 2001).
Several studies have found a positive link between trust and loyalty
in both offline (Garbarino & Johnson, 1999; Stathopoulou & Balabanis,
2016) and online (Pascual-Miguel, Agudo-Peregrina, & Chaparro-
Peláez, 2015) markets. One notable limitation of these studies is that
they use cross-sectional survey data to measure the variables at the
same point in time, resulting in bias from common method variance
(Podsakoff et al., 2003) and carryover and backfire effects (Bickart,
1993). Furthermore, these studies commonly measured loyalty as cus-
tomers' loyalty intentions (Kim, Ferrin, & Rao, 2009). Although this is
theoretically reasonable, the predictive validity of loyalty intention
measures could vary widely depending on business context, study de-
sign, and time frame (Morwitz, Steckel, & Gupta, 2007). Another lim-
itation of the previous literature is that while there are antecedents of
loyalty beyond trust, not many of them have been studied. For example,
a recent meta-analysis by Pan, Sheng, & Xie (2012) suggests the im-
portance of product- and customer-related perceptual factors in
building loyalty. One problem with using perceptual factors is that
firms have to collect survey data regularly to understand the drivers of
loyalty. Typically, constructs such as commitment, perceived value, and
reputation (Pan, Sheng, & Xie, 2012) can only be obtained through
surveys. Although many firms are constantly collecting transaction and
non-survey data in the process of acquiring customers, responding to
their queries, selling products, and processing returns, limited research
has studied the impact of these activities on loyalty and retention, and
even less so in emerging markets.
In this research, we use a unique set of longitudinal transaction
data, supplemented with survey data, to study the impact of trust and
other relevant relationship and transactional characteristics on

⁎
Corresponding author.
E-mail addresses: akjaiswal@iima.ac.in (A.K. Jaiswal), rkn10@case.edu (R. Niraj), cpark@binghamton.edu (C.H. Park),
agarwal@binghamton.edu (M.K. Agarwal).

https://doi.org/10.1016/j.jbusres.2018.07.007
Received 30 June 2017; Received in revised form 5 July 2018; Accepted 7 July 2018
0148-2963/ © 2018 Published by Elsevier Inc.
A.K. Jaiswal et al. Journal of Business Research 92 (2018) 25–35

customer retention in emerging online markets. We choose character- developing and developed markets (Ashraf, Razzaque, & Thongpapanl,
istics that have a strong theoretical or managerial importance and for 2016).
which online retailers are likely to have data in normal course of In the next section, we present the conceptual background and de-
business. Specifically, in addition to trust, we examine the role of the velop our hypotheses. Thereinafter, we present our empirical metho-
mode of customer acquisition, relationship length, service-related dology and describe our research site. We then present our analysis and
communications, product return activity, and the type of products results. The final section provides brief conclusions, study limitations,
purchased. Using a latent attrition model (Fader, Hardie, & Lee, 2005; and some directions for future research.
Schmittlein, Morrison, & Colombo, 1987), we probabilistically assess
customer retention from individuals' transactional patterns and ex-
2. Conceptual development
amine the impact of trust and the aforementioned relationship and
transactional characteristics on retention.
Extant marketing literature on customer retention has extensively
In our analysis, we find strong effects of four such characteristics, in
examined the relationship between satisfaction and loyalty (Szymanski
addition to trust, on retention behavior. Our results reveal that re-
& Henard, 2001; Verhoef, 2003). Loyalty is believed to be composed of
lationship length, service communication experience, product return
attitudinal (capturing intentions to rebuy and recommend) and beha-
activity, and the type of products purchased significantly affect cus-
vioral (capturing actual retention and usage) components (Chaudhuri &
tomer retention. In particular, we note that, contrary to our expecta-
Holbrook, 2001; Russell-Bennet et al., 2007). In their literature review,
tions, product return experience can have a positive impact on reten-
Kumar, Pozza, and Ganesh (2013) indicate that customer satisfaction
tion. Furthermore, customers acquired through referrals do not exhibit
explains only a small portion of variance in loyalty and does not sig-
higher retention rates. Prior studies have suggested that people exhibit
nificantly enhance customer retention. Several researchers have ex-
varying propensity to trust depending on their demographic and be-
amined other potential antecedents affecting retention and loyalty, in-
havioral characteristics (Mayer, Davis, & Schoorman, 1995). We also
cluding the impact of loyalty programs (Bolton, Kannan, & Bramlett,
conduct split-sample analysis (Hitt, Hoskisson, & Kim, 1997) in which
2000), commitment (Gustafsson, Johnson, & Roos, 2005), short-term
we estimate our latent attrition model separately for different levels of
promotions (Lewis, 2004), and service quality (Venetis & Ghauri,
three variables: customer income, breadth of buying, and size of
2004), among others. Authors have incorporated relationship variables,
transactions. We find that the income level and transaction size mod-
including trust (Augustin & Singh, 2005; Chiou & Droge, 2006), and
erate the impact of trust on retention behavior.
some have called for research to consider additional variables (Kumar,
Our research makes several contributions. We propose and study the
Pozza, & Ganesh, 2013).
impact of a broad set of relationship and transactional characteristics
The focus of our research is on examining the impact of trust and a
on customer retention behavior, some of which are examined for the
broader set of other relevant characteristics on customer retention in
first time in this context (viz. trust, mode of customer acquisition, and
non-contractual relationships using longitudinal data. We propose that
type of products purchased). Our finding that return experience can
customer retention can be affected by two broad sets of factors: re-
actually increase retention is both novel and unique. This and our other
lationship characteristics and transactional characteristics. Within the
findings suggest that researchers should include characteristics such as
first set, we consider trust as an overall indicator of the strength of the
the type of products purchased, service communications, and product
customer–firm relationship (from the customer's point of view). We also
return activity—all of which can be found readily in company trans-
record how the relationship started by looking at the mode of customer
action records—as important drivers of retention. We are the first to
acquisition (whether the customer was acquired through a personal
show that trust improves retention behavior, not just loyalty intentions.
referral); we include this factor to study the value of referred customers
We investigate the trust–retention relationship by combining survey
to firms, as some research has suggested that referred customers have a
data on consumer trust with longitudinal data of actual retention be-
stronger relationship with the firm in the long run (Schmitt, Skiera, &
havior. This contrasts with prior studies, which have used a cross-sec-
Van den Bulte, 2011). In addition, we also look at the length of the
tional survey design or modeled retention as loyalty intentions. We also
customer's relationship with the retailer (Bolton, 1998). Within the
find that customers' acquisition through referrals does not affect their
second set, we examine three transactional characteristics of relation-
retention. Furthermore, managerially, the results of our split-sample
ships: customers' service communications, product return activity, and
analysis provide guidelines regarding which customer segments in de-
type of products purchased. For the conceptual model tested in our
veloping markets firms should target for their retention-enhancing ac-
research, see Fig. 1. We discuss these factors in more detail next.
tivities. We also find that the role of trust in enhancing retention be-
havior is similar to developed markets in some respects, but different in
others. This emphasizes the need for more theoretical research ex- 2.1. Trust
ploring the similarities and differences in consumer behaviors between
Among different customer relationship constructs, trust has been

Relaonship Characteriscs
- Trust
- Acquired by Referral
- Relaonship Length

Customer Retenon
Transaconal Characteriscs
- Service Communicaon
- Product Return Acvity
- Types of Products

Fig. 1. Conceptual model.

26
A.K. Jaiswal et al.
identified as a driving force behind customer retention (Agustin &
Singh, 2005; Garbarino & Johnson, 1999). In our research, we focus on
trust because we believe that the importance of trust in retaining cus-
tomers increases significantly in online markets of emerging economies.
Compared with offline transactions, purchasing online involves sig-
nificantly higher risk for customers. The physical separation between
buyers and sellers creates additional risks that customers may “[fall]
prey to a site that posts low prices but is proficient only in charging
credit cards, not delivering the goods” (Brynjolfsson & Smith, 2000,
page 578). In addition, other risks include misuse of credit cards and
breach of personal data (Jaiswal, Niraj, & Venugopal, 2010).
