1 Introduction

Service providers are using various relationship marketing strategies to maintain good relationships with consumers to cultivate loyal consumers which in turn leads to increased sales and profits (Olavarría-Jaraba et al. 2018). To foster long-term relationships with the customers and to mitigate the negative consequences after a service failure, companies that are consumer-oriented and have a long-term perspective generally undertake service recovery actions to increase customer satisfaction and promote customer loyalty and other positive behavioral outcomes, such as word of mouth (Cai and Qu 2018; Kenesei and Bali 2020). However, literature reported contradictory findings regarding relationship strength in service failure encounters. Customers in high-quality relationships are tolerant of service failure and recovery encounters known as the “buffering effect” (Ha and Jang 2009). Recent studies challenged this well-established notion, suggesting that there might be a “backfiring” effect. Customers in a strong relationship with the service provider feel betrayed and seek revenge after a service failure (Grégoire et al. 2009).

Some other studies investigated the moderating role of commitment type on perceived justice and recovery satisfaction. For instance, Ozkan Tektas (2017) found that while emotional commitment mitigates the impacts of procedural and interactional justice on recovery satisfaction, calculative commitment amplifies the influence of distributive and procedural justice. Further, scholars suggested that the effectiveness of relationship marketing investments depends on wide range of factors, such as relational characteristics (Balaji 2015). Therefore, it appears that customer’s perceptions of service failure and recovery outcomes vary considerably, even though their evaluation of the relationship is highly reliable.

Further, recent research indicates that customers’ interactions with businesses are shaped by relationship norms that are the expectations that relationship partners (e.g., customers and businesses) have of one another about how they should act (Aggarwal and Larrick 2012). In the present environment, customers’ perceptions of brands are shaped by their conformity to or violation of these relationship standards after service failure experiences. To fill the above-mentioned gap, this study identifies two kinds of relationships: communal relationships which are characterized by mutual trust and offering help in need and exchange relationships that are centered on quid pro quo.

Although satisfaction with service recovery has been studied extensively in service recovery literature, the relationship between satisfaction and loyalty is more complex than initially thought (Mittal 2016; Mittal and Kamakura 2001). Research shows that the relationship between satisfaction and loyalty is tenuous as satisfied customers need not be loyal or dissatisfied customers need to be disloyal (White et al. 2013). For example, Reichheld (1993, p. 71) holds that “between 65 and 85 percent of customers who defect say they were satisfied or very satisfied with their former supplier.” Therefore, scholars have called for a fresh stream of research to address these issues on alternative emotional-mediating mechanisms, such as discrete positive emotions between service recovery strategies and behavioral outcomes (Lastner et al. 2016). However, most research focused on a composite form of emotions (i.e., positive affect) instead of discrete positive emotions (Chebat and Slusarczyk 2005; So et al. 2015; Kranzbühler et al. 2020). Although this limitation has been partially addressed by Schoefer and Diamantopoulos (2008), their proposed scale measured dimensions such as involvement and concern but did not measure specific emotions. As the traditional relationship marketing activities offered little differentiation, current research has turned their attention to understanding discrete emotional mechanisms, such as gratitude (Fazal e Hasan et al. 2014).

Although researchers are investigating gratitude for a considerably long time, scholars from multiple disciplines started studying the role of gratitude in social and business contexts. Bridger and Wood (2017) attempted to show that the psychological mechanism of gratitude in the consumer context is similar to how it operates in individual relationships and argued for the inclusion of gratitude in relationship marketing to influence positive behavioral responses. Therefore, gratitude has been recognized as an important construct and attracted considerable attention from scholars from diverse fields and needs to be further investigated.

The purpose of this study is to bridge the above-identified gaps by drawing upon appraisal theory. We propose that customers’ perceptions of the recovery efforts depend not only on the relationship length but also on relationship type. Also, the feelings of gratitude manifest because of outcome desirability and controllability of the other party’s actions, but the type of relationship (exchange vs communal) moderates this outcome. Further, we investigate the role of gratitude in mediating the recovery effort and word of mouth. Essentially, our work adds to the above-mentioned literature and seeks to address the following two research problems:

R1

Are customers in a communal relationship more sensitive to perceived effort than customers in exchange relationship?

R2

Does gratitude mediates the recovery effort and the desired behavioral intentions?

Through this study, we contribute to the literature and practice in several ways. We extended cognitive appraisal theory by relationship type, showing that customers in a communal relationship are more sensitive to perceived effort than exchange relationship customers, especially in high outcome desirability. Second, we also heed the calls from several researchers for studying positive discrete emotions such as gratitude in service recovery encounters.

