The effects of low satisfactory
recovery strategy combinations
following an apparel product defect
in an online context
Student: Wenjing Cai
Student number: 10602941
Date and version: 12/08/2014 Final draft of Master Thesis
MSc in Business Studies – Marketing Track
Institution: FEB UvA
Content
ABSTRACT ... 4
1. INTRODUCTION ... 4
2. LITERATURE REVIEW ... 9
2.1CUSTOMER COMPLAINTS AND SERVICE FAILURE ... 9
2.2SERVICE RECOVERY STRATEGY ... 12
2.3CLOTHING PRODUCTS ... 17
2.4GENDER, AGE, EDUCATION LEVEL, AND PAST ONLINE SHOPPING FREQUENCY ... 18
2.5CUSTOMER SATISFACTION AND PURCHASE INTENTION ... 20
2.6GAP AND RESEARCH QUESTION ... 22
2.7HYPOTHESES ... 24
3. DATA AND METHODOLOGY ...28
3.1RESEARCH DESIGN ... 28 3.2SAMPLE ... 31 3.3INDEPENDENT VARIABLES ... 32 3.4DEPENDENT VARIABLES ... 34 3.5CONTROL VARIABLES ... 36 3.6EXPERIMENTAL PROCEDURE ... 37 4. RESULTS ... 9 4.1RELIABILITY ANALYSIS ... 39 4.2DESCRIPTIVE STATISTICS ... 40 4.3CORRELATION ANALYSIS ... 43 4.4NORMALITY ANALYSIS ... 45 4.5HYPOTHESES TESTING ... 47
4.5.1INDEPENDENT SAMPLE T-TESTS ... 47
4.5.2REGRESSION ANALYSIS... 50
4.6LIMITATIONS ... 51
5. DISCUSSION ...52
5.1DISCUSSION OF RESULTS ... 53
5.1.2 Answers to research question ... 55 5.2IMPLICATIONS ... 56 6. CONCLUSION ...58 6.1SUMMARY ... 58 6.2FUTURE RESEARCH ... 60 APPENDIX QUESTIONNAIRES ...62
APPENDIX TABLES AND GRAPHS ...68
Abstract
This study extends previous research by studying the effects that low-satisfaction
recovery strategy combinations have on perceived customer satisfaction and purchase
intention following an apparel product defect in an online shopping context. 266
respondents are randomly divided into five treatment groups (one control group and
four with different recovery strategy combinations) to compare the effects of different
low-satisfaction recovery strategy combinations. Due to the nature of the experiment,
independent sample t-tests are used to test the significance of the difference of the
dependent variables. The results reveal that, the combination of rapid response and
replacement and the combination of apology and replacement reveal higher ability in
increasing purchase intention than the combination of rapid response, apology, and
replacement. The findings are beneficial for start-up e-retailers and companies that
want to have cost-benefit in the long run.
has emerged and rapidly grew worldwide. In recent years, online shopping has
become a new trend in almost every industry and revealed enormous potential to
overrule offline shopping. More people start shopping online instead of shopping at
brick-and-mortar stores. According to Forrester Research (March, 2010), online sales
in the US will grow to $250 billion in 2014, and are expected to increase by 8-10%
annually in the future.
Along with the emerging online shopping trend, service failure has become an
important research topic attracting scholars’ attention. Service failure, as one of the
many issues that emerge within this revolution, has been studied by many scholars
(Forbes et al. 2005; Holloway and Beatty, 2003; Kelley et al., 1994; Kuo et al., 2011).
Among all the service failures studied by previous research, product defect is one of
the major service failures that customers encounter during the purchase process,
possessing the feature of being both severe and frequent (Kelley et al., 1994; Kuo et
al., 2011). Therefore, it would be an ideal and representative of service failure for
studying the influence of service failure has on customers’ perception.
compares anticipated consequences with rewards and costs (Bolton and Drew, 1991;
Maxham, 2001; Oliver, 1980; Yi, 1990). A service failure would influence the
purchase outcome and then reflect through customer satisfaction. Apart from
customer satisfaction, purchase intention is another important indicator of customer
retention. Previous studies have suggested that an effective recovery strategy will
make purchase intention increase or at least stabilize, whereas a poor recovery
strategy will reduce customer’s purchase intention from the same retailer in the future
(Chen, 2012; LaBarbera and Mazursky, 1983; Maxham, 2001). Therefore, customer
satisfaction and purchase intention were used in this research as the dependent
variables in this study of effects of recovery strategies.
However, when service failure occurs, effective service recovery strategy can
help recover the relationship between customers and e-retailers, increase customer
satisfaction and purchase intention (Bejou & Palmer, 1998; Blodgett, Hill, & Tax,
1997; Chang & Wang, 2012; Goodwin & Ross, 1992; Maxham, 2001; McCollough,
Berry, & Yadav, 2000; Mohr & Bitner, 1995). High-satisfaction recovery strategies
Wang, 2012). Low-satisfaction recovery strategies are considered ineffective
comparing with high and moderate-satisfaction recovery strategies (Forbes et al.,
2005; Kelley et al., 1994; Kuo et al., 2011; Maxham, 2001). Nevertheless, the
low-satisfaction recovery strategies, such as Rapid Response, Apology and
Replacement, cost much less than high- and moderate-satisfaction recovery strategies
(e.g., refund, discount). For start-up e-retailers and e-retailers that have financial
troubles, it is the comparatively costless service recovery strategies that they prefer.
Therefore, this research examines the effectiveness of the combinations of
low-satisfaction yet costless recovery strategies (e.g., Apology, Replacement and
Rapid Response) on customer satisfaction and purchase intention after product defect
failure.
Due to the massive transaction volume of clothing products according to
McKinsey study (2011) and its ongoing growth (Hung, 2012), the chance of making
service failures in apparel category would be much higher than that of other
categories.
strategies combination perform versus other strategy combinations in terms of
customer satisfaction and purchase intention following an apparel product defect in an
online context.
Since this study is in an online context, the author collects data through online
surveys. Respondents are randomly divided into five treatment groups (one control
group and four different recovery groups). In order to examine the perceived
differences across five treatment groups after apparel product defect, independent
sample t-tests are used to prove the significance of mean differences.
