16
http://jsr.sagepub.com/ Journal of Service Research http://jsr.sagepub.com/content/6/2/193 The online version of this article can be found at: DOI: 10.1177/1094670503257048 2003 6: 193 Journal of Service Research Tom DeWitt and Michael K. Brady Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure Published by: http://www.sagepublications.com On behalf of: Center for Excellence in Service, University of Maryland can be found at: Journal of Service Research Additional services and information for http://jsr.sagepub.com/cgi/alerts Email Alerts: http://jsr.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: http://jsr.sagepub.com/content/6/2/193.refs.html Citations: What is This? - Nov 1, 2003 Version of Record >> at Bartin Universitesi on November 14, 2014 jsr.sagepub.com Downloaded from at Bartin Universitesi on November 14, 2014 jsr.sagepub.com Downloaded from

Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

Embed Size (px)

Citation preview

Page 1: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

http://jsr.sagepub.com/Journal of Service Research

http://jsr.sagepub.com/content/6/2/193The online version of this article can be found at:

 DOI: 10.1177/1094670503257048

2003 6: 193Journal of Service ResearchTom DeWitt and Michael K. Brady

Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure  

Published by:

http://www.sagepublications.com

On behalf of: 

  Center for Excellence in Service, University of Maryland

can be found at:Journal of Service ResearchAdditional services and information for    

  http://jsr.sagepub.com/cgi/alertsEmail Alerts:

 

http://jsr.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://jsr.sagepub.com/content/6/2/193.refs.htmlCitations:  

What is This? 

- Nov 1, 2003Version of Record >>

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 2: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

Rethinking ServiceRecovery StrategiesThe Effect of Rapport on ConsumerResponses to Service Failure

Tom DeWittMichael K. BradyFlorida State University

Researchers and practitioners have invested heavily inidentifying effective complaint management strategies.However, most of the strategies identified to date occuronly after a service failure occurs. This article proposesthat antecedent states such as an ongoing rapport withservice employees can also provide service recovery bene-fits. Four independent studies test this approach and indi-cate that an existing rapport between the customer andservice provider results in increased postfailure customersatisfaction, increased repatronage intentions, and de-creased negative word of mouth. Yet results for complaintintentions suggest that rapport does not increase the pro-pensity for customers to complain about poor service.These results are discussed in terms of their implicationsfor research and practice.

Keywords: service recovery; rapport; customer rela-tions; services

It is now believed that retaining current customers maybe more profitable than obtaining new customers(Reichheld 1996). Accordingly, scholars and practitionersare focused on managing customer retention. This focus isevident in the prominence of customer relationship man-agement (CRM) strategies in practice and in the researchattention given to the study of customer asset management

(e.g., Berger et al. 2002; Hogan, Lemon, and Rust 2002),customer equity management (e.g., Hogan, Lemon, andRust 2002; Rust, Zeithaml, and Lemon 2001), and rela-tionship marketing (e.g., Lemon, White, and Winer 2002;Morgan and Hunt 1994). It is also evident in the attentiongiven to managing service failure and recovery (e.g.,Kelley and Davis 1994; Tax and Brown 1998). Becauseservice failure is difficult or impossible to eliminate, theobjective of complaint management is to lessen or elimi-nate any damage done and, ultimately, to retain a once-dissatisfied customer. Unfortunately, widespread successin this pursuit has been limited.

Keaveney’s (1995) study on switching behavior indi-cates that service failures and failed recoveries are a lead-ing cause of customer switching behavior in serviceorganizations. At least part of this problem stems from thefact that 70% to 95% of dissatisfied customers do notbother to complain to the service provider (Harari 1992).However, of the customers that do complain, more thanhalf of attempted recovery efforts only reinforce dissatis-faction (Hart, Heskett, and Sasser 1990). It is our conten-tion that this poor success rate is due in part towhentypicalservice recovery efforts are initiated. Most recommendedrecovery strategies are initiated only after the dissatisfyingevent has occurred. For example, a timely response, anapology, empathy, and recompense are indeed effective re-covery strategies, but they are reactive rather than

Journal of Service Research, Volume 6, No. 2, November 2003 193-207DOI: 10.1177/1094670503257048© 2003 Sage Publications

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 3: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

proactive strategies. It may be that reactive actions are ini-tiated too late to mitigate all of the perceived damage doneto the consumer. Carrying this logic further, there may berecovery benefits to developing a positive antecedent stateamong service consumers. Indeed, Lemon, White, andWiner’s (2002) study indicates that expected future useand anticipated regret, which are antecedent states in a ser-vice encounter, positively influence satisfaction and cus-tomer retention.

This study examines the recovery benefits of fosteringrapport between the service provider and customer. Al-though the literature has examined the personal interac-tions between customers and employees in serviceencounters (Bettencourt and Gwinner 1996; Gremler andGwinner 2000; Gremler, Gwinner, and Brown 2001;Jones, Mothersbaugh, and Beatty 2000), little attentionhas been focused on the role that such interactions play inthe service recovery and customer complaint process. As aresult, there is evidence that strategies designed to fosterrapport yield satisfaction benefits and increase positiveword of mouth (Gremler and Gwinner 2000). What is notyet known is whether rapport also provides recovery bene-fits. The results of four independent studies address this is-sue and lead to important strategic implications forrecovery management and complaint solicitation.

BACKGROUND

Service practitioners have devoted considerable re-sources to building rapport with customers. Service com-panies such as Disney and Southwest Airlines havedeveloped extensive training programs to teach employeeshow to be perceived as friendly, helpful, and open. Corpo-rate slogans such as United Airline’s “Fly the FriendlySkies” and State Farm’s “Like a Good Neighbor” suggestsimilar investments. However, scholars have only recentlybegun to investigate the benefits of rapport.

Gremler and Gwinner (2000) provided an exhaustivereview of the rapport literature to date. Their findings sug-gest that two components of rapport, enjoyable interactionand personal connection, are perceived by both customersand employees as being important in the development ofrelationships in service contexts. Enjoyable interaction re-fers to “an affect-laden, cognitive evaluation of one’s ex-change with a contact employee” (p. 92). Personalconnection is a reflection of the customer’s perception of abond between the two parties in the dyad. A personal con-nection is present when there is a “strong affiliation withthe other person based on some tie” (p. 92). Based on thisresearch, the following working definition of rapport is ad-vanced: “Rapport is a customer’s perception of having anenjoyable interaction with a service provider employee,

characterized by a personal connection between the twointeractants” (p. 92). Empirical analysis of the relation-ships between rapport and service outcome measures indi-cates that rapport is positively linked to satisfaction,loyalty, and positive word of mouth. Thus, existing evi-dence indicates that rapport is a beneficial strategy whenservice is delivered without failure.

In view of the variable nature of service delivery(Zeithaml, Parasuraman, and Berry 1985) and the consid-erable effort to better understand service recovery (Kelleyand Davis 1994; Smith and Bolton 1998; Smith, Bolton,and Wagner 1999; Tax and Brown 1998) and complaintsolicitation (Richins 1981; Singh 1990; Stephens andGwinner 1998; Tax, Brown, and Chandrashekaran 1998),perhaps a more intriguing issue is whether rapport alsoprovides benefits when service is poor. Theory drawnfrom relationship marketing indicates that this view mayhave merit. It is suggested that if customers are receivingimportant relational benefits,1 they may remain in a rela-tionship even if they perceive the core service attributes tobe less than superior (Gwinner, Gremler, and Bitner1998). This suggests that relational benefits serve as aswitching barrier. Unfortunately, this effect has not beentested, and little thought has been given to how this mayoccur.

CONCEPTUAL MODEL ANDRESEARCH HYPOTHESES

At first glance, the effects of rapport on service out-comes appear straightforward. A positive, engaging rela-tionship with service customers should yield benefits tothe customer (e.g., companionship, a sense of belonging)and the firm (e.g., customer retention). Indeed, prior re-search supports this perspective (Gremler and Gwinner2000). However, these relationships become less clear inthe presence of a dissatisfying service encounter. Buildingon Tax, Brown, and Chandrashekaran’s (1998) researchon prior experience and complaint handling, positive priorexperiences may influence customer judgments in twocontrasting ways. The first is a negative perspectivewhereby the positive prior relationship heightens expecta-tions, which in turn increases negative disconfirmationfollowing the dissatisfying experience. That is, rapportmay widen the negative gap between performance andprior expectations. The second, positive perspective infersthat the positive prior relationship diminishes the negativeeffects of a service failure. This suggests that rapport actsas a buffer against negative service outcomes (e.g., dissat-

194 JOURNAL OF SERVICE RESEARCH / November 2003

1. Relational benefits are positive exchange outcomes that come froman ongoing, as opposed to a transaction-based, relationship (cf. Bolton1998),which is the typical condition inwhich rapport is suggested toexist.

