Loyal customers’ tipping points of spending for services: a reciprocity perspective

Abstract

Purpose

Controversy exists about the shape of the relationship between loyalty and profitability. This paper aims to address the possibly nonlinear effects of behavioral loyalty (BLOY) on customer spending (as a proxy for profitability). Building on social exchange theory and the norm of reciprocity, it examines the asymmetries between BLOY and customer spending and the moderating influence of personal communication (PCOMM) as a social reward and dispositional positive reciprocity as process evidence.

Design/methodology/approach

Study 1a (n = 309) gathered customer data from four restaurants and Study 1b (n = 252) data from hotel guests after they checked out. Study 2 is an experimental study with two manipulated factors (BLOY and PCOMM). In total, 295 participants from a large German online panel completed the study.

Findings

The results indicate an inverted-U shaped relationship between BLOY and customer spending: after reaching a turning point, customers gradually curb spending as their BLOY further increases. High PCOMM acts as a reciprocal response while triggering additional customer spending particularly at higher levels of behavioral loyalty; positive reciprocity adjusts the differences in customer spending when social rewards such as PCOMM are present.

Research limitations/implications

The asymmetric relationship between BLOY and customer spending is tested only for hedonic service settings.

Practical implications

Not all loyal customers spend more – companies need to meet their reciprocal obligations before they can benefit from increased customer spending.

Originality/value

The present research re-considers the nature of the relationship between BLOY and customer spending and reveals an inverted-U shaped relationship, with a turning point beyond which greater customer loyalty decreases customer spending. It finds converging process evidence for the mechanism of reciprocity underlying this relationship. This study also details the financial impact of BLOY on the firm by investigating actual customer spending.

Citation

Teichmann, K. (2021), "Loyal customers’ tipping points of spending for services: a reciprocity perspective", European Journal of Marketing, Vol. 55 No. 13, pp. 202-229. https://doi.org/10.1108/EJM-10-2019-0781

1. Introduction

Business models of the service industry evaluate efforts to increase customer spending per transaction as central for a service firm’s success (Forbes, 2012). Restaurants, for instance, benefit from skilled servers that engage in conversations, thereby making customers spend more than they originally planned (WebstaurantStore, 2018). A customer’s increased spending may emerge from customer loyalty which comes at a cost for a company though. A customer engages in repeat purchase (i.e. behavioral loyalty, BLOY) in exchange for equitable rewards (Liu, 2007) seeking reciprocity with the company (Schultz and Bailey, 2000). Traditionally, loyalty programs establish such reciprocity (Kumar and Shah, 2004). Some companies cannot or do not want to launch loyalty programs though as these programs often fail to perform as expected (Bombaij and Dekimpe, 2020; Wang and Lalwani, 2019). Loyalty programs have also been criticized to potentially infringe on customer privacy (Stourm et al., 2020). Different to such institutionalized programs, this study explores alternatives for rewarding customers understanding that customer-company relationships involve social exchanges based on reciprocity (Henderson et al., 2011). In particular, this study considers moderating influences of personal communication (PCOMM) as a social reward and customers’ endorsed positive reciprocity between BLOY and customer spending. PCOMM captures a consumer’s perception of how kind, personal and warm conversations are with a company (i.e. service employee; De Wulf et al., 2001; Metcalf Lynn et al., 1992). Positive reciprocity is an individual’s tendency to exchange positive treatment for positive treatment (Fehr and Gächter, 2000). To test these predictions, two field studies and one experimental study conducted in two service contexts address the following questions:

RQ1.

What is the relationship between BLOY and customer spending?

RQ2.

To what extent do social rewards (i.e. PCOMM) and positive reciprocity influence the relationship between BLOY and customer spending?

In the widely accepted satisfaction–loyalty–profitability chain, customer satisfaction should increase customer loyalty (Anderson et al., 1994; Rust et al., 1995), which then leads to profitability (Homburg et al., 2009) or on the customer level, to increased customer spending per purchase (Reibstein, 2002) as a proxy for firm profitability (Bolton et al., 2004). Less than 10% of the studies discussing the satisfaction–loyalty–profitability chain actually examine the loyalty–profitability link (Watson et al., 2015) though. Recent research has started to revisit the proposed linear relationships in the loyalty–spending chain but came up with conflicting results (Thakur, 2019; Umashankar et al., 2017). For example, compared with less loyal customers, loyal customers may demand more price discounts (Wieseke et al., 2014) or assign more importance to price ( Umashankar et al., 2017), thus finding negative relationships between loyalty and profitability. Such conflicting results imply the possibility of a nonlinear relationship between BLOY and firm profitability. This study addresses this gap in the literature. In particular, whereas existing studies predominantly focus on the firm level (Petersen et al., 2018) and customer reward programs from large organizations (Homburg et al., 2009), the present research responds to calls for more research at the customer level and therefore proposes a nonlinear link between BLOY (as defined by repeat purchase behavior; Dick and Basu, 1994; Umashankar et al., 2017) and customer spending (as defined by amount spent per purchase).

This study contributes to the extant literature in several ways. First, the investigation of the nature of the relationship between BLOY and customer spending predicts an inverted-U shaped relationship, with a turning point beyond which greater customer loyalty decreases customer spending. This proposition is in line with social exchange theory (Blau, 1964; Emerson, 1976) and the reciprocity principle (Gouldner, 1960), which predict that if an exchange is not mutually rewarding and reciprocal, an imbalance emerges such that behaviorally loyal customers curb spending. This research demonstrates empirically that the relationship between BLOY and customer spending is negative curvilinear (inverted-U shaped) and thereby adds to emerging research re-addressing the consequences of customer loyalty and its contingencies (Cambra-Fierro et al., 2020; Umashankar et al., 2017). Second, it establishes that social rewards (i.e. PCOMM) moderate the amount of customer spending produced by different levels of customer loyalty. Through PCOMM, companies restore balance in the exchange process resulting in higher customer spending at higher levels of BLOY. This study thus ties in with recent research focusing on relational benefits in customer-company relationships (Gremler et al., 2020; Güntürkün et al., 2020). Third, this study provides evidence for the process of reciprocity shaping the relationship between BLOY and customer spending. It adds to relationship marketing literature considering the principle of reciprocity as dominating factor in exchange relationships (Bagozzi, 1995; De Wulf et al., 2001). The research at hand further demonstrates empirically that in the presence of social rewards, customers who strongly endorse the norm of positive reciprocity spend equally high amounts of money across all levels of BLOY. Finally, this study details the financial impact of BLOY on the firm by investigating actual customer spending. A recent meta-analysis of customer loyalty research, involving more than 160 studies, reveals that almost two-thirds of studies examine loyalty as an end outcome, which is a vague proxy for firm profitability (Watson et al., 2015). By drawing on objective customer spending data, this study satisfies calls for more objective measures of profitability and their proxies (Cooil et al., 2007) such as actual customer spending.

2. Reciprocity in customer-company relationships

Social, reciprocal exchanges lie at the core of customer-company relationships (Henderson et al., 2011). Social exchange theory establishes that social exchanges are “two-sided, mutually contingent and mutually rewarding processes involving ‘transactions’” (Emerson, 1976, p. 336). Resources of social exchanges are transacted through a process of reciprocity: one party undertakes some positive initiating actions and the other party repays in kind by engaging in more positive (and/or fewer negative) reciprocating responses (Gouldner, 1960). The norm of reciprocity “evokes obligation toward others on the basis of their past behavior” (Gouldner, 1960, p. 168). With that, an obligation to return an equal resource emerges to balance the exchange (Fiske, 1991). Reciprocating responses can be relational (e.g. loyalty, trust) or behavioral (Cropanzano et al., 2017) and the value of reciprocal obligations are enforced through cultural and normative standards of behavior (Cropanzano and Mitchell, 2005). Resources of exchange can also be concrete, such as money, goods and services (Fiske, 1991).