The rest of the paper is organized as follows. The next section presents a brief review of the literature on gratitude and its antecedents based on cognitive appraisal theory and the related hypotheses. The ‘methods’ section presents our experimental study designed to test the hypotheses. The findings are discussed in the ‘results’ section. Subsequently, the results are discussed in the ‘discussion’ section. Finally, the ‘conclusion’ section presents the contributions and acknowledges the limitations and scope for future research.

2 Review of literature and hypotheses development

According to Lawler (2001), emotions are generated when people exchange goods and services and these emotions experienced by the participants in the current exchange show significant effects on the future exchanges and the relationships. Suppose the exchange has caused positive-valanced feelings, such as gratitude. In that case, it is reasonable to assume that the relationship becomes stronger and leads to future exchanges. If the exchange has caused negative-valanced feelings, such as anger or disappointment, their willingness to participate in future exchanges will come down and the relationship will become weaker. Therefore, considering the emotions generated during exchanges is a potential ground to predict the participants’ future behavioral outcomes.

2.1 Cognitive appraisal theory

Scholars suggested that cognitive appraisal theory can investigate the role of emotions in marketing contexts (Bagozzi et al. 1999). Cognitive appraisal theory (hereafter CAT) explains the coping responses of individuals during stressful incidents. It posits that emotions are elicited when individuals appraise the environment related to their well-being (Lazarus 1991; Folkman and Moskowitz 2004). Put it differently, it is the individuals’ assessment of the event that elicits the emotions, not the event itself. According to CAT, a person’s response to stimuli goes through a “cognitive–emotion–behavior” sequence, which may be applied to service failure and recovery incidents. Emotions arise when customers evaluate the actions taken by the service provider to rectify the failure are fair or not (Xu et al. 2019). Appraisal theory can also explain the positive discrete emotions generated during the service recovery process as the interplay between different attributions leads to discrete emotions (Lastner et al. 2016). Discrete emotions are evoked when outcome desirability is combined with other appraisal dimensions. Although CAT is widely applied, most studies focus on studying discrete negative emotions after service failures (Luo and Mattila 2020).

Although scholars disagreed on the number of dimensions in an appraisal process (Watson and Spence 2007), some commonly recurring appraisals are identified, such as outcome desirability, controllability, certainty, and agency. “Outcome desirability refers to the initial cognitive appraisal of whether the outcome of a situation is good or bad (positive or negative) with respect to personal well-being” (Watson and Spence 2007, pg. 491). Agency refers to who is responsible for the outcome. Suppose an employee is responsible for the desired outcome, feelings of gratitude manifest. If a customer perceives that his actions are responsible for the outcome, feelings of pride arise. Controllability refers to the outcome received by the customer is in the control of the service employee or not. A customer may receive superior recovery after a failure due to organizational policies rather than a deliberate employee effort.

2.2 Gratitude

Gratitude is defined as “the emotion that arises when an individual (beneficiary) perceives that an exchange partner (benefactor) (e.g., person or organization) has intentionally acted to improve the beneficiary’s well-being” (Raggio et al. 2014, p. 4). Gratitude is positive-valanced emotion that correlates positively with other positive emotions and occurs when a person receives something and is prompted to reciprocate. In marketing, customer gratitude is defined as a customers’ obligation to reciprocate the benefits received by the seller (Wetzel et al. 2014). Besides, gratitude is high when a beneficiary feels that the benefit received by him is costly to the giver and the giver acted beyond the call of duty (Tsang 2006). Reciprocity is a topic of substantial interest for relationship marketing researchers and research demonstrates that reciprocity, expressed as a desire to repay a benefactor, is a significant component of conceptualizing gratitude in the consumer experience (Bock et al. 2016).

Gratitude is known to mediate the firm’s relationship marketing investments and positive behavioral outcomes of customers. Morales (2005) points out that customers recognize the firms’ efforts and exhibit feelings of gratitude through the willingness to pay more. Soscia (2007) found that gratitude is positively related to repurchase intention and positive word of mouth. Palmatier et al. (2009) showed that gratitude mediates marketing efforts and desirable behavioral outcomes such as word of mouth in a similar vein. Wetzel et al. (2014) found that gratitude is a critical element in a firm’s prioritization strategies in enhancing sales and profit. In the context of virtual health communities, Mpinganjira (2019) found that gratitude mediates social capital and willingness to reciprocate. Provided that feelings of gratitude are elicited through the benefits received and controllable behaviors of the other party, Service recovery incidents are ideal ground for studying this emotion. As stated by researchers, a buyer’s heightened need gets converted into a breeding ground for gratitude, and service recovery intentions lead to the development of gratitude.