The rest of the paper is structured as follows. In section 2, the author reviews
relevant literature, elaborates the research gap as well as the research question, and
then lists the key hypotheses. In Section 3, data are collected and general
methodology is outlined. Section 4 contains the results of the analysis and tests the
hypotheses. Section 5 discusses the thesis and a summary of managerial implications
is revealed, along with suggestions for future research. The summary of the study is in
2. Literature review
2.1 Customer complaints and service failure
Due to the web-based attribute of online shopping, when shopping online, there
is no necessity for “face-to-face” interaction between customers and products (Wu,
2012). For this reason, online shoppers, compared to offline shoppers, inherently and
inevitably shop with more uncertainty (Wu, 2012). With this inherent feature,
online-shopping consumers have relatively more possibilities to encounter incidents
and service problems that can lead to consumer complaint. Consumer complaint is an
action coming from customer dissatisfaction (Rogers et al., 1992) and is an outcome
of service failure. Service failure refers to the incident that makes customer
dissatisfied during a service delivery (Maxham, 2001). Since service failure has
already penetrated to every service business and it is inevitable yet important (Chung
and Hoffman, 1998), studies of service failure are numerous (Andreassen 1999;
Chung and Hoffman, 1998; Harris et al. 2006; Holloway and Beatty 2003; Kuo et al.
2011; Maxham, 2001; Wu 2012). Study suggests that negative effects caused by the
Therefore, how to address service failure properly is an issue that needs to be
concerned by e-retailers and organizations. Compared to traditional shopping, online
shopping is more likely to generate problems during service delivery because of its
lack of “face-to-face” interaction between customers and products (Harris et al. 2006;
Wu, 2012). Therefore, scholars have begun to study service failure and recovery
strategy in order to better understand online shopping behavior (Andreassen 1999;
Holloway and Beatty 2003; Kuo et al. 2011; Wu 2012).
Kuo et al. (2011) conduct survey in their study and their respondents report 867
incidents of service failures in terms of online shopping. They classify the 867
incidents into three main groups, namely service delivery system failures, buyer needs
and requests and unprompted and unsolicited seller actions. Service system failures
refer to the failures that make customers dissatisfied during the delivery of essential
services (Kuo et al., 2011) and the number of incidents from this group accounts for
76.3% of the total number of incidents. The failure of fulfilling buyer needs and
requests happens due to customer dissatisfaction when service provider could not
failure accounts for 13.2% of the total number of incidents. The failure unprompted
and unsolicited seller actions refers to the customer dissatisfaction caused by the
service attitude and service behavior they encounter (Kuo et al., 2011) and is not
associated with the other two groups. The number of incidents from this group
accounts for the smallest portion, which is 0.5%. In the biggest group (service
delivery system failures, 76.3%), Kuo et al. (2011) classify the failures into nine
different categories, which are packaging problem, slow or unavailable service,
product defect, out of stock, bad information, alterations and repairs, hold disasters,
pricing failure, policy failure. Different service failures may differ in frequency and
severity (Kelley and Davis, 1994). The specified findings of Kuo et al. (2011) are
shown in Table 1 (on Page 68).
From their findings, it is easy to recognize that slow or unavailable service,
product defect, and packaging problem are the three major failure categories, which
make up 58.4% of all online shopping service incidents. According to another finding
from Kuo et al. (2011), among the three failures, the failure product defect has the
problem make up 7.1% and the failure slow or unavailable service make up 6.9% (see
Table - Severity of service failure of the service delivery system failures group on Page
68).
In fact, back in 1994, Kelley et al. (1994) classified 661 critical incidents into
fifteen different types of retail failures in general retailing context. In their findings,
the failure product defect, which accounts for 33% of total service failure incidents
(220 out of 661 cases), is the most frequently occurred retail service failure. Sixteen
years later, the study from Kuo et al. (2011) proves that product defect in the online
shopping context still deserves attention of scholars. In both contexts, product defect
is considered a frequently encountered service failure by customers and should be
paid attention to by scholars, e-retailers and organizations. Therefore, product defect
is included in this research as an independent variable representing service failure.
2.2 Service recovery strategy
When a consumer experiences a loss due to a service failure, the service provider
makes up for the mistake by a service recovery strategy (Smith et al. 1999; Grewal et
recover the relationship between customers and e-retailers and increase customers’
repurchase intention (Bejou & Palmer, 1998; Blodgett, Hill, & Tax, 1997; Chang &
Wang, 2012; Goodwin & Ross, 1992; Maxham, 2001; McCollough, Berry, & Yadav,
2000; Mohr & Bitner, 1995) and lead to “behavioral outcomes” such as customer
satisfaction, positive word-of-mouth (Maxham, 2001) and loyalty (Kim et al., 2010).
Among these scholars, Maxham proposes that high-satisfaction service recovery
strategies may increase customer satisfaction and purchase intention while
low-satisfaction service recovery strategies may influence customers to discontinue
service with such e-retailers. With low-satisfaction recovery strategies, e-retailers may
lose revenue because customers are reluctant to return after the service failure. Since
it costs considerably more to attract a new customer than to keep a current one (Hart
et al., 1990), it is essential to address service failures by effective service recovery
strategies (Blodgett et al., 1993, 1995; Hart et al., 1990; Hoffman et al., 1995;
Snellman & Vihtkari, 2003; Tax et al., 1998).
In previous studies, service recovery strategies such as correction, correction plus,
strategies; strategies such as apology and nothing are ranked as low-satisfaction
recovery strategies (Forbes et al., 2005; Kuo et al., 2011). Correction refers to simple
and uncomplicated fixing of the mistakes such as by increasing repair speed without
extra compensation while correction plus refers to not only make up for the mistake,
but also compensate the customer in some additional way. Store credit is a credit
applied to future purchase(s). In fact, given an apparel product defect context, the way
to implement correction strategy is to replace the product, which is the same as
implementing strategy replacement (replace the defective product with a new one).
However, when being tested as an individual recovery strategy, replacement is not
favored by customers and is considered a low-satisfaction recovery strategy (Kuo et al.,
2011). The reason could be the diverse situations and solutions which correction
includes (such as providing reasonable explanations, exchanging wrong information,
and replacing misdelivered products), which satisfy customers in many other ways.
Comparing with correction, replacement is only about replacing a faulty or out of stock
product with a good one whose price equaled that of the originally ordered one. In the
Moreover, the level of customer satisfaction and purchase intention for this recovery
strategy is 3.4 and 3.5, respectively, in a 10-point scale (refer to Table – categories of
online auction service recovery strategies, Page 71).