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 4: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

isfaction, negative word of mouth). Empirical support foreither perspective is limited. We therefore rely on relevanttheory and any available prior research to examine the in-fluence that rapport has on (a) customer satisfaction, (b)repatronage intentions, (c) negative word-of-mouth inten-tions, and (d) complaint intentions, after a dissatisfyingmarketplace experience.

Theory appears to support both a positive and a nega-tive relationship between rapport and service outcomes, asdiscussed by Tax, Brown, and Chandrashekaran (1998).However, the weight of existing evidence supports a posi-tive effect. For instance, the relationship marketing litera-ture suggests that customers will resist switching in thepresence of relational benefits. Gwinner et al. (1998) sug-gested that consumers obtain social and psychologicalbenefits from their relationships with service personnelthat go beyond satisfaction with the core service. Gremlerand Brown (1998) asserted that interpersonal bonds arestrong predictors of service loyalty. Jones, Mothersbaugh,and Beatty (2000) found that the social benefits derivedfrom interpersonal relationships between customers andservice personnel encouraged customers to remain withthe service provider. These results suggest that rapportmay have an “inoculation effect” on service failures thatdiminishes the negative effects of a dissatisfying serviceencounter.

From a complaint behavior context, Singh (1990)found exit behaviors to be negatively influenced by theperceived responsiveness of the supplier. Moreover, Tax,Brown, and Chandrashekaran’s (1998) study supports apositive relationship between prior experience and percep-tions of complaint handling, which indicates that customersatisfaction and loyalty strategies can be built around rela-tional benefits. Thus, to the extent that rapport strengthensthe relationship between customers and employees in apositive service encounter, we expect a positive relation-ship between rapport and customer satisfaction andrepatronage intentions in a service failure condition. Thisleads to the first and second hypotheses:

Hypothesis 1:Following a dissatisfying service experi-ence, those who are engaged in a high level of rap-port with the service provider will be more satisfiedthan those who are engaged in low levels of rapport.

Hypothesis 2:Following a dissatisfying service experi-ence, those who are engaged in a high level of rap-port with the service provider will be more likely toexhibit repatronage intentions than those who areengaged in low levels of rapport.

If rapport does indeed influence satisfaction andrepatronage intentions as argued, then diminished levels

of negative word of mouth are a logical extension of thepositive influences associated with rapport. Research inthe complaint behavior literature supports this assertion,as Singh (1990) found negative word-of-mouth responsesto be less common when the firm is viewed as being recep-tive to complaints. Customers therefore seem to defer neg-ative word of mouth when there is a high probability thatservice providers will redress their complaint. This leadsto the third hypothesis:

Hypothesis 3:Following a dissatisfying service experi-ence, those who are engaged in a high level of rap-port with the service provider will be less likely toexhibit negative word-of-mouth communicationsthan those who are engaged in low levels of rapport.

Regardless of operating practices, dissatisfying mar-ketplace experiences are bound to occur, resulting in somelevel of customer dissatisfaction. The manner in whichmanagement deals with customer dissatisfaction can havean important influence on customer retention, word ofmouth, and firm profitability (Spreng 1995; Tax andBrown 1998). Although the number of complaints in re-sponse to dissatisfaction varies, it is generally acceptedthat the incidence of complaints is much lower than thenumber of dissatisfying events. Best and Andreasen(1977) referred to this as the tip-of-the-iceberg problem orwhat Singh (1990) termed “invisible” customer complaintbehaviors (i.e., customer exit or negative word of mouth).

Given the inclination of dissatisfied customers to notcomplain when dissatisfied, the onus for generating com-plaints rests solely on the firm. Learning theories indicatethat as customers learn about the mechanisms and optionsof complaining (e.g., consumer rights, complaint chan-nels, and knowledge of unfair practices) and realize thatthese actions yield positive outcomes, they develop posi-tive attitudes toward complaining (Richins 1981). The ma-jority of organizational initiatives recommended byresearchers to encourage complaints have apparently beenbuilt around this theory (Spreng 1995; Stephens andGwinner 1998; Tax and Brown 1998). Tax and Brown(1998) proffered the following approaches: setting perfor-mance standards for employees, communicating the im-portance of service recovery within the organization,training customers in how to complain, and using techno-logical support offered through customer call centers andthe Internet. Various researchers offer alternative meth-ods, such as 100% satisfaction guarantees (Spreng 1995;Stephens and Gwinner 1998) and the use of “costless”communication channels like 800 numbers and postage-paid cards (Oliver 1997). Although perhaps conforming tolearning theory, the inherent weakness in these approaches

DeWitt, Brady / SERVICE RECOVERY 195

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 5: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

is that they depend largely on the personal initiative of theconsumer, without taking the consumer’s emotions and in-dividual coping strategies into full consideration.

Day (1984) proposed that dissatisfaction serves as themotivation to consider complaint behavior, but the ulti-mate complain/don’t complain decision is based on a com-parison of the costs and benefits that are affected bysituational and personal factors. Of particular relevance isthe personal control that customers feel they possess overthe complaint process. A related concept is balance ofpower, which refers to how much control consumers per-ceive companies have in comparison to themselves. Cus-tomers who perceive the firm to possess an unfavorablebalance of power will exhibit self-control as an emotion-focused coping tactic in which they simply resist the urgeto voice a complaint (Lazarus and Folkman 1984).

Research demonstrates that perceived control can beincreased and the balance of power can be improved whencustomers and employees form relationships. Ping’s(1993) study of the effects of dissatisfaction in supplier-client relations found a positive relationship between rela-tionship satisfaction and complaint behavior. Oliver(1997) provided two reasons for this phenomenon: (a) It isin the best interest of satisfied clients to voice concernswhen problems arise so that problems do not occur in thefuture, and (b) a single dissatisfying event can occurwithin basically satisfying relations and not upset the bal-ance of positive encounters. Thus, when a customer has anongoing rapport with the service provider, customersshould exhibit higher levels of complaint behavior. Thisleads to the fourth hypothesis:

Hypothesis 4:Following a dissatisfying service experi-ence, those who are engaged in a high level of rap-port with the service provider will be more likely tocomplain than those who are engaged in low levelsof rapport.

RESEARCH METHOD

Four independent studies were used to test the four re-search hypotheses. The first study employed a scenario-based experimental design (cf. Smith and Bolton 1998)and a student sample. The second study was intended toassess the external validity of the Study 1 findings by test-ing the four hypotheses in two new industries and using anonstudent sample. The third and fourth studies were de-signed to further explore some of the results found inStudies 1 and 2. The third study was intended to uncoverstrategies that can be used to encourage complaint behav-ior under conditions of high rapport, whereas the fourthstudy tested whether the strategies uncovered in Study 3were viable.

Study 1

The Study 1 sample was composed of 320 marketingstudents at a large university. Data were collected using in-dividually completed questionnaires in groups of 20 to100 participants. Each participant had a random chance ofreceiving either a high-level-of-rapport or low-level-ofrapport scenario. A total of 29 questionnaires were dis-carded due to multiple missing values, leaving 291 usablequestionnaires. The gender of the participants was nearlyequally distributed, with 148 male students and 143 fe-male students participating in the study.

The research method involved a between-subjects de-sign in which the magnitude of rapport was manipulated.Participants were asked to evaluate a written scenario inthe context of a service failure at a well-known fast-foodrestaurant (refer to Appendix A for a description of thescenario). The setting of the scenario was deemed appro-priate given the demographics of the sample group andtheir familiarity with fast-food restaurants. A scenario-based study was chosen because of the biases often associ-ated with retrospective self-reports, such as memorylapses, rationalization tendencies, and consistency factors(Smith, Bolton, and Wagner 1999). Moreover, scenario-based studies have been used successfully in a service con-text to evaluate service encounters involving both failureand recovery (Smith and Bolton 1998; Smith, Bolton, andWagner 1999).