According to Bart et al. (2005, p. 134), “online trust includes con-
sumer perceptions of how the site would deliver on expectations, how
believable the site's information is, and how much confidence the site
commands.” Trust has been found to be positively related to satisfaction
and loyalty intentions in offline (Singh & Sirdeshmukh, 2000) and on-
line (Kim, Ferrin, & Rao, 2009; Malhotra, Sahadev, & Purani, 2017)
environments. Therefore, we also expect that in our study, all else being
equal, those with higher trust toward the online retailer are more likely
to be retained. Because the trust–retention relationship has been ex-
tensively studied in prior literature, we do not formulate an explicit
hypothesis for this relationship. However, we include trust in our latent
attrition model as a major determinant of the attrition probability.
2.2. Acquired by referral
In the online world, loyal consumers can spread word-of-mouth
very easily and frequently through online tools such as email, blogs,
and reviews. Several studies on offline marketing have shown that sa-
tisfied and loyal customers spread positive word of mouth (Richins,
1983). This relationship has been established in online environments as
well (Bansal et al., 2004). Gu, Park, and Konana (2012) find that ex-
ternal (not hosted by retailer) electronic word of mouth (e-WOM) af-
fects consumers' information search behavior and e-retailers' sales. The
key motives for consumers to seek out e-WOM are to economize time
spent on decision making and to improve purchase decisions (Hennig-
Thurau & Walsh, 2003).
While it is clear that word of mouth is important for gaining new
customers, there is almost no research on the loyalty behavior of these
referred customers. In one of the few studies on this topic, Schmitt,
Skiera, and Van den Bulte (2011) report that referred bank customers
have higher margins and retention rates. They suggest that having a
family member or friend as a customer of the same bank should in-
crease a person's trust and emotional bond with that bank, leading to
higher commitment and attachment. Similarly, we argue that for cus-
tomers who are referred to an online store, there is a transference of the
referrer's positive feelings and attitudes to the receiver of the word of
mouth. This mitigates customers' perceived risk of purchasing online
and increases their willingness to buy (Garbarino & Strahilevitz, 2004),
thereby helping retailers retain customers longer. In comparison, cus-
tomers acquired through means such as advertising, search engine
marketing, or promotional offers might switch the moment they find
convenient or attractive offers from other retailers. Retailers might not
have enough time to develop and strengthen the relationships with
these customers. Thus, we expect that, all else being equal, customers
acquired through referral are more likely to be retained longer than
those acquired through other methods.
H1. Customers acquired through referrals are more likely to be
retained.
2.3. Relationship length
Relationship length indicates the duration of time for which custo-
mers have maintained a buying relationship with the e-retailer (Cooil
et al., 2007; Sabiote & Román, 2009). The customer–firm relationship is
27
Journal of Business Research 92 (2018) 25–35
often fragile in the early stages of the interaction. As firms gain more
experience with customers, they develop a better understanding of
customer needs, expectations, and preferences and, thus, serve their
customers better (Reichheld & Sasser, 1990). Accordingly, over time,
customers' purchase volume often increases and a mutually beneficial
relationship gradually takes shape. Coulter and Coulter (2002) show
that the impact of different marketing actions on trust changes with
tenure.
Prior research has indicated that customer retention depends on the
length of the customer's relationship with the firm (Bolton, 1998; Ittner
& Larcker, 1998; Reichheld & Sasser, 1990; Verhoef, 2003). In other
studies, the relationship age was found to be positively associated with
the number of services purchased (Verhoef, Franses, & Hoekstra, 2002)
as well as repurchase visits and spending (Seiders et al., 2005). By her
behavior, a longstanding customer has indicated that she finds a re-
lationship with the firm worthwhile and has returned to the firm re-
peatedly over time. In our context, the length of the relationship with
the online retailer is therefore expected to be positively linked with
customer retention. Accordingly, we propose the following hypothesis:
H2. Customers who have a longer relationship with the online retailer
are more likely to be retained.
2.4. Service communication
Customers often reach out to the firm through website features,
email, or phone to inquire about products before a purchase, ask for the
status of an order, or complain and communicate problems faced during
or after purchase. These activities relate to the cultivation factor pro-
posed by Srinivasan, Anderson, and Ponnavolu (2002), one of their
“8Cs” constructs that affect e-loyalty. Furthermore, a high degree of
customer–firm communication is often observed in a strong and mature
relationship that is likely to continue (Jaiswal & Niraj, 2011; Zeithaml,
Berry, & Parasuraman, 1996). We expect that the presence of customer-
initiated service communications will positively influence retention.
Therefore, we propose the following hypothesis:
H3. Customers who have initiated a service communication with the
online retailer are more likely to be retained.
2.5. Product return activity
Customers expect retailers to provide timely delivery of the right
merchandise. Accordingly, an accurate fulfillment of purchase orders
positively affects customer satisfaction and retention (Francis & White,
2002; Wolfinbarger & Gilly, 2003). Satisfactory fulfillment can enhance
the perception of the “care” factor discussed by Srinivasan, Anderson,
and Ponnavolu (2002). Care refers to the attention that an e-tailer pays
to customer interface activities, including returns. It is reflected both in
the quality of service and in how well the firm resolves a problem. If an
online retailer is not delivering right products or is delivering them in
poor condition, customers may end up returning the products. This
could result in their dissatisfaction and, in turn, lead to defection. Prior
research using cross-sectional survey data has shown that online cus-
tomer dissatisfaction can cause lower repurchase intentions and revisit
intentions (Cronin & Taylor, 1992; Loiacono, Watson, & Goodhue,
2007) as well as lower website stickiness (Bansal et al., 2004). If cus-
tomers have returned a product, it suggests that they were dissatisfied
with their purchase. Therefore, we propose the following hypothesis:
H4. Customers who had product return activity during one of the
previous purchase encounters are less likely to be retained.
2.6. Type of products purchased
Online retailers offer a wide range of products that include both
A.K. Jaiswal et al.
common and unique items. Srinivasan, Anderson, and Ponnavolu
(2002) suggest that e-tailers that offer higher product variety can en-
gender loyalty because they enable customers to save time when
searching for the right product. They term this as a “choice” factor in
their 8Cs framework of factors that affect e-loyalty. The increasing
availability of recommendation agents and shopping search engines
also facilitates consumers' product search (Häubl & Trifts, 2000). To
take advantage of this, some online retailers offer unique merchandise
as a means of differentiating themselves (Alba & Lynch, 1997). We
argue that the types of products in a customer's purchase history may
indicate the strength of her relationship with the e-retailer. If a cus-
tomer buys widely available goods at a particular retailer, this may be a
strong indication of her volition to sustain her relationship with the
retailer despite the option of buying from competing retailers. In
comparison, purchasing unique products, which are more difficult to
find elsewhere, may involve a compulsory element in the relationship,
as customers do not have other comparable choices. When other op-
tions are found, the relationship can be easily broken and the customer
might switch. Thus, we propose the following hypothesis:
H5. Customers who primarily buy products that are widely available
from multiple retailers are more likely to be retained.
3. Research methodology
We utilize two sets of data, obtained from an online retailer in an
emerging economy, that include a customer survey to measure trust
(and some other variables) and the customers' actual purchase trans-
action data from before and after the survey. Our customer transaction
data is non-contractual in nature: customers may become inactive at
any point in time, unobserved by the retailer. In such contexts, custo-
mers can easily switch to other options. They do business with the firm
only when they desire and do not need to inform the firm when they
terminate the relationship. Thus, we do not know whether a customer is
active or has defected after her last transaction in the data. To address
this latency, we employ a latent attrition model (Schmittlein, Morrison,
& Colombo, 1987; Fader, Hardie, & Lee, 2005) that infers the attrition
of customers in a probabilistic manner on the basis of their transac-
tional patterns. We then use the inferences to test the hypotheses pro-
posed in the previous section. We begin by discussing our data and
measures and follow this with a description of the latent attrition
model.