According to Chang and Wang (2012), compensation, response speed, apology
and contact channel are four key service recovery strategies most commonly adopted
by companies. Compensation refers to refund, discount or other monetary
reimbursement the service provider makes after a service failure; response speed is
the speed with which e-retailers response to service failures in order to resolve the
mistake; contact channel is the channel by which e-retailers communicate with
customers (Chang and Wang, 2012). In terms of cost, response speed, apology and
contact channel are costless recovery strategies compared to compensation. Response
speed has three time levels, including: within 24 hours, between 1 and 3 days, and
more than 3 days. The study of Chang and Wang demonstrates that customers thank a
response within 24 hours, and consider the other two as negative responses. This
result indicates that customers are expecting a response within 24 hours after
compensation, response speed or apology in the perception of customers (see Table –
Relative importance and Part-worth utilities of strategies, Page 71).
It is certain that high-satisfaction recovery strategies have better performance in
increasing the level of customer satisfaction and purchase intention than
low-satisfaction ones (Maxham, 2001), nevertheless, these high service recovery
strategies are monetary reimbursement in general and can be seen as a loss in revenue
when service failure happens frequently. Since e-retailers cannot control the
occurrence of service failure, especially when it is in an online shopping situation,
addressing service failure with monetary reimbursement becomes fatal for start-up
e-retailers and e-retailers that have financial troubles. However, even for other
companies and organizations, it would be economical in the long run to use costless
service recovery strategies. From previous studies, the costless or low-cost service
recovery strategies are mostly rated as low-satisfaction recovery strategies (Forbes et
al., 2005; Kelley et al., 1994; Kuo et al., 2011). Nevertheless, until now we know
nothing about the effectiveness of the combination of these costless strategies such as
costless to e-retailers. If this new strategy would work in a way similar to other
monetary compensational strategies, it would be beneficial to those e-retailers who
operate in an economical way yet want to compensate when service failure occurs.
Therefore in this study, the author focused on the effect of the low-satisfaction service
recovery strategy combinations in order to propose another option to those start-up
e-retailers and small companies that have financial troubles, and those companies that
want to lower their costs in the long term.
2.3 Clothing products
With the development of IT technology, items that are thought to be only salable
in brick-and-mortar environment, such as apparel and jewelry, are gaining more sales
globally (Kim and Kim, 2004). More customers have joined the community of online
shopping and experienced the convenient lifestyle brought to them by Internet
technology. At present, China is undoubtedly the largest flourishing marketplace in
the globe. Taobao.com, the biggest online mall in China, has claimed to be the largest
e-commerce website in the world (Chee Sing, 2010). As stated in Internet Business
49% of the total transaction volume in the Chinese online retail market. According to
a 2011 McKinsey study, clothing products consist of the biggest portion for
e-commerce in China and 36 percent of online spending is for clothing products.
From this viewpoint, apparel is a representative sample of online products.
Transaction volume in apparel, compared to other categories, is large and continues to
grow fast (Hung, 2012). Therefore, the chance of encountering service failures is
higher than that of other categories. Moreover, the typical acquisition cost per
customer in apparel is the lowest and the profit per customer is the largest in apparel
among the main categories (Reichheld and Schefter, 2000). The recent integration of
apparel manufacturers into direct online selling, and the ongoing trend of
brick-and-mortar retailers going into the online channel, has accelerated the clothing
surge (Kim & Kim, 2004; Schaeffer, 2000). Therefore, the author conducted the
research in a clothing product defect context.
2.4 Gender, age, education level, and past online shopping frequency
Previous studies reveal that female customers potentially have lower levels of
Harris, 2003; Van Slyke et al., 2002). In particular, male customers exhibit higher
levels of trust and self-efficacy towards online shopping than female customers.
In terms of age, younger customers only measure the perceived benefits of
shopping and report more hedonic and utilitarian benefits of online shopping than
older customers (Dholakia and Uusitalo, 2002). However, older customers are more
likely to purchase online than younger customers, even though they have less positive
attitudes towards online shopping (Donthu and Garcia, 1999). It is interesting to note
that the positive correlation of age versus the likelihood of shopping online found by
Donthu and Garcia: a research conducted by Joines et al. (2003) revealed that the
likelihood of online shopping by younger customers is higher than that of older ones.
The education level of subjects was included in this research following the study
of Lightner (2003). This analysis indicates the significant and positive correlation
between education and satisfaction. The result also indicates that preferences for
security, sensory impact, price, information quantity, comparison-shopping and other
site characteristics are dependent on education. One can conclude that as education
In addition, past online shopping frequency is found as significant predictor
regarding customer potential online shopping intention (Jain and Jain, 2011). In this
study, the probability of online shopping is significantly higher among customers who
more frequently use e-share trading.
The results suggested that gender, age, education level, and past online shopping
frequency are four indicators that are highly correlated with online shopping behavior.
Therefore, they were included as control variables in this research.
2.5 Customer satisfaction and purchase intention
Customer satisfaction is defined as the favorable and cumulative evaluation a
person subjectively derived from consuming a product (Cronin and Taylor 1994;
Maxham, 2001; Westbrook, 1980). To some extent customer satisfaction represents
the outcome of comparing anticipated rewards and costs after a purchase (Bolton and
Drew, 1991; Cronin and Taylor 1994; Maxham, 2001). Thus customer satisfaction is
a valid indicator of customer’s perception of the quality of certain service. Service
failure would influence the evaluation of the purchase outcome and reflect the
positive impact on customer satisfaction (Chen, 2012); that is to say, positive online
shopping experience can increase the level of customer satisfaction from customer’s
perception. Therefore, when evaluating the quality of service from a service provider,
customer satisfaction can be used as a global judgment. As such, customer
satisfaction was used in this paper to examine customer’s perception of online service
after service failure and service recovery strategy.
It has been supported that customer satisfaction has a positive impact on
purchase intention (Chen, 2012; LaBarbera and Mazursky, 1983; Maxham, 2001),
which means that an increase in the level of customer satisfaction will lead to an
increase in the level of purchase intention, whereas a decrease in the level of customer
satisfaction will lead to a decrease in the level of purchase intention. Since obtaining a
new customer costs five times higher than retaining a current customer (Hart et al.,
1990), purchase intention of current customers is thus very important for e-retailers.
Scholars have proved that retailers are able to maintain customer retention through
recovering service failures in “a fair manner” (Blodgett et al., 1993; Maxham, 2001).
increase or at least stabilize, whereas a poor recovery strategy will reduce customer’s
purchase intention from the same retailer in the future (Maxham, 2001). Therefore,
purchase intention is considered a parameter of online service and was included in
this research to test the effectiveness of low-satisfaction service recovery strategy
combinations.
In summary, according to the findings above, service recovery strategies may
have a vital influence on the level of customer satisfaction and purchase intention
from customers’ perceptions. Therefore, customer satisfaction and purchase intention
were used in this research to test the effectiveness of low-satisfaction recovery
strategy combinations following apparel defects in online context.