The rapport element of the scenario was developed us-ing the two components of rapport (enjoyable interactionand personal connection) conceptualized by Gremler andGwinner (2000). On the basis of results from pretesting, aproduct defect (an improperly prepared menu item) waschosen as a representative outcome failure, and inattentiveservice was chosen as a representative process failure(failure of coworkers to assist the service counter atten-dant). To guard against potential gender bias due to thespecific name and gender of the key actors in the high-rapport scenario, 50% of the participants in the high-rapport condition were given a scenario in which the ser-vice counter attendant was female (Mary), and 50% weregiven a scenario in which the service counter attendantwas male (Mark). The scenarios were later tested for pos-sible gender bias.

The format of the questionnaire used for both scenarioswas identical. Participants began by reading the scenario,followed by a series of items measuring satisfaction,repatronage intentions, negative word of mouth, com-plaint intentions, and evaluations of the service provider (amanipulation check for magnitude of rapport) (AppendixB). Customer satisfaction and repatronage intentions weremeasured using items adapted from Gremler andGwinner’s (2000) scales, with origins in Oliver (1997).

196 JOURNAL OF SERVICE RESEARCH / November 2003

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 6: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

Negative word of mouth was measured using itemsadapted from Blodgett, Hill, and Tax (1997). Complaintintentions were measured using items developed for use inthis survey. Finally, the rapport manipulation was mea-sured using items from Gremler and Gwinner’s (2000)scale. All items were rated on 7-point Likert-type scales,ranging from 1 (strongly disagree/very unlikely) to 7(strongly agree/very likely).

STUDY 1 RESULTS

All of the scale items were submitted to an iterative pu-rification process recommended by Churchill (1979), con-sisting of exploratory factor analysis, reliability analysis,and confirmatory factor analysis (CFA). This process ledus to retain a total of 19 items, which are provided in Ap-pendix B. A comprehensive confirmatory factor analysiswas conducted wherein each item was constrained to loadon its respective factor. The CFA resulted in a chi-square of241.2 (df = 142,p < .001). Although this is a significantvalue, there are potential problems with this statistic dueits strict assumptions and sensitivity to sample size(Bentler and Bonett 1980). Emphasis was therefore placedon evaluation of the fit indexes which are more robust mea-sures of model fit (Hu and Bentler 1999). The standardizedroot mean square residual was .052, the root mean squareerror of approximation was .05, the Comparative Fit Indexwas .99, and the Incremental Fit Index was .99. All con-struct reliability estimates were near or above .70 (cf. Peter1979). Specifically, they ranged from a low of .69 (com-plaint intentions) to a high of .93 (rapport).

Convergent and discriminant validity was assessednext using the procedure recommended by Fornell andLarcker (1981). Convergent validity was assessed by cal-culating the average variances-extracted estimates and byinvestigating the significance oft-values representing therelationships between the items and their latent constructs.The average variances extracted were at or above .50 forthree of the five scales, the exceptions being the ComplaintIntentions and Repatronage Intentions scales, which were.43 and .45, respectively. Allt-values were significant (p<.01). Discriminant validity was assessed by comparing theaverage variance extracted for each measure with theshared variances between each variable in the model. Theresults indicate discriminant validity, as the average vari-ance extracted by each of the measures was greater thanthe shared variance between the construct and all othervariables.

A manipulation check using the rapport scale indicatedthat the rapport manipulation was successful. An inde-pendent samplest test showed a significant main effect forthe manipulated independent variable (t = 19.23,p< .001).

As expected, participants felt a significantly greater senseof rapport with the service attendant when exposed to thehigh-rapport scenario (M = 5.10) than participants ex-posed to the low-rapport scenario (M = 2.60). An inde-pendent samplest test was also conducted to test forpossible gender bias across the rapport scenarios. Resultsshowed no significant difference in satisfaction (t = .563,p= .574), repatronage intentions (t = .193,p= .847), nega-tive word of mouth (t = 1.19,p= .236), or complaint inten-tions (t = .794,p = .428).

A MANOVA with customer satisfaction, repatronageintentions, negative word of mouth, and complaint inten-tions was used to test the four hypotheses. MANOVA anal-ysis assumes the observations in each group areindependently sampled from a multivariate normal distri-bution with equal covariance matrices over the groups. Vi-sual inspection of stem and leaf plots indicated nodepartures from normality. An analysis of the variance ho-mogeneity assumption resulted in fail to reject decisionsfor all of the dependent variables, a result consistent withthe assumption that the variances were equal over thegroups. Because the resulting Wilks’s statistic was signifi-cant (Λ = .593,df = 279,p < .001), subsequent ANOVAswere conducted.

The ANOVA results corresponding to the hypothesestests are presented in Table 1. Power indexes were derivedusing SPSS and an alpha of .05. Observed power for eachmain effect was well above .50 (cf. Murphy and Myors1998), with the exception of complaint intentions (.292).Consistent with Hypothesis 1, there was a significant maineffect of rapport on customer satisfaction,F(1, 285) =173.90,p <.001. Satisfaction was significantly higher inthe high-rapport condition (M = 3.66) than in the low-rapport condition (M = 2.13). These results suggest that in-dividuals who have developed a high level of rapport witha service provider will be more satisfied with the firm aftera dissatisfying marketplace experience than those whohave a low level of rapport with the service provider.

Consistent with Hypothesis 2, there was a significantmain effect of rapport on repatronage intentions,F(1, 285)= 128.56,p< .001). Repatronage intentions were found tobe significantly higher in the high-rapport condition (M =3.70) than in the low-rapport condition (M = 2.24). Theseresults infer that individuals who have developed a highlevel of rapport with a service provider will be more likelyto repatronize the firm after a dissatisfying marketplaceexperience than those who have a low level of rapport withthe service provider.

Consistent with Hypothesis 3, there was a significantmain effect of rapport on negative word of mouth,F(1, 285)= 74.32,p < .001. Intentions to exhibit negative word ofmouth were significantly lower in the high-rapport sce-

DeWitt, Brady / SERVICE RECOVERY 197

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 7: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

nario condition (M = 3.70) than in the low-rapport sce-nario (M = 5.16). These results suggest that after adissatisfying marketplace experience, those individualswho have developed a high level of rapport with the ser-vice provider will be less likely to tell their friends and rel-atives about their dissatisfying experience than individualswho have a low level of rapport with the service provider.

Contrary to Hypothesis 4, there was not a significantdifference in complaint intentions for low- and high-rapport groups,F(1, 285) = 2.16,p = .143. In fact, themean response for complaint intentions in the low-rapportcondition was slightly higher (M = 3.68) than those in thehigh-rapport condition (M = 3.45). Therefore, rapportwith the service provider appears to have no significant in-fluence on an individual’s intention to complain after adissatisfying marketplace experience. Given the lowpower of theF statistic, these results should be acceptedwith caution. The power of a statistical test is a function ofits sensitivity, the effect size, and the criteria used to testthe statistical hypothesis (Murphy and Myors 1998). Sen-sitivity is defined in terms of the degree to which samplingerror introduces imprecision into a study. The sensitivityof the F test may have been compromised by the poorpsychometric properties of the Complaint Intentionsscale.

The results of Study 1 thus supported three of the fourresearch hypotheses, with the exception being the ex-pected positive effect of rapport on complaint intentions.In view of the limited nature of the Study 1 sample, a sec-ond study was devised to determine whether the resultscould be generalized beyond the fast-food industry and astudent sample. A second objective for Study 2 was to im-prove the Complaint Intentions and Repatronage Inten-tions scales, as their performance was marginal in Study 1.

Study 2

The same four hypotheses proposed for Study 1 weretested in Study 2 using a sample of nonstudent service con-sumers who were introduced to scenarios from one of twoservice industries (quick lube oil change and haircuttingoutlet). Like the scenarios used in Study 1, a product de-fect was chosen as a representative outcome failure, andinattentive service was chosen as a representative processfailure (Appendix A). Each participant had a randomchance of receiving either a high-level-of-rapport or a low-level-of-rapport scenario. Participants were recruited bytrained student assistants at multiple locations in a me-dium-sized metropolitan area. To ensure that the samplewould represent the regional population from which thesample was drawn, the assistants were provided with spe-cific solicitation instructions and demographic quotas(e.g., age, gender, education). Participants were only al-lowed to complete one questionnaire, and 10% of the totalparticipants were contacted after data collection to verifytheir responses. A total of 5 questionnaires were discardeddue to multiple missing values, leaving 148 usable ques-tionnaires. The sample was equally distributed between thetwo service industries (n = 74) and two treatment condi-tions (n = 37). The participants in the sample were 51% fe-male and ranged between the ages of 18 and 86 years, witha mean age of 43 years. The sample demographicsmatched the regional population very well on all of themeasures collected.