3.1. Measures and data
As a distinctive feature of our data, we combine and utilize two
distinct data sets: survey data on customers' attitudes toward a major e-
commerce site in India and longitudinal transaction data on the sur-
veyed customers' purchase behavior.1 For the collection of the survey
data, we employed a leading professional marketing research firm. We
obtained the list of all customers who made a purchase at least once
within the 12 months before the time of the survey. Out of this list, we
randomly selected approximately 600 people and forwarded their
contact details to the research firm with a goal of retaining approxi-
mately 500 customers. The research firm trained interviewers for this
research, and this training was partly supervised by one of the study's
co-authors. Customers were primed by an email from the business
schools with which some of the authors were affiliated, and the inter-
viewers also carried a letter jointly signed by those authors highlighting
our academic connections and intent. A remarkable response rate of
more than 85% was achieved without even going through the whole
list, though multiple attempts were made to contact many customers.
Interviews were conducted face-to-face, typically at customers' homes.
1
For confidentiality reasons, we cannot reveal the name of the firm.
28
Journal of Business Research 92 (2018) 25–35
A total of 491 customers fully completed the survey. With the full
cooperation of the firm, we were able to obtain these customers' entire
transaction history going back to their first purchase, which ranged
from the previous six years to just days before we generated the cus-
tomer list for the survey. We were provided with the full log and details
of transactions, which included transaction date, stock-keeping units
(SKUs) of purchased items, their quantities, and price paid. The data
also included information about customers' return activity, matched
with their original purchase. Separately, we obtained the records of all
customer-initiated contacts and service communications. The contacts
were typically made by email, and occasionally by phone. The primary
reason for customers to contact the firm was to inquire about delivery
issues, followed by price and availability inquiries. We further tracked
these customers' purchase behavior for nine quarters (27 months) after
the survey date and collected their transaction information over that
period. Merging the transaction with the survey data enables us to link
the expressed attitudes (trust) and behavioral characteristics of custo-
mers to their continued engagement with the retailer. We measured
trust using a scale adapted from previous studies (Ganesan, 1994;
Morgan & Hunt, 1994). Trust was operationalized by four-item, ten-
point semantic differential scales (“very undependable/very depend-
able,” “very incompetent/very competent,” “very low integrity/very
high integrity,” and “very unresponsive to customers/very responsive
to customers”). The Cronbach's alpha value of the four trust items was
0.88, which is well above the recommended threshold of 0.75, and we
use an average of the four items as our measure of trust. We chose to use
the summated score because of the lack of interpretability of factor
scores. Hair et al. (1998) suggest using a summated score as a com-
promise between using a surrogate variable and factor score ap-
proaches. Potential disadvantages of the summated score approach
(compared with taking a factor score) include complications from
multicollinearity and the possibility that measures may not be ortho-
gonal. Hair et al. suggest that as long as (1) Cronbach's alpha values for
the summed items are high and (2) they load on a single dimension in
factor analysis—both true in our case—this approach may be used
because it can minimize measurement error, is easily replicable across
different research contexts, and allows many different facets of the
construct to be represented.
We present some descriptive statistics about customer demo-
graphics and behavioral characteristics in Table 1. The median level of
reported trust is 7.75, with substantial variation ranging from below 6
to above 9 (10th and 90th percentile values, respectively). Customers in
our data were recorded as being acquired through a variety of methods.
We note that the largest proportion of customers was acquired by
search engine advertising. Of particular interest to us is the proportion
of customers who were acquired by word-of-mouth referral (approxi-
mately 18%). We observe that customers varied significantly in how
long they have been with the firm, with the median age of account
being about a year. Consumers generally did not initiate service com-
munications as a matter of routine, given that the median number of
communications was zero. The median customer had one product re-
turn experience. The retailers sold items belonging to more than three
dozen different broad categories, but individual customers generally
bought items from a very limited number of categories, the median
being just one category. We coded these categories with the help of
experienced shoppers and placed those in two bins: commonly available
offline and difficult to find comparable selection offline. On one end,
there were grocery and other common household items that were
judged to be very commonly available offline. On the other end, there
were books, specialty gift items, and so on for which the online retailer
had very wide selections and it was difficult to find comparable selec-
tions offline. We calculate the percentage of commonly available items
in shoppers' historical, cumulative shopping baskets and report the
percentile values in Table 1.
In the marketing literature, recency, frequency, monetary value
(RFM) variables have been widely used to characterize customers' past
A.K. Jaiswal et al. Journal of Business Research 92 (2018) 25–35

Table 1 time tij and ti, j+1 is then given by

Descriptive statistics.
f (ti, j + 1 | tij ) = λi exp{ −λi (ti, j + 1 − tij )},
Variable Unit Percentile
where parameter λi is assumed to follow a log-normal distribution such
10th 25th Median 75th 90th
that logλi~N(μλ, σλ2). The log-normality ensures that the purchase rate
Trust 10-point scale 6.00 6.75 7.75 8.50 9.00 is nonnegative. Next, the survival function for the right-censored period
Acquired by referral Binary 0 0 0 0 1 (tixi, T] (i.e., the probability that the customer remains active but made
Relationship length Months 10 12 13 22 66 no purchase in the time interval) is given by
Service communication Frequency 0 0 0 3 10
Product return activity Frequency 0 0 1 5 11 S (T | tixi ) = exp{ −λi (T − tixi )}.
Purchasing commonly Percentage 0 10 67 100 100
available products Our model assumes that, after making a purchase, customer i may
Recency Days 43 79 236 319 376 become inactive with a probability of pi, as per Fader, Hardie, and Lee's
No. of purchases Frequency 1 2 5 13 31
(2005) BG/NBD model. We model pi as a function of covariates of in-
Monetary value of Indian rupees 593 911 1441 2956 6687
transactions terest to capture their effect on the attrition probability (Abe, 2009;
Schweidel & Knox, 2013)3:

transaction behavior (Fader & Hardie, 2009; Schweidel & Knox, 2013).
The RFM variable indicates recency (how recently a customer made her
last purchase), purchase frequency (the total number of purchases), and
purchase amount (the value of purchases made), respectively. Research
has found that RFM variables are associated with customer lifetime
value (Fader, Hardie, & Lee, 2005), and database marketers use these
variables when making targeting decisions (Hughes, 1996). Following
the literature, to describe our customer transaction data, we report the
summary measures of the RFM variables over the nine-quarter period
after the survey date. Customers' recency (i.e., the number of days be-
tween an individual's last transaction and the end of the data period)
ranges from just a few days to more than a year. The median customer
placed 5 orders during the data period, and more than 10% customers
placed 30 or more orders. The median customers spent 1441 Indian
rupees per purchase transaction.
3.2. Latent attrition model
Inferring customer attrition has been one of the most important
tasks in analyzing and assessing customer base in non-contractual
businesses (Fader & Hardie, 2009).2 To account for the latency of
customer attrition, marketing researchers have developed a class of
probability models, known as latent attrition models or “Buy 'til You
Die” models (Fader, Hardie, & Lee, 2005; Schmittlein, Morrison, &
Colombo, 1987). This class of models considers that consumers do
business with a firm while they are active (alive) at the firm, and at
some unknown time point they may become inactive (dead) for good.
Latent attrition models have been extensively employed in prior studies
that predict customers' transactional behavior and lifetime values in a
non-contractual setting (Abe, 2009; Park, Park, & Schweidel, 2014;
Reinartz & Kumar, 2000). We adopt this modeling framework to infer
customer attrition following this stream of the literature. In particular,
we employ a well-regarded modeling approach introduced by Fader,
Hardie, and Lee (2005), who developed the beta geometric negative
binomial distribution (BG/NBD) model as an alternative to the classic
Pareto/NBD model (Schmittlein, Morrison, & Colombo, 1987) for ease
of model estimation.
To specify the model, consider the time period (0, T], where 0 is the
beginning of the model calibration period and T denotes the censoring
point that corresponds to the end of the period, during which we ob-
serve customer i making xi transactions at times ti1, ti2, …, tixi. We as-
sume that while the customer has an active relationship with the re-
tailer, her inter-purchase times are exponentially distributed with a
purchase rate of λi. The probability density function for customer i's
interpurchase time between the jth and (j + 1)th purchases occurring at
2
Consistent with the literature on customer relationship management
(Pfeifer & Carraway, 2000), we define customer attrition as a situation in which
customers who are not retained are considered lost for good.
exp{β0 + β1 Trust i + β2 Referrali + β3 Lengthi + β4 Comi
+ β5 Returni + β6 UniqProdi }
pi = .