2.6 Gap and Research question
Most papers referred to in this thesis were focused on either single service
recovery strategy, such as the effect of correction, refund, discount individually
(Forbes et al., 2005; Kelley et al., 1994; Kuo et al., 2011), or focused on the
combinations of satisfaction service recovery strategies at high, medium and low
to the customer, showing empathy, apologizing, and resolving problems (Chang and
Wang, 2012; Maxham, 2001). But for start-up e-retailers and small companies that
have financial troubles, and companies that want to lower costs in the long term,
costless recovery strategies are preferable. Nevertheless, customers usually consider
those costless strategies, such as Rapid Response, Apology and Replacement,
unfavorable according to the results of previous studies (Chang and Wang, 2012;
Forbes et al., 2005; Kelley et al., 1994; Kuo et al., 2011). Therefore, the
low-satisfaction recovery strategy combinations shown in this research were the
combinations of those costless strategies with economical (to e-retailers) yet valuable
(to customers) attributes, such as apology, rapid response, and replacement. As
reasoned above, due to the high severity and frequency of product defect and the
massive volume of clothing products, the author selected apparel defect and the whole
research was set in an online context. The research question is how well the
combinations of certain low-satisfaction service recovery strategies perform versus
other recovery strategy combinations in terms of customer satisfaction and purchase
2.7 Hypotheses
According to the literature reviewed, customer satisfaction represents the
outcome of a purchase, whereby consumers compares anticipated consequences with
rewards and costs (Bolton and Drew, 1991; Maxham, 2001; Oliver, 1980; Yi, 1990).
A service failure would influence the purchase outcome and reflect in the evaluation
of the purchase through customer satisfaction. When a service failure occurs, the
service recovery strategy may influence the relationship between customers and
e-retailers. A service failure would influence not only customer satisfaction but also
purchase intention. Previous studies indicate that customer satisfaction has a positive
impact on purchase intention (Chen, 2012; LaBarbera and Mazursky, 1983; Maxham,
2001). An increase in the level of customer satisfaction will lead to an increase in the
level of purchase intention. In the research of Kuo et al., the satisfaction rate after
apology (3.7) is higher than that after replacement (3.4), while the purchase intention
rate after apology (3.4) is lower than that after replacement (3.5) (refer to Table –
categories of online auction service recovery strategies, Page 71), therefore the author
Rapid Response and Apology than after the combination of Rapid Response and
Replacement.
H1b. Level of purchase intention will be higher after the combination of Rapid
Response and Replacement than after the combination of Rapid Response and
Apology.
According to the study of Chang and Wang (2012), response speedis considered
more important than apology (refer to Table – Relative importance and Part-worth
utilities of attributes, Page 72). These two strategies were included separately in
recovery strategy C and D, in which Replacement was combined respectively with
Rapid Response and Apology. Given that the perceived importance of recovery
strategy is positively related to customer satisfaction and purchase intention (Maxham,
2001), the author hypothesizes:
H2a: level of customer satisfaction will be higher after the combination of Rapid
Response and Replacement than after the combination of Apology and Replacement.
H2b: level of purchase intention will be higher after the combination of Rapid
Comparing with Rapid Response, Replacement as a recovery strategy can be
seen as a way of compensation. According to the findings of Chang and Wang,
compensation (38.25%) is perceived more important than response speed (26.02%)
(Refer to Table – Relative importance and Part-worth utilities of strategies, Page 72).
It can be concluded that customers probably consider Replacement in a recovery
strategy more important than Rapid Response. Thus the author hypothesizes:
H3a: level of customer satisfaction will be higher after the combination of
Apology and Replacement than after the combination of Rapid Response and Apology.
H3b: level of purchase intention will be higher after the combination of Apology
and Replacement than after the combination of Rapid Response and Apology.
The study of Chang and Wang (2012) suggests that customers perceive different
levels of recovery differently. Along with the increase of the level of service recovery,
post-failure levels of customer satisfaction and purchase intention significantly
increase as well (Maxham, 2001). As the number of recovery strategies increases, the
perceived quality of recovery strategy will increase correspondingly. Since a recovery
perceived as a multiplied-recovery strategy comparing with a recovery with only two
combinations (such as recovery strategy B: the combination of Rapid Response and
Apology; recovery strategy C: the combination of Rapid Response and Replacement;
recovery strategy D: the combination of Apology and Replacement), thus the author
hypothesizes:
H4a: level of customer satisfaction after the combination of Replacement, Rapid
Response and Apology is higher than that of the combinations of Rapid Response and
Apology.
H4b: level of purchase intention after the combination of Replacement, Rapid
Response and Apology is higher than that of the combinations of Rapid Response and
Apology.
H4c: level of customer satisfaction after the combination of Replacement, Rapid
Response and Apology is higher than that of the combinations of Rapid Response and
Replacement.
H4d: level of purchase intention after the combination of Replacement, Rapid
Replacement.
H4e: level of customer satisfaction after the combination of Replacement, Rapid
Response and Apology is higher than that of the combinations of Apology and
Replacement.
H4f: level of purchase intention after the combination of Replacement, Rapid
Response and Apology is higher than that of the combinations of Apology and
Replacement.
3. Data and methodology
In the following chapter the research method and design are explained.
3.1 Research design
As the use of scenarios to examine service encounters enabled the author to
gather all the required responses and permitted examination of the most concerned
variables (Weiner, 2000), data in this research were gathered using a scenario-based
experimental online survey through a survey website.
service.
The author constructs specified scenario in terms of product defect to manipulate
dependent variables (customer satisfaction and purchase intention) under different
combinations of low-satisfaction recovery strategies. The specified service recovery
strategies in this study were the different combinations of Replacement, Apology and
Rapid Response (recovery strategy B: the combination of Rapid Response and
Apology; recovery strategy C: the combination of Rapid Response and Replacement;
recovery strategy D: the combination of Apology and Replacement; recovery strategy
E: the combination of Apology, Rapid Response and Replacement). The purpose of
the study is to examine the effectiveness of the combinations of low-satisfaction
recovery strategies by comparing the result of each dependent variable.
To ensure that the recovery strategies are low-satisfaction, the author developed
an overall scenario. In this scenario, several single service recovery strategies from
different satisfaction levels were listed (i.e., discount, correction, replacement,
apology, refund, store credit, and rapid response). In total 20 respondents ranked
low, 2 = moderate, 3 = high). As a result, all 20 respondents precisely perceived
Apology, Replacement and Rapid Response as low-satisfaction service recovery
strategies. These pre-test scenarios can be viewed in Appendix A on Page 62.
