Exploring the Impacts of Service Gaps and Recovery Satisfaction on Repurchase Intention: The Moderating Role of Service Recovery in the Restaurant Industry

Abstract

This study integrates Expectation–Confirmation Theory (ECT) and the existing literature on service failure to propose a novel service gaps model. This model categorizes service gaps into three dimensions: Gap 1—the service delivery system gap; Gap 2—the customer needs and requests gap; and Gap 3—the unprompted and unsolicited service behaviors gap. This study examines the effects of these service gaps and recovery satisfaction on repurchase intention, as well as the moderating role of service recovery in the relationship between recovery satisfaction and repurchase intention. Based on the findings, this study offers practical recommendations to address service gaps, enhance recovery satisfaction, and ultimately improve customers’ repurchase intentions in the restaurant industry.

1. Introduction

Over the past three years, the COVID-19 pandemic has significantly influenced consumer behavior, with people adhering to social distancing measures and reducing their frequency of dining out. In contrast, online shopping and food ordering have increased considerably, exerting a substantial impact on the restaurant industry. In response, many restaurants have introduced takeout and delivery services and developed new product formats—such as frozen meal kits and ready-to-eat foods—to meet evolving customer demands (Taiwan Trend Research, 2022). To further expand their market reach, many business operators have partnered with major logistics providers and joined e-commerce platforms to diversify their sales channels and enhance revenue.

According to statistics from the Department of Statistics, Ministry of Economic Affairs, R.O.C. (2023), the restaurant industry reported a turnover of TWD 76 billion in November 2022, representing an annual growth rate of 13.3%. This increase was partly driven by promotional events such as the “11.11” shopping festival and the 2022 FIFA World Cup in Qatar. During this period, the annual growth rates of restaurants and beverage shops reached 13.3% and 9.8%, respectively. Additionally, the increase in air travel passenger volume stimulated demand for in-flight catering services, resulting in a 24.9% annual growth rate in the catering and group meal sectors (News Radio, 2022). Despite this post-pandemic recovery, service-related complaints in the restaurant sector have become increasingly prevalent. These complaints typically arise when customers are dissatisfied with service quality or feel that their concerns have not been addressed fairly. In many cases, such grievances are shared publicly on social media or formally submitted to consumer protection agencies.

Bitner et al. (1990) analyzed service failure incidents in hotels, airlines, and restaurants, categorizing them into three types: (1) employee responses to service delivery system failures, (2) employee responses to customer needs and requests, and (3) unprompted and unsolicited employee actions. Their classification reveals that the service delivery process is often influenced by a range of uncontrollable factors, including unexpected circumstances, fluctuating customer expectations, and a high degree of human interaction. The tourism and hospitality industry, in particular, is highly vulnerable to such failures due to its dependence on intensive interpersonal interactions and customer participation. Customers frequently evaluate service quality based on these interactions, making effective management of customer contact points essential. If a business fails to respond to a service failure with appropriate recovery efforts, even minor issues can escalate into serious reputational damage (Piercy & Archer-Brown, 2014; Sarkar Sengupta et al., 2015). As customer expectations continue to rise in parallel with improved living standards and heightened awareness of consumer rights, addressing service failures through timely and effective recovery has become a strategic imperative for restaurant operators.

Given these challenges, avoiding service failures and implementing effective recovery strategies are now top priorities. In recent years, research in hospitality and tourism has increasingly focused on service failure and recovery (Koc, 2019; Ogbeide et al., 2017; Shams et al., 2021). However, the extent to which service recovery efforts—along with customer satisfaction with those efforts—can promote repeat patronage remains insufficiently explored (Owusu et al., 2025). In particular, the interrelationships among service recovery, recovery satisfaction, and repurchase intention have not been thoroughly addressed in the hospitality marketing literature. To address these gaps, the present study investigates the impact of recovery satisfaction on repurchase intention and examines the moderating role of service recovery in this relationship. Additionally, this study integrates Expectation–Confirmation Theory (ECT) and prior research on service failure to develop a novel service gaps model tailored to the restaurant industry. This model further explores the influence of various service gaps on repurchase intention. Finally, this study offers practical recommendations to bridge service gaps and enhance recovery satisfaction, providing actionable insights for restaurant operators.