The format of the questionnaire and the scales used inStudy 2 were identical to those in Study 1, with the excep-tion of the Repatronage Intentions and Complaint Inten-tions scales. Several items in these scales were modified orremoved to improve psychometric performance. For ex-

198 JOURNAL OF SERVICE RESEARCH / November 2003

TABLE 1Results for Studies 1 and 2

Low Rapport High Rapport

Dependent Variable M SD M SD F Significance η2 Power

Study 1Customer satisfaction 2.13 0.78 3.66 1.14 173.90 .000 .381 1.000Repatronage intentions 2.24 0.93 3.70 1.21 128.56 .000 .312 1.000Negative word of mouth 5.16 1.36 3.70 1.47 74.32 .000 .208 1.000Complaint intentions 3.68 1.54 3.45 1.26 2.16 .143 .007 0.292

Study 2: Quick lube oil changeCustomer satisfaction 2.04 1.54 3.22 1.03 15.05 .000 .173 0.969Repatronage intentions 1.92 1.10 3.39 1.70 19.38 .000 .212 0.991Negative word of mouth 5.47 1.39 4.06 1.63 16.29 .000 .184 0.978Complaint intentions 5.67 1.34 4.97 1.59 4.21 .044 .055 0.526

Study 2: Haircutting outletCustomer satisfaction 2.11 1.08 3.66 1.41 28.32 .000 .282 0.999Repatronage intentions 2.06 1.10 3.73 1.44 51.95 .000 .287 1.000Negative word of mouth 5.28 1.55 3.53 1.36 26.54 .000 .269 0.999Complaint intentions 5.48 1.33 4.51 1.69 7.51 .008 .094 0.771

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 8: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

ample, a potential problem with the Complaint Intentionsitems used in Study 1 was that they were too specific.Three items in that scale asked respondents whether theywere likely to complain to (a) the frontline service em-ployee directly, (b) the manager in person, and (c) on thetelephone once they returned home. Logic suggests that,even if a respondent was likely to opt for one of these com-plaint options, there is no need to use all three avenues.This may account for the relatively low interitem correla-tions and poor performance of the scale in Study 1. Thewording of the Complaint Intentions scale was thereforemodified to refer to complaint intentions more generally,without making reference to specific complaint avenues.A complete list of the items used in Study 2 appears in Ap-pendix B.

STUDY 2 RESULTS

The same scale analysis procedures as in Study 1 werefollowed in Study 2. The comprehensive CFA resulted in achi-square of 227.27 (df= 109,p< .001). The standardizedroot mean square residual was .059, the root mean squareerror of approximation is .06, the Comparative Fit Indexwas .99, and the Incremental Fit Index was .99. All con-struct reliability estimates were well above .70 in Study 2,which suggests the scale improvements were successful.Specifically, the construct reliability estimate for com-plaint intentions improved from .69 in Study 1 to .88 inStudy 2, and the construct reliability estimate forrepatronage intentions improved from .70 in Study 1 to .86in Study 2. The remainder of the construct reliability esti-mates ranged from .81 (customer satisfaction) to .96 (rap-port). The measures also passed the convergent anddiscriminant validity tests. The average variances ex-tracted ranged from .59 (customer satisfaction) to .85 (rap-port), and none of the shared variances exceeded theaverage variances extracted.

An independent samplest test showed a significantmain effect for the manipulated independent variable forboth the quick lube oil change (t = 7.17 ,p < .001) andhaircutting (t = 9.38 ,p< .001) scenarios. Participants felt asignificantly greater sense of rapport with the customer-contact employee when exposed to the high-rapport con-dition than participants exposed to the low-rapport condi-tion for both the quick lube oil change (M = 5.41 versus2.98) and haircutting (M = 5.20 versus 2.85) scenarios.

A MANOVA with customer satisfaction, repatronageintentions, negative word of mouth, and complaint inten-tions was used to test the hypotheses. Visual inspection ofstem and leaf plots suggested some departures from nor-mality, but none were severe. Levene’s test was employedto assess the assumption of variance homogeneity andagain resulted in fail to reject decisions for all dependent

variables, which is consistent with the assumption that thevariances were equal across the groups. Because the re-sulting Wilks’s statistic was significant (Λ = .024,df= 139,p < .001), subsequent ANOVAs were conducted.

The ANOVA results corresponding to the hypothesestests are presented in Table 1. Consistent with Hypothesis1, there was a significant main effect of rapport on cus-tomer satisfaction for both the quick lube oil change,F(1, 72) = 15.05,p < .001, and haircutting,F(1, 72) =28.32, p < .001, scenarios. Customer satisfaction wasfound to be significantly higher in the high-rapport condi-tion than in the low-rapport condition for both the quicklube oil change (M = 3.22 versus 2.04) and haircutting (M= 3.66 versus 2.11) scenarios. These results suggest thatindividuals who have developed a high level of rapportwith a service provider will be more satisfied with the firmafter a dissatisfying marketplace experience than thosewho have a low level of rapport with the service provider.

Consistent with Hypothesis 2, there was a significantmain effect of rapport on repatronage intentions for boththe quick lube oil change,F(1, 72) = 19.38,p < .001, andhaircutting, F(1, 72) = 51.95,p < .001, scenarios.Repatronage intentions were found to be significantlyhigher in the high-rapport condition than in the low-rapport condition for both the quick lube oil change (M =3.39 versus 1.92) and haircutting (M = 3.73 versus 2.06)scenarios. These results infer that individuals who havedeveloped a high level of rapport with a service providerwill be more likely to repatronize the firm after a dissatis-fying marketplace experience than those who have a lowlevel of rapport with the service provider.

Consistent with Hypothesis 3, there was a significantmain effect of rapport on negative word of mouth for boththe quick lube oil change,F(1, 72) = 16.29,p < .001, andhaircutting,F(1, 72) = 26.54,p < .001) scenarios. Inten-tions to exhibit negative word of mouth were significantlylower in the high-rapport scenario condition than in thelow-rapport scenario for both the quick lube oil change(M = 4.06 versus 5.47) and haircutting (M = 3.53 versus5.28) scenarios. These results suggest that individualswho have developed a high level of rapport with the ser-vice provider will be less likely to tell their friends and rel-atives about their dissatisfying experience thanindividuals who have a low level of rapport with the ser-vice provider.

Contrary to Hypothesis 4, high rapport did not increasecomplaint intentions. Rather, participants exposed to thelow-rapport scenario condition were more likely to com-plain than their high-rapport counterparts in both the quicklube oil change,F(1, 72) = 4.21,p = .044, and haircutting,F(1, 72) = 7.51,p = .008, scenarios. These results suggestthat individuals who have developed a high level of rap-port with the service provider will belesslikely to com-

DeWitt, Brady / SERVICE RECOVERY 199

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 9: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

plain than individuals who have a low level of rapport withthe service provider.

In view of the discouraging statistics on complaint be-havior in practice (i.e., up to 95% of dissatisfying incidentsare unreported) and the complaint intentions results fromStudies 1 and 2, a better understanding of the effect of rap-port on complaint intentions was needed. Managers needhelp in identifying ways to improve the probability ofvoiced or formal complaints after a dissatisfying serviceexperience. This is a critical element of customer reten-tion. With this in mind, two studies were developed to fur-ther explore the relationship between rapport andcomplaint intentions, with particular emphasis on identi-fying ways to increase complaint intentions in the pres-ence of high rapport between the customer and serviceprovider.

Studies 3 and 4

Studies 3 and 4 were conducted with three objectives.First, at a general level, we hoped to develop a better un-derstanding of the complaint intentions results fromStudies 1 and 2. Second, we wanted to uncover strategiesthat can be used to encourage complaint behavior underconditions of high rapport. Third, we wanted to testwhether the strategies uncovered were viable. Two relatedstudies (3 and 4) were undertaken to accomplish these ob-jectives. An exploratory qualitative study (Study 3) wasfirst conducted to help elucidate why service customersappear unwilling to complain in the presence of high rap-port and to unearth potential complaint-generating strate-gies that can be employed to improve the likelihood ofcomplaint behavior. Then, a scenario-based empiricalstudy (Study 4) was conducted to test the viability of thestrategies uncovered in Study 3.