In this research, there were five treatment groups in total (see Table 2, Page 72).
Five different scenarios were randomly sent to respondents (introductory scenario: no
recovery; manipulation scenario 1: the combination of Rapid Response and Apology;
manipulation scenario 2: the combination of Rapid Response and Replacement;
manipulation scenario 3: the combination of Apology and Replacement; manipulation
scenario 4: the combination of Apology, Rapid Response and Replacement).
This experiment asked respondents to read one of the five hypothetical scripts
randomly and use the information provided in the script to respond to the question,
which would reveal their perception of customer satisfaction and purchase intention.
This is a frequently used “role-playing” approach in social science research
(Carlsmith et al., 1976) and can have a high degree of realism (Brown, 1962; Kelman,
1968; Schultz, 1969).
other scenarios, respondents imagine they are facing the low-satisfaction service
recovery strategy combination mentioned in that particular scenario, and then answer
questions in terms of their perception of customer satisfaction and purchase intention.
These five scenarios can be viewed at Appendix B, Page 63.
3.2 Sample
Considering the number of treatment groups, an experimental online survey was
distributed worldwide through a survey website. It is reasonable to choose online
survey as the method to collect data because the subject of this research was online
shopping. Those who are capable of using the Internet possess the basic knowledge of
Internet technology and are the potential customers who can shop online. In the end,
the author collected a total of 266 valid responses by convenient sampling. (E.g.
sending it to a friend after filling the questionnaire)
The five groups in this study were approximately same in size, ranging from 50
to 59， which are large enough to proceed with the experimental research. Group A
(control group) has 50 respondents; group B (Rapid Response and Apology) has 50
group D (Apology and Replacement) has 50 respondents, and group E (Rapid
Response, Apology and Replacement) has 59 respondents.
3.3 Independent variables
In a service recovery process, high-satisfaction service recovery strategies can
increase the level of customer satisfaction (Maxham, 2001), restore customer
purchase intention (Bejou & Palmer, 1998; Blodgett, Hill, & Tax, 1997; Chang &
Wang, 2012; Goodwin & Ross, 1992; Maxham, 2001; McCollough, Berry, & Yadav,
2000; Mohr & Bitner, 1995), whereas low-satisfaction recovery strategy can further
annoy the already upset customers. However, high-satisfaction service recovery
strategies are mostly monetary reimbursement or tangible benefits from customers’
perspectives (Chang and Wang, 2012) while low-satisfaction recovery strategies
comparatively costless from e-retailers’ perspectives. Since service failure in online
shopping inevitably happens more frequently than in offline shopping, due to its high
level of uncertainty, it would be beneficial for start-up e-retailers to recover service
failure with costless recovery strategies.
discount, store credit, and refund are considered as high-satisfaction recovery
strategies; strategies such as apology and nothing are ranked as low-satisfaction
recovery strategies (Forbes et al., 2005; Kuo et al., 2011). Although replacement is part
of the correction strategy, customers perceive it as a low-satisfaction recovery strategy
in online auction environment in the study of Kuo et al. possibly due to its unvarying
solving method. Besides, according to the study of Chang and Wang (2012),
customers also consider compensation, response speed, apology and contact channel
to be key service recovery strategies. Among these strategies, response speed,
apology and contact channel are comparatively costless recovery strategies; response
speed is perceived much more important than contact channel. Among the three levels
of response speed, response within 24 hours is most favored by customers while
response between 1 and 3 days and more than 3 days are perceived as negative
responses.
Therefore, in order to find economical yet effective recovery strategy
combinations, the author selected Rapid Response, Apology and Replacement in the
recovery strategy combinations. Together with one control group followed with no
recovery strategy, Table 3 (on Page 73) shows the 5 recover strategy combinations in
this research.
3.4 Dependent variables
As discussed earlier, customer satisfaction is the outcome from comparing
between anticipated rewards and costs after a purchase (Bolton and Drew, 1991;
Cronin and Taylor 1994; Maxham, 2001), and thus a valid indicator of customer’s
perception of certain service. A two-item customer satisfaction measure was
developed for this study in an online shopping situation (Fishbein and Ajzen, 1975;
Maxham, 2001) and was rated on a 7-point Likert scale ranging from not at all
satisfied (1) to very satisfied (7).
Since obtaining a new customer costs five times higher than retaining a current
customer (Hart et al., 1990), purchase intention of current customers is thus a very
important indicator for e-retailers. Purchase intention with the e-shop was measured
using a two-item scale derived from previous research (e.g., Bitner, 1990; Cronin and
Likert scale ranging from not strongly disagree (1) to strongly agree (7).
All items in this research can be viewed in Appendix C, Page 65.
Once respondents read the introductory scenario, respondents from different
manipulation groups were led to manipulation scenarios. After the scenario was read,
all respondents were instructed to answer the four items. The purpose of the
measurement was to examine the extent to which customers perceive the
low-satisfaction service recovery strategy combinations and response to the service
recovery effort through customer satisfaction and purchase intention.
In order to examine whether between-strategy difference in terms of customer
satisfaction and purchase intention existed, the means of each strategy were compared.
The author chose independent sample t-tests over one-way ANOVAs and mixed
design ANOVAs to test hypotheses because the measurement of the dependent
variables was only repeated once in this research whereas mixed design ANOVAs
requires the variables being repeated more than once, and because the hypotheses
only involved testing of one strategy < another strategy thus one-tailed testing was
conclusion, a one-tailed independent sample t-tests was carried out for the change of
both dependent variables.
3.5 Control variables
The study used gender, age, level of education, experience of using the Internet,
years of online shopping, and frequency of online shopping as control variables.
Gender was included as a control variable based on the findings that female
customers potentially have lower levels of trust and self-efficacy towards online
shopping than male customers (Cho and Jialin, 2008; Rodgers and Harris, 2003; Van
Slyke et al., 2002). Age was used as another control variable in this research because
previous findings reveal that older customers are more likely to purchase online than
younger customers, even though they have less positive attitudes towards online
shopping (Donthu and Garcia, 1999). This might lead to different perception of the
quality of different service recovery strategy. In terms of education level, Lightner
(2003) indicates the significant and positive correlation between education and
satisfaction. In addition, in the study of Jain and Jain (2011), the probability of online
trading.