2. Conceptual Development and Research Hypotheses

2.1. Conceptual Development

2.1.1. Service Failure

Rao et al. (2011) defined service failure as service performance that is below customer expectations. Thus, a service failure occurs when a customer perceives the service received at service touchpoints or during service delivery as lower than initially expected. Service failures can often occur during service delivery, and service failure typically affects customers’ service experience (Steyn et al., 2011). Moreover, service failure may cause customers to seek out competitors to satisfy their needs. This results in lost revenue and increased costs for the original company and hurts employee morale (Wu et al., 2018). Bitner et al. (1990) utilized the Critical Incident Technique (CIT) to analyze service failures in airlines, hotels, and restaurants. They categorized these failures into three groups: employee responses to service delivery system failures, employee responses to customer needs and requests, and unprompted and unsolicited employee actions. Based on Bitner et al.’s (1990) classification, Kelley et al. (1993) further categorized service failures into 15 categories in three main groups for the retail industry. Similarly, Hoffman et al. (1995) classified service failures into 11 categories in three main groups for restaurants.

2.1.2. Service Recovery

Service recovery refers to companies’ actions to compensate customers who experience dissatisfaction or inappropriate services (i.e., service failures) and bring customers back to the company satisfactorily (C. H. Choi et al., 2014). Chou (2015) found that service recovery is an action taken by a company in response to service failures, whereby customers typically expect that the losses they have incurred will be fairly remedied during the process. In other words, service recovery is a process in which a company attempts to correct mistakes. The goal of the recovery effort is to move a customer from a state of dissatisfaction to a state of satisfaction and, in turn, build a strong relationship with the customer (Azemi et al., 2019; Kelley et al., 1993). In studying service recovery, justice theory is a representative theory used to explain people’s responses to conflict situations. This pertains to people’s evaluation of how fairly they have been served during service recovery (Chang et al., 2012). Justice theory can be further categorized into distributive justice, interactional justice, and procedural justice. Distributive justice refers to how customers evaluate the fairness of service recovery outcomes. It typically includes discounts, coupons, free meals, changing restaurant layouts, etc. Interactional justice refers to the extent to which customers evaluate the fairness of the service provider’s attitude during the service recovery process. This often includes treating customers with dignity and respect or providing appropriate explanations for service failures. Procedural justice refers to the degree of fairness in the service recovery procedures and policies. It typically includes wait times, responsiveness, and flexibility in the recovery process (Ozkan-Tektas & Basgoze, 2017; Smith et al., 1999).

2.1.3. Recovery Satisfaction and Repurchase Intention

Recovery satisfaction is a positive status of emotion perceived by customers in the process and result of recovering the failed service (T. Kim et al., 2009). That means recovery satisfaction refers to customers’ overall satisfaction with the secondary service (remedial action) of a service provider after a service failure; it is different from customers’ satisfaction with the first service encounter (Kuo & Wu, 2012). Intention is an individual’s subjective probability or possibility of engaging in a specific behavior. Intention also pertains to an individual’s subjective judgment of possible future behavior (Fishbein & Ajzen, 1975; Schiffman & Kanuk, 2007). In contrast, repurchase intention refers to an individual’s subjective tendency or possibility to purchase a product or service repeatedly. Chuah et al. (2022) found that repurchase intention not only reflects customers’ willingness to purchase products or services repeatedly but also acts as a behavioral expression of loyalty. If a company can establish and maintain customer loyalty, it will help increase its revenue and sustainable competitive advantage (Latif et al., 2020).

2.1.4. Expectation–Confirmation Theory (ECT)

ECT was proposed by Oliver (1980) to explain that individuals have certain expectations for a product or service before purchasing it. After experiencing the product or service, they form a new perception of its performance. Subsequently, they compare this perceived performance with their initial expectations to determine the level of confirmation. Positive disconfirmation occurs when the perceived performance exceeds initial expectations, simple confirmation occurs when it meets initial expectations, and negative disconfirmation occurs when it falls below them (Oliver, 1980; J.-N. Wang et al., 2018). Ultimately, the level of disconfirmation influences a person’s satisfaction and their intention to make repeat purchases of the product or service (Bhattacherjee, 2001; Veeramootoo et al., 2018).