Relationship marketing research frequently uses the norm of reciprocity (Gouldner, 1960) as the rule of exchange (Bagozzi, 1995). Customers seek reciprocity with the company from which they buy (Schultz and Bailey, 2000) and studies in that context have mostly focused on behaviors with economic consequences (Perugini et al., 2003). For example, reciprocity has served to explain increased restaurant tipping when servers made an extra effort (Rind and Strohmetz, 1999). A company’s positive treatment (i.e. providing rewards) of its customers evokes an obligation which – based on the reciprocity principle – eventually entails positive actions on the customer’s side (De Wulf et al., 2001).

Research focusing on the satisfaction–loyalty–profitability chain follows a similar logic. It proposes that BLOY leads to increased spending based on a positively evaluated exchange (e.g. a satisfying service experience). However, despite the widespread acceptance of a positive relationship between customer loyalty and customer spending, empirical support is rare ( Evanschitzky et al., 2012; Reinartz and Kumar, 2003). Beyond the majority of studies that examine loyalty as an end outcome, a recent meta-analysis of over 160 studies implies that about one-third of them consider loyalty as an exogenous construct, but only one-third of those note any direct link between BLOY and profitability (Watson et al., 2015). In addition, findings indicate a controversial relationship (Umashankar et al., 2017). For instance, research shows that as customers’ BLOY increases, their claiming for rewards also increases (Liu, 2007; Wieseke et al., 2014).

These findings point toward the need for a more pronounced perspective on reciprocal social exchanges because how these exchanges between the customer and the company unfold remains unclear. More specifically, how BLOY translates into customer spending and what role additional social rewards (e.g. PCOMM) offered by the company play is not well-understood. Customers who are committed to strong, enduring social relationships with a company take a long-term approach, with the progression of reciprocity over time determining the perceived balance in exchanges (Blau, 1964). Thus, based on the norm of reciprocity (Gouldner, 1960), it is expected that when customers develop BLOY, their inputs into the exchange relationship increase and consequentially they will expect increasing inputs from the company balancing the relationship into an equitable one.

The following sections more closely unpack the nature of these reciprocal, social exchanges and derive hypotheses capturing the relationships between BLOY, PCOMM and customer spending. As not all customers are alike in terms of their orientation toward reciprocity (Cropanzano et al., 2017), these individual differences will also be taken into account.

2.1 The nonlinear relationship between customer loyalty and customer spending

Customer loyalty encompasses attitudinal and behavioral elements (Dick and Basu, 1994; Oliver, 1999). Oliver (1999, p. 34) defines customer loyalty as “a deeply held commitment to rebuy or repatronize a preferred product/service consistently in the future, thereby causing repetitive same-brand or same brand-set purchasing, despite situational influences and marketing efforts having the potential to cause switching behavior.” In this definition, attitudinal loyalty (commitment) entails BLOY, also known as repeat patronage (Dick and Basu, 1994). This study defines BLOY as repeat purchase behavior (Dick and Basu, 1994; Umashankar et al., 2017). Repeat purchase behavior may emerge as consumers want to increase their status, follow their habits or result from “forward-looking desires to maintain a strong, committed social relationship based on trust and reciprocity with an organization” (Henderson et al., 2011, p. 272). This study focuses on the relational components of BLOY and investigates the exchange relationship between the customer and the company.

Service research identifies both positive and negative relationships between customer loyalty and profitability. Loyalty increases the firm’s financial performance when competitive intensity is high (Homburg et al., 2009) and in high-contact service industries characterized by competitiveness, customer spending similarly is positively influenced by customer loyalty (Yee et al., 2010). Yet a recent study in the airline industry reveals that customer spending decreases with greater customer loyalty unless those loyal customers experience service convenience (Umashankar et al., 2017). Destination-loyal visitors also spend less at a destination than first-time visitors (Alegre and Juaneda, 2006), implying a negative relationship between loyalty and profitability. The likely nonlinear relationship between customer loyalty and profitability might arise in highly competitive service industries, where barriers to switching providers are low and where rewards such as loyalty programs are absent. In this case, customer spending likely is low when they have little BLOY because they are not familiar with the firm, have limited information and suffer from limited learning and decision-making capabilities (Han et al., 2008). As their loyalty grows though, customer spending should reach an optimum (Homburg et al., 2009), beyond which increasing BLOY even further does not necessarily expand customer spending anymore.

The reciprocity principle (Gouldner, 1960) embedded in social exchange theory (Blau, 1964; Emerson, 1976) predicts a potential nonlinear relationship between BLOY and customer spending when exchanges between the company and the customer are imbalanced (i.e. not reciprocal). In other words, if the company fails to return its reciprocal obligations toward the customer (i.e. the company does not repay the customer’s favorable behavior), customers might compensate a lack of reciprocity from the company by reducing their amount of resources (i.e. their spending) in the exchange relationship with the company. Thus, in the absence or in the case of unsatisfying loyalty programs, the research at hand predicts and demonstrates that if a company does not react to increased BLOY with additional rewards, customers will most likely diminish their spending. When customers do not receive rewards for their increased loyalty – as what would be typical in well-designed loyalty programs – they will perceive an imbalance caused by their high efforts (i.e. their strong BLOY) in return for low (or no) reward leading to curbed spending. Thus, customers increase spending as their loyalty increases but then beyond a point where they perceive a lack of reciprocity (e.g. no loyalty rewards offered) they reduce spending. Consequently, following the reciprocity principle (Gouldner, 1960) embedded in social exchange theory (Blau, 1964; Emerson, 1976), the initial positive relationship between BLOY and customer spending likely turns after some point into a negative one as BLOY further increases. The relationship takes the shape of an inverted-U. Therefore,

H1.

In the inverted U-shaped relationship between BLOY and customer spending, BLOY increases customer spending up to a certain point, after which it decreases customer spending.

2.2 Moderating effect of personal communication

Customer loyalty research classifies communication as a social reward, which the customer receives from the company apart from the core service/product and which make people feeling closer to each other (Paul et al., 2009). This study defines PCOMM as a consumer’s perception of the extent to which a service provider (i.e. service employee) interacts with the customer in a kind and personal way (De Wulf et al., 2001; Metcalf Lynn et al., 1992). PCOMM involves engaging in friendly conversations and exhibiting personal warmth (Crosby et al., 1990) which the customer enjoys (Paul et al., 2009). Social rewards govern customer-company exchanges through “personal connections and informal practices to encourage desirable behavior, as well as build and reinforce mutual relational norms” (Heirati et al., 2019, p. 2). Customers perceive special attention or good treatment as social rewards received from the company (Melancon et al., 2011). Morales (2005) finds that customers demonstrate reciprocity when employees direct extra efforts toward them individually; customers then reward companies (i.e. further engage in reciprocating response) by increased willingness to pay.

According to the principle of reciprocity (Gouldner, 1960) rooted in social exchange theory (Blau, 1964; Emerson, 1976), customers expect reciprocal exchanges in their relationships with companies (Schultz and Bailey, 2000), so behaviorally loyal customers likely demand more from the company than less behaviorally loyal customers (Liu, 2007), reflecting what they have already given the company (Slater, 1997). That is, these customers evaluate the relationship between what they have given up and what the firm has provided differently than do less behaviorally loyal customers as the former devoted considerable resources to maintaining their relationship with a company (e.g. spending; Umashankar et al., 2017). PCOMM constitutes a reward for such BLOY (Heirati et al., 2019). Highly loyal customers will consider such PCOMM as a reciprocal reaction of the service provider balancing the exchange relationship. Thus, PCOMM will most likely influence customer spending by mitigating the potential decrease proposed in H1 resulting from increased BLOY. Formally,

H2.

Through more PCOMM, the negative slope of the inverted U-shaped relationship between BLOY and customer spending is attenuated (i.e. less negative).