1 + exp{β0 + β1 Trust i + β2 Referrali + β3 Lengthi + β4 Comi
+ β5 Returni + β6 UniqProdi }
In the previous equation, Trusti is the level of trust customer i has
for the retailer, measured on a 1–10 scale; Referrali is a binary variable
indicating whether the customer was acquired by personal word-of-
mouth referral; Lengthi is the number of months since the customer
made her first purchase at the retailer's website; Comi is a binary
variable indicating whether the customer has ever contacted the com-
pany for a service-related issue; Returni is a binary variable indicating
whether the customer has ever returned a product purchased; and
UniqProdi is a binary variable indicating whether the customer pri-
marily (based on median split) bought products not commonly avail-
able elsewhere in her past transactions.4
Because customer attrition is latent, it is probabilistically considered
and assessed on the basis of people's interpurchase times (ti, j+1 − tij)
and the length of the censored period after their last transaction
(T − tixi). For a frequent shopper with short interpurchase times, it is
less likely to observe a long right-censored period of no purchases if she
is active. In contrast, it is more likely that an infrequent shopper ex-
hibits a long period of no purchases while she is active. Combining the
modeling components to reflect these points, the likelihood function for
customer i's transactional patterns is formulated as
x −1
⎧ i ⎫
Li = ∏ f (ti,j+1 | tij ) (1 − pi ) ⎬ {pi + (1 − pi ) S (T | tixi) },
⎨ j=0
⎩ ⎭
where ti0 is defined to be 0. In the likelihood function, the expression in
the first set of curly braces provides the likelihood for the customer's
interpurchase times and of her being active at least until her (xi − 1)th
transaction. The expression in the second pair of curly braces combines
the two possibilities after the last observed transaction as it considers
(1) the likelihood of the customer's attrition after the last transaction
(pi) and (2) the likelihood of the complementary event (1 − pi) multi-
plied by the probability that the customer is active and just has not
made a purchase in the right-censored period.
Finally, the overall likelihood function of the model is obtained by
considering customer heterogeneity and multiplying Lis across custo-
mers:
3
One could also model λi as a function of covariates to examine their impact
on the purchase rate. Given that our research focuses on retention, we did not
pursue this direction.
4
We define Comi and Returni as binary variables because most of their ob-
served values are 0 and 1. We estimated the model with the frequency values
for Comi and Returni, instead of their binary values, and found that the sig-
nificance and direction of our results remained the same.

29
A.K. Jaiswal et al.
N
L= ∏ ∫ Li dF (λi),
i=1
where N is the number of customers and dF(λi) denotes the probability
density function of the log-normal distribution for the purchase rate λi.
3.3. Estimation approach
We take a Bayesian approach and use the Markov chain Monte Carlo
(MCMC) method to estimate our latent attrition model. The execution
of the method is relatively simple in WinBUGS, a publicly available
software application for Bayesian estimation. To complete the Bayesian
specification of the model, we assign non-informative conjugate priors
to the parameters. For aggregate-level parameters β0, β1, …, β6 and
mean parameter μλ, we use a diffuse normal density prior, N(0,100).
For variance parameter σλ2, we assume that the prior follows an in-
verse-gamma distribution, IG(0.01,0.01).
4. Results
We used sample draws generated from the MCMC chain to calculate
summary values of model fit measures and parameter estimates. The
reported findings were obtained from the MCMC draws for 30,000
iterations, with a burn-in period of 30,000 iterations.
4.1. Model validation
To validate our latent attrition model, we forecast customers' pur-
chase behavior during the nine-quarter data period. Panel A of Fig. 2
compares the distribution of the observed and predicted number of
purchases across customers. As another means to assess the model, we
compare the predicted values of the cumulative number of purchases,
aggregated across customers, to the corresponding observed values in
Panel B of the figure. At the end of the data period, the model predicts
the cumulative number of purchases at a 1.8% error rate. These results
demonstrate the model's capability to capture customers' transactional
patterns. The model's log marginal density and deviance information
criterion, commonly used model fit measures in Bayesian estimation
(Abe, 2009; Schweidel & Knox, 2013), are computed to be −25,537.3
and 51,375.9, respectively.
4.2. Parameter inferences
We estimate our latent attrition model and use the estimates of
model parameters to test our hypotheses. Table 2 reports the posterior
means and 95% posterior intervals for the parameter estimates.5 As the
table shows, the coefficients for five out of six variables included in the
model are significant at a 0.05 level or lower, as their respective 95%
posterior intervals do not contain zero. The effect of four varia-
bles—trust, relationship length, service communications, and pur-
chasing unique products—on customer retention are significant and in
the expected direction. Thus H2, H3, and H5 are supported. In addition,
the role of trust is reconfirmed in our non-contractual context in which
retention is inferred using actual transaction data instead of a survey of
repurchase intention, as in almost all prior studies. The effect of product
return experience (H4) is significant, however, in the reverse direction,
while the mode of customer acquisition (H1) has no significant effect on
retention.
Starting with relationship characteristics, our first result pertains to
the mode of customer acquisition. In H1, we expected a negative
coefficient for the referral variable, signifying that customers acquired
through referral are more likely to be retained. However, in our analysis
5
The 99% posterior intervals for the parameter estimates are available in
supplementary online material.
30
Journal of Business Research 92 (2018) 25–35
of the data, we find that mode of customer acquisition has no significant
effect on customer retention. Thus, H1 is not supported. This could be
because, in the online world, a referral does not require face-to-face
interaction but is often made simply by a few clicks and therefore is
considered a weaker endorsement than its offline counterpart. We re-
turn to this topic in the “Discussion and implications” section. We note
that the parameter for the length of relationship is negative and sig-
nificant, indicating that the longer a customer has been with the re-
tailer, the lower her probability of dropping out from the customer
base. Thus, H2 is supported. This finding suggests that it is worthwhile
for the firm to make early-stage investments and efforts to retain cus-
tomers, because the chance of customer defection decreases with the
length of their relationship with the firm.
With regard to transactional characteristics, the coefficient for our
first variable characterizing customer–retailer interactions, service
communications, is negative and significant at a 0.05 level, in support
of H3. As we expected, those who exchange a service-related commu-
nication with the firm appear to become more engaged with the firm,
and thus they are slower to defect. Assuming that all other covariates
take their average value, we find that customers who do not have a
service communication with the retailer defect with a probability of
0.085 after each purchase transaction. In comparison, the attrition
probability of customers who have the interaction experience is 34%
(= 0.085 )
0.085 − 0.056
smaller, at 0.056.
Next, we find that the parameter for the product return variable is
significant. However, unlike our expectation in H4, the sign of the
parameter is negative, indicating that those who have had a return
experience at the retailer are less likely to stop doing business with the
firm. For an average customer, all else being equal, a product return
experience decreases the attrition probability by 68%, from 0.151 to
0.048. We conjecture that customers who successfully returned a pro-
duct might have become more engaged with the retailer and thereby
chosen to continue the relationship, compared with those who never
had to return a product. In India, returning a product to a brick-and-
mortar retailer is usually a hassle, and thus the ability to return a
product easily can lead to more trust in the online retailer. We further
elaborate on the implications of this finding in the next section.
We find a higher attrition probability for customers who are pri-
marily buying unique, difficult-to-find products, compared with those
who buy products that are commonly available. This finding supports
H5. Customers who primarily buy products that are not commonly
available are probably maintaining their relationship with the online
retailer because of this need and because of the substantial advantage
provided by this retailer (and online retailers in general). Thus, their
continuing to do business with the retailer may be less attributed to the
inherent strength of their relationship with the retailer. The attrition
probability for the two groups of customers is 0.086 versus 0.054.
Finally, we note that trust has a negative and significant impact on
the attrition probability. This indicates that customers who reported a
high level of trust have a lower probability of attrition—or equiva-
lently, are more likely to be retained. Thus, our data and analysis
provide empirical support for the trust–customer retention linkage
when retention is inferred using transaction data. The finding suggests
that making efforts and investments in enhancing the perceptions of
competence, integrity, dependability and the responsiveness to custo-
mers' needs—all constituent elements of trust—are worthwhile for on-
line retailers to increase customer retention. At the average, a one-point
increase in the trust measure decreases a customer's attrition prob-
ability by 9%, from 0.069 to 0.063.