ECT has been widely used to study satisfaction. For example, Zehrer et al. (2011) applied ECT to assess international travelers’ satisfaction with hotels. They showed that when international travelers feel negative disconfirmation of hotel products or services, it has a greater impact on global traveler satisfaction than positive disconfirmation. Belarmino et al. (2021) discussed the motivation for using online meal delivery platforms (such as Uber Eats) before and during COVID-19 quarantine periods. They found that whether it was before or during the quarantine period, the level of consumer confirmation belief significantly influenced satisfaction (Belarmino et al., 2021). X. Wang et al. (2023) explored the impact of passengers’ emotions (generated from disconfirmation) on their intention to recommend an airline during COVID-19. They found that positive disconfirmation of passengers’ emotions significantly positively affected their satisfaction.

This study applies ECT and synthesizes the existing literature on service failure to construct a research framework tailored to the restaurant industry, as illustrated in Figure 1. The model consists of two main components: the expectation–confirmation process and the repurchase intention process. In the expectation–confirmation process, this study draws on the service failure classification framework proposed by Bitner et al. (1990, 1994) and applies Expectation–Confirmation Theory (ECT) to assess disconfirmation by comparing customer expectations with perceived service performance. First, customers evaluate disconfirmation related to the service delivery system; when expectations exceed actual performance, Gap 1—the service delivery system gap—emerges. Second, customers assess disconfirmation concerning their needs and requests. If their expectations are not met, Gap 2—the customer needs and requests gap—occurs. Third, disconfirmation is evaluated in relation to unprompted and unsolicited employee behaviors. When performance falls short of expectations, Gap 3—the unprompted and unsolicited service behaviors gap—arises. These three service gaps reflect distinct dimensions of service failure in the restaurant industry. In the repurchase intention process, this study investigates how these gaps, along with recovery satisfaction, influence customer repurchase intention. Furthermore, it examines the moderating role of service recovery in the relationship between recovery satisfaction and repurchase intention.

2.2. Research Hypotheses

Service failures occur when customers perceive a discrepancy between the service they expect and the service actually delivered, often leading to dissatisfaction (Weitzl et al., 2018). These failures can result from a variety of factors, including staff errors, customer behavior, or unforeseen incidents. Harrison-Walker (2019) identified common types of service failures, such as excessive wait times, rude or slow staff responses, incorrect billing, and poor maintenance. These failures may lead to negative customer outcomes, including complaints, negative word-of-mouth, and emotional responses such as unhappiness, discomfort, or dissatisfaction (Liang et al., 2025). Mandl and Hogreve (2020) emphasized the detrimental impact of service failures on customers’ repurchase intentions. Similarly, J. Zhang et al. (2024) found that service failures negatively influence users’ willingness to reuse AI chatbot services.

In the context of dining, the service process typically includes multiple stages: arrival and seating (through reservation or walk-in, waiting area, and being guided to a table), ordering food and beverages, enjoying the meal, optional dessert ordering, and finalizing the experience through billing, payment, and tipping. Disruptions at any stage can lead to dissatisfaction and reduced likelihood of customer return. This aligns with findings by (S. Choi & Mattila, 2008), who observed a negative relationship between the frequency or magnitude of service failures (i.e., service gaps) and repurchase intention. Accordingly, this study posits that if customers perceive delays or absence of service, procedural errors, or mistakes in pricing and packaging, the perceived service gap will widen. Similarly, if the restaurant staff fails to promptly accommodate special requests or preferences, the service gap will increase. Additionally, failures in proactive service behaviors—such as errors in food delivery or unprocessed orders—further exacerbate the perceived gap. In summary, the greater the gap between expected and actual service performance, the lower the customer’s satisfaction and likelihood of repurchase. Based on this understanding, the following hypotheses is proposed:

H_1. 

The service delivery system gap (Gap 1) negatively impacts customer repurchase intention within the restaurant industry.

H_2. 

The customer needs and requests gap (Gap 2) negatively impacts customer repurchase intention within the restaurant industry.

H_3. 

The unprompted and unsolicited service behaviors gap (Gap 3) negatively impacts customer repurchase intention within the restaurant industry.