2.3 Moderating effect of positive reciprocity

Social exchange research identifies the social norm of reciprocity as strongly shaping human behavior but not all people endorse this norm to the same degree (Cropanzano et al., 2017). The norm of reciprocity is defined as an internalized social norm prescribing that “one should help those who have helped him/her in the past and retaliate against those who have been detrimental to his/her interests” (Perugini et al., 2003, p. 252). People who approve of this norm are high in exchange ideology and more likely to reciprocate than those who adhere less to this ideology (Eisenberger et al., 1987). Reciprocity can be both: positive and negative. An individual’s positive reciprocity orientation entails a tendency to exchange positive treatment for positive treatment; a negative reciprocity orientation entails a tendency to retaliate negative treatment for negative treatment (Fehr and Gächter, 2000). Negative reciprocity may be more likely to be obstructed due to constraints such as rules and regulations though (Cropanzano et al., 2017). Importantly, both orientations are only modestly correlated (Eisenberger et al., 1986), implying that a person can endorse high positive reciprocity and low negative reciprocity at the same time (and vice versa).

Relationship marketing literature stresses the dominance of reciprocal exchange in customer-company exchange (Bagozzi, 1995; De Wulf et al., 2001). In this study, the company’s social rewards (i.e. PCOMM) are traded for the outcomes received, such as customer spending and loyalty. Consequently, this study uses the norm of positive reciprocity (Perugini et al., 2003) as the individual difference to examine its influence on the relationship between BLOY and customer spending. If a mechanism of reciprocity indeed shapes the relationship between BLOY, positive communication and customer spending, participants who endorse strong positive reciprocity should be more susceptible to higher spending when PCOMM is high (vs low). For the study at hand, this means that the moderating influence of positive communication (see H2) on the inverted-U shaped relationship between BLOY and customer spending (see H1) is stronger for individuals with strong dispositional positive reciprocity compared to little dispositional positive reciprocity. Due to the nature of the relationship (inverted-U shaped), the strongest effects of higher levels of positive reciprocity are expected to become visible in situations where PCOMM is high and BLOY is either low or high. When customers have little BLOY, they have limited experience with the service provider and exchange processes (Han et al., 2008). Less behaviorally loyal customers are expected to repay positive treatment (i.e. high PCOMM) in form of increased customer spending the more they endorse positive reciprocity as receiving these social rewards may evoke gratitude and pleasure, resulting in reciprocal customer spending (Palmatier et al., 2009). Similarly, high behaviorally loyal individuals who endorse strong positive reciprocity will acknowledge that the firm has devoted considerable resources (i.e. PCOMM) to rewarding the customer in return for their loyalty (Umashankar et al., 2017) and react with increased spending. Consequently, the positive moderating effect described in H2 will be more accentuated. As this study’s setting does not involve any negative treatment of customers, no such effects are expected for dispositional negative reciprocity. Formally,

H3.

The attenuating effect of high PCOMM on the inverted U-shaped relationship between BLOY and customer spending (see H2) is stronger the more customers endorse positive reciprocity.

3. Overview of studies

The tests of these hypotheses rely on two cross-sectional studies, conducted as part of a larger study and one experimental study in two service encounter settings marked by close social exchanges with highly customized service solutions (Bowen, 1990). Service encounters are “social” encounters where customers and companies (i.e. service employees) directly interact (Bitner et al., 1990). Service encounters of high contact services such as restaurants and hotels are thus suitable settings to study dynamic, reciprocal processes between exchanging partners. Study 1a investigates the negative curvilinear effects between BLOY and perceptional customer spending (H1), using field data from four restaurants. Study 1b tests the moderating effects of PCOMM (H2) using field data from a hotel. To gain process evidence of reciprocity actually shaping the relationship between BLOY, PCOMM and customer spending, Study 2 follows recommendations of Achar et al. (2020) and uses a moderation approach with measured positive reciprocity as moderating the relationship between BLOY and customer spending in a controlled setting. All three studies seek to link behavior and spending at the customer level (Bolton et al., 2004). Furthermore, Study 1b demonstrates the financial impact of BLOY by measuring customers’ actual spending and thus drawing on objective data (Cooil et al., 2007).

3.1 Study 1a: negative curvilinear effects between behavioral loyalty and perceptional customer spending

3.1.1 Survey context and sample.

The context of Study 1a spanned four restaurants that offer full table service of food and drinks, at similar prices. A pretest with five restaurant guests led to minor adjustments to the questionnaire to clarify the wording. The finalized questionnaires were distributed to guests after they paid. The data collection lasted for about three weeks. The respondents received a voucher for a free drink (around 5 Euros) at the bar as compensation for participating in the study. Of 390 questionnaires distributed, 309 usable questionnaires were received. The majority of respondents were men (61.8%) and in terms of age, 23.6% were between 21 and 30 years, 18.8% between 31 and 40 years, 21.7% between 41 and 50 years, 24.3% between 51 and 60 years and 11.6% 61 years or older. They were mainly Austrian in nationality (67.6%), followed by German (18.1%), Swiss (3.6%) and Italian (2.3%).

3.1.2 Measures

Dependent variable.

The customer spending measure assessed respondents’ self-reports of the number of Euros they spent for services (e.g. drinks, dining) at the restaurant per person (M = 30.97, SD = 14.81) (Venkatesan and Kumar, 2004).

Independent variable.

BLOY was measured as respondents’ self-reports of their repeat patronage of the restaurant (i.e. “How often do you visit this restaurant?”). A seven-point scale was used from 1 = not very often to 7 = very often (M = 3.14, SD = 1.42).

Control variables.

Demographics (i.e. gender: coded men = 0 and women = 1; age) and dummy variables for the four different restaurants were included as control variables. Table 1 presents the correlations of the measures.

3.1.3 Data analysis: negative curvilinear effects.

The proposed negative curvilinear relationship implies that the incremental effects of BLOY on customer spending decrease with a one-unit change of BLOY, as detailed in the following function:

(1) Spendingi= b0+ b1BLOYi+ b2BLOY2i+ bkCki+ ei,

where b₀ represents the intercept, b₁ is the effect of BLOY on spending, b₂ indicates diminished spending through a downward effect on the slope of b₁, C_k encompasses the control variables and e_i is the error term.

To test for the negative curvilinear effect of BLOY, this study applies a hierarchical polynomial regression analysis (Cohen et al., 2003; Frazier et al., 2004). Model 1 includes the control variables only. Model 2 integrates the linear effect (i.e. BLOY) and Model 3 adds the quadratic effect (i.e. BLOY²). As Edwards and Parry (1993) suggest the predictor variables were mean-centered to facilitate interpretations.

3.1.4 Results.

Table 2 contains the results of the three hierarchical regressions for customer spending. To test H1, the interpretation of the coefficients includes all the terms in equation (1) (Cohen et al., 2003). The variance inflation factors (VIF) for the explanatory variables did not exceed the recommended value of 3 (O’Brien, 2007). As the Model 3 results show, gender and age had no significant effects on customer spending. Only one of the restaurants type dummy variables had a significant negative effect on customer spending(b = −12.31, t = −2.89, p < 0.01).

The linear effect of BLOY (b = −0.53, t = −0.65, ns) on customer spending was not significant. The significant and negative quadratic effect of BLOY on customer spending(b = −0.88, t = −2.49, p < 0.05) implies that BLOY had a curvilinear (i.e. negative asymmetric) effect on customer spending. Thus, the findings support H1 and suggest an inverted U-shaped effect of BLOY on customer spending. After reaching a BLOY optimum, customer spending decreased as BLOY increased (see Figure 1 Panel A). To calculate the BLOY optimum (i.e. where customer spending is the highest), the first derivation has to be set to zero, resulting in a tipping point at the value of BLOY = 3.02 (mean-centered = −0.30) and a peak value for customer spending at 37.89 Euros. The result implies that restaurants benefit most from customers with average levels of repeat patronage.

3.2 Field Study 1b: moderation by personal communication for actual customer spending

3.2.1 Survey context and sample.

The study setting was a three-star hotel that provides guests with leisure and recreational activities. The study was conducted in a hotel with international guests, so the questionnaire was available in English, German and Italian. Back-translation ensured vocabulary, idiomatic, grammatical and syntactic equivalence (Hofstede et al., 1999). Then the data collection spanned five months. Hotel employees at the reception desk distributed the questionnaire to 301 guests after they checked out and 252 usable questionnaires were received. The majority of respondents were men (65.1%) and 47.09 (SD = 12.54) years old on average. Most of the respondents were German (28.2%), followed by Italian (17.3%), Austrian (13.7%), USA (9.3%), British (8.9%) and Swiss (7.3%). The average number of days they stayed at the hotel was 4.10. The hotel provided objective data on actual customer spending.