4.3. Split-sample analysis on the trust–retention relationship
While previous literature has documented the importance of trust in
retaining customers (Agustin & Singh, 2005; Garbarino & Johnson,
1999), it would be managerially useful to find segments that show
A.K. Jaiswal et al.
Panel A: Distribuon of purchase frequency
Panel B: Cumulave transacons
Cumulave no. of purchases
6,000
4,500
3,000
1,500
0
0 3
stronger or weaker links between trust and retention. A manager could
use these findings when allocating marketing budgets and resources to
improve dimensions of segments in which they are likely to positively
influence retention. Demographic variables have been used in some
prior studies in the direct marketing and customer profitability mea-
surement literature (Lee & Turban, 2001; Reinartz & Kumar, 2003).
These demographic variables affect shoppers' trust propensity to pur-
chase from online stores (Lee & Turban, 2001). Massey and Dawes
(2007) studied the effect of personal characteristics on trust formation.
Drawing on previous research and available data, we examine three
potential segmentation variables: income (Homburg & Giering, 2001;
Reinartz & Kumar, 2003), breadth of buying (Srinivasan, Anderson, &
Ponnavolu, 2002), and transaction value (McCarty & Hastak, 2007). We
use a split-sample approach that is often employed to explore con-
tingencies of relationships (Hitt, Hoskisson, & Kim, 1997; Miller,
1987).6 We use a median split on the variables to create segments and
6
A more formal way to conduct this analysis is by including the split
Journal of Business Research 92 (2018) 25–35
Observed Predicted
6 9 12 15 18 21 24 27
Month
Fig. 2. Model validation.
then estimate our latent attrition model to determine whether the
trust–retention relationship differs across these segments. MacCallum
et al. (2002) also report that many studies use this approach to find
differences across dichotomous or dichotomized groups. Table 3 reports
the coefficient for trust in our latent attrition model, calibrated on sets
of split-sample data. The online supplementary material provides full
estimation results for each split-sample analysis.
4.3.1. Customer income
In extant literature, Homburg and Giering (2001) showed that
customer income moderates the impact of satisfaction on loyalty.
Seiders et al. (2005) reported that income moderates the effect of sa-
tisfaction on repurchase visits and spending. From the estimation
(footnote continued)
variables and their interaction terms with trust in specifying the attrition
probability pi. However, we found that this approach was not feasible in our
research because of the multi-collinearity problem arising from high correla-
tions (larger than 0.9) between the main and interaction terms.
31
A.K. Jaiswal et al. Journal of Business Research 92 (2018) 25–35

Table 2 product assortment, quality offered, and systems of dealing with cus-
Parameter estimates. tomers (Reichheld & Teal, 2001). However, in terms of the trus-
Parameter Description Posterior 95% posterior t–retention relationship, we find no significant difference between
mean interval customers who buy a narrow range of products versus those who buy a
large variety of products from the retailer in our study. It may be that
μλ Mean parameter for log(λi) −4.421 [−4.522,
breadth of buying interacts with other variables to affect customer re-
−4.318]
σλ2 Variance parameter for log(λi) 0.814 [0.676, 0.960]
tention. For instance, Dawes (2009) found that product breadth has a
β0 Intercept for the attrition −0.066 [−0.779, 0.656] positive impact on retention only among newer customers.
probability
β1 Trust −0.105 [−0.193, 4.3.3. Monetary value of transactions
−0.021]
We also split customers on the basis of the average monetary value
β2 Acquired by referral −0.080 [−0.393, 0.221]
β3 Relationship length −0.044 [−0.057, of their transactions. We find that trust decreases the attrition prob-
−0.033] ability only for customers who made lower average purchases. One
β4 Service communication −0.448 [−0.710, possible explanation for this result could be the correlation of purchase
−0.184]
amount with income. It is expected that, all else being equal, customers
β5 Product return activity −1.268 [−1.561,
−0.969]
with lower income purchase less than those with higher income. We
β6 Purchasing products not 0.486 [0.214, 0.777] also find a positive correlation between purchase amount and income in
commonly available our data. In this regard, the results of the split-sample analysis for the
size of transactions are in line with our finding that trust significantly
Notes: Boldface indicates that the 95% posterior interval excludes zero. increases the retention of lower-income customers, but not higher-in-
come ones.
Table 3
Coefficients for trust in split-sample analysis.
5. Discussion and implications
Split Variable Description Posterior 95% Posterior
Mean Interval In this research, we examine the effect of online trust and other
Customer income Higher income −0.092 [−0.188, 0.005]
relevant relationship and transactional characteristics on customer re-
Lower income −0.250 [−0.492, tention in an emerging market. We use a latent attrition model to assess
−0.015] customer retention in a probabilistic manner, based on individuals'
Breadth of buying Purchasing a large −0.056 [−0.208, 0.100] transactional patterns. We then use the inferred attrition probability to
variety of products
establish the role of trust and other factors in promoting retention. We
Purchasing a narrow −0.101 [−0.229, 0.034]
range of products find that trust enhances the likelihood of retention. Among relationship
Monetary value of Larger monetary −0.055 [−0.210, 0.093] characteristics, the length of customer tenure is positively associated
transactions value with future retention, whereas how the relationship started does not
Smaller monetary −0.167 [−0.293, seem to matter. With regard to transactional characteristics, customers'
value −0.050]
service communications and product return activity with the online
Notes: Boldface indicates that the 95% posterior interval excludes zero. retailer increase the likelihood of retention, while customers who pre-
dominantly buy difficult-to-find products are less likely to be retained.
results of the split-sample analysis for customer income, we find an In our split-sample analysis, which aims to understand con-
insignificant coefficient for trust among customers with higher income. tingencies of the relationship between trust and retention, we find that
In contrast, for those with lower income, the parameter for trust is es- consumers' income and purchase amount affect the importance of trust
timated to be negative and significant, indicating that trust increases in influencing retention. These findings can help online retailers iden-
the retention of the lower-income customers. Prior research has sug- tify a segment of customers for whom trust would play a more im-
gested that higher income generally increases the number of shopping portant role in their retention and, accordingly, allocate marketing re-
options available to customers and thus results in a lower level of sources and communications to enhance the trust level of customers in
loyalty (Homburg & Giering, 2001). Reinartz and Kumar (2003) argue the segment.
that affluent customers have greater opportunity costs of their time; The key contributions of our research can be summarized as follows.
therefore, they may switch to other retailers to save time. Trust is a risk- Using both perceptual and behavioral data to examine the trus-
mitigation mechanism, and thus higher-income consumers have less t–retention relationship in a longitudinal framework, we study the
need for such a mechanism because their perceived risk in purchasing impact of a broad set of relationship and transactional characteristics
remotely is lower than that of low-income consumers. Our result sup- on retention behavior. Our data and modeling approach enable us to
ports the idea that the online retailer should focus its trust-enhancing probabilistically infer customer retention using actual customer trans-
activities on lower-income consumers. action behavior, rather than relying on customers' stated intent of
loyalty. As such, by linking trust to the probability of attrition, we are
able to show that trust positively affects actual loyalty behavior, and
4.3.2. Breadth of buying not just loyalty intentions. Our findings also show that relationship and
Although many e-commerce sites offer a large collection of pro- transactional characteristics that firms can easily track using data re-
ducts, customers may buy only a limited range of products from each cords have a systematic impact on retention. Our finding that return
site. For example, a customer may prefer to purchase books from one experience (discussed in more detail next) can actually increase reten-
site, apparel from another, and electronic products from a yet another. tion is quite noteworthy. The findings thus emphasize the point made
Other customers, however, may prefer to buy most of these products by Kumar, Pozza, and Ganesh (2013): that researchers should broaden
from just one online store. In the latter case, the scope of the re- their focus to other variables beyond satisfaction and trust when
lationship is much broader than the former case. The primary benefit studying loyalty outcomes.
for customers who purchase products across different categories from Scholars (Burgess & Steenkamp, 2006; Sheth, 2011) have contended
one online store is the reduction in opportunity costs of search time that a key area of concern in the marketing literature is that the extant
(Alba & Lynch, 1997). In offline environments, advantages of a broader body of knowledge is based solely on studies carried out in North
relationship for customers include their having a better idea of a store's America and Europe. Our research helps fill this gap in the literature.