As stated above, gender, age, education level and past online shopping frequency
- as four indicators highly correlated with online shopping behavior - were included as
control variables in this research. Data on the control variables were collected through
an experiment combined with a survey questionnaire (refer to Appendix D on Page 66,
Demographic questions, question 1, 2, 4, and 8). The operationalization of the control
variables is outlined in Table 4, Page 73.
3.6 Experimental procedure
Prior to exposure to experimental conditions, the experimental survey supplied
subjects with an introductory scenario. This introductory scenario briefly explained
the hypothetical overall service history with an online shop and the hypothetical
service failure, respondents were facing. The introductory scenario (shown in
Appendix A, Page 62) was offered to each of the five groups in order to build
“baseline” levels for the dependent variables that were equal across all five groups.
All respondents were asked to read the “introductory scenario” first. For the control
respondents with four service recovery strategy combinations (group B: the
combination of Rapid Response and Apology; group C: Rapid Response and
Replacement; group D: the combination of Apology and Replacement; group E: the
combination of Apology, Rapid Response and Replacement). After reading the
scenario, respondents answered the questions to evaluate their perception of customer
satisfaction and purchase intention.
4. Results
The following section reports the results of this research. To begin with, the
reliability of the variables was reported. The Cronbach’s alpha coefficient was
analyzed for the dependent variable: customer satisfaction and purchase intention.
Secondly, descriptive statistics were given to provide an overview of the data. A
correlation analysis was later carried out and the most significant correlations were
reported. In order to make sure the normal distribution of the scores on the dependent
variables in each scenario were satisfactory, the results of the normality analysis were
sample t-tests. Since independent sample t-tests do not allow testing with control
variables, a regression analysis was carried out at the end to validate the results of the
independent sample t-tests and reveal possible effects of the control variables in this
research.
4.1 Reliability analysis
For the purpose of the research, two dependent variables were measured using
multiple questions.
The dependent variables customer satisfaction and purchase intention were
respectively captured using 2 questions and tested under the manipulation of no
recovery, the combination of Rapid Response and Apology, the combination of Rapid
Response and Replacement, the combination of Apology and Replacement, and the
combination of Apology, Rapid Response and Replacement (refer to Appendix C and
Appendix D, Page 65 - 66). In all five scenarios, the values of the Cronbach’s alpha
coefficient were between 0.704 and 0.975 (see Table 5, Page 74), which were all
above 0.7. The result indicates that the dependent variables had good internal
negative value in the Inter-Item Correlation Matrix, which means that all items were
always measuring the same underlying characteristic.
To summarize, all 2-item dependent variables had good internal validity and can
be recoded into individual variables.
4.2 Descriptive statistics
For the purpose of this research, ten categorical variables were used: independent
variables – product defect, strategy A (no recovery), strategy B (Rapid Response and
Apology), strategy C (Rapid Response and Replacement), strategy D (Apology and
Replacement), and strategy E (Apology, Rapid Response and Replacement), and the
control variables – gender, age, education level, and past online shopping frequency.
Although age and past online shopping frequency are not genuine categorical
variables, for the purpose of this analysis, they were classified as categorical variables.
For range options used in each variable, a median number is used as the categorical
option (less than 20 = 15, between 20 and 30 = 25, between 30 and 40 = 35, between
A description of all categorical variables is outlined in Table 6 – Frequencies for
Categorical Variables, Page 75.
From a total of 266 respondents, 266 respondents read the scenario of certain
product defect. 50 people received strategy A, containing no recovery. Another 50
people received strategy B (Rapid Response and Apology). 57 people received
strategy C (Rapid Response and Replacement). 50 respondents encountered strategy D
(Apology and Replacement) and another 59 respondents encountered strategy E
(Apology, Rapid Response and Replacement).
Of these 266 respondents, 163 (61.3%) were female and 103 (38.7%) were male.
The majority of respondents were aged 20 to 40 (74.1%). Respondents below 20 years
old accounted for 6%, respondents aged from 20 to 30 accounted for 39.1%,
respondents aged from 30 to 40 accounted for 35%, respondents aged 40 to 50
accounted for 11.3%, and respondents above 50 years old accounted for 12.4%.
In terms of education level, 126 (47.4%) had Bachelor degree, 59 (22.2%)
College degree, 52 (19.5%) had Master degree, 20 (7.5%) had high school
backgrounds. Most respondents in this research were highly educated.
Regarding past shopping online frequency, 99 respondents (37.2%) shop online 1
to 10 times a year, 66 respondents (24.8%) shop online once a month, 48 respondents
(18%) shop online more than once a week, 40 respondents (15%) shop once a week, 8
(3%) never shop online, and 5 (1.9%) shop online less than once a year.
Apart from the categorical variables, two continuous variables were used,
namely the dependent variables customer satisfaction and purchase intention. The
author used seven-point Likert-type scales to measure all the items; therefore
descriptive analysis was used for summary statistics, such as mean, median and
standard deviation (Pallant, 2010).
The result of descriptive statistics is shown at Table - Descriptive Statistics for
Continuous Variables.
Table - Descriptive Statistics for Continuous Variables
Variable Level N Min Max Mean SD
Customer satisfaction
No recovery 50 1 7 3.87 1.51
Rapid Response and Apology 50 1 7 4.78 1.20 Rapid Response and Replacement 57 2 7 5.09 1.29
Apology and Replacement 50 1 7 5.00 1.32
Apology, Rapid Response and Replacement
Purchase intention
No recovery 50 1 7 4.52 1.66
Rapid Response and Apology 50 1 7 4.91 1.34 Rapid Response and Replacement 57 2 7 5.22 1.18
Apology and Replacement 50 1 7 5.32 1.07
Apology, Rapid Response and Replacement
59 1 7 5.03 1.70
4.3 Correlation analysis
Table 7 (on Page 77) outlines the results from the bivariate correlation analysis
of the variable. The most significant correlations have been marked with an asterisk
(*). The relationships were investigated using Pearson product-moment correlation
coefficient.
For the purpose of this research, age and past online shopping frequency, two of
the categorical variables, were transformed into continuous variables by taking the
average value of each of the levels of the variables. Since dummy variable approach
has advantage in terms of the consistency, flexibility and transparency of variable
parameterization, the rest of the categorical variables, namely gender and education
level, were transformed into dummy variables by assigning 1 and 0 for each of their
College’s Level (EL-C), Education Level – Bachelor’s Level (EL-B), Education Level
– Master’s Level (EL-M), Education Level – Doctorate’s Level (EL-D), Education
Level – Others (EL-O), Past online shopping frequency (FQC). Other variables in 5
strategy groups were: customer satisfaction after strategy A (SAT-A), customer
satisfaction after strategy B (SAT-B), customer satisfaction after strategy C (SAT-C),
customer satisfaction after strategy D (SAT-D), customer satisfaction after strategy E
(SAT-E), purchase intention after strategy A (INT-A), purchase intention after strategy
B (INT-B), purchase intention after strategy C (INT-C), purchase intention after
strategy D (INT-D), and purchase intention after strategy E (INT-E).