3.2.2 Measures

Dependent variable.

Customer spending was measured as the number of Euros spent by the customer per day and per person (Venkatesan and Kumar, 2004) for services (e.g. bar, dining) other than the hotel room (M = 11.04, SD = 12.44) as the latter is fixed in advance. This measure was retrieved from the hotel’s database after customers checked out.

Independent variable.

As in Study 1a, the measure of BLOY used respondents’ self-reported repeat patronage of the hotel (i.e. “How many times have you visited this hotel before?”; open-ended question), drawn from Kumar and Shah (2004), Reinartz and Kumar (2000), Shah et al. (2014) (M = 2.53, SD = 2.67).

Moderator.

PCOMM consisted of six items (e.g. “Talking to the front desk employees was positive for me”) that addressed customers’ perceptions of their interpersonal communication with the hotel employees, measured on seven-point, disagree–agree, Likert-type scales (1 = “strongly disagree,” 7 = “strongly agree”). After confirming the factor loadings, composite reliability and average variance extracted (AVE) in confirmatory factor analysis (CFA), the construct’s items were averaged into a single-item vector (see Tables 3 and 4 for construct measures, correlations and reliability tests). Discriminant validity was confirmed following propositions from Fornell and Larcker (1981): all construct correlations are lower compared to the square root of the construct’s AVEs (Table 4).

Control variables.

The demographic items from the survey (i.e. gender: codedmen = 0 and women = 1; age) were included as control variables. Customer spending for the hotel room may influence customer spending for other services, so it was included as a control variable, measured as the number of Euros spent by the customer for the room per day and per person (M = 57.93, SD = 35.56). This measure was retrieved from the hotel’s database. Value-for-money (i.e. “The hotel offered value for money”; M = 5.31, SD = 1.64) and advocacy (i.e. “In the past, I was able to convince others to stay at the hotel”; M = 5.06, SD = 1.85) may affect a customer’s spending behavior and thus their effects were controlled for.

3.2.3 Data analysis: negative curvilinear and moderating effects.

As in Study 1a, equation (1) describes the proposed negative curvilinear relationship. In equation (2) PCOMM enters as a moderator:

(2) Spendingi= b0+ b1BLOYi+ b2PCOMMi+ + b3(BLOYiPCOMMi) + + b4BLOY²i+ b5(BLOY²iPCOMMi) + + bkCki+ ei.

To test for the negative curvilinear effects of BLOY and PCOMM on customer spending, again a hierarchical polynomial regression analysis was applied (Cohen et al., 2003; Frazier et al., 2004). Model 1 includes the control variables only. Model 2 integrates the linear effects (i.e. BLOY and PCOMM), Model 3 adds the two-way interaction effects (i.e. BLOY × PCOMM and BLOY²) and Model 4 finally integrates the three-way interaction effect (i.e. BLOY²× PCOMM). The predictor and moderating variables were mean-centered (Edwards and Parry, 1993) and the effects were shown in a standardized three-dimensional space (i.e. surface plot) using MATLAB.

3.2.4 Results.

The coefficients were interpreted in the presence of all the terms from equation (2) (Cohen et al., 2003). The VIF for the explanatory variables in the model did not exceed 3 (O’Brien, 2007). Gender, age, value-for-money and advocacy had no significant effects on customer spending (Table 5). The control variable reflecting the spending for the hotel room (b = 0.19, t = 6.40, p < 0.01) had a significant positive effect on customer spending; the more customers spent for their room, the more money they spent on additional services. The linear effect of BLOY exerted a marginally significant, positive effect on customer spending (b = 1.88, t = 1.83, p < 0.10). Its quadratic effect had a significant negative effect (b = −1.28, t = −2.34, p < 0.01), indicating an inverted U-shaped effect, in support of H1. With increasing BLOY, customer spending increases up to some point, beyond which it decreases (see Figure 1 Panel B). Again, the first derivation was set to zero to calculate the BLOY optimum in terms of the highest customer spending. The tipping point occurred at BLOY = 3.30 (mean-centered = 0.73) with the peak customer spending for additional services of 16.05 Euros per person per day. The result implies that customers who have visited the hotel more than three times before spent the highest amount of Euros for hotel services other than the hotel room. The linear effect of PCOMM was also marginally significant (b = 4.01, t = 1.67, p < 0.10), indicating PCOMM was an independent and positive predictor of customer spending.

In support of H2, both the linear and linear by quadratic interaction of PCOMM and BLOY on customer spending were significantly (the latter marginally significantly) positive. Specifically, the linear interaction of PCOMM and BLOY affected customer spending (b = 3.19, t = 2.00, p < 0.05), as did the linear by quadratic interaction of PCOMM and BLOY (b = 1.52, t = 1.86, p < 0.10). Figure 2 depicts the positive asymmetric effect of PCOMM with increasing BLOY, such that its downward effect on customer spending converts into an upward effect at high levels of BLOY. High levels of PCOMM thus overcome the decrease in customer spending among behaviorally loyal customers.

3.3 Study 2: moderation by positive reciprocity as process evidence

3.3.1 Research design and experimental procedure.

The study had two manipulated between-subjects factors (BLOY: low, moderate and high; PCOMM: low and high) yielding six different conditions. In total, 295 participants (55.9% men, mean age = 37.57 years) from a large German online panel agency, completed the study. Participants were randomly assigned to one of the six conditions. They read a scenario about having dinner in a mid-price restaurant in which they have been before and then responded to questions. In the low BLOY condition, participants read that they would have dinner once a year in this restaurant, in the moderate BLOY condition several times a year and in the high BLOY condition, they read that they would have dinner there several times a month. The second factor, PCOMM, was manipulated at two levels. As defined earlier, PCOMM involves friendly, kind and enjoyable conversations between the customer and the service employee (De Wulf et al., 2001; Metcalf Lynn et al., 1992). As PCOMM is a social reward which the customer receives apart from the core service product (Paul et al., 2009), the service level (e.g. being welcomed and served, questions answered) was held constant. In the low PCOMM condition, participants read that they were welcomed when entering the restaurant. They were immediately served, but the staff was not open to conversation. Participants’ questions were sufficiently answered but the staff was uninvolved and did not take the time for personal suggestions. The interaction with the staff was purely formal and no personal conversation was established. In the high PCOMM condition, participants instead read that they were warmly welcomed when entering the restaurant. They were immediately served and the staff was open to conversation. The staff enjoyed taking the time to answer participants’ questions and made personal suggestions. In no time, a cordial level of conversation was established and personal information was exchanged spontaneously.

3.3.2 Pretest.

A pretest with 104 participants (60 men and 44 women; mean age = 34.26 years) was conducted to test the effectiveness of the manipulation. Participants were randomly assigned to one of the six conditions. To confirm that the manipulation of BLOY (low, moderate and high) was successful, two items focusing on the perception of repeat purchasing were used (i.e. “I visit this restaurant frequently,” “I rarely visit this restaurant,” reverse-coded; items were averaged into a single-item vector; M = 4.62, SD = 2.09; α = 0.88). A significant main effect emerged for BLOY [F(2,101) = 62.99, p < 0.001]. As expected, post hoc comparisons using a Bonferroni correction showed that participants perceived the high BLOY condition (M = 6.23, SD = 0.73) as significantly more frequent purchasing than the moderate condition (M = 5.36, SD = 1.24, p < 0.05) or low BLOY condition (M = 2.67, SD = 1.89, p < 0.001). To test the effectiveness of the PCOMM (low, high) manipulation, four items from the scale of Study 1b and two additional items (M = 4.85, SD = 1.88; α = 0.97) were adapted and used which focused on customers’ positive perception of the PCOMM with the employees/staff (items see Table 3; items were averaged into a single-item vector). The main effect of PCOMM was significant [F(1,102) = 182.27, p < 0.001] and participants perceived the high PCOMM condition (M = 6.24, SD = 1.04) significantly more positive than the low (M = 3.23, SD = 1.04, p < 0.001) condition. Thus, both manipulations were successful.