32
A.K. Jaiswal et al.
While our research examines the effect of trust on customer retention
using data from an Indian retailer, our results can have implications for
other emerging economies where trust is an essential issue. For ex-
ample, in Thailand, in which merely 10% of 25 million Internet users
purchase online, the Thai E-Commerce Association claims that “trust is
still a key stumbling block for online shopping growth as Thai con-
sumers in general still lack confidence in making transactions via
electronic commerce” (Phusadee, 2013). Related research by Forrester
Consulting on customers in Asia-Pacific countries (e.g., China) reports
that consumers are not keen to buy from small-sized online retailers,
primarily due to lack of trust in e-commerce. Along this line, trust-re-
lated reasons were cited by 46% of customers in this region to explain
why they did not shop online, compared to the global average of just
21% (Philippines Daily Inquirer, 2015). Thus, it appears that trust issues
are highly important in many emerging economies, and marketers in
these countries can benefit from our findings.
One interesting finding from our research is that consumers with
prior return activity have a higher, rather than lower, likelihood of
retention. The literature on complaining behavior and service recovery
(Clark, Kaminski, & Rink, 1992) has indicated that the customer–firm
relationship is stronger for customers who experienced an adverse
service event in which the marketer effectively performed service re-
covery, compared with those who did not experience an adverse service
event in the first place. This condition has also been referred to in the
literature as the service recovery paradox (SRP) effect—when customer
satisfaction is higher after service recovery than before the service
failure (McCollough & Bharadwaj, 1992). However, extant empirical
studies on SRP have shown mixed findings. The meta-analysis of SRP
studies by Matos, Henrique, and Rossi (2007) showed that SRP has a
positive and significant effect on satisfaction, yet its effect on loyalty
intentions is insignificant. Our research contributes to the service re-
covery literature, as our results—in contrast to the meta-analysis—-
show that the SRP enhances customer retention. In a direct marketing
context, Reinartz and Kumar (2003) argue that customers who make
more purchases tend to return more merchandise. These frequent
buyers are comfortable undertaking the actions required to return
products. They handle the process effectively and perceive returns as
part and parcel of buying from direct marketers. Although customers'
product returns lead to additional reverse logistic costs and negatively
affect firm profits, when managed well, they can have a positive impact
on future purchase behavior and thus contribute in increasing long-
term customer values (Petersen & Kumar, 2009). In making decisions
related to customer relationship management and the deployment of
marketing resources, managers often ignore product return behavior or
treat it as a cost that needs to be reduced. However, our finding suggests
that product returns by customers can actually be advantageous to
firms if they are managed appropriately and to customers' satisfaction.
This is consistent with recent research in developed markets indicating
that firms can enhance their short- and long-term profitability when
they consider the perceived risk and costs involved in product returns
(Petersen & Kumar, 2015).
Although we had hypothesized that customers acquired through
referrals are more likely to be retained, our results indicate that mode of
customer acquisition—whether it is from e-WOM, advertising, or search
engines—has no effect on customer retention in our data. There are
several reasons that could explain our findings. First, consumers using
social media outlets such as Twitter and Facebook can spread e-WOM
instantly to a very large audience. Thus, the sheer quantity of e-WOM
has increased exponentially in recent years (Hennig-Thurau et al.,
2004), making it difficult to evaluate the content and credibility of the
communication (Levy & Gvili, 2015). Second, in many cases, referrals
are associated with some monetary benefits to the referrer. Because
companies often provide financial incentives to promote e-WOM, this
introduces potential artificiality into e-WOM and could, in turn, reduce
the credibility and effectiveness of referrals in this context (Cheung
et al., 2009), resulting in no differences in the retention rates of referred
33
Journal of Business Research 92 (2018) 25–35
customers. This finding needs to be studied further in other countries
and product domains.
In terms of implications for practice, our result underscores the
important role of trust for the success of online retailers. Firms should
prioritize marketing and operational activities that enhance customers'
perceptions of them as being dependable and having integrity. This
would include appropriate communications to customer inquiries,
third-party certifications on the website, consistent product assortment,
and a high degree of product availability. In addition, our finding on
return behavior suggests that the return process should be made easy
for customers because this seems to further bolster the customer–firm
relationship and reduce churn.
6. Limitations and directions for future research
Our research has certain limitations. We use data from a single re-
tailer and do not consider the effect of competition among retailers.
Future research probing customer behavior across multiple retailers
would enhance our understanding of the role of trust in customer re-
tention. Our study uses data from India, and findings may not ne-
cessarily be generalizable to other emerging markets. However, given
that PricewaterhouseCoopers has projected India to be the world's
second-largest economy in 2050 (Hawksworth & Audino, 2017), our
study makes important contributions to research on a major emerging
market. Future research could collect data from other important
emerging markets (e.g., China, Indonesia) to examine their similarities
and differences. Future research could also investigate the effect of trust
and relationship characteristics on other relevant outcomes, such as
customers' share of wallet. Prior studies have shown that enhancing
customers' share of wallet and increasing customer retention may
warrant different strategies (Verhoef, 2003). It would also be fruitful to
investigate the psychological process through which trust is engendered
as well as the factors that affect the trust-building mechanism and how
they differ between emerging and developed markets. Finally, future
research could study how trust and other relationship characteristics
affect customers' profitability, future behavior, and lifetime value.
Acknowledgements
The first author acknowledges support from the Center for Retailing
at Indian Institute of Management, Ahmedabad. The second author
acknowledges support from USC-Marshall School of Business research
support fund. We thank Piyush Kumar Sinha, former faculty at Indian
Institute of Management Ahmedabad, for help in connecting with the
company that enabled us to collect data for this research and for useful
suggestions during early stages of the project. We thank seminar par-
ticipants at Kent State University and conference session attendees at
the 6th NASMEI-Great Lakes Conference in Chennai, India in 2012 and
China-India Customer Insights Conference in London Business School in
2016.
Appendix. Additional Estimation Results
Supplementary material to this article can be found online at
https://doi.org/10.1016/j.jbusres.2018.07.007.
References
Aaker, D. A. (1992). The value of brand equity. Journal of Business Strategy, 13(4), 27–32.
Abe, M. (2009). Counting your customers one by one: A hierarchical Bayes extension to
the Pareto/NBD model. Marketing Science, 28(3), 541–553.
Agustin, C., & Singh, J. (2005). Curvilinear effects of consumer loyalty determinants in
relational exchanges. Journal of Marketing Research, 42(1), 96–108.
Alba, J., & Lynch, J. (1997). Interactive home shopping: Consumer, retailer, and manu-
facturer incentives to participate in electronic marketplaces. Journal of Marketing,
61(3), 38–53.
Ashraf, A. R., Razzaque, M. A., & Thongpapanl, N. T. (2016). The role of customer reg-
ulatory orientation and fit in online shopping across cultural contexts. Journal of
A.K. Jaiswal et al.
Business Research, 69(12), 6040–6047.
Ashraf, A. R., Thongpapanl, N. T., & Auh, S. (2014). The application of the technology
acceptance model under different cultural contexts: The case of online shopping
adoption. Journal of International Marketing, 22(3), 68–93.
Bansal, H. S., McDougall, G. H. G., Dikolli, S. S., & Sedatole, K. L. (2004). Relating e-
satisfaction to behavioral outcomes: An empirical study. Journal of Services Marketing,
18(4), 290–302.
Bart, Y., Shankar, V., Sultan, F., & Urban, G. L. (2005). Are the drivers and role of online
trust the same for all web sites and consumers? A large-scale exploratory empirical
study. Journal of Marketing, 69(4), 133–152.
Bickart, B. A. (1993). Carryover and backfire effects in marketing research. Journal of
Marketing Research, 30(1), 52–62.
Bolton, R. N. (1998). A dynamic model of the duration of the customer's relationship with
a continuous service provider: The role of satisfaction. Marketing Science, 17(1),
45–65.