The sequence of constructs satisfaction and gratitude occurred differently in many studies. Kim and Lee (2013) showed that satisfaction and gratitude are parallel mediators. In a qualitative study conducted by Raggio et al. (2014) respondents are differed in their opinion. “However, we find mixed results regarding its temporal relationship with satisfaction; nearly half believe gratitude precedes satisfaction, while the others believe satisfaction precedes gratitude. Following is an example of the latter.” Lastner et al. (2016) tested their model with satisfaction as a consequent to gratitude. We operationalized gratitude in our model as a consequence of specific combinations of antecedents.

2.3 Outcome desirability

Outcome desirability in the service recovery context is how close the desired outcome is reached. When customers perceive that the outcome they wanted is not achieved, they experience a wide range of emotions and complain to the service provider. If the service recovery is successful, they may experience feelings of gratitude and show positive behavioral responses. According to appraisal theorists, customers experience higher levels of gratitude if the need to achieve the desired outcome is high and met (Soscia 2007).

Although outcome desirability and distributive justice appear to be related and represent the same construct, significant differences exists between the two as these two constructs are originated from two different theories: Cognitive appraisal theory (CAT) (Weiner 2000) and Justice theory (Adams 1965). To explain the specific emotions felt in service encounters, CAT is widely used. Here, outcome desirability is the primary appraisal. People first check the outcome is in favor or not. Then they go for secondary appraisals. Based on the different combinations of the appraisals, discrete emotions are felt by the people. On the other hand, distributive justice deals with the monetary compensation offered by the firm in the process of service recovery. The majority of studies in service recovery investigated emotions based on valence: positive emotions and negative emotions.

Outcome desirability is considered critical in generating feelings of gratitude. This dimension is considered the most important of all appraisals and, in some studies, account for up to 88% of the variance (Ruth et al. 2002). Customers experience gratitude when the outcome is in the expected direction of their expectations. Therefore, we hypothesize that,

H1

Outcome desirability (low vs. high) has a positive influence on customer’s gratitude.

2.4 Perceived employee effort

Dominant framework in compliant handling is Justice theory (Adams 1965), which focuses on distributive justice (i.e., compensation), interactional justice (i.e., how employees treat customers), and procedural justice (i.e., policies and procedures). Although perceived effort is considered as a part of procedural justice, scholars have recognized this as a distinct construct (Cambra-Fierro et al. 2015; Huang et al. 2017; Achabou 2020). Perceived effort is defined as “the amount of energy an observer believes an actor has invested in a behavior” (Mohr and Bitner 1995, p. 240). A distinct feature of effort is persistence. Someone persisting in achieving something for a long time is perceived as effortful. According to attribution theory, people often consider ability and effort while explaining success or failure (Weiner 2000). In the short term, ability is considered an uncontrollable factor, whereas effort is considered controllable. Therefore, attributing positive outcomes to controllable factors lead to positive emotions such as gratitude, pride, and negative outcomes to controllable factors, leading to negative emotions, such as anger and shame (Soscia 2007).

In general, people have feelings of gratitude and reward those who put extra effort in fulfilling their responsibilities and penalize those who failed to uphold them. During the service recovery process, human interactions (i.e., between customer and service provider) are crucial, and perceived employee effort is an essential element in customers’ evaluation of service recovery. Maxham (2001) shows that perceived effort is a crucial antecedent in the complaint handling process and influences customer satisfaction. Therefore, companies must take enough care to ensure that the customer’s effort during complaint handling is felt (Mattila and Patterson 2004). Customer’s perceptions of the genuine employee effort exerted during the service recovery process lead to favorable fairness judgments. The research found that the customer’s perceptions of employee effort and justice perceptions are directly related (McQuilken et al. 2013).

Although, perceived effort is a crucial factor in generating feelings of gratitude, several studies have shown that consumer’s perception of employee effort depends on a wide array of factors. Mohr and Bitner (1995), using the data from their critical incident technique, found that the relationship between satisfaction and perceived effort is malleable. Although perceived effort explains large part of the variance in satisfaction, there exist contrarian cases where perceived effort is high but satisfaction is low. For instance, if an employee is taking an unusual amount of time to accomplish a task, consumers may perceive that the employee is inept. Similarly, consumers may attribute that the employee has not exerted adequate effort if the outcome is not in their favor.

Customers appreciate staff who expend extra effort to solve problems (Johnston 1995). When less desirable outcomes are received, customers try to appraise the event to understand the failure (Duhachek 2005). Effort indicates motivation and motivation is considered favorably by consumers (Mohr and Bitner 1995). Bell and Zemke (1987) observed that customers attribute good intentions to perceived effort, although the problem is not fixed. In a similar vein, Mohr and Bitner (1995) observed that if customers notice that the employee is underpowered to fix the problem, they may at least value the employee’s dedicated effort. Customers notice the effort put in by the employees even if the desired outcome is not in favor. Consider the following excerpt from the exploratory study conducted by (Raggio et al. 2014, p. 16).