3.3.3 Measures.

Participants responded to a series of questions capturing customer spending, positive and negative reciprocity and control variables. Finally, participants’ demographic information was collected.

Dependent variable.

Customer spending was measured using the amount participants were willing to spend for the described restaurant (“How much would you be willing to spend for food and beverages at the described restaurant?”; M = 26.53, SD = 14.32) in EUR per person.

Moderators.

Five items assessed positive reciprocity (e.g. “I am ready to undergo personal costs to help somebody who helped me before”) encompassing an individual’s tendency to reward positive treatment. Negative reciprocity consisted of four items (e.g. “If I suffer a serious wrong, I will take my revenge as soon as possible, no matter what the costs”) that addressed an individual’s tendency to retaliate negative treatment. The items were measured on seven-point, disagree–agree, Likert-type scales (1 = “strongly disagree,” 7 = “strongly agree”) and taken from Perugini et al. (2003)’s Personal Norm of Reciprocity scales. Table 3 presents the construct measures and factor loadings; Table 6 shows correlations and composite reliability values and establishes discriminant validity following Fornell and Larcker (1981). The items were averaged into single-item vectors.

Control variables.

Gender (coded men = 0 and women = 1), age and income were included as control variables. Additionally, compensating behavior for many restaurant visits may influence customer spending, so it was included as a control variable (“All in all, I do not spend more money at restaurants, even though I often eat out,” M = 4.39, SD = 1.67) using a seven-point, disagree–agree, Likert-type scale (1 = “strongly disagree” to 7 = “strongly agree”).

3.3.4 Manipulation checks.

The analysis of variance yielded a significant main effect of BLOY on the manipulation checks for BLOY [F(2,292) = 179.55, p < 0.001; same items as in the pretest averaged to a single-item vector]. Post hoc comparisons using a Bonferroni correction revealed that participants perceived the high BLOY condition (M = 6.00, SD = 1.12) as significantly more frequent purchasing than the moderate (M = 4.81, SD = 1.49, p < 0.001) and low condition (M = 2.49, SD = 1.38, p < 0.001). The main effect of PCOMM on the manipulation checks for PCOMM (same items as in the pretest, construct measures and reliability tests (Table 3); items were averaged into a single-item vector after confirming the factor loadings, composite reliability and AVE in a CFA) was also significant [F(1, 293) = 907.12, p < 0.001]. Participants in the high (M = 6.42, SD = 0.78) PCOMM condition detected more positive conversation with the staff than in the low condition (M = 2.98, SD = 1.15, p < 0.001), implying successful manipulations.

3.3.5 Results.

This study applied a hierarchical and a bootstrapped regression using Model 3 in PROCESS v3 (Hayes, 2018) with 5,000 bootstrap samples indicating participants’ customer spending as the dependent variable, BLOY, PCOMM, positive and negative reciprocity, eight two-way and four three-way interaction terms and controls as predictors. All variables for constructing products were mean-centered. Three treatments (rather than two) of BLOY (low–moderate–high) allow to detect potential turning points in customer spending. As BLOY is a multicategorical, ordinal variable, sequential coding for BLOY was applied. Sequential coding compares the mean of the dependent variable for one level (e.g. low BLOY) of a categorical variable to the mean of the dependent variable in the next (one ordinal position higher; e.g. moderate BLOY) level of the categorical variable (Darlington and Hayes, 2017). The model was significant [F(21,273) = 3.48, p < 0.001], with two significant coefficients for the control variables gender (b = 3.79, t = 2.32, p < 0.05) and age (b = −0.24, t = −2.96, p < 0.01; Table 7). A marginally significant negative difference emerged between moderate and high BLOY (cf. coefficient X2; b = −3.25, t = −1.70, p < 0.10), indicating a downward trend of customer spending from moderate to high BLOY, in partial support of H1. As in Study 1b, PCOMM was an independent and significant predictor of customer spending (b = 11.79, t = 4.41, p < 0.001). The significant interactions of BLOY and PCOMM (X1: b = −10.52, t = −2.74, p < 0.01; X2: b = 7.91, t = 2.04, p < 0.05) showed that by more PCOMM, the negative slope of the inverted U-shaped relationship between BLOY and customer spending (cf. X2) was indeed attenuated (i.e. not negative but significantly positive), in partial support of H2. Though not predicted, by more PCOMM, customer spending was significantly higher at low compared with moderate levels of BLOY. Yet, for full interpretation, the conditional effects need to be inspected.

Significant coefficients emerged also for the interaction between PCOMM and positive reciprocity (b = −7.16, t = −2.75, p < 0.05). The interaction between PCOMM and negative reciprocity was marginally significant (b = −4.54, t = −1.92, p < 0.10). As predicted, no other interactions with negative reciprocity were significant. Significant coefficients emerged for the three-way interaction between the X1 coefficient of BLOY, PCOMM and positive reciprocity (b = 11.75, t = 2.71, p < 0.01), indicating that PCOMM and positive reciprocity conditionally moderated the relationship between BLOY and customer spending (Figure 3).

The three-way interaction was broken down using a spotlight analysis completed at the 16th, 50th and 84th percentiles of positive reciprocity for low and high levels of PCOMM. Recall that a negative coefficient for X1 implies that customer spending is higher for low BLOY compared with moderate BLOY; a positive X1 coefficient means that customer spending is lower for low BLOY compared with moderate BLOY. A negative coefficient for X2 means that customer spending is higher for moderate BLOY than for high BLOY and a positive X2 coefficient implies that customer spending is lower for moderate BLOY compared with high BLOY. Table 8 shows that in low PCOMM settings, the difference between low (M = 21.72) vs moderate BLOY (M = 29.29; cf. X1) on customer spending was positive and significant (b = 7.57, t = 2.34, p < 0.05). This difference became smaller and nonsignificant as positive reciprocity increased. The difference between moderate (M = 29.29) vs high BLOY (M = 22.54; cf. X2) on customer spending was negative and marginally significant at low positive reciprocity (b = −6.75, t = −1.96, p < 0.10) and high positive reciprocity (M_moderate = 22.34 vs M_high = 14.61; b = −7.73, t = −1.96, p < 0.10); this difference was negative and significant at moderate levels of positive reciprocity (M_moderate = 25.82 vs M_high = 18.58; b = −7.24, t = −2.67, p < 0.01). Thus, the negative difference became larger with stronger positive reciprocity, implying that when PCOMM was low, the downward effect at high levels of BLOY became accentuated with increasing positive reciprocity. In other words, the inverted U-shape between BLOY and customer spending persists in low PCOMM settings across different levels of positive reciprocity: customer spending increases up to moderate BLOY and then customer spending decreases. The difference in customer spending between low and moderate levels of BLOY was not significant the more strongly customers endorsed positive reciprocity. However, with stronger positive reciprocity, customers having high BLOY react to a company’s negative treatment (i.e. low PCOMM) in form of diminishing spending.

Regarding the effect of high levels of PCOMM when customers endorse stronger positive reciprocity (H3), support was found for the prediction. H3 predicted that by more PCOMM and stronger positive reciprocity, the negative slope of the inverted U-shaped relationship between BLOY and customer spending is attenuated. The results showed that across all positive reciprocity levels (low: b = −0.52, t = −0.14, n.s.; moderate: b = 0.85, t = 0.31, n.s.; high: b = 2.23, t = 0.59, n.s.), customer spending did not drop at high compared with moderate levels of BLOY (M_high = 26.73 vs M_moderate = 27.25; M_high = 28.30 vs M_moderate = 27.44; M_high = 29.87 vs M_moderate = 27.64).