Bolton, R. N., Kannan, P. K., & Bramlett, M. D. (2000). Implications of loyalty program
membership and service experiences for customer retention and value. Journal of the
Academy of Marketing Science, 28(1), 95–108.
Brynjolfsson, E., & Smith, M. D. (2000). Frictionless commerce? A comparison of Internet
and conventional retailers. Management Science, 46(4), 563–585.
Burgess, S. M., & Steenkamp, J.-B. E. M. (2006). Marketing renaissance: How research in
emerging markets advances marketing science and practice. International Journal of
Research in Marketing, 23(4), 337–356.
Chaudhuri, A., & Holbrook, M. B. (2001). The chain of effects from brand trust and brand
affect to brand performance: The role of brand loyalty. Journal of Marketing, 65(2),
81–93.
Cheung, M. Y., Luo, C., Sia, C. L., & Chen, H. (2009). Credibility of electronic word-of-
mouth: Informational and normative determinants of on-line consumer re-
commendations. International Journal of Electronic Commerce, 13(4), 9–38.
Chiou, J.-S., & Droge, C. (2006). Service quality, trust, specific asset investment, and
expertise: Direct and indirect effects in a satisfaction-loyalty framework. Journal of
the Academy of Marketing Science, 34(4), 613–627.
Clark, G. L., Kaminski, P. F., & Rink, D. R. (1992). Consumer complaints: Advice on how
companies should respond based on an empirical study. Journal of Consumer
Marketing, 9(3), 5–14.
Cooil, B., Keiningham, T. L., Aksoy, L., & Hsu, M. (2007). A longitudinal analysis of
customer satisfaction and share of wallet: Investigating the moderating effect of
customer characteristics. Journal of Marketing, 71(1), 67–83.
Coulter, K. S., & Coulter, R. A. (2002). Determinants of trust in a service provider: The
moderating role of length of relationship. Journal of Services Marketing, 16(1), 35–50.
Cronin, J. J., Jr., & Taylor, S. A. (1992). Measuring service quality: A reexamination and
extension. Journal of Marketing, 56(3), 55–68.
Dawes, J. (2009). The effect of service price increases on customer retention: The mod-
erating role of customer tenure and relationship breadth. Journal of Service Research,
11(3), 232–245.
Fader, P. S., & Hardie, B. G. S. (2009). Probability models for customer-base analysis.
Journal of Interactive Marketing, 23(1), 61–69.
Fader, P. S., Hardie, B. G. S., & Lee, K. L. (2005). “Counting your customers” the easy way:
An alternative to the Pareto/NBD model. Marketing Science, 24(2), 275–284.
Francis, J. E., & White, L. (2002). PIRQUAL: a scale for measuring customer expectations
and perceptions of quality in Internet retailing. Proceedings of the 2002 American
marketing association winter Educators' conference: Marketing theory and applications
(pp. 263–270). Chicago: American Marketing Association.
Ganesan, S. (1994). Determinants of long-term orientation in buyer-seller relationships.
Journal of Marketing, 58(2), 1–19.
Garbarino, E., & Johnson, M. S. (1999). The different roles of satisfaction, trust, and
commitment in customer relationships. Journal of Marketing, 63(2), 70–87.
Garbarino, E., & Strahilevitz, M. A. (2004). Gender differences in the perceived risk of
buying online and the effects of receiving a site recommendation. Journal of Business
Research, 57(7), 768–775.
Gu, B., Park, J., & Konana, P. (2012). Research note: The impact of external word-of-
mouth sources on retailer sales of high-involvement products. Information Systems
Research, 23(1), 182–196.
Gustafsson, A., Johnson, M. D., & Roos, I. (2005). The effects of customer satisfaction,
relationship commitment dimensions, and triggers on customer retention. Journal of
Marketing, 69(4), 210–218.
Hair, J. F., Tatham, R. L., Anderson, R. E., & Black, W. (1998). Multivariate data analysis
(5th ed.). Upper Saddle River, NJ: Prentice Hall.
Häubl, G., & Trifts, V. (2000). Consumer decision making in online shopping environ-
ments: The effects of interactive decision aids. Marketing Science, 19(1), 4–21.
Hawksworth, J., & Audino, H. (2017). The world in 2050. PWC Globalhttps://www.pwc.
com/gx/en/issues/economy/the-world-in-2050.html, Accessed date: 28 June 2018.
Hennig-Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word-
of-mouth via consumer-opinion platforms: What motivates consumers to articulate
themselves on the Internet? Journal of Interactive Marketing, 18(1), 38–52.
Hennig-Thurau, T., & Walsh, G. (2003). Electronic word-of-mouth: Motives for and
consequences of reading customer articulations on the Internet. International Journal
of Electronic Commerce, 8(2), 51–74.
Hitt, M. A., Hoskisson, R. E., & Kim, H. (1997). International diversification: Effects on
innovation and firm performance in product-diversified firms. Academy of
Management Journal, 40(4), 767–798.
Homburg, C., & Giering, A. (2001). Personal characteristics as moderators of the re-
lationship between customer satisfaction and loyalty—An empirical analysis.
Psychology and Marketing, 18(1), 43–66.
Hughes, A. M. (1996). Boosting responses with RFM. 3(3), Marketing Tools4.
Internet Live Stats (2016). Internet users by country. https://www.internetlivestats.com/
34
Journal of Business Research 92 (2018) 25–35
internet-users-by-country/, Accessed date: 17 October 2016.
Ittner, C. D., & Larcker, D. F. (1998). Are nonfinancial measures leading indicators of
financial performance? An analysis of customer satisfaction. Journal of Accounting
Research, 36(3), 1–35.
Jaiswal, A. K., & Niraj, R. (2011). Examining mediating role of attitudinal loyalty and
nonlinear effects in satisfaction-behavioral intentions relationship. Journal of Services
Marketing, 25(3), 165–175.
Jaiswal, A. K., Niraj, R., & Venugopal, P. (2010). Context-general and context-specific
determinants of online satisfaction and loyalty for commerce and content sites.
Journal of Interactive Marketing, 24(3), 222–238.
Kim, D. J., Ferrin, D. L., & Rao, H. R. (2009). Trust and satisfaction, two stepping stones
for successful e-commerce relationships: A longitudinal exploration. Information
Systems Research, 20(2), 237–257.
Kumar, V., Pozza, I. D., & Ganesh, J. (2013). Revisiting the satisfaction–loyalty re-
lationship: Empirical generalizations and directions for future research. Journal of
Retailing, 89(3), 246–262.
Lee, M. K. O., & Turban, E. (2001). A trust model for consumer internet shopping.
International Journal of Electronic Commerce, 6(1), 75–91.
Levy, S., & Gvili, Y. (2015). How credible is e-word of mouth across digital-marketing
channels? Journal of Advertising Research, 55(1), 95–109.
Lewis, M. (2004). The influence of loyalty programs and short-term promotions on cus-
tomer retention. Journal of Marketing Research, 41(3), 281–292.
Loiacono, E. T., Watson, R. T., & Goodhue, D. L. (2007). WebQual: An instrument for
consumer evaluation of web sites. International Journal of Electronic Commerce, 11(3),
51–87.
MacCallum, R. C., Zhang, S., Preacher, K. J., & Rucker, D. D. (2002). On the practice of
dichotomization of quantitative variables. Psychological Methods, 7(1), 19–40.
Malhotra, N., Sahadev, S., & Purani, K. (2017). Psychological contract violation and
customer intention to reuse online retailers: Exploring mediating and moderating
mechanisms. Journal of Business Research, 75(6), 17–28.
Martinsons, M. G. (2008). Relationship-based e-commerce: Theory and evidence from
China. Information Systems Journal, 18(4), 331–356.
Massey, G. R., & Dawes, P. L. (2007). Personal characteristics, trust, conflict, and effec-
tiveness in marketing/sales working relationships. European Journal of Marketing,
41(9/10), 1117–1145.
Matos, C. A.d., Henrique, J. L., & Rossi, C. A. V. (2007). Service recovery paradox: A meta-
analysis. Journal of Service Research, 10(1), 60–77.
Mayer, R. C., Davis, J. H., & Schoorman, F. D. (1995). An integrative model of organi-
zational trust. Academy of Management Review, 20(3), 709–734.