For example, my insurance rates just went up [. . .]. The 20 percent increase I just got, [my broker] can’t control that, but he tried to run the numbers another way, tried several things [. . .]. But he could not get it to work out any way he ran it [. . .]. I was grateful, and I expressed gratitude for trying.

Based on the above discussion, we hypothesize that,

H2

Perceived effort (low vs. high) has a positive influence on customer’s gratitude.

2.5 Exchange and communal relation type

Scholars argue that customers vary in their needs to form relationships with firms and a firm’s relationship marketing efforts should match their preferences (Jones et al. 2015). Recent research (Wan et al. 2011; Aggarwal and Larrick 2012) in consumer psychology shows that two types are relationships—communal and exchange regulated by precise norms and helps explain consumer behavior in relationship marketing. Communal relationships are characterized by genuine care for the partner’s needs and do not expect any return for the favors given. Therefore, the norm in this case is “... is to give benefits in response to needs or to demonstrate a general concern for the other person. In [these] relationships, the receipt of a benefit does not change the recipient’s obligation to respond to the other’s needs.” On the other hand, exchange relationships are functional and based on “quid pro quo” (Clark and Mills 1979; Clark and Mils 1993; Wan et al. 2011). That is, benefits are given “with the expectation of receiving a comparable benefit in return or as repayment for a benefit received previously” (Clark and Mils 1993, p. 684). While persons involving community relationships usually return the favors they got, their reciprocity is usually driven by feelings of appreciation instead of feelings of obligation (Wan et al. 2011).

Thus, the norms which govern these two relationships differ from each other. Impersonal norms rule exchange relationships. The customers’ expectations in this relationship are that the services they purchase should have equivalence in value and be compensated reasonably for failed services. Similarly, customers who stay in hotels expect to have good hospitality and leave a tip during their stay. In contrast, individuals in communal relationships keep track of each other’s needs and attend to their needs independent of the benefits they got in the past (Clark et al. 1987). Violating these norms is considered objectionable and leads to adverse reactions.

Customers in exchange relationships focus on the input and output ratios, and all they expect is a desirable outcome (Hur and Jang 2016). Therefore, we hypothesize that exchange customers are not as sensitive as the communal customers to the employee effort when outcome desirability is high. Contrarily, communal customers elicit less gratitude in low effort scenario than exchange customers as they perceive that the outcome is not because of the employee’s determined effort. Therefore, based on the above discussion,

H3a

When outcome desirability is low, no significant difference is found between customers’ feelings of gratitude in communal and exchange relationships.

H3b

When outcome desirability is high, customers in communal relationship have lesser feelings of gratitude than exchange relationship customers in low effort condition.

2.6 Word of mouth and mediating role of gratitude

In the consumption context, word of mouth is defined as “information a customer provides to others concerning the consumption of a product or service” (Leisen Pollack 2017, p. 514). Word of mouth may be particularly damaging if someone thinks the service provider wronged them. Despite this, customers who had excellent service recovery are more inclined to recommend to others. Individuals often feel high emotions during consumption events and these events are likely to have a substantial influence on the emotional link that consumers have with the organization. As a result, customers whose emotional ties become stronger as a result of service recovery, indicating increased customer attachment, are more likely to engage in positive word of mouth. In contrast, customers whose emotional ties weaken as a result of insufficient compliant handling are more likely to engage in negative word of mouth (Choi and Choi 2014).

Watkins et al. (2006) posited that gratitude has a positive valence since it is positively associated with many other positive emotions and negatively associated with negative emotions. Gratitude is defined as a sense of respect or admiration for a person for anything (such as actions, gifts) (Bock et al. 2016). One way for appreciative consumers to demonstrate thanks to a service provider is to spread positive word of mouth. Studies have shown that gratitude promotes positive word-of-mouth behavior (Soscia 2007; Xie et al. 2015; Bock et al. 2016). Further, in a recent study, Kim and Park (2020) showed that gratitude reduces customers’ willingness to take vindictive actions such as spreading negative word of mouth against the wronged business. Consumers who have received help from the employees of an organization display their gratitude by continuing their relationship with the firm and acting as champions for the organization through positive word of mouth. However, in a study conducted in a compliant handling scenario, the relationship between gratitude and word of mouth is insignificant (Simon et al. 2015).

The meta-analysis by Orsingher et al. (2010) showed that recovery efforts of the firms influence subsequent downstream behavioral outcomes, such as word of mouth. In addition, Morales (2005) showed that gratitude mediates firm efforts and consumer rewards. Therefore, given the positive relationship between gratitude and word of mouth, we hypothesize that the indirect effect of recovery efforts on word of mouth is expected through gratitude. Therefore, to further investigate this relationship, we hypothesize that,

H4

Gratitude will mediate the word of mouth and the three-way interaction between outcome desirability, perceived effort, and relationship type.