Though not predicted, a negative, significant difference emerged between low (M = 38.68) and moderate levels of BLOY (M = 27.25; b = −11.43, t = −3.05, p < 0.01; cf. X1) when PCOMM was high, implying a “surprising effect.” Customers having little BLOY seemed surprised by the positive treatment and reacted with increased customer spending. However, moderate positive reciprocity shrank this “surprising effect” at low levels of BLOY (M_low = 34.56 vs M_moderate = 27.44, b = −7.12, t = −2.60, p < 0.01) and customer spending finally dropped at low (M = 30.45) compared with moderate levels of BLOY (M = 27.64) when positive reciprocity was high (b = −2.81, t = −0.71, n.s.). In sum and in support of H3, the results showed when PCOMM was high (i.e. when customers experienced a favorable treatment), stronger positive reciprocity leveled out the differences in customer spending at low, moderate and high levels of BLOY.

4. General discussion

This research set out to provide deeper insights into the loyalty–spending link by assessing the potential nonlinear effects of BLOY on customer spending. Building on social exchange theory (Blau, 1964; Emerson, 1976) and the norm of reciprocity (Gouldner, 1960), three hypotheses predicted asymmetries between BLOY and customer spending and the influence of social rewards (i.e. PCOMM) and positive reciprocity on spending by customers who vary in their BLOY. In turn, this research provides four key contributions.

First, in two cross-sectional studies and one experimental study across two different service sectors, this research demonstrates that if an imbalance of reciprocity exists (e.g. absence of loyalty programs), the relationship between BLOY and customer spending is negative curvilinear in nature. The initial effects of growing BLOY on customer spending are increasing, but after a peak, customer spending gradually decreases as BLOY increases more. Thus, BLOY does not always benefit a company’s financial performance. The asymmetric relationship between BLOY and customer spending emerges when the company offers no social rewards for the customer’s loyalty efforts. These results provide some clarity for the conflicting findings in recent studies, such as Yee et al.’s (2010) finding of only positive linear effects of customer loyalty on customer spending vs indications that loyal customers pay less for airline services (Umashankar et al., 2017) or spend less money at a tourism destination (Alegre and Juaneda, 2006). The underlying reason for these conflicting results may stem from the nonlinear relationship between BLOY and customer spending. Precisely, the current study specifies an inverted U-shaped relationship of customer loyalty and spending.

Second, this study demonstrates the power of social rewards to augment customer spending and restore perceived reciprocal equity at different levels of BLOY. High PCOMM (social reward) triggers customer spending at all levels but is particularly effective at higher levels of BLOY. A positive PCOMM with the customer is a suitable means to reward more behaviorally loyal customers and thereby evoke customers’ reciprocal responses in form of increased customer spending. As BLOY grows though, customers demand more in return for what they have given (Slater, 1997). With increasing loyalty, customers attribute more value to these exchange aspects of their relationship with the firm (Beatty et al., 1996). Thus, when loyal customers receive social rewards, such as positive PCOMM, they likely spend more with the firm. PCOMM thereby shifts the negative slope of the inverted U-shaped relationship toward a positive slope.

Third, this study finds converging process evidence for the mechanism of reciprocity underlying the relationship between BLOY and customer spending. When social rewards (i.e. PCOMM) are present, customers high in dispositional positive reciprocity repay the company’s positive treatment (i.e. high PCOMM) by increasing spending. Positive reciprocators are particularly appreciative to positive treatment at low and high levels of BLOY. When customers are not familiar with a firm and have limited information, they are surprised and grateful for kind behavior (Palmatier et al., 2009) and repay the company’s efforts by increasing their spending. At moderate BLOY, customers are more experienced with the company and have invested some efforts in the relationship. They are less surprised if they are treated well but expect some level of good treatment in return for their efforts. When customers have devoted considerable efforts to the company in terms of loyalty and they strongly endorse positive reciprocity, they reciprocate high PCOMM. Their reciprocity expectation appears balanced and in return for positive social actions, such as high PCOMM during the exchange process, customers with a strong endorsement of positive reciprocity increase their spending.

This study also reveals one surprising finding. Interestingly, PCOMM has the highest impact on customer spending for low behaviorally loyal customers who endorse little positive reciprocity. A reason for this unexpected finding could be that individuals who do not strongly adhere to reciprocity norms (Perugini et al., 2003) evaluate exchanges with companies according to an alternate logic. For instance, there is evidence that reciprocal exchanges are different to economic exchanges that trade costs against benefits (Mitchell et al., 2012). Applying an economic exchange logic to the findings of this study would mean that the high spending of low behaviorally loyal customers who endorse little positive reciprocity is an indicator of their valuation of PCOMM as a benefit of a service encounter one is willing to pay good money for. This is an interesting finding calling for further research investigating how different attitudes and dispositions at the individual level toward exchange relationships (e.g. social vs economic exchange) impact customer spending behavior. Finally, the findings indicate that negative reciprocity does not affect the relationship between BLOY and customer spending. This is not surprising as low PCOMM does not imply negative treatment. However, further research might seek to investigate negative reciprocity more into depth.

Forth, this research ties in with emerging literature revisiting the relationship between satisfaction and loyalty (Thakur, 2019) and it adds to limited knowledge on potential relationships between customer loyalty and actual customer spending. Extant research mainly infers positive effects on customer spending from loyalty as the end outcome (Watson et al., 2015). Only a few studies test the relationship between customer loyalty and objective customer spending (Stein and Ramaseshan, 2019; Stokburger-Sauer et al., 2016). The current research empirically demonstrates the financial impact of BLOY on the firm and the power of social rewards (PCOMM) for boosting customers’ actual spending.

4.1 Theoretical implications

This study ties in with recent research revisiting linearity in the satisfaction–loyalty–spending link (Otto et al., 2020; Thakur, 2019; Umashankar et al., 2017) and finds support for a negative curvilinear relationship between BLOY and customer spending. Enduring customer-company relationships are mutually rewarding exchange processes (Blau, 1964) characterized by reciprocity (Bagozzi, 1995; Houston and Gassenheimer, 1987) in which customers trade loyalty for a company’s equitable rewards. If the reciprocity principle holds, the exchanging partners enter a cycle of returning reciprocal obligations in which an initiator’s positive action entails a positive response of the partner causing again a positive action on the initiator’s side et cetera (Gouldner, 1960). This study demonstrates that if this reciprocal exchange process is disturbed, customers put less effort in their relationship with the company and retaliate against lack of rewards in form of reduced customer spending. With that, an inverted-U shaped relationship between BLOY and customer spending emerges.

This research supports De Wulf et al.’s (2001) concept of perceived relationship investment which holds that customers evoke obligation toward the company which treats them well and provides value. With PCOMM on the company’s side, customers feel an obligation to return something to the company, that is increased customer spending. The personal norm of positive reciprocity further corroborates the existence of the reciprocity principle in the loyalty–spending link: the more strongly customers endorse positive reciprocity, the more equal are spending levels across different levels of BLOY when social rewards (i.e. PCOMM) are present vs absent.

This study demonstrates the power of reciprocal dynamics in consumer research. As Moon (2000) argued, reciprocal dialogues emerge first and foremost in one-to-one customer-company exchanges. Extant research and meta-analyses exist on the “one-to-many” or cumulative rewards such as loyalty programs (for extensive overviews see Kumar and Shah, 2004; Otto et al., 2020; Palmatier et al., 2006; Stourm et al., 2020; Watson et al., 2015) and their effectiveness for customer metrics, such as share-of-wallet or profitability. This study adds to relationship marketing literature by showing that situational customer spending can be triggered by dynamic, reciprocal exchanges between the customer and the company (i.e. employee). It shows that in the exchange process, customer spending depends on past BLOY plus the characteristics of the current exchange process, such as customer treatment and customers’ reciprocity dispositions.

Further, this study provides evidence that specific customers might not endorse reciprocity norms and that these might apply distinct principles to evaluate customer-company exchanges leading to different spending behavior. However, overall this study supports Henderson et al. (2011) conclusion that customer-company relationships are socially induced.