Past studies have confirmed that recovery satisfaction positively impacts customers’ repurchase intentions. For example, Tax et al. (1998) claimed that customer satisfaction with service recovery has a positive effect on a customer’s revisit intention. Holloway et al. (2005) observed that customers who experience satisfactory service recovery following a failure are more likely to engage in positive word-of-mouth communication and exhibit stronger repurchase intentions. Conversely, Collier and Bienstock (2006) reported that dissatisfaction with recovery efforts negatively influences customers’ future behavioral intentions, such as switching to competitors or spreading negative word-of-mouth. Similarly, T. Kim et al. (2009) identified a significant relationship between recovery satisfaction and revisit intentions in the context of upscale hotels. Extending this understanding to the restaurant industry, Alenazi (2021) demonstrated that satisfaction with service recovery has a positive and significant impact on repurchase intentions among restaurant customers in Saudi Arabia. Prior studies have consistently established that satisfaction with service recovery is a critical determinant of repurchase intentions. Building on this foundation, the current study formulates the following hypothesis:

H_4. 

Recovery satisfaction positively affects customer repurchase intention within the restaurant industry.

Maxham and Netemeyer (2003) found that both distributive and procedural justice positively influence post-recovery satisfaction. Gautam (2011) showed that customer-perceived justice was positively related to recovery satisfaction, and the strength of its effect was in the order of distributive, interactional, and procedural justice. Similarly, Ding and Lii (2016) found that distributive justice, such as discounts and refunds, leads to the highest customer satisfaction after service recovery. Chen et al. (2018) explained that “high recovery” refers to recovery efforts that exceed consumers’ expectations, leading to positive disconfirmation and increased satisfaction. Owusu et al. (2025) surveyed online shoppers who experienced service failures followed by subsequent recovery efforts from retailers. Their findings revealed that consumers’ perceptions of the quality of retailers’ recovery policies significantly influence consumer satisfaction and repurchase intentions. These findings demonstrate that positive perceived justice can eliminate the emotional unfairness and ambivalence customers feel after service failures, and it has the potential to further increase customer satisfaction (Ozkan-Tektas & Basgoze, 2017). Furthermore, even though companies cannot eliminate service failures, companies can work on service recovery and effectively address such failures to increase customer trust and perhaps even improve customer satisfaction and loyalty in the future (Steyn et al., 2011). In other words, service recovery and complaint handling represent meaningful opportunities that provide chances for companies to communicate with their customers (Bougoure et al., 2016). Based on the above literature review, it is evident that if service recovery can meet customer expectations, it may lead to better outcomes and higher satisfaction and further increase customer repurchase intention (Chen et al., 2018; L. Wang et al., 2024). Therefore, the present study posits that when a service failure occurs in a restaurant, if it is remedied fairly, it will likely improve or increase customer satisfaction with that service provider and further increase customer repurchase intention. Therefore, this study proposes the following hypothesis:

H_5. 

Service recovery moderates the relationship between recovery satisfaction and customer repurchase intention within the restaurant industry.

3. Methodology

3.1. Research Design

This study applied a survey questionnaire method for data collection. First, this study applies Expectation–Confirmation Theory (ECT) and draws on the work of Bitner et al. (1990, 1994) to define three service gaps: Gap 1—the service delivery system gap; Gap 2—the customer needs and requests gap; and Gap 3—the unprompted and unsolicited service behaviors gap. In addition, following Bitner et al. (1990, 1994), this study develops measurement items to assess the differences between customers’ service expectations prior to receiving service and their perceived service performance afterward, based on the approach suggested by Rao et al. (2011). Second, this study defined service recovery as the actions taken by restaurant operators to compensate customers who have experienced service failures (C. H. Choi et al., 2014). Furthermore, this study referred to justice theory (distributive justice, interactional justice, and procedural justice) to develop the service recovery measurement items (Chang et al., 2012; Ozkan-Tektas & Basgoze, 2017; Smith et al., 1999). Third, this study defined recovery satisfaction as an overall evaluation of the level to which customers perceive the performance of services provided by the restaurant to be consistent with their expectations (Oliver, 1980). The measurement items were developed based on Eveleth et al. (2015). Finally, this study defined repurchase intention as the possibility or subjective tendency that customers are willing to consider revisiting a restaurant in the future (Fishbein & Ajzen, 1975; Schiffman & Kanuk, 2007). The measurement items were developed based on Hsu et al. (2014) and Pee et al. (2018).

