1 Introduction

Service and product failures are major issues for organizations of every type. According to the 2020 Customer Rage Survey, 66% of all US households encounter sub-optimal service performance, with US$494 in lost sales forecasted as a consequence of such service failures (Customer Care Measurement and Consulting [CCMC], 2020). Against this backdrop, the effects of various recovery tactics to mitigate these adverse effects and improve customer satisfaction, loyalty, and profitability have been widely studied (Béal et al., 2019; Knox & Van Oest, 2014; Smith et al., 1999). One commonly used recovery tactic is firm-offered compensation—monetary or non-monetary reimbursement for problems caused by the firm—which is considered the most potent means of offsetting disgruntled customers (Gelbrich & Roschk 2011; Grewal et al., 2008). Several studies exploring the effects of this recovery tactic have shown it to be effective if performed under the right conditions, depending on who is responsible for the problem and the effectiveness of the recovery procedure, including how long after the failure the customer is reimbursed (Grewal et al., 2008; Hogreve et al., 2017; Wirtz & Mattila, 2004).

Nevertheless, recent literature highlights a need for further research on the role of social interaction during recovery situations, and whether employee behaviors during the service recovery encounter affect the utility of firm-offered compensation (Gelbrich et al., 2016). The present study examines the link between the presence (or absence) of firm-offered compensation and customers’ emotional responses, and the downstream effect of those emotions on customer bad-mouthing behavior—that is, telling others about specific problems with a service or product. Customer bad-mouthing behavior is different from negative word-of-mouth (NWOM) in that bad-mouthing distinctively refers to customer communication of a specific service or product failure, whereas NWOM is a more generic description of all negative customer-to-customer communications (Swan & Oliver, 1989). As such, although these phenomena represent interrelated constructs, bad-mouthing adds nuance to the WOM literature, as called for by several scholars (He et al., 2019; Sweeney et al., 2020).

Drawing on the emerging literature on customer participation in service recovery encounters (Van Vaerenbergh et al., 2018), we discuss the process of collaborative recovery, in which customers and front-line employees interact and engage in organized activities to recover a failed situation for mutual benefit. Our findings make two critical contributions to the service recovery literature. First, responding to Grewal et al. (2008), we examine the effects of compensation in a real-world (retail) setting. We confirm that the mitigating effect of firm-offered compensation on bad-mouthing behavior is mediated by customers’ negative emotional responses. Second, our empirical exploration of the novel concept of collaborative recovery confirms that compensation is not effective in isolation. Rather, our findings document that, for almost a third of customers, compensation does not mitigate negative emotional responses, as the service recovery process is often characterized by a lack of collaboration. The demonstrable importance of collaboration in mitigating negative emotional responses and customer bad-mouthing confirms that compensation without collaboration can have negative financial implications, as this amounts to spending money for nothing.

2 Conceptual Framework and Hypotheses

2.1 The mitigating effect of compensation on customer bad-mouthing behavior

Service and marketing research has examined the effects of firm-offered compensation on customer responses in critical situations, and this is commonly viewed as one of the most potent recovery tactics to offset disgruntled customers (Gelbrich & Roschk, 2011; Grewal et al., 2008). In service recovery research, compensation is conceptualized as the monetary or non-monetary reimbursement customers receive for problems caused by the firm (Jung & Seock, 2017; Mattila & Patterson, 2004a; Wirtz & Mattila, 2004). While monetary compensations can take various forms (e.g., store voucher, store credit, or discount on future or additional purchases), non-monetary compensation is commonly understood as replacement of goods or repetition of a service procedure (Roschk & Gelbrich, 2014). On that basis, compensation is conceptualized here as a monetary or non-monetary reimbursement for any inconvenience caused by the firm.

Compensation as a recovery tactic has demonstrable positive effects on different marketing metrics, including customer satisfaction, repurchase behavior, and NWOM (Grewal et al., 2008; Harris et al., 2006; Smith et al., 1999). While we agree that NWOM is a critical measure that warrants attention in the context of failure and recovery, it is necessary to discern different types of WOM communication to avoid confusion and better understand consumer responses (He et al., 2019; Sweeney et al., 2020). In the service recovery context, NWOM is largely assumed to be generic in the sense that a single recovery encounter will color customers’ entire perception of the company in question, prompting them to talk negatively about it (Blodgett et al., 1993; Casidy & Shin, 2015; Hocutt et al., 2006).