4.2 Managerial implications

The findings bear important practical implications for service managers. If managed correctly, customer loyalty is a critical source of competitive advantage for service firms, with potentially powerful impacts on firm performance through increased customer spending (Reichheld and Sasser, 1990; Anderson and Mittal, 2000). The current results challenge conventional beliefs about the inherently positive relationship between customer loyalty and spending though. Not all loyal customers spend more – companies have to get their reciprocity right before they can benefit from increased customer spending. Social rewards such as PCOMM are a viable means for companies to balance an exchange and thereby return a reciprocal obligation to behaviorally loyal customers. With that, behaviorally loyal customers enter a reciprocity cycle where they repay the company’s efforts by increased spending.

Managers should take asymmetries into account to find the optimum level of BLOY at which customer spending is highest. In addition to identifying this optimum level of BLOY, managers should listen to loyal customers to discover their habits. Behaviorally loyal customers are more knowledgeable and seem to make more efficient choices, based on their better information. Service employees are urged to personally recommend new products and services to customers. For instance, servers are advised to engage in conversations, explain items from the menu to the customers and make personal recommendations such that the customer eventually spends more than he/she originally planned (WebstaurantStore, 2018). Moreover, servers can thereby significantly increase the amount of tips received (Fernandez et al., 2019).

Large service organizations have incorporated customer loyalty rewards into their business strategies, but smaller firms often lack the capacity or resources to implement complex customer relationship management systems – nor do they necessarily require them. Differentiation through service or quality seems more effective from a customer view (Palmer, 2013). In highly competitive service industries, social rewards are significant tactics to trigger additional spending. Particularly in service settings such as hotels, marked by close interactions between the customer and the service provider, PCOMM can evoke additional customer spending and a lack thereof can have detrimental effects for the company. Customers who endorse the social norm of positive reciprocity repay a company’s positive treatment such as high PCOMM. Positive reciprocators are grateful for kind treatment (Perugini et al., 2003) and might respond with increased spending. Most likely, these customers might not only increase spending at the purchase momentum but also in the future. Not only the monetary consequences might benefit the company though. Positive reciprocators with a positive experience might also potentially engage in other favorable behavior, such as positive word-of-mouth.

Getting the communication right throughout the service delivery process and managing the interactions are of crucial importance. How the customer perceives the employee and the interaction with him/her can shape the customer’s spending (Echchakoui, 2017). However, what sounds simple can be problematic for firms struggling with excessive employee turnover. In the Austrian hospitality sector for example, for every existing job, 1.4 resignations occur each year (Biehl and Kaske, 2011). Such turnover creates challenging implications for customer loyalty because PCOMM is essential for success (Webster and Sundaram, 2009). Service managers must recognize that their greatest assets are their employees; they should spare no effort to retain them. Furthermore, managers are well-advised to create a service environment in which PCOMM can occur naturally (Yoo, 2019).

5. Limitations and further research

This study has several limitations that suggest avenues for further research. First, a central finding of this study pertains to the asymmetric relationship between BLOY and customer spending. Research calls for managerially relevant outcomes such as customers’ actual behavior. This study uses customer spending as one behavioral outcome. Behavior encompasses all different kinds of actions that customers can engage in such as purchase quantities, length of stay or actual (not intended) electronic word-of-Mouth generation (Hulland and Houston, 2021). Further research is needed to more thoroughly explore the asymmetric impact of BLOY on alternative, also nonmonetary behavioral outcomes. Moreover, it might be interesting to examine the relationship between BLOY and behavioral outcomes taking different business sectors, including potentially less hedonic service settings (e.g. health care, financial services) into account. Testing for asymmetries in a business-to-business context also would extend the current findings in a customer context.

Second, the present study identifies PCOMM as a remedy to overcome decreased spending beyond the tipping point of spending. PCOMM is a social reward that helps to establish reciprocity in an exchange relationship. It involves kindness and warmth between exchange partners to establish personal connections. For future research, it would be interesting to extend the concept of personal connections in relationships and involve artificial intelligence (AI) agents. AI agents can support relational bonds and add value to customer experience (Moffett et al., 2021); they can even satisfy a customer’s need for personal (emotional) connection in technology-mediated services (Gelbrich et al., 2021). In their traditional function, AI agents are effective in delivering accurate, unambiguous information (e.g. facts about ingredients, cooking processes or wine; Thomaz et al., 2020), which might be needed in early and later relationship stages. Marketing scholars and practitioners might want to learn more about how integrating AI agents in customer-company interactions adds to a mutually rewarding and reciprocal relationship.

Third, other moderating effects, such as amenities of the service environment or customer-to-customer communication (Patti et al., 2020), similarly might be useful for exploring the asymmetries between BLOY and customer spending. Service settings can do both encourage and distract from additional consumption as customers enjoy an atmosphere or struggle with noise, to give examples. It would be interesting for future research to test potential delighters or distractors of spending when customers perceive a relationship with the company to be rewarding and reciprocal. Also, this study does not investigate further contextual constraints of spending. For instance, customers might spend money with a company out of habit or because alternatives are missing. Additional research is needed to get a better understanding of how perceived switching costs impact on the asymmetric relationship between BLOY and customer spending.

In all three studies, the majority of respondents were men. Although the studies controlled for effects of gender a potential bias cannot be ruled out entirely. Additionally, the studies did not involve any similarity measures between the interacting parties. Based on the homophily principle, people connect more with people to whom they feel similar (McPherson et al., 2001). Future studies should thus incorporate employees’ age and gender to gain a broader perspective on how reciprocal social exchanges unfold taking into account (dis-)similarity between the interacting parties.

Finally, for Study 1a and Study 1b, only a limited number of demographic variables was available. For example, customer spending might be dependent on additional factors such as income. Although in Study 2, no significant influence of income was found, potential biases through missing control variables cannot be ruled out entirely. Future research might want to address these shortcomings, too.

Figures

Figure 1.

Negative curvilinear effect of behavioral loyalty on customer spending

Figure 2.

Moderating effect of personal communication on behavioral loyalty and customer spending (Study 1 b)

Figure 3.

Moderating effects on behavioral loyalty and customer spending (Study 2)

Table 1.

Correlations (Study 1a)

Variables	1	2	3	4	5	6	7
Gender	1
Age	−0.14*	1
Dummy 1	0.13	−0.01	1
Dummy 2	0.10	0.15**	−0.27**	1
Dummy 3	−0.17**	−0.20**	−0.30**	−0.77**	1
BLOY	−0.08	0.02	0.14**	−0.40**	0.32**	1
Customer Spending	0.02	0.13*	0.26**	0.24**	−0.44**	−0.21**	1

Notes:

BLOY = behavioral loyalty;

**significant at p < 0.01; *significant at p < 0.05 (two-tailed)

Table 2.

Hierarchical polynomial regressions for customer spending in Study 1a

	Model 1	Model 2	Model 3
Independent variables	Control effects	Linear effect	Quadratic effect
Intercept (b0)	37.72**	37.77**	37.73**
Gender	−1.56	−1.74	−1.97
Age	0.40	0.53	0.66
Dummy 1	4.80	5.71	6.68
Dummy 2	−3.12	−4.05	−2.34
Dummy 3	−14.54**	−13.70**	−12.31**
BLOY		−1.59*	−0.53
BLOY²			−0.88*
R² (%)	21.3	22.9	24.7
Adjusted R² (%)	19.8	21.1	22.6
F (df)	13.92 (5)**	12.66 (6)**	11.96 (7)**
F-change (df)	--	5.21 (1)*	6.20 (1)*

Notes:

Unstandardized coefficients. BLOY = behavioral loyalty;

**

p < 0.01; *p < 0.05

Table 3.