In the following sections, the most important correlations with significant
coefficients of 0.05 or less are discussed. The significant correlations between
dependent variable and the control variables are outlined.
Regarding to correlations between the dependent variable and the control
variables, customer satisfaction after strategy B (Rapid Response and Apology) was
negatively correlated with High school’s level of Education (Pearson correlation
Bachelor’s level of education (Pearson correlation coefficient = 0.294; N = 50; Sig. =
0.038). Dependent variable purchase intention after strategy E (Apology, Rapid
Response and Replacement) was negatively correlated with College’s level of
Education (Pearson correlation coefficient = -0.390; N = 59; Sig. = 0.002). In
addition, customer satisfaction, after strategy B (Rapid Response and Apology), was
positively correlated with Past online shopping frequency (Pearson correlation
coefficient = 0.354; N = 50; Sig. = 0.012). When followed by strategy C (Rapid
Response and Replacement), customer satisfaction was negatively correlated with the
variable past online shopping frequency (Pearson correlation coefficient = -0.268; N
= 57; Sig. = 0.044).
The full set of correlations of variables of five groups in terms of customer
satisfaction and purchase intention is presented in Table 7 (refer to Page 77).
4.4 Normality analysis
A normality test was conducted to assess whether the dependent variables
In terms of customer satisfaction following five strategies, the dependent
variable had mean values ranging from 3.87 to 5.00 and a 5% trimmed mean ranging
from 3.85 to 5.07 (refer to Table 8 – Normality analysis, Page 78). This indicated that
the more extreme scores did not have a substantial influence on the mean. Moreover,
the negative skewness and the positive kurtosis values in each of the five scenarios
suggested that the distribution of the scores of customer satisfaction were mostly
rather peaked, clustered with some extreme cases in the center of the graph (Table 8,
Page 78, and Graphs 1, 2, 3, 4, and 5, Page 82 - 86).
With respect to purchase intention following five strategies, this dependent
variable had mean values ranging from 4.52 to 5.32, and a 5% trimmed mean ranging
from 4.57 to 5.40 (refer to Table 8 – Normality analysis, Page 78), which indicated
that more extreme scores did not have a substantial influence on the mean.
Furthermore, the values of skewness and kurtosis in each of the five scenarios
indicated that the distributions were peaked and the extreme cases were clustered to
the right (Table 8, Page 78, and Graphs 6, 7, 8, 9, and 10, Page 87 - 91).
the distribution of scores, were outlined in Table 9 (refer to Page 102). Except
customer satisfaction with no recovery, the values of both dependent variables in
other scenarios were significant. Normal probability plots for each of the independent
variables were outlined in Graph 11 to 20 (refer to Page 92 – 101). There was a
reasonably straight line in all scenarios, suggesting a relatively normal distribution of
the scores on customer satisfaction and purchase intention.
4.5 Hypotheses testing
To check whether the between strategy difference was significant, one-tailed
independent sample t-tests were carried out primarily and hierarchical multiple
regression analysis was carried out to validate the result.
4.5.1 Independent sample t-tests
H1 to H4 all predict the between-strategy difference perceived by customers in
terms of customer satisfaction and purchase intention.
H1 predicts that customers perceive higher level in terms of customer
Response and Apology) rather than after strategy C (the combination of Rapid
Response and Replacement) when encountering service failure. The results of the
independent sample t-tests can be found in Table 10 (refer to Page 103). Table 10
revealed that the between-strategy difference between strategy B and strategy C was
not significant at the 0.05 level for both dependent variables. Therefore the results of
Table 10 cannot provide evidence for H1a and H1b.
H2 was to test that customers perceive higher level of customer satisfaction and
purchase intention after strategy C (the combination of Rapid Response and
Replacement) than after strategy D (the combination of Apology and Replacement).
According to the results of the independent sample t-tests in Table 11 (refer to Page
103), the between-strategy difference between strategy C and strategy D was not
significant at the 0.05 level in terms of customer satisfaction and purchase intention.
Therefore no evidence supported H2a and H2b.
In H3, it was predicted that customers perceive higher level of customer
satisfaction and purchase intention after strategy D (the combination of Apology and
However, as shown in Table 12 (refer to Page 103), no significant between-strategy
difference was found in terms of customer satisfaction (one-tailed p-value = 0.777)
and purchase intention (one-tailed p-value = 0.077). Therefore the results of Table 12
cannot support H3a and H3b.
The results of the independent sample t-tests between strategy E and other three
strategies (i.e. strategy B, strategy C, and strategy D) were shown in Table 13, Table
14, and Table 15 (refer to Page 104). H4 predicted that customers perceive higher
level of customer satisfaction and purchase intention after strategy E than after
strategy B, strategy C, or strategy D. In Table 13, the between-strategy difference
between strategy E and strategy B was not significant at the 0.05 level in terms of
customer satisfaction and purchase intention. The results in Table 14 showed that
there was significant difference between strategy E and strategy C in terms of
purchase intention, but no significant difference was found in customer satisfaction.
According to the Mean difference, purchase intention in strategy C had higher score
than that in strategy E, therefore supported the counterpart of H4d. Table 15 revealed
purchase intention was significant. However, no significant difference between these
two strategies was found at the 0.05 level in terms of customer satisfaction. From the
results of Mean difference, purchase intention in strategy D had higher score than that
in strategy E, therefore supported the counterpart of H4f. Overall, the results of
independent sample t-tests only provided evidence to testify H4d and H4f and the
counterpart of H4d and H4f was supported.