STUDY 3

For Study 3, a qualitative survey was developed anddistributed to 40 graduate and undergraduate business stu-dents at two geographically dispersed universities. Thesurvey contained the high-rapport scenario used in the firststudy (Appendix A), the complaint intentions items, andseveral open-ended questions. The open-ended questionsasked respondents to explain why they were likely or notlikely to complain in the scenario. They were then asked torecommend specific strategies that the restaurant coulduse to improve the likelihood of complaints. In takingthese steps, we hoped to gain a better understanding ofwhy service customers are seemingly less likely to com-plain in the presence of high rapport (Study 2). We also in-tended to identify strategies that service managers can

employ to help diminish the apparent negative effect ofrapport on complaint intentions.

Two independent coders content analyzed the data.Coder disagreements were discussed, and if a resolutioncould not be met, the incident counted against the reliabil-ity assessment (Kassarjian 1977). The interrater reliabilityestimate was 89%.

Analysis of the open-ended responses suggests a pri-mary reason for not complaining was that a relationshipexists between the customer and service provider and thatcomplaining was perceived as a possible disruption to thisrelationship (25%). One respondent said, “I wouldn’t wantto destroy our relationship and I’d like to reward him (bynot complaining) for being so friendly to me in the past.”Many respondents indicated that they were unlikely tocomplain because they did not want to bother the server orbring on added stress to a person regarded as a friend orally. One respondent said, “I would not complain to Markbecause I see that he is doing the best job he knows howand that he is probably stressed out already.” Such a re-sponse suggests that rapport can act as a complaint deter-rent in the presence of high rapport and that customersmay view not complaining as a “reward” for a relationship.That is, customers view complaining as a negative behav-ior that can stress a friendly employee or even jeopardizethe employee’s job. The act of not complaining about a dis-satisfying service encounter is therefore seen as a benefitawarded to the employee for the relationship. These quali-tative findings begin to explain the results from Studies1and 2 and suggest a need to understand how to motivatecustomers to complain in the presence of high rapport.

With a view to identify strategies to promote com-plaints in the presence of rapport, respondents were alsoasked to list actions that would increase the likelihood ofcomplaining to both the employee and the manager. Theresults for the employee were straightforward. Fifty per-cent of the sample suggested that the employee needs toactively encourage complaint behavior, either in generalor about the specific incident. “She should ask for feed-back the next time that I come into the restaurant. It takesthe customer out of the position of feeling bad to com-plain.” An underlying premise is that encouraging feed-back makes customers aware that suppressing complaintsshould not be viewed as a reward for the relationship. “Sheshould let me know that she is open to criticism and that it’snot personal.” Twelve percent of the sample indicated thatemployees should monitor the performance of other em-ployees and ask for help when needed. “Mark could haveasked his fellow associates for help and maybe broughtthis situation to his manager.” This implies that some cus-tomers are unwilling to accept the responsibility of com-plaining but feel that employees should collaborate in

200 JOURNAL OF SERVICE RESEARCH / November 2003

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 10: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

minimizing service failures. Some respondents indicatedthat there was no need for any changes. Seventeen percentof the sample indicated that the employee was already do-ing enough and therefore should not have to do more to en-courage complaints. “I don’t feel Mary has anything moreshe can do. If it (the service failure) were bad enough, Iwould complain to her anyways.”

The results for the manager yielded two distinct strate-gies that would encourage complaints. The first is the pres-ence of an anonymous feedback system. Thirty-onepercent of the sample suggested a strategy that called for anonconfrontational or anonymous feedback system. “Therestaurant could have a complaint box, or they could attacha survey at the bottom of the receipt asking how the servicewas.” The second recommended strategy for increasingcomplaints to the manager was to make sure that the man-ager is visible and engaging. This was mentioned by 32%of the sample. “If the manager was walking around or eas-ily summoned, I would be more likely to complain.” Otherstrategies indicated by respondents included providingvisible signage about their complaint policy (10%) andproviding better service (9%).

It is notable that the three most common recommenda-tions given by the sample are frequently used in practice toencourage complaint behavior. Although service custom-ers seem to believe that these strategies are effective com-plaint mechanisms, the motivating strength of thesestrategies has not been empirically tested.

STUDY 4

An empirical study was developed to test each of thecomplaint-motivating strategies uncovered in Study 3. Asin Studies 1 and 2, low- and high-rapport scenarios wereused in the analyses. However, the high-rapport scenarioswere extended in Study 4 to include the presence of en-couraged feedback by the server, an anonymous feedbacksystem, a visible and engaging manager, or some combi-nation of these strategies. Other than the modificationsneeded to manipulate the three strategies, the low- andhigh-rapport scenarios used in Study 4 were identical tothose in Study 1 (refer to Appendix A). The survey wasalso identical, with the exception of the addition of a scaleto measure attitude toward complaining. Because a smallportion of the qualitative sample had either a strong orweak attitude toward complaining, this was added as acovariate in Study 4.

The scenarios and attached survey were distributed to126 marketing students at a medium-sized university. Thepresence of rapport, the gender of the employee, and thecomplaint strategy were randomly distributed to groups of30 to 50 students. Minimum cell size for the manipulations

was 31. A manipulation check supported the realism of thescenarios (M = 5.5/7).

The results are reported in Table 2. The findings againindicate an insignificant difference in complaint intentionsacross the four scenarios,F(3, 121) = 1.28,p= .284), evenwhen attitude toward complaining was added as acovariate. The complaint intentions means for the high-rapport, encouraged general feedback and the high-rapport, visible and engaging manager scenarios were notsignificantly higher than the low-rapport scenario (M =4.72 and 4.86 versus 4.24). There was virtually no meandifference for the high-rapport, anonymous feedback sys-tem scenario versus the low-rapport scenario. Thus, theStudy 4 results reinforce the complaint intentions findingsfrom Studies 1 and 2. Service customers appear unlikely tocomplain about a dissatisfying service encounter in thepresence of high rapport, regardless of any strategies em-ployed to encourage complaint behaviors.

GENERAL DISCUSSION

As competition in service industries increases, reflect-ing a new age of service commoditization (Pine andGilmore 1998), managers are looking for new ways to dif-ferentiate their service products. They are also looking forcost-effective customer retention strategies. Rapport is amechanism with the potential to accomplish both of theseobjectives. The idea is that a personal relationship,whether in the form of friendship or simple camaraderie,gives service customers a reason to return. Prior researchsupports this strategy in the context of a satisfying experi-ence (Gremler and Gwinner 2000). Our findings extendthis benefit to the realm of a dissatisfying experience. Theresults indicate that rapport does play a positive role in re-covery. They also support the assertion that positive ante-cedent states can help mitigate the negative effects ofservice failure.

Study 1 examined the effects of rapport on outcomesassociated with service failures. Student participants ex-posed to a scenario depicting rapport with a service em-ployee in a negative fast-food encounter were moresatisfied, were less likely to spread negative word-of-mouth communications, and were more likely to remain inthe relationship with the firm (repatronage intentions).However, participants exposed to high rapport were notmore likely to voice complaints to the service provider.

Study 2 examined the effects of rapport on outcomesassociated with service failures using a nonstudent sampleacross divergent service settings. Given a dissatisfyingmarketplace experience, participants exposed to a situa-tion depicting rapport with a service employee were again

DeWitt, Brady / SERVICE RECOVERY 201

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 11: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

more satisfied, less likely to spread negative word ofmouth, and more likely to remain in the relationship withthe firm (repatronage intentions). However, participantsexposed to high rapport wereless likely to voice com-plaints to the service provider.

Study 3 was intended to better understand why dissatis-fied service customers seem unwilling to complain whenrapport is present. A second objective for Study 3 was toidentify strategies that could increase complaint behavior.The qualitative study revealed that customers may viewnot complaining as a reward for the personal bond be-tween the customer and employee. In effect, some custom-ers view a complaint as a potential problem or hassle forthe employee, and they would rather avoid causing a prob-lem for someone that they view as a friend or ally. Thismay clarify why service customers were more likely tocomplain in the low-rapport setting in Study 2. The strate-gic implication from this finding is straightforward: In thepresence of rapport, it appears that managers need to find amethod to promote complaint behavior that is not per-ceived by the customer as damaging to the employee.