3 Method

To test the proposed hypotheses, we conducted an online experiment with a 2 (relationship norm types: exchange vs. communal) × 2 (perceived effort: low vs. high) × 2 (outcome desirability: low vs. high) between-subjects factorial design. The experiment was designed in Qualtrics software, and we sent a survey link to 600 students and alumni of a large eastern Indian university after taking necessary approvals. This university runs various graduate programs, executive programs, and certificate courses catering to a wide range of students and professionals from the industry. Conducting experiments using student samples is widespread in psychology and marketing domains (Refer Ashraf and Merunka (2017) for an insightful discussion). Further, the university charges a higher tuition fee compared to other universities. Therefore, the majority of students studying in this university apply for education loans. Another reason for choosing the respondents is sample and stimuli fit. Ashraf and Merunka (2017) suggested choosing an appropriate sample so that there are no validity issues. The reason behind selecting students and alumni is that students often apply for study loans with banks and alumni also apply for bank loans for various purposes. Most of the students have work experience, and the average experience of students taking admission for MBA here is 3–4 years. These students were part of the industry before taking admission. They have had sufficient experience as customers of the banking industry and would be well versed in banking services. Many of them avail of loan facilities for funding their MBA education. The differences in lifestyle would not differ dramatically to influence the results of the study. Further, to ensure equality of variance across the experimental groups, we conducted Levene’s test to establish homogeneity of variances.

We have chosen banking industry in this study as the competition in this industry has reached new heights with rising customer acquisition costs, heightened customer expectations, and a high rate of customer switching (YuSheng and Ibrahim 2019). As the customers’ interactions with the banks have increased, a deeper understanding of bank–customer relationships is vital to bank success (Durkin and Howcroft 2003; Kaur et al. 2012). Further, past studies have investigated the role of emotions in banking sector (Loureiro and Sarmento 2018).

All participants were randomly assigned to each of the eight experimental conditions. Of the received responses, we dropped 12 respondents who failed to pass attention check and left with 408 completed observations with a response rate of 70%.

The demographic details of the participants are shown in Table 1.

Table 1 Demographic details
Full size table

3.1 Experiment design and pre-test

There are a total of 8 cells [2 (communal vs. exchange) × 2 (outcome desirability low vs. high) × 2 (perceived effort low vs. high)]. In an experimental setting, these relationship norms, communal or exchange, could be made salient through priming technique [e.g., experimental participants read a scenario that describes how a consumer interacts with a brand, and they are asked to assume the role of the consumer in the scenario (Aggarwal and Larrick 2012)].

We conducted several pre-tests to check whether the scenarios are working as intended. First, a pre-test has been conducted to eliminate any possible confounds concerning relationship type manipulation. Fifty participants who are not part of the main study participated in this study with an average age of 23.2 years (SD = 3.4) and 45% are female. Participants are randomly assigned to the relationship norm scenarios and answered several manipulation check questions used in the main experiment. We reverse coded the exchange relationship type items and combined them with the first two items, and averaged all four items to get a net relationship score. The results indicate successful priming of participants. Specifically, participants in the communal relationship (M = 4.19, SD = 1.1) are significantly different to exchange relationship participants (M = 3.12, SD = 1.22), F(1, 49) = 11.37, p < 0.001. Further, the manipulation did not show any impact on the scores of perceived service quality (4.8 vs. 4.6, p = 0.62), service involvement (5.12 vs. 4.99, p = 0.71), and mood (5.8 vs. 5.6, p = 0.78), ruling out any potential confounding effects.

Second, we interacted with two loan officers at a local bank to find reasonable loan sanction amounts in low and high outcome desirability situations. We were told that, in general, customers get 80% of the total loan amount they applied for. We fixed 80% as the high outcome desirability condition and 50% as the low outcome desirability condition. Other inputs given by the pre-test participants are used to improve the experimental design.

3.2 Measures and manipulations

In the experiment, we told the respondents to imagine that they are customers of a fictitious bank, “Right Bank.” We used this name to remove any confounding effects from participants’ prior experience with banks. We exposed participants to one of the carefully constructed relationships (exchange vs communal) scenarios to manipulate the relationship type. The scenarios were adopted from extant literature (Aggarwal 2004; Aggarwal and Larrick 2012) and necessary changes were made to fit the banking scenario. In exchange relationship type, the bank was depicted as “an ideal business partner” that “provides you a fair value for your money.” In communal relationship type, the bank was described as “a caring community member and partner… who has a mission to help people, build relationships, and promote mutual trust.”

After that, the participants answered manipulation check questions intended to measure relationship type. Items for communal relationship type are “help in times of need” and “you’d miss them if you moved away” (Cronbach’s alpha = 0.846) and for exchange relationship are “good value for money” and “you get money’s worth” on a seven-point scale (Cronbach’s alpha = 0.834).