Surprisingly, the concept of bad-mouthing has received little attention and is often used interchangeably with general NWOM (Blodgett et al., 1993; Swanson & Kelley, 2001). We define customer bad-mouthing as a particular form of NWOM in which the customer expresses dissatisfaction with a specific service or product failure, making this construct more relatable and precise regarding a typical service recovery situationFootnote 1. Based on prior evidence of the link between firm-offered compensation and key customer outcomes, including NWOM intentions (Casidy & Shin, 2015; Noone, 2012), we hypothesize:

  • H1: The presence (vs. absence) of firm-offered compensation reduces customer bad-mouthing behavior.

2.2 The mediating role of emotions in the compensation–bad-mouthing link

Service recovery research has devoted considerable attention to understanding how compensation is linked to various post-recovery attitudes and behaviors. Grounded in justice theory (Orsingher et al., 2010), compensation is often related to customer evaluation of the fairness of service recovery encounters (Mattila, 2001; Mattila & Patterson, 2004b; Smith et al., 1999), with such evaluations linked to customer satisfaction and loyalty, including repurchase and WOM intentions (Maxham & Netemeyer, 2002; McColl-Kennedy & Sparks, 2003; Sparks & McColl-Kennedy, 2001).

However, recent conceptual studies highlight a need for alternative explanations of why different recovery tactics elicit differing customer responses following a service recovery (Van Vaerenbergh & Orsingher, 2016; Van Vaerenbergh et al., 2019). One possible explanation for the positive effect of compensation on firm-relevant outcomes is the emotional responses that the compensation evokes (Valentini et al., 2020; Vázquez-Casielles et al., 2012). Certainly, recovery situations can be characterized as emotional, as they are out of the ordinary and often result in strong affective responses that shape customer attitudes and loyalty-linked behaviors (Tronvoll, 2011). In particular, a recent meta-analysis documented a link between compensation and its potency in mitigating negative emotions (Valentini et al., 2020) and research has established a positive link between customers’ emotional responses and their loyalty intentions (DeWitt et al., 2008; Otterbring et al., 2021). However, with a few notable exceptions (Mattila, 2010; Valentini et al., 2020; Vázquez-Casielles et al., 2012), surprisingly few studies have examined the link between compensation and customers’ emotional responses. Considering the evidence that compensation mitigates negative emotional responses, and the linkage between such emotions and NWOM (Gelbrich, 2010), we hypothesize:

  • H2: The impact of firm-offered compensation on customer bad-mouthing behavior is mediated by customers’ negative emotional responses, such that customers who are offered compensation by an employee respond less negatively than those who are not and, consequently, are less likely to engage in bad-mouthing behavior.

2.3 The moderating effect of collaborative recovery

Although the literature has documented positive effects of compensation on various key metrics, there is extensive support for the notion that this tactic has certain limitations in recovery situations. The effect of compensation on customer response varies as a function of several factors. Research on the appropriate amount of compensation indicates that overcompensation can induce feelings of guilt and shame (Gelbrich et al., 2016; Roschk & Gelbrich, 2014). Timing is also important, with evidence of a “goldilocks zone” within which compensation is the most effective cure for service failures (Hogreve et al., 2017).

Following calls for further research to examine how different employee behaviors contribute to the effectiveness of compensation (Gelbrich et al., 2016), the present study uses the concept of collaborative recovery. Grounded in the service recovery literature on collaboration in service recoveries (Arsenovic et al., 2019), we define this construct as the social process in which customers and companies interact and engage in organized activities to recover a failed situation for mutual benefit. In such situations, both the customer and the service provider take the responsibility for having the problem solved (Arsenovic et al., 2019). This kind of collaboration recently received considerable attention (Park & Ha, 2022; Van Vaerenbergh et al., 2018) and has been referred to as co-creation (Roggeveen et al., 2012), customer participation (Dong et al., 2008), or customer involvement (Guo et al., 2016). Interestingly, there is conflicting evidence as to whether firms should introduce collaboration to improve customer responses to recovery encounters (Van Vaerenbergh et al., 2018). Collaboration during the service recovery process can be harmful if there is a mismatch between the level of collaboration in service delivery and service recovery (Heidenreich et al., 2015), but can play a crucial role in improving customers’ beliefs regarding an outcome (Dong et al., 2008; Hazée et al., 2017; Roggeveen et al., 2012). Despite these mixed findings, several studies have shown that collaboration in service recoveries is positively linked to customers’ emotional responses, both in terms of increased positive post-recovery responses and mitigation of the negative emotional responses commonly associated with service failures (Gohary et al., 2016; Park & Ha, 2016; 2022). Thus, as depicted in our conceptual model (Figure 1), we hypothesize:

  • H3: The mediating influence of emotional responses on the relationship between compensation and bad-mouthing behavior is moderated by the degree of collaborative recovery in the service encounter, such that customer emotional responses mediate the compensation–bad-mouthing behavior link in service encounters characterized by high but not low levels of collaborative recovery.