The qualitative sample was also asked to report whatthe service provider could do to improve the likelihood ofcomplaint behavior when rapport is present. Top re-sponses were in line with the objective to protect thefriendly employee. The sample recommended having theemployee actively solicit complaints. This raises aware-ness of the importance of complaints to the service pro-vider while also making the customer aware that there willbe no negative consequences to the friendly employee. Asecond strategy mentioned by the qualitative sample wasto have a visible and engaging manager present in the ser-vice environment. This strategy effectively excludes thefriendly employee from having to deal with the unpleasantsituation and places this responsibility on management. Athird strategy mentioned was the presence of an anony-mous feedback system. Again, this strategy removes thefriendly employee from the complaint process and avoidsthe unpleasant task of complaining to a friendly recipientor the manager.

Study 4 provided an empirical test of the recommendedstrategies uncovered in Study 3. Despite the promising re-

sults of Study 3, little, if any, improvement to complaint in-tentions was found. The results suggest that customerswho have developed a high level of rapport with the ser-vice provider are unlikely to complain regardless of thecomplaint-generating strategy introduced by the firm.These findings are important because they demonstratethe strong influence that relational bonds have on cus-tomer complaint intentions. Considered collectively withthe results of Study 3, it can be implied that customersplace a high value on the relationship that they have devel-oped with the service provider and appear unwilling to po-tentially damage that relationship by complaining either tothe employee or management, even when given the oppor-tunity or having been encouraged to do so. Managersshould therefore approach the development of rapport as aservice recovery strategy with caution. For although rap-port appears to have a positive influence on the negativeoutcomes often associated with service failure, the resultsof this research demonstrate that it apparently does so atthe cost of consumer intentions to complain.

IMPLICATIONS AND DIRECTIONSFOR FUTURE RESEARCH

This research suggests that rapport may improve thelikelihood of a satisfactory recovery. Rapport was not pre-viously considered a recovery mechanism, ostensibly be-cause it is developed before a service failure occurs. Extantservice recovery research is focused only on postfailurestrategies. Studies 1 and 2 nonetheless demonstrate thatdeveloping rapport with customers improves postfailureoutcome measures. Although this research was not in-tended to uncover why this may occur, the qualitative re-search suggests service customers are more forgiving offirms for which they have an established, friendly relation-ship with employees. That is, rapport seems to have an in-oculation effect on service failures wherein goodwill thatis developed from a positive relationship with an em-ployee is later applied to the interpretation of a failed ser-vice encounter. Enacting strategies that promote suchpositive antecedent states appears to be an excellent way to

202 JOURNAL OF SERVICE RESEARCH / November 2003

TABLE 2Results for Study 4

High Rapport High RapportHigh Rapport

Visible and AnonymousLow Encouraged Engaging Feedback

Rapport Feedback Manager System

Dependent Variable M SD M SD M SD M SD F Significance η2 Power

Complaint Intentions 4.24 1.43 4.72 1.09 4.86 1.41 4.26 1.28 1.28 .284 .031 .361

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 12: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

improve postfailure customer retention and reduce nega-tive word of mouth. These outcomes are particularly sa-lient as service products evolve and as service marketsbecome increasingly competitive (Pine and Gilmore1998).

For managers focused on rapport, the findings indicatethat in addition to enhancing perceptions of service qualityand customer satisfaction, rapport can have a positive ef-fect on the negative outcomes often associated with ser-vice failures. Thus, providing the foundation for thedevelopment of rapport with customers can hold a varietyof benefits. To reap such benefits, managers may need tohelp certain employees develop rapport with customers.Indeed, for many employees, the ability to build rapport isa function of their innate personalities. However, for oth-ers, this aptitude must be developed and encouraged by thefirm and the contact employee’s work group.

Research suggests that building rapport can be alearned behavior and that employees should practice do-ing so (Hollman and Kleiner 1997). Techniques such asconveying empathy, making the customer feel important,and mirroring a customer’s pace of speech are helpful inbuilding rapport. However, key to the success of such aninitiative is a customer-centric business strategy that fos-ters employee satisfaction and reinforces the importanceof developing rapport with customers. In doing so, thishelps to open the door for firms to take a proactive stance inaddressing service failures, regardless of whether the firmis given an opportunity to enact a service recovery.

For service recovery, the results of Studies 1 and 2 mayhelp to improve the reported 50% recovery failure rate(e.g., Hart, Heskett, and Sasser 1990). Under the worst-case scenario, if only 5% of dissatisfied customers com-plain (e.g., Harari 1992) and a firm effectively recoversfrom only half of those complaints, more than 95% of dis-satisfied customers will remain dissatisfied. Under thebest-case scenario presented by Harari (1992), this figureis still 85% (i.e., 30% complain, and half of those result inrecovery). Given the complaint behavior statistics re-ported in the literature and supported in Study 4, mishan-dling a recovery opportunity should be a particularconcern for managers.

For complaint behavior, results of all four studies reit-erate that finding effective strategies to increase com-plaints remains an elusive goal. This difficulty may be areflection of the nature of service industries themselves inthat they tend to be more interactive and personal thanphysical goods products. In such circumstances, consum-ers have difficulty delivering bad news in the form of acomplaint. Our four studies indicate that this problem maybe exacerbated by the presence of rapport.

The results of this study suggest that consumers viewthe relationship they have developed with the service pro-

vider as a benefit that they are unwilling to sacrifice in theface of a dissatisfying marketplace experience. Stephensand Gwinner (1998) elaborated on the nature of the cost-benefit analysis and extended the conceptualization by ex-plicitly considering the role of emotion and the identifica-tion of various coping alternatives. The authors suggestthat compassion moderates the relationship between an in-dividual’s emotional reaction to a dissatisfying market-place experience and his or her coping strategies. Theirresearch demonstrates that because consumers empathizeand understand a service provider’s situations or feelings,they are likely to remain silent in the face of marketplaceproblems. Taken collectively, the results of this study sug-gest that consumers who have developed rapport with theservice provider are less likely to complain because theyare empathic toward the situations these individuals workunder and are unwilling to possibly damage their relation-ship by complaining.

Given this circumstance, service practitioners shouldconsider alternative methods to identify service failure.Technology may offer some assistance in this pursuit. Forexample, the Internet offers customers an accessible, im-personal, and convenient medium from which to submit acomplaint. Perhaps providing a convenient and imper-sonal outlet that allows customers to complain when theincident is fresh in their minds will increase complaint be-havior. In many service industries, managers are activelytracking waiting time and looking for ways to decrease ac-tual or perceived waits. Perhaps this technology could beused to issue an automatic apology or recovery strategy if await exceeds a maximum threshold. This way, customerswould not have to complain for the recovery to be enacted.

While eliminating errors in a service company is likelyan impossible goal, fostering a philosophy of zero defectsis not. Service companies should resist the urge to find anemployee responsible for a service failure and instead fo-cus on removing the source of the problem. These problemsources are often operational defects that can be correctedif failures are properly tracked over time. Indeed, this re-search underscores the importance of being proactive to-ward service failure. It may be that waiting to addressservice failure after it occurs is too late to mitigate damage.The poor recovery rate in practice and the improbabilitythat dissatisfied customers will complain supports thisperspective.

Future research should be directed toward studyingcomplaint behaviors with the aim of identifying strategiesto increase propensity to complain. This is particularly im-portant in the presence of rapport because our results sug-gest customers are less likely to complain in a high-rapportcondition. It might be that customer characteristics play arole in complaint behavior. Certain individuals may bemore likely to complain about poor service, and these

DeWitt, Brady / SERVICE RECOVERY 203

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 13: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

differences might emanate from demographic or psycho-graphic factors. Or it might be that those who feel stronglyabout complaining have educated themselves about theimportance of complaints and the resulting benefits withrespect to service improvements and restitution. It is alsopossible that service customers are simply reluctant tocomplain due to the interpersonal nature of service en-counters. If this is true, then future research should exam-ine the applicability of alternative methods to identifyservice failure.

Additional research is also needed to further explorethe costs and benefits of rapport. The findings reportedhere indicate a positive return on rapport generation, but insome service industries, there may be risks associated withsuch a strategy. Customers sometimes respond negativelyto professional service providers that are perceived as toofriendly. This may be especially prevalent in serviceswhere the service provider is expected to hold a position ofesteem or expertise, such as in psychology, accounting, orin situations where service customers simply prefer ano-nymity. Research is therefore needed to better understandwhen customers may not welcome rapport and which cus-tomers might feel this way.