We developed scenarios adopted from the previous research to manipulate the perceived effort (low vs high). In the low effort scenario, the employees are shown as a little unconcerned and asked the customer to complain about the complaint management system. In the high effort scenario, the employees are described as very concerned about the problem, and they involve personally to take necessary actions. Then participants answered two manipulation check items that are “the employee exerted a lot of energy” and “the employee was very persistent” on a seven-point scale (Cronbach’s alpha = 0.861).

Then participants are shown two levels of their loan approval amounts (50% and 80%) and answered a single item manipulation check “the resolution to my complaint regarding loan approval is” on a seven-point scale (1 = Not at all satisfactory ~ 7 = highly satisfactory).

Word of mouth is measured using a three item scale with items such as “I will recommend this bank to someone who seeks advice,” “I will encourage friends and relatives to do business with this bank,” and “I will say positive things about this bank to other people” on a seven-point scale adapted from extant literature (Moliner-Velázquez et al. 2015) (Cronbach’s alpha = 0.812).

Gratitude is measured using three item scale adapted from (Xia and Kukar-Kinney 2013) with items “I feel grateful to the bank,” “I feel thankful to the bank,” and “I feel appreciative of the bank” on a seven-point scale.

Also, participants answered four realism measures, and the results (M = 5.3) confirmed that they perceived the scenarios are realistic (7-point scale) (Gelbrich et al. 2015) (Cronbach’s alpha = 0.81).

3.3 Manipulation checks

We conducted three analysis of variance (ANOVA) tests to check whether the intended manipulations are successful or not. We submitted the manipulation scores as dependent variables and relationship type, perceived effort, and outcome desirability as independent variables. As expected, participants in the communal relationship (M = 4.46, SD = 1.36) scored higher than exchange relationship [M = 3.75, SD = 1.79; F(1, 400) = 18.66, p < 0.01].

To check if perceived effort manipulation is successful, we conducted another three-way analysis of variance (ANOVA) with perceived effort manipulation check as dependent variable and relationship type, perceived effort, and outcome desirability as independent variables. As expected, respondents in the low effort (M = 2.73, SD = 0.98) reported low perceived effort than the respondents in the high effort condition [M = 4.41, SD = 1.56; F(1, 400) = 166.99, p < 0.05]. Other main effects and interactions are not significant.

Next, to check whether outcome desirability manipulation is successful, we ran another three-way analysis of variance with outcome desirability manipulation check score as dependent variable and relationship type, perceived effort, and outcome desirability as independent variables. As expected, the respondents who received 50% of the loan (M = 2.68, SD = 0.84) reported low desirability compared to the respondents who received 80% of the loan amount explanation [M = 4.38, SD = 0.86, F(1, 400) = 413.27, p < 0.01]. Other main effects and interactions are not significant.

4 Results

Levene’s test of equality of variances has been conducted before the ANOVA to ensure homogeneity between the groups and found to be insignificant F(7, 400) = 1.93, p = 0.064. Therefore, homogeneity rule is not violated. Also, Confirmatory factor analysis has been conducted using Amos 22 software to check reliability and validity of the scales.

To test the hypotheses related to the effects of relationship norm, perceived effort, and outcome desirability, the two dependent variable, gratitude, and a three-way analysis of variance (ANOVA) were used. Table 2 presents the mean differences for the independent variables. The results of the ANOVA test are shown in Table 3.

Table 2 Measurement model
Full size table
Table 3 Mean differences
Full size table

In a study conducted by Agarwal (2004), main effect for relationship type on brand evaluation is not significant when customers are charged some fee for the services provided by the bank. They found that interaction effect is significant as customers in communal relationship viewed as it is a violation of norms. In a study conducted to assess the effectiveness of the presence of emoji, Smith and Rose (2020) found that neither relationship type nor emoji has main effect on affective reactions but interaction is significant. Therefore, as expected ANOVA on the main effect of relation type is insignificant. p = 0.478 (Refer Table 4).

Table 4 ANOVA results
Full size table

A three-way ANOVA on gratitude revealed a main effect for perceived effort [F(1, 400) = 60.81, p < 0.01]—Gratitude is higher when employees exert high effort (M = 4.86, SD = 1.26) than low effort (M = 3.99, SD = 1.3). Main effect for Outcome desirability is significant [F(1,400) = 289.31, p < 0.01] and Gratitude is higher when outcome desirability is high (M = 5.04, SD = 1.25) than low outcome desirability (M = 3.81, SD = 1.19). Thus, the formulated hypotheses H1 and H2 are supported. However, no such effect was observed with respect to the main effect of relationship type on gratitude. The three-way interaction is significant [F(1, 400) = 12.99, p < 0.05]—the pattern of one of the significant two -ay interactions is different at two levels of outcome desirability. Contrast analysis reveals that relation type × effort is significant at high outcome desirability [F(1, 400) = 13.40, p < 0.01] but not at low outcome desirability condition F(1, 400) < 1.