3 Empirical Study

3.1 Method

After pre-testing and fine-tuning of questionnaire items in a pilot study of university students (N = 188), we initiated a collaboration with a global retailer and distributed an online survey through its website. In this cross-sectional study, countries served as a strategic sample, representing a diverse set of customers. To ensure a representative global sample, customer data were gathered from Belgium, China, Croatia, Germany, Italy, Japan, Russia, South Korea, and the UK.

Customers who visited the retailer’s website were asked whether they had experienced an unfavorable event during the last six months (no/yes), and whether they had voiced this directly to the firm (no/yes), enabling the provider to respond to the service failure. The six-month timeframe reflects the interval typically adopted in prior service recovery studies (Voorhees et al., 2006; Wallin-Andreassen, 2000), and the retrospective approach is also typical in this research area (Grégoire and Fisher, 2006). If customers responded yes to both questions (i.e., voiced the complaint and went through the recovery process), they were invited to complete the full survey. This yielded a final sample of 867 customers who had had an unfavorable experience and voiced their complaint to the provider. The sample was split between male (41%) and female (55.2%), customers, along with 3.8% who preferred not to indicate their gender. Demographically, most respondents were employed (78.9%) and were in their thirties (36.3%) or forties (27.1%).

To increase the internal validity of the study, customers were first asked to indicate the level of severity the service failure caused them. We adapted three items measuring severity (Grégoire and Fisher, 2006; Grégoire, et al., 2009; Maxham & Netemeyer, 2002) using 7-point semantic differential scales regarding the extent of problem, inconvenience, and aggravation stemming from the service failure (α = .89). Next, participants indicated whether they had received any compensation during the recovery encounter (no/yes), followed by items measuring collaborative recovery and negative emotional responses. Collaborative recovery was measured using nine items (α = .94) inspired by the discourse on collaborative consumption (Ozanne & Ballantine, 2010; Möhlmann, 2015) and grounded in a discussion between the researchers and the firm. Items ranged from 1 (strongly disagree) to 7 (strongly agree); see Table 1 for details. Using a similar response format (1 = not at all; 7 = very much), negative emotional responses were measured through three well-established negative arousal items (α = .93) frustration, annoyance, and anger (Russell 1980), which we adapted from Schoefer and Ennew (2005). Finally, to capture the outcome variable of bad-mouthing behavior, participants indicated whether they had told their friends and/or family about the service failure (no/yes).

4 Results and Discussion

To test whether the presence (vs. absence) of compensation was associated with bad-mouthing behavior (H1), we performed a Pearson’s χ2-analysis using 2 (compensation: no, yes) × 2 (bad-mouthing: no, yes) crosstabs. The analysis revealed a statistically significant association between the variables, χ2 (1, N = 867) = 8.41, p = .004, V = .10. In line with H1, the level of bad-mouthing behavior was significantly lower among customers who had received compensation (76.8%) than among those who did not (85.7%). Compensation group descriptors are provided in Table 2.

To examine the H2 and H3 regarding customers’ emotional responses and the role of collaboration in the recovery process, we conducted a moderated mediation analysis (PROCESS Model 7; Hayes, 2017). We used compensation (no/yes) as the predictor; the index of customers’ emotional responses as the mediator; the index of collaborative recovery as the moderator; and customer bad-mouthing behavior (no/yes) as the dependent variable. According to H2, customers’ emotions serve as a mediator between compensation and bad-mouthing behavior.Footnote 2 The findings support this hypothesis by demonstrating that the presence (vs. absence) of firm-offered compensation has a mitigating effect on customers’ emotional responses to the service recovery (β = -. 82, t = 4.40, p < .001), and that these responses are significantly linked to bad-mouthing behavior (β = .43, Z = 9.24, p < .001). These results confirm that compensation diminishes customer bad-mouthing behavior through mitigated negative emotional responses.