McCarty, J. A., & Hastak, M. (2007). Segmentation approaches in data-mining: A com-
parison of RFM, CHAID, and logistic regression. Journal of Business Research, 60(6),
656–662.
McCollough, M. A., & Bharadwaj, S. G. (1992). The recovery paradox: An examination of
consumer satisfaction in relation to disconfirmation, service quality, and attribution
based theories. In C. T. Allen, (Ed.). Marketing theory and applications (pp. 119).
Chicago: American Marketing Association.
Miller, D. (1987). Strategy making and structure: Analysis and implications for perfor-
mance. Academy of Management Journal, 30(1), 7–32.
Morgan, R. M., & Hunt, S. D. (1994). The commitment-trust theory of relationship mar-
keting. Journal of Marketing, 58(3), 20–38.
Morwitz, V. G., Steckel, J. H., & Gupta, A. (2007). When do purchase intentions predict
sales? International Journal of Forecasting, 23(3), 347–364.
Pan, Y., Sheng, S., & Xie, F. T. (2012). Antecedents of customer loyalty: An empirical
synthesis and reexamination. Journal of Retailing and Consumer Services, 19(1),
150–158.
Park, C. H., Park, Y.-H., & Schweidel, D. A. (2014). A multi-category customer base
analysis. International Journal of Research in Marketing, 31(3), 266–279.
Pascual-Miguel, F. J., Agudo-Peregrina, Á. F., & Chaparro-Peláez, J. (2015). Influences of
gender and product type on online purchasing. Journal of Business Research, 68(7),
1550–1556.
Petersen, J. A., & Kumar, V. (2009). Are product returns a necessary evil? Antecedents
and consequences. Journal of Marketing, 73(3), 35–51.
Petersen, J. A., & Kumar, V. (2015). Perceived risk, product returns, and optimal resource
allocation: Evidence from a field experiment. Journal of Marketing Research, 52(2),
268–285.
Pfeifer, P. E., & Carraway, R. L. (2000). Modeling customer relationships as Markov
chains. Journal of Interactive Marketing, 44(2), 43–55.
Philippines Daily Inquirer (2015, March 6). Selling goods online? Trust is keyhttp://
business.inquirer.net/187897/selling-goods-online-trust-is-key, Accessed date: 28
June 2018.
Phusadee, A. (2013, January 10). Campaign aims to boost trust in online shopping.
Bangkok Post.
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method
biases in behavioral research: A critical review of the literature and recommended
remedies. Journal of Applied Psychology, 88(5), 879–903.
Reichheld, F. F., & Sasser, W. E., Jr. (1990). Zero defections: Quality comes to services.
Harvard Business Review, 68(5), 105–111.
Reichheld, F. F., & Teal, T. (2001). The loyalty effect: The hidden force behind growth, profits,
and lasting value. Boston: Harvard Business School Press.
Reinartz, W. J., & Kumar, V. (2000). On the profitability of long-life customers in a
noncontractual setting: An empirical investigation and implications for marketing.
Journal of Marketing, 64(4), 17–35.
Reinartz, W. J., & Kumar, V. (2003). The impact of customer relationship characteristics
on profitable lifetime duration. Journal of Marketing, 67(1), 77–99.
Richins, M. L. (1983). Negative word-of-mouth by dissatisfied consumers: A pilot study.
Journal of Marketing, 47(1), 68–78.
A.K. Jaiswal et al.
Russell-Bennett, R., McColl-Kennedy, J. R., & Coote, L. V. (2007). Involvement, sa-
tisfaction, and brand loyalty in a small business services setting. Journal of Business
Research, 60(12), 1253–1260.
Sabiote, E. F., & Román, S. (2009). The influence of social regard on the customer–service
firm relationship: The moderating role of length of relationship. Journal of Business
and Psychology, 24(4), 441–453.
Schmitt, P., Skiera, B., & Van den Bulte, C. (2011). Referral programs and customer value.
Journal of Marketing, 75(1), 46–59.
Schmittlein, D. C., Morrison, D. G., & Colombo, R. (1987). Counting your customers: Who
are they and what will they do next? Management Science, 33(1), 1–24.
Schweidel, D. A., & Knox, G. (2013). Incorporating direct marketing activity into latent
attrition models. Marketing Science, 32(3), 471–481.
Seiders, K., Voss, G. B., Grewal, D., & Godfrey, A. L. (2005). Do satisfied customers buy
more? Examining moderating influences in a retailing context. Journal of Marketing,
69(4), 26–43.
Sheth, J. N. (2011). Impact of emerging markets on marketing: Rethinking existing per-
spectives and practices. Journal of Marketing, 75(4), 166–182.
Singh, J., & Sirdeshmukh, D. (2000). Agency and trust mechanisms in consumer sa-
tisfaction and loyalty judgments. Journal of the Academy of Marketing Science, 28(1),
150–167.
Srinivasan, S. S., Anderson, R., & Ponnavolu, K. (2002). Customer loyalty in e-commerce:
An exploration of its antecedents and consequences. Journal of Retailing, 78(1),
41–50.
Stathopoulou, A., & Balabanis, G. (2016). The effects of loyalty programs on customer
satisfaction, trust, and loyalty toward high- and low-end fashion retailers. Journal of
Business Research, 69(2), 5801–5808.
Szymanski, D. M., & Henard, D. H. (2001). Customer satisfaction: A meta-analysis of the
empirical evidence. Journal of the Academy of Marketing Science, 29(1), 16–35.
Venetis, K. A., & Ghauri, P. N. (2004). Service quality and customer retention: Building
long-term relationships. European Journal of Marketing, 38(11/12), 1577–1598.
Verhoef, P. C. (2003). Understanding the effect of customer relationship management
efforts on customer retention and customer share development. Journal of Marketing,
67(4), 30–45.
Verhoef, P. C., Franses, P. H., & Hoekstra, J. C. (2002). The effect of relational constructs
on customer referrals and number of services purchased from a multiservice provider:
Does age of relationship matter? Journal of the Academy of Marketing Science, 30(3),
202–216.
35
Journal of Business Research 92 (2018) 25–35
Wolfinbarger, M., & Gilly, M. C. (2003). eTailQ: Dimensionalizing, measuring and pre-
dicting eTail quality. Journal of Retailing, 79(3), 183–198.
Zeithaml, V. A., Berry, L. L., & Parasuraman, A. (1996). The behavioral consequences of
service quality. Journal of Marketing, 60(2), 31–46.
Anand Kumar Jaiswal is Associate Professor of Marketing and Chairperson of Marketing
Area at Indian Institute of Management, Ahmedabad, India. His research interests include
bottom of the pyramid markets, services management, customer satisfaction and ecom-
merce. He has published papers in the Long Range Planning, Journal of Interactive
Marketing, Organization & Environment, Journal of Services Marketing, Journal of
Consumer Marketing, Innovations, Journal of Business Ethics, Managing Service Quality
and Decision, among others.
Rakesh Niraj is Associate Professor, Design & Innovation and faculty director of the
MSM-Business Analytics program at the Westherhead School of Management, Case
Western Reserve University, Cleveland. His research deals with both analytical and em-
pirical modeling of marketing strategy in areas like information sharing, price targeting
and customer relationship management (CRM). He has published papers in the Journal of
Marketing, Journal of Marketing Research, Marketing Science, Management Science and
Journal of Interactive Marketing, among others.
Chang Hee Park is Assistant Professor of Marketing at the School of Management,
Binghamton University, SUNY. His research interests include customer relationship
management, digital marketing, and probability models. His work has been published in
Marketing Science, International Journal of Research in Marketing Journal of Retailing,
and Journal of the Royal Statistical Society: Series A, among others.
Manoj K. Agarwal is Professor of Marketing and Associate Dean, School of Management,
Binghamton University, SUNY. His research interests are in the area of marketing-finance
interface, impact of user interaction design on consumers, branding and brand equity
measurement, and conjoint analysis. His branding research has received the prestigious
Robert D. Buzell Best Paper award from Marketing Science Institute in 2005. His research
has appeared in Journal of Marketing, Journal of Consumer Research, Journal of
Marketing Research and Marketing Science, among others.