Therefore, we further probed the significant interaction. When outcome desirability is high, effort is highly significant for communal relation type F(1, 400) = 54.38, p < 0.01 than exchange relationship type F(1, 400) = 6.34, p < 0.05. In low outcome desirability condition, no significant difference is found between communal and exchange relationship respondents in high effort condition [F(1, 400) < 1] supporting H3a. When outcome desirability is low, effort is significant for both communal F(1, 400) = 32.76, p < 0.01 and exchange types F(1, 400) = 4.38, p < 0.01. In case of high outcome desirability and low effort, difference in gratitude for exchange relation type (M = 4.86, SD = 1.01) and communal relationship type (M = 4.15, SD = 1.17) is significant [F(1, 400) = 10.18, p < 0.01] supporting H3b (Fig. 1).

Fig. 1
figure 1

Gratitude as a function of relationship type, perceived effort, and outcome desirability

Full size image

4.1 Moderated mediation analysis

A moderated mediation analysis was conducted to determine gratitude mediates the three-way interaction of independent variables and word of mouth. We also seek to find whether the relationship between gratitude and word of mouth is significant or not (Fig. 2).

Fig. 2
figure 2

The results of a moderated mediation analysis

Full size image

First, we regressed word of mouth on all combinations of interactions between the independent variables and found that the three-way interaction coefficient was statistically significant [β = 1.08, t(400) = 2.01, p < 0.05]. Then we used the SPSS macro developed by Hayes (2012) to conduct a moderated mediation analysis with 5000 bootstrap samples. The relation between gratitude and word of mouth [β = 0.62, t(400) = 11.93, p < 0.05] is significant. Also the three-way interaction on gratitude [β = 1.19, t(400) = 2.67, p < 0.01] is found to be statistically significant. The three-way interaction on word of mouth lost its statistical significance, indicating full mediation (β = 0.34 p = 0.46). Further, the index of moderated mediation = 0.74, BootSE = 0.28, and the CI [0.2, 1.28] did not contain zero supporting hypothesis H4.

5 Discussion

Extant literature has identified the need for examining the role of positive emotions in influencing the relationship between service recovery strategies and customer behavior (Kranzbühler et al. 2020). Past research has focused primarily on the composite form of emotions and has excluded discrete positive emotions (Chebat and Slusarczyk 2005). Encouraged by the need to study the role of discrete positive emotions such as gratitude after service recovery, a scenario-based experiment was conducted to test the research hypotheses empirically. Besides this, the research also attempted to address the contradictory results about relationship strength’s role in buffering the negative impacts after service failures. It appears that customer’s perceptions of recovery outcomes vary considerably, even though their evaluation of the relationship is highly reliable.

The findings of the experimental study supported our hypotheses. The perceived effort has a significant influence on gratitude. The results of our study are in line with the earlier studies which recognized importance of perceived effort. Raggio et al. (2014) mentioned that the feelings of gratitude manifest when customers recognize the organizational efforts. However, when customers perceive that the efforts are self-serving and customers are forced to reciprocate then customer become cynical toward the firms. Soscia (2007) found that word of mouth and other behavioral intentions are predicted by gratitude rather than happiness. Kim and Lee (2013) found that relational benefits offered by the firm invokes feelings of gratitude. Their study focused on confidence, social, and special treatment benefits offered by the firm to the long-term customers. Our study broadens this idea that the customer who are in long relationship with the firm may be of two types and their perception of benefits and outcomes from the service encounters are determined by whether they are in exchange or communal relationship. Similarly, outcome desirability was found to have a positive impact on gratitude. Xia and Kukar-Kinney (2013) found that when penalties levied by the bank are lifted customers expressed gratitude toward the bank but perceptions of fairness moderate this relationship. However, Exchange and communal customers differ in their perceived effort in low and high desirability situations. According to attribution theory, when outcome desirability is low, customers search for the reasons that caused the outcome. Customers forgive the firms when they perceive that the employees exerted effort to avoid the negative outcome. Therefore, the perceived effort is significant for both outcomes on gratitude. When outcome desirability is high, but employees did not exert adequate effort, customers in communal relationships felt betrayed and their feelings of gratitude lessened. This is because they perceive that the outcome is not by a controllable factor but by chance. Therefore, their feelings of gratitude are lesser when compared to exchange relation customers, although outcome desirability is high. Using the appraisal theory as the theoretical base, we addressed the research gap by examining specific antecedents, leading to the discrete emotion—gratitude. The findings demonstrate that gratitude appears from outcome desirability and perceived effort, but the type of relationship (exchange vs. communal) moderates this outcome. The present study investigated the impact of different types of positive relationships exhibited by the customer in a bank’s context, responding differently to the bank’s recovery efforts.