Construct measures

	Study 1b	Study 2
	CFA	CFA
Constructs	FL	FL
Personal communication (adapted from Jha et al., 2017, van Dolen et al., 2007)	M = 6.33 SD = 0.99	M = 4.69 SD = 1.98
Talking to the staff in the restaurant/cafeteria was positive for me	0.86	0.96
The communication with the employees was reliable (Study 2: enjoyable).	0.94	0.96
The employees were friendly	0.94	0.89
The employees readily answered my questions	0.94	0.92
Talking to the front desk employees was positive for me	0.82
Talking to the host was positive for me	0.81
The staff was easy to talk with		0.92
The conversation created a sense of well-being		0.88
Negative reciprocity (adapted from Perugini et al., 2003)		M = 3.22 SD = 1.30
If I suffer a serious wrong, I will take my revenge as soon as possible, no matter what the costs		0.78
If somebody puts me in a difficult position, I will do the same to him/her		0.79
If somebody offends me, I will offend him/her back		0.77
If somebody is impolite to me, I become impolite		0.61
Positive reciprocity (adapted from Perugini et al., 2003)		M = 5.92 SD = 0.91
I am ready to undergo personal costs to help somebody who helped me before		0.80
If someone does a favor for me, I am ready to return it		0.84
If someone is helpful with me, I am pleased to help him/her		0.80
I am ready to do a boring job to return someone’s previous help		0.62
I go out of my way to help somebody who has been kind to me before		0.73

Notes:

CFA = confirmatory factor analysis; FL = standardized factor loadings;

All factor loadings are significant at p < 0.001

Table 4.

Correlations and reliability tests (Study 1b)

Variables	CR	1	2	3	4	5	6	7	8
Gender	N/A	N/A
Age	N/A	−0.19**	N/A
Room spending	N/A	−0.00	0.13	N/A
Value-for-money	N/A	−0.07	0.02	0.14	N/A
Advocacy	N/A	−0.11	0.07	0.09	0.74**	N/A
BLOY	N/A	0.01	−0.03	−0.11	0.18**	0.18**	N/A
PCOMM	0.96	0.03	0.01	0.12	0.52**	0.50**	0.22**	0.88
Customer spending	N/A	0.05	0.02	0.49**	0.02	−0.00	0.04	0.17*	N/A

Notes:

BLOY = behavioral loyalty, PCOMM = personal communication.

CR = construct reliability, AVE = average variance extracted.

Diagonal numbers (in italics) are the square root of AVE values and off-diagonal numbers are correlations;

*

**significant at p < 0.01; significant at p < 0.05 (two-tailed)

Table 5.

Hierarchical polynomial regressions for customer spending in Study 1b

	Model 1	Model 2	Model 3	Model 4
Independent variables	Control effects	Linear effects	Two-way interaction effects	Three-way interaction effect
Intercept (b0)	12.68***	12.56***	13.61***	15.36***
Gender	−0.14	−0.27	−0.43	−0.82
Age	−0.09	−0.10	−0.14	−0.16
Room spending	0.17***	0.17***	0.18***	0.19***
Value-for-money	0.14	−0.41	−0.98	−0.94
Advocacy	−3.22	−0.71	−0.72	−0.91
BLOY		0.49	2.23**	1.88*
PCOMM		3.25**	5.62**	4.01*
BLOY × PCOMM			2.08	3.19**
BLOY²			−0.27***	−1.28**
BLOY² × PCOMM				1.52*
R² (%)	26.2	30.8	39.6	42.0
Adjusted R² (%)	22.0	25.2	33.2	35.0
F (df)	6.24 (5)***	5.48 (7)***	6.13 (9)***	6.02 (10)***
F-change (df)	--	2.91 (2)**	6.10 (2)***	3.45 (1)*

Notes:

Unstandardized coefficients. BLOY = behavioral loyalty, PCOMM = personal communication.

***p < 0.01; **p < 0.05. *p < 0.10

Table 6.

Correlations and reliability tests (Study 2)

Variables	CR	1	2	3	4	5	6	7	8	9	10
Gender	N/A	N/A
Age	N/A	−0.04	N/A
Income	N/A	0.02	0.05	N/A
Compensation behavior	N/A	0.10	−0.04	0.01	N/A
BLOY X1	N/A	0.03	−0.07	−0.02	0.00	N/A
BLOY X2	N/A	−0.01	−0.07	0.02	0.03	0.51**	N/A
PCOMM	0.97	0.04	0.04	0.08	0.06	−0.00	0.01	0.92
PREC	0.83	0.07	0.04	0.05	0.13*	0.06	0.03	0.10	0.73
NREC	0.87	−0.17**	−0.16**	−0.06	0.13*	−0.01	0.01	−0.01	−0.10	0.76
Customer spending	N/A	0.10	−0.11	0.07	−0.08	−0.09	−0.13*	0.26**	−0.04	−0.07	N/A

Notes:

BLOY = behavioral loyalty, PCOMM = personal communication, PREC = positive reciprocity, NREC = negative reciprocity.

CR = construct reliability, AVE = average variance extracted.

Diagonal numbers (in italics) are the square root of AVE values and off-diagonal numbers are correlations;

**significant at *p < 0.01; significant at p < 0.05 (two-tailed)

Table 7.

Hierarchical regressions for customer spending in Study 2 with personal communication, positive and negative reciprocity as moderating variables

Independent variables	Model 1 Control effects	Model 2 Linear effects	Model 3 Two-way interaction effects	Model 4 Three-way interaction effects
Intercept (b0)	32.54***	40.49***	35.76***	37.76***
Gender	3.06*	2.49	3.33**	3.79**
Age	−0.17**	−0.20**	−0.23***	−0.24**
Income	0.70	0.51	0.46	0.44
Compensation behavior	−0.82	−0.78	−0.80	−0.76
BLOY X1		−1.47	3.44	−2.26
BLOY X2		−3.12	−6.85**	−3.25*
PCOMM		7.63***	11.94***	11.79***
PREC		−0.89	−0.70	−1.55
NREC		−0.72	−0.03	0.12
BLOY X1 × PCOMM			−10.45***	−10.52***
BLOY X2 × PCOMM			7.75*	7.91**
BLOY X1 × PREC			−1.93	−0.51
BLOY X2 × PREC			0.93	0.55
PCOMM × PREC			0.99	−7.16**
BLOY X1 × NREC			1.15	0.53
BLOY X2 × NREC			−1.04	−0.95
PCOMM × NREC			−2.61**	−4.54*
BLOY X1 × PCOMM × PREC				11.75***
BLOY X2 × PCOMM × PREC				2.33
BLOY X1 × PCOMM × NREC				2.89
BLOY X2 × PCOMM × NREC				0.05
R² (%)	3.7	12.9	17.3	21.1
Adjusted R² (%)	2.4	10.2	12.2	15.0
F (df)	2.78 (4)**	4.69 (9)***	3.41 (17)***	3.48 (21)***
F-change (df)	--	6.03 (5)***	1.85 (8)*	3.27 (4)**

Notes:

Unstandardized coefficients. BLOY = behavioral loyalty, PCOMM = personal communication, PREC = positive reciprocity, NREC = negative reciprocity.

***

p < 0.01; **p < 0.05; *p < 0.10

Table 8.

Conditional effects of behavioral loyalty on customer spending (Study 2)

Personal communication	Positive reciprocity	Coefficient BLOY -> customer spending		T-value	95% CI LL	95% CI UL
Low	Low	X1	7.57**	2.34	1.19	13.96
Low	Moderate	X1	2.48	0.92	−2.83	7.80
Low	High	X1	−2.61	−0.70	−9.96	4.75
Low	Low	X2	−6.75*	−1.96	−13.53	0.03
Low	Moderate	X2	−7.24***	−2.67	−12.59	−1.89
Low	High	X2	−7.73*	−1.96	−15.52	0.05
High	Low	X1	−11.43***	−3.05	−18.79	−4.06
High	Moderate	X1	−7.12***	−2.60	−12.51	−1.72
High	High	X1	−2.81	−0.71	−10.58	4.96
High	Low	X2	−0.52	−0.14	−7.89	6.86
High	Moderate	X2	0.85	0.31	−4.53	6.24
High	High	X2	2.23	0.59	−5.20	9.65

Notes:

Unstandardized coefficients. BLOY = behavioral loyalty, CI = confidence interval, LL = lower limit, UL = upper limit;

***

p < 0.01; **p <0 .05; *p < 0.10