Importantly, the link between compensation and negative emotional responses was moderated by the level of recovery collaboration (b = -.30, t = -3.12, p = .002). A floodlight analysis (Spiller et al., 2013) was performed to gain further insights into the nature of this moderation. Mediation by negative emotional response becomes statistically significant at a mean-centered value (commonly referred to as the Johnson-Neyman point) of -1.08 (t = -1.96, p = 0.050; see Figure 2). This means that compensation is useful only when the mean-centered value of collaborative recovery is equal to or above -1.08, corresponding to a mean value of M = 3.37 on the 7-point scale used. In the present case, the company failed to mitigate the negative emotional responses of almost a third (29.5%) of the customers who received compensation without sufficiently high levels of collaboration (i.e., when the collaboration mean was lower than M = 3.37). To assess the moderated mediation effect, we used a bootstrap procedure that generated a sample size of 5,000. Consistent with H3, the index of the moderated mediation was significant (95% CI = [-.24, -.03]). In other words, high but not low levels of collaboration during the recovery process diminished customers’ negative emotional responses, which ultimately reduced the likelihood of bad-mouthing behavior, thus confirming the adverse consequences of failing to emphasize collaboration in recovery contexts.Footnote 3

5 Theoretical and Managerial Implications

This study makes two critical contributions to service recovery research. First, by using real customer data, it addresses the effects of compensation under ecologically valid conditions as called for by Grewal et al. (2008). Our data source allowed us to capture emotional responses more realistically than in scenario-based studies, where customers only imagine themselves in a situation (Grégoire and Fisher, 2006). While confirming that compensation mitigates customer bad-mouthing behavior, a key finding is that this link is mediated by the customer’s negative emotions.

Second, building on the emerging literature on collaboration in service recovery situations (Arsenovic et al., 2019), this study measures the actual level of collaboration rather than manipulating collaboration as in earlier studies (Dong et al., 2008; Hazée et al., 2017; Roggeveen et al., 2012). Our alternative approach thus enabled us to determine the level of collaboration at which compensation as a recovery tactic effectively mitigates negative customer responses, thereby reducing bad-mouthing behavior.

The current findings demonstrate that the mitigating effect of compensation on customers’ negative emotional responses in recovery situations depends on collaboration. While compensation can influence key customer outcomes (Wirtz & Mattila, 2004), we show that compensation should not be used as “quick-fix” solution to recover customers in distress. Rather, we find that compensation has no effect on a large chunk of customers (almost a third) in terms of their emotional responses and their subsequent bad-mouthing behavior because the level of collaboration was deemed insufficient. These findings highlight the need for firms to view recovery situations as a social process, in which employees and customers interact and engage in organized activities to recover a failed situation for mutual benefit. The combination of compensation and collaboration is prerequisite for mitigating negative consumer responses. The use of customer bad-mouthing behavior as a proxy for recovery effectiveness enabled us to contextualize the NWOM construct. Unlike most related research that has relied on customer intention to disseminate negative attitudes regarding the company (Casidy & Shin, 2015; Hocutt et al., 2006), the concept of bad-mouthing behavior is arguably a more contextually appropriate measure that does not reflect customers’ overall evaluation of a brand, but only their evaluation of a particular recovery encounter as manifested in an actual behavioral response.

In conclusion, this article reveals a link between firm-offered compensation and customer bad-mouthing behavior and documents the emotional process behind this relationship. These findings have both theoretical and managerial implications in terms of the importance of customer–employee collaboration in service recovery encounters. An appropriate level of collaboration is needed if compensation is to be effective as a recovery tactic, as low levels of collaboration appear to undermine the mitigating effects of firm-offered compensation. Thus, employees need to focus on interpersonal aspects in service recovery situations, as a failure of collaborating with customers may make compensation both costly and counterproductive.

6 Limitations and Future Research

The research approach used to analyze the impact of compensation on service recovery has some limitations. As the study design limits causal inferences of the relationships between our theorized paths, future research should manipulate both compensation and collaboration by employing experimental studies in field, online, or laboratory settings. Nevertheless, our collaboration with a global international retailer ensured ecologically valid conditions, as called for by several scholars (Grewal et al., 2008; Otterbring, 2021). In addition, while scenario-based experiments are common in recovery research (Roggeveen et al., 2012; Dong et al., 2008; Hazée et al., 2017), studying real customers enabled us to capture emotions more accurately than in scenario-based experimental designs (Grégoire and Fisher, 2006; Otterbring, 2017). Finally, as the present study did not distinguish between different types or levels of compensation, future research should take such aspects into account. Some studies suggest that too much compensation can have a negative effect (Gelbrich et al., 2016; Gelbrich & Roschk, 2011); however, it would be useful to explore how the type and amount of compensation as well as the level of collaboration may serve to optimize the effectiveness of service recovery efforts.