6 Conclusion and implication

The present study contributes to research by examining the relationship type as a moderating variable to examine the relationship between perceived effort and outcome desirability and gratitude toward the service provider. We observed that relationship strength does not directly elicit positive emotions (e.g., gratitude), but plays a vital role in eliciting it through perceived effort and outcome desirability. The findings of the paper offer few implications for service providers. Most often than not managers are apprehensive about the impact of any service failure incident on customer’s future behavior with the firm and their loyalty. The results show that customers do not always look at service failure as a stand-alone incident. They tend to evaluate the overall situation before making any judgment toward the service provider.

6.1 Theoretical contributions

We contribute to the literature in three ways. First, we heed the call by several scholars to study positive discrete emotions after service recovery. The study has examined the influence of antecedent variables that develop gratitude in the customer’s mind in the banking industry. Most significantly, the present study establishes that recovery strategies can generate positive emotions that advance the relationship between the customer and the service provider.

Second, we extended the cognitive appraisal theory by including relationship type and demonstrating that relationship type differences play a vital role in eliciting differences in emotions. Specifically, we have shown that customers who are in a communal relationship with the service provider are more sensitive to perceived effort and lack of effort leads to lesser gratitude despite outcomes are in favor.

Third, we shed new light on the relationship marketing literature, showing that instead of considering the length of the relationship, differentiating customers based on exchange and relationship type helps to gain a deeper understanding of the relationships.

6.2 Managerial contributions

Service failure can be seen as a range of small and big incidents. For some, a loan that is not approved can be a service failure, whereas others’ even 20 percent rejection of loan can be a service failure incident. However, as observed in the present study, customers also look at an employee’s perceived effort. Whether or not the service provider and their employees have put sufficient effort in helping the customer. This outlook toward perceived effort can elicit gratitude among the customers even though the outcome might not exceed their expectations. Usually, individuals elicit feelings of gratitude and reward the service providers they see putting extra effort into providing customer service. However, they also punish them when they observe the employees not giving their best in serving the customers.

Managers must ensure that every employee in the organization gives their best while providing service to the customer. Since interaction between customer and service provider is crucial and perceived employee effort becomes an important element in customers’ evaluation of service recovery. Managers must ensure that customers can notice and feel the employees’ effort while providing service and complaint handling. On the flip side of displaying effort, managers must ensure that the customer should not feel that an employee’s extra effort is due to the employee’s incompetence or the employees are not adequately trained by the company. The same perceived effort can generate a negative trickle effect, and instead of eliciting positive emotion, it can elicit negative emotion toward the company.

Another significant aspect that managers need to care about is outcome desirability. Managers need to train their frontline service employees about setting up customer expectations. Whenever the employees communicate with the customers, they need to set expectations that they can fulfill later. Setting very high customer expectation and not fulfilling it, later on, can result in dissatisfaction among the customers. The results of the study note that customers will not develop the feeling of gratitude if the service provider initially sets very high unrealistic expectations.

Another aspect that the managers need to consider the relationship norms. What type of relationship is being exhibited—exchange relationships or communal relationships. The managers should strategize their recovery actions based on the type of relationship being exhibited. Exchange relationships are ruled by impersonal norms, while communal relationships track each other’s needs and attend to their needs independent of the benefits they got in the past. Frontline employees are to be trained sufficiently to identify the relationship type perceived by the customers. As the first point of contact, frontline employees provide valuable information to the managers about the customers. During the service recovery scenarios, displaying more empathy and genuine effort helps alleviate the customer's concerns who thought that they are in a communal relationship.

Notwithstanding the outcome of the recovery process, they tend to show feelings of gratitude and subsequent repurchase intentions. On the other hand, customers who perceive themselves in the exchange relationship care about the outcomes. According to Ringberg et al. (2007), customers applying the utilitarian cultural model are rational and expect compensation for the failure. Exchange customers usually fall into this head of cultural models’ theory; therefore, an appropriate compensation strategy helps to reduce the negativity associated with failure and helps in building long-term relationships with the firm.

7 Limitations and future research

This study has conducted a scenario-based experiment that has lesser external validity. So future studies can employ field studies and surveys. Second, we considered only gratitude as our dependent variable, but future studies should also consider other discrete emotions like hope. Third, future studies should also consider moderating personality factors such as chronic empathy and self-esteem to gain a deeper understanding of relations between customers and service provider. Fourth, we considered outcome desirability and perceived effort as antecedents to gratitude, and future studies should consider other variables, such as certainty.