In recognizing the benefits that rapport affords a firmafter a service failure, research should also explore thestrategies that firms can use to assist employees in their ef-forts to cultivate rapport with customers. A qualitativestudy conducted by Gremler, Rinaldo, and Kelley (2002)identified rapport-building strategies commonly used bycustomer-contact employees in service encounters. Theirclassification scheme revealed five groups of rapport-building behaviors: sharing or gathering of information,common grounding, connecting behavior, uncommon at-tention, and demonstrated values. Future research shouldaddress the organizational factors that both encourage andlimit the display of these behaviors. A related inquiry is aninvestigation of the role of rapport in customer acquisition.Rapport has been studied for customers who are in the ser-vice process, and this study examines the influence of rap-port after service provision. The strength of a rapportstrategy in acquiring customers remains unknown.

The results presented here also have future research im-plications for service recovery. Although postfailure re-covery is currently a popular topic in service research,these efforts have been directed at uncovering strategiesthat respond to service failure. Our findings indicate thatprefailure conditions can influence perceptions of recov-ery. Further work is needed to identify the factors that af-fect recovery, both before and after a failure occurs.

APPENDIX AScenario for Study 1

FAST FOOD: LOW RAPPORT

You are driving in your car and decide to stop for breakfast at afast-food restaurant that is part of a well-known national chain.When entering the restaurant, you notice that only one of the fiveavailable service stations is open. A long line has formed as a sin-gle service counter attendant struggles to operate the cash regis-ter and assemble orders. Meanwhile, two other servers are busychatting near one of the closed service stations, seemingly un-aware of their workmate’s dilemma. The outlet manager appearsto be in her office. After waiting 7 minutes, you finally reach thecounter and order a breakfast sandwich, only to be told that theyare out of eggs, thus limiting your breakfast selection. However,in view of the present situation, the restaurant has begun to servelunch menu items as well. You settle on a hamburger withketchup only, french fries, and soda but have to wait an extra 3minutes as they’ve just begun to prepare the french fries. The ser-vice counter attendant assembles your order, thanks you, andmoves on to the next customer. You then take a seat and unwrapyour hamburger. Upon taking the first bite of your hamburger,you discover that they’ve neglected to prepare it with ketchuponly, having included mustard, pickles, and onions as well.

Scenarios for Study 2

QUICK LUBE OIL CHANGE: LOW RAPPORT

Your car is due for its regularly scheduled 3,000-mile oilchange, so you decide to take it to a quick lube oil change outletlocated near your home. When driving in the entrance, you noticethat only one of three available service bays is open. A line of twocars has formed behind the car in the single bay as a single me-chanic struggles to take down customer information on a clip-board, check and refill the fluid levels for each car, and receivepayment for each customer transaction. Meanwhile, two othermechanics are busy chatting near one of the closed service bays,seemingly unaware of their workmate’s dilemma. The outletmanager is visible working behind the service counter adjacentto the set of service bays. After waiting longer than you would ex-pect, the mechanic directs your car into the single bay. You waitin your car while the mechanic works under the hood and anothermechanic drains the oil from beneath your car. He then closes thehood and collects your credit card for payment. As the mechanictakes the credit card from your hand, he mentions that he was un-able to fill your windshield washing fluid, as the outlet had notyet received their regularly scheduled shipment. You leave theoutlet, return home, and park your car. As you approach your carthe following day, you notice that a pool of oil has formed be-neath your car, suggesting that the drain plug in your oil pan wasnot securely fastened during your oil change.

204 JOURNAL OF SERVICE RESEARCH / November 2003

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 14: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

HAIRCUTTING OUTLET: LOW RAPPORT

You are due for a haircut, so you decide to go to a haircuttingoutlet located near your home. When walking in the entrance,you notice two customers sitting in the waiting area while a sin-gle hair stylist works feverishly to cut hair, answer the phone, andcheck new customers in. Meanwhile, two other hairstylists arebusy chatting near a small table on the other side of the outlet, ap-parently eating their lunch and seemingly unaware of theirworkmate’s dilemma. The outlet manager can be seen workingbehind a desk on the other side of a partition that is out of view ofthe haircutting area. After waiting longer than you would expect,it is your turn, and the hairstylist asks you to take a seat in thechair. She works hurriedly, not taking the time or attention thatshe normally would and nicks your ear with the haircuttingshears in the process. As there are other customers waiting to gettheir hair cut, she also neglects to provide you with the opportu-nity to view your haircut in the mirror. Upon returning home, youlook in the mirror and examine your haircut and find that it wasn’tcut to your satisfaction.

HIGH-RAPPORT SCENARIOS

The high-rapport scenarios are identical to the low rapport,but with the following attributes of rapport integrated:

Enjoyable interaction

• The customer frequents the service firm several times aweek.

• The customer regularly engages in conversation with thecustomer-contact employee.

• The customer contact employee smiles and waves at thecustomer.

Personal connection

• The customer-contact employee is from the same home-town and shares the same interests.

• The customer has developed a close relationship with thecustomer-contact employee.

• The customer knows the employees of the service firm ona first name basis.

APPENDIX BThe Measures

Construct and Scale Items

Customer satisfactionBased on what I read about the firm, I am very satisfied with the ser-

vices it provides.

I would have been happy with my decision to go to this firm.My overall evaluation of the services provided by this firm wouldnot

be very good. (reverse coded)a

I think I did the right thing when I decided to patronize this firm formy particular needs.

Repatronage intentionsI would intend to continue doing business with this firm over the next

few years.I wouldnotbe very likely to recommend this firm to a friend. (reverse

coded)a

As long as the present service continues, I doubt that I would switchfirms.a

It is very likely that I would continue to patronize this firm in thefuture.b

It is very unlikely that I would do most of my future business with thisfirm. (reverse coded)a

Negative word-of-mouth communicationIf this had happened to me, I would make sure to tell my friends not to

patronize this firm.If this had happened to me, I would make sure to tell my relatives not

to patronize this firm.If this had happened to me, I would complain to my friends and rela-

tives about this firm.Complaint intentions

Given the circumstances, I would complain to the frontline serviceemployee.a

Taking everything into consideration, I would return home and com-plain to the firm by telephone.a

Given the circumstances, I would ask to see the manager, so that Icould voice my dissatisfaction with the poor service.b

Given the circumstances, I would inform the firm of my problem.b

Overall, if this had happened to me, I would be very likely to complainto the firm.b

If this had happened to me, I would be very likely to voice my dissatis-faction to the firm.b

RapportIn thinking about my relationship with the frontline service em-

ployee, I would enjoy interacting with this employee.I would look forward to seeing the frontline service employee when I

visit this restaurant.The frontline service employee relates well to me.I strongly care for the frontline service employee.a

I am comfortable interacting with the frontline service employee.The frontline service employee has taken a personal interest in me.

NOTE: Frontline service employee: The type of frontline service em-ployee (service counter attendant, mechanic, and hairstylist) was used inplace of the termfrontline service employeedependingon which scenariowas administered to the respondent. Firm: The type of service firm (fast-food restaurant, quick lube oil change, and haircutting salon) was used inplace of the termfirm, depending on which scenario was administered tothe respondent.a. Removed during the scale purification process.b. Added for Study 2.

DeWitt, Brady / SERVICE RECOVERY 205

APPENDIX B (continued)

Construct and Scale Items

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 15: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

REFERENCES

Bentler, Peter M. and Douglas G. Bonett (1980), “Significance Tests andGoodness-of-Fit in the Analysis of Covariance Structures,”Psycho-logical Bulletin, 88, 588-606.

Berger, Paul D., Ruth N. Bolton, Douglas Bowman, and Elten Briggs(2002), “Marketing Actions and the Value of Customer Assets: AFramework for Customer Asset Management,”Journal of ServiceResearch, 5 (1), 39-54.

Best, Arthur and Alan R. Andreasen (1977), “Consumer Response to Un-satisfactory Purchases: A Survey of Perceiving Defects, VoicingComplaints, and Obtaining Redress,”Law & Society Review, 11(Spring), 701-42.

Bettencourt, Lance A. and Kevin Gwinner (1996), “Customization of theService Experience: The Role of the Frontline Employee,”Interna-tional Journal of Service Industry Management, 7 (2), 3-20.

Blodgett, Jeffrey G., Donna J. Hill, and Stephen S. Tax (1997), “The Ef-fects of Distributive, Procedural, and Interactional Justice onPostcomplaint Behavior,”Journal of Retailing, 73 (2), 185-210.