4.5.2 Regression analysis
In respect of customer satisfaction, in the hierarchical multiple regression
analysis, model 1 contained the control variables (age, gender, education level, and
past online shopping frequency), and model 2 contained the control variables and
independent variables (low-satisfaction recovery strategies). After the regression
analysis, no variable significantly contributed to explaining the variation of customer
satisfaction (refer to Table 16 in Page 105). The results of the regression analysis
validated that the results from independent sample t-tests do not support H1a, H2a,
H3a, H4a, H4c, and H4e.
analysis, model 1 contained the control variables (age, gender, education level, and
past online shopping frequency), and model 2 contained the control variables and
independent variables (low-satisfaction recovery strategies). The analysis revealed
that only the control variable gender (one-tailed p-value = 0.023) and Replacement
(one-tailed p-value = 0.005) significantly contributed to explaining the variance of
purchase intention (refer to Table 17 in Page 106). In the second run of regression
analysis that included gender in Model 1 and replacement in Model 2, the results in
Table 18 (refer to Page 106) revealed that even though gender and replacement
contributed to explaining the variance of purchase intention, the significant
contribution was small: only 1.5% to Model 1 and 3.5% to Model 2. The result is in
line with the result of independent sample t-tests for purchase intention, which
validated that the results from independent sample t-tests can provide evidence related
to H4d and H4f and support the counterpart of H4d and H4f, whereas validated that
no evidence can support H1b, H2b, H3b, and H4b in either side.
4.6 Limitations
First of all, the results from the study may be limited due to the convenient
nature of the samples. As described in the respondent profile, 91.7% of the
respondents were from Asia. Therefore the results found here may be biased and more
representative of Asian consumers than of other consumers.
Secondly, the service failure setting in this research was only apparel product
defect in online context. The results may be markedly different given another service
failure setting or in another product category.
Thirdly, external validity may be detracted due to the somewhat artificial nature
of the online survey setting of the study.
Despite these limitations, the research here contributes to the service recovery
literature by offering an experimental study, interesting results and some valuable
managerial implications.
5. Discussion
The previous chapter presented the results of testing the hypotheses. In this
limitations of this research were described in the following section. The
recommendations for future research were given in the final section of this chapter.
5.1 Discussion of results
The research question of this research was to examine how well the
combinations of certain low-satisfaction service recovery strategies perform versus
other recovery strategy combinations in terms of customer satisfaction and purchase
intention after product defect failure in online apparel context. The hypotheses
according to this research question was shown as follows:
To answer the research question, an overall research design that allowed making
between-strategy comparisons quantitatively tested all these hypotheses. In the
following part, the results were discussed and the research question was answered.
5.1.1 Differences between low-satisfaction recovery strategy combinations
According to the results from this research, it is testified that there were
significant differences between strategy E and strategy C, and between strategy E and
strategy D. However, this research proved the counterpart of corresponding
of Replacement, Rapid Response and Apology is higher than that of the combinations
of Rapid Response and Replacement, and that level of purchase intention after the
combination of Replacement, Rapid Response and Apology is higher than that of the
combinations of Apology and Replacement. Then why could a combination of two
recovery strategies have better performance in increasing purchase intention than a
combination of three recovery strategies? Previous study of Maxham (2001) reveals
that the post-failure level of purchase intention will significantly increase when the
level of service recovery increases. Apparently, comparing with two-recovery
strategies such as strategy B, strategy C and strategy D, strategy E offered customers
with additional compensation. But this increased recovery strategy did not necessarily
be perceived as an increase in service recovery level. When additional compensation
is offered after a service failure, the corresponding purchase intention is not
automatically enhanced (Kwon and Jang, 2012). A customer with high RL (i.e.,
shared interaction with a company) and a longer time horizon is less likely to need to
restore equity to the relationship after a service failure. Therefore, compensation
have a high RL. This might explain why two-recovery strategies received higher level
of purchase intention than three-recovery strategy.
According to the results, no significant difference in customer satisfaction was
found across all comparisons. Why didn’t customer satisfaction show similar
tendency as purchase intention did? Theoretically, greater customer satisfaction
should be linked to improved purchase intention (Mittal and Kamakura, 2001). But by
reviewing the literature, this link is significantly lowered when the satisfaction feeling
is uncertain (Chandrashekaran et al., 2007; Tudoran et al., 2012). Therefore the
explanation to that question could be that the service failure may have increased the
uncertainty in the attitude of customers towards this e-retailer.
5.1.2 Answers to research question
To sum up the above and answer the research question, current research had no
evidence to testify the effect of low-satisfaction recovery strategy combinations in
terms of customer satisfaction. Moreover, the strategy containing Rapid Response and
effective than the other recovery strategy combination of this research in increasing
purchase intention after apparel product defect in an online context.
5.2 Implications
From the study it can be concluded that, although it is perceived as a low-satisfaction
recovery strategy when implemented individually, replacement plays an important
role in service failure recovery process when combined with other recovery strategies
(i.e., apology and rapid response). Moreover, unlike individual low-satisfaction
recovery strategies that further upset customers (Maxham, 2002), these
low-satisfaction recovery strategy combinations do not decrease the level of customer
satisfaction and purchase intention (refer to Table - Descriptive Statistics for
Continuous Variables, Page 76). More specifically, these low-satisfaction recovery
strategy combinations can be beneficial in increasing customer purchase intention.
The results from this research are undoubtedly beneficial to e-retailers and
companies. Since customers are seeking substantial and tangible compensations for
the damage that service failure caused (Maxham, 2001), using high-satisfaction
that have financial troubles. From discussion it is revealed that, combination of rapid
response and replacement, and the combination of apology and replacement, can to
some extent achieve the purpose of increasing customer retention, and have almost no
extra monetary compensation for implementing recovery strategy. Therefore, although
high-satisfaction service recovery strategies are effective in increasing customer
purchase intention, there is still an alternative for start-up e-retailers and e-retailers
that looking for long-term cost-benefit to choose low-satisfaction recovery strategy
combinations studied in this research to win their customers back. Comparing with
high-satisfaction service recovery strategies, these two low-satisfaction recovery
strategy combinations are costless and easy to implement. When a service failure
occurs, an economical resolution is to offer replacement within 24 hours. If the
e-retailer fails in responding in time, try replacing the item with sincere apology.
Another implication the results bring is the proactive consideration for e-retailers
and managers to carry out employee training. Employee is the person who works
between customers and the manager. This nature of position gives employees the
managers. When there is nowhere to find the manager while the service failure occurs,
and employees are trained to use replacement with one of the other two recovery
strategies (i.e., apology and rapid response) to address different incidents, it is very
likely to successfully retain upset customers and recover the service failure.
However, this study didn’t succeed in finding significant difference in customer
satisfaction between recovery strategy combinations. Another translation for this
result is that when addressing apparel product defect, it may to better to use high or
moderate-satisfaction service recovery strategies since they may have strategic
benefits in making customers satisfied (Maxham, 2001).
6. Conclusion
6.1 SummaryIn previous literatures, numerous researchers have recognized the severity of
product defect as a service failure (Andreassen 1999; Holloway and Beatty 2003;
Kelley et al., 1994; Kuo et al., 2011; Wu 2012); many have classified service recovery