Bolton, Ruth N. (1998), “A Dynamic Model of the Duration of the Cus-tomer’s Relationship with a Continuous Service Provider: The Roleof Satisfaction,”Marketing Science, 17 (1), 45-65.

Churchill, Gilbert A., Jr. (1979), “A Paradigm for Developing BetterMeasures of Marketing Constructs,”Journal of Marketing Research,16 (February), 64-73.

Day, Ralph L. (1984), “Modeling Choices among Alternative Responsesto Dissatisfaction,” inAdvances in Consumer Research, Vol. 1.Thomas C. Kinnear, ed. Provo, UT: Association for Consumer Re-search, 496-99.

Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equa-tions Models with Unobservable Variables and Measurement Error,”Journal of Marketing Research, 18 (February), 39-50.

Gremler, Dwayne and Stephen Brown (1998), “Service Loyalty: Ante-cedents, Components, and Outcomes,” in1998 AMA Winter Educa-tors’Conference: Marketing Theory and Applications, D. Grewal andC. Poechmann, eds. Chicago: American Marketing Association, 165-66.

and Kevin Gwinner (2000), “Customer-Employee Rapport inService Relationships,”Journal of Service Research, 3 (August), 82-104.

, , and StephenBrown (2001), “GeneratingPositive Word-of-Mouth Communication through Customer-Employee Relation-ships,” International Journal of Service Industry Management, 12(1), 44-59.

, Shannon Rinaldo, and Scott W. Kelley (2002), “Rapport-Building Strategies Used by Service Employees: A Critical IncidentStudy,” in 2002 AMA Summer Educators’ Conference: EnhancingKnowledge Development in Marketing, W. J. Kehoe and J. H.LindgrenJr., eds. Chicago: AmericanMarketing Association, 73-74.

Gwinner, Kevin, Dwayne Gremler, and Mary Jo Bitner (1998), “Rela-tional Benefits in Services Industries: The Customer’s Perspective,”Journal of the Academy of Marketing Science, 26 (Spring), 101-14.

Harari, Oren (1992), “Thank Heaven for Complainers,”Management Re-view, 81 (1), 59-60.

Hart, Christopher W. L., James L. Heskett, and W. Earl Sasser (1990),“The Profitable Art of Service Recovery,”Harvard Business Review,68 (July/August), 148-56.

Hogan, John E., Katherine N. Lemon, and Roland T. Rust (2002), “Cus-tomer Equity Management: Charting New Directions for the Futureof Marketing,”Journal of Service Research, 5 (1), 4-12.

Hollman, Wayne A. and Brian H. Kleiner (1997), “Establishing Rapport:The Secret Business Tool to Success,”Managing Service Quality, 7(4), 194-97.

Hu, Li-tze and Peter M. Bentler (1999), “Cutoff Criteria for Fit Indexes inCovariance Structure Analysis: Conventional Criteria versus NewAlternatives,”Structural Equation Modeling, 6 (1), 1-55.

Jones, Michael A., David L. Mothersbaugh,and Sharon E. Beatty (2000),“Switching Barriers and Repurchase Intentions in Services,”Journalof Retailing, 76 (2), 259-74.

Kassarjian, Harold H. (1977), “Content Analysis in Consumer Re-search,”Journal of Consumer Research, 4 (1), 8-18.

Keaveney, Susan M. (1995), “Customer Switching Behavior in ServiceIndustries: An Exploratory Study,”Journal of Marketing, 59 (April),71-82.

Kelley, Scott W. and Mark A. Davis (1994), “Antecedents to CustomerExpectations for Service Recovery,”Journal of the Academy of Mar-keting Science, 22 (1), 52-61.

Lazarus, Richard S. and Susan Folkman (1984),Stress, Appraisal, andCoping. New York: Springer.

Lemon, Katherine N., Tiffany Barnett White, and Russell S. Winer(2002), “Dynamic Customer Relationship Management: Incorpo-rating Future Considerations into the Service Retention Decision,”Journal of Marketing, 66 (January), 1-14.

Morgan, Robert M. and Shelby D. Hunt (1994), “The Commitment-TrustTheory of Relationship Marketing,”Journal of Marketing, 58 (3), 20-38.

Murphy, Kevin R. and Brett Myors (1998),Statistical Power Analysis.Englewood Cliffs, NJ: Lawrence Erlbaum.

Oliver, Richard A. (1997),Satisfaction—A Behavioral Perspective on theConsumer. New York: McGraw-Hill.

Peter, J. Paul (1979), “Reliability: A Review of Psychometric Basics andRecent Marketing Practices,”Journal of Marketing Research, 16(February), 6-17.

Pine, B. Joseph and James H. Gilmore (1998), “Welcome to the Experi-ence Economy,”Harvard Business Review, 76 (July/August),97-105.

Ping, Robert A., Jr. (1993). “The Effects of Satisfaction and StructuralConstraints on Retailer Exiting, Voice, Loyalty, Opportunism, andNeglect,”Journal of Retailing, 69 (Fall), 320-52.

Reichheld, Frederick (1996), “Learning from Customer Defections,”Harvard Business Review, 74 (2), 56.

Richins, Marsha (1981), “An Investigation of Consumer’s Attitude to-ward Complaining,”Advances in Consumer Research, 9, 502-6.

Rust, Roland T., Valarie A. Zeithaml, and Katherine N. Lemon (2001),“Where Should the Next Marketing Dollar Go?”Marketing Manage-ment, 10 (3), 24-28.

Singh, Jagdip (1990), “Voice, Exit and Negative Word-of-Mouth Behav-iors: An Investigation across Three Service Categories,”Journal ofthe Academy of Marketing Science, 18 (Winter), 1-15.

Smith, Amy K. and Ruth N. Bolton (1998), “An Experimental Investiga-tion of Customer Reactions to Service Failure and Recovery Encoun-ters: Paradox or Peril?”Journal of Service Research, 1 (August), 65-81.

, , and Janet Wagner (1999), “A Model of Customer Satis -faction with Service Encounters Involving Failure and Recovery,”Journal of Marketing Research, 36 (August), 356-72.

Spreng, Richard A. (1995), “International Consumer Behavior: Its Im-pact on Marketing Strategy Development,”Journal of InternationalMarketing, 3 (3), 118-20.

Stephens,Nancy and Kevin P. Gwinner (1998), “Why Don’t Some PeopleComplain? A Cognitive-Emotive Process Model of Consumer Com-plaint Behavior,”Journal of the Academy of Marketing Science, 26(3), 172-89.

Tax, Stephen S. and Stephen W. Brown (1998), “Recovering andLearning from Service Failure,”Sloan Management Review, 40(Fall), 75-88.

and Murali Chandrashekaran (1998), “Customer Evaluations ofService Complaint Experiences: Implications for Relationship Mar-keting,” Journal of Marketing, 62 (April), 60-77.

Zeithaml, Valarie A., A. Parasuraman, and Leonard L. Berry (1985),“Problems and Strategies in Services Marketing,”Journal of Mar-keting, 49 (Spring), 33-46.

206 JOURNAL OF SERVICE RESEARCH / November 2003

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from

Page 16: Rethinking Service Recovery Strategies: The Effect of Rapport on Consumer Responses to Service Failure

Tom DeWittis a doctoral candidate in the marketing departmentat Florida State University. His current research interests are inservices marketing, particularly in understanding aspects of theemployee-customer encounter, customer complaint behavior,and service recovery issues. His research has appeared in theJournal of Marketing-Theory and Practiceand in numerous con-ference proceedings.

Michael K. Brady is an assistant professor of marketing atFlorida State University. His research interests are in the areas ofmanaging the service decision-making process, analyzing ser-

vice recovery issues, and in the strategic ramifications of brand-ing for service firms. His research has been published in theJournal of Service Research, theJournal of Marketing, theJour-nal of Retailing,Psychology & Marketing, theJournal of Busi-ness Research, theInternational Journal of Service IndustryManagement, the Journal of Services Marketing, and in otheroutlets. He has won the M. Wayne Delozier Award for Best Con-ference Paper at the Academy of Marketing Science Conferenceand the Steven J. Shaw Award for Best Conference Paper at theSociety for Marketing Advances Conference.

DeWitt, Brady / SERVICE RECOVERY 207

at Bartin Universitesi on November 14, 2014jsr.sagepub.comDownloaded from