The development of the measurement items in each of the study constructs was based on support from the relevant literature. The measurements were adapted and reorganized according to the context of the restaurant industry. Before formally distributing the questionnaires, this study was repeatedly discussed with three experts in hospitality management and service quality management. We carefully revised the items and wording based on the experts’ feedback to confirm the thoroughness and appropriateness of the questionnaire content. Furthermore, this study corrected unclear or misleading items through a pilot test to improve the questionnaire’s overall accuracy. The first part of the questionnaire asked respondents about demographic variables, including gender, age, occupation, education level, marital status, and the average number of meals eaten out in a week. Furthermore, the demographics (gender and age) were included as control variables as they may affect restaurant choice (E. Kim et al., 2017). The second part of the questionnaire included the items for each research construct. All items are measured using a seven-point Likert scale (1 = completely disagree; 7 completely agree). The measurement items of each construct are presented in

Table 1. The questionnaire items and related references.

Constructs		Items
Bitner et al. (1990); Bitner et al. (1994); Kuo et al. (2011);
Service gaps	Gap 1—the service delivery system gap	SDS1: During my meal, the service provided by the restaurant (e.g., food delivery, water service, etc.) was not as good as I had originally expected. SDS2: The server was less friendly than I had originally expected. SDS3: The restaurant failed to inform me in advance about dishes that would not be available that day. SDS4: The restaurant failed to provide detailed information about the ingredients used in the dishes (e.g., whether onions, garlic, eggs, pork, or beef were included).
	Gap 2—the customer needs and requests gap	CNR1: The quality of the ingredients was not as good as I had originally expected (e.g., meals had an odor, ingredients were not fresh, etc.). CNR2: The cost-performance (quality-to-price ratio) of the meal ingredients was not as high as I had originally expected. CNR3: The portion size of the meal I received was slightly different from what I had originally expected. CNR4: The meal I received was different from the pictures I had previously seen on the restaurant’s official website or menu.
	Gap 3—the unprompted and unsolicited service behaviors gap	UPS1: The server was not sufficiently informed about the restaurant’s policies, resulting in me having to pay additional unexpected fees. UPS2: The personal behavior of the server (e.g., chatting, using their phone, etc.) was not as professional as I had originally expected. UPS3: The server was not sufficiently familiar with the menu to assist me in completing my order. UPS4: The server was not sufficiently familiar with the checkout system to assist me in completing the payment process accurately and efficiently.
Ding and Lii (2016); Ha and Jang (2009); Ozkan-Tektas and Basgoze (2017); Smith et al. (1999)
Service recovery		SR1 In solving my problem, the restaurant gave me what I needed. SR2 The compensation I received from the restaurant was fair. SR3 I have been compensated for the inconvenience or any loss caused by restaurant. SR4 I think the restaurant put a lot of effort into solving my problem. SR5 I think the restaurant considered my question from my point of view. SR6 I think the restaurant was very courteous in communicating with me to solve my problem. SR7 I think the restaurant tried to solve my problem as quickly as possible. SR8 The restaurant showed adequate flexibility in dealing with my problem. SR9 I think restaurants have fair policies and practices to deal with problems.
Chang et al. (2012); Chen et al. (2018); Ding and Lii (2016);
Recovery satisfaction		RS1 I was satisfied with the way the restaurant resolved the problem. RS2 I was satisfied with how the restaurant handled my problem. RS3 For me, the restaurant provided a satisfactory solution to my problem.
Ding and Lii (2016); Hsu et al. (2014); Pee et al. (2018)
Repurchase intention		RI1 I plan to eat at this restaurant again in the future when we need to have a gathering. RI2 I will probably still visit this restaurant when I want to dine next time. RI3 I expect to visit this restaurant again soon when I want to dine.

.

3.2. Data Collection and Sampling

This study distributed the questionnaire via the Internet to investigate customer responses from those who have experienced service failure and recovery within the restaurant industry. This study used purposive sampling as the sampling design, allowing the researchers to determine suitable respondents for the questionnaire (M. Zhang et al., 2020). Regarding data collection, we used an online questionnaire platform (SurveyCake), and the questionnaire’s hyperlinks were posted through online communities to invite and guide customers who had unpleasant (or unsatisfactory) dining experiences to complete the questionnaire. The questionnaires were distributed for one week starting from 14 August 2023, and 365 valid questionnaires were collected. Finally, the questionnaires were statistically analyzed.

Among the 365 valid samples collected, 58.6% of respondents were women, 47.7% of respondents were between 31 and 40 years old, and 25.8% were between 41 and 50 years old. Regarding marital status, 54% of the respondents were married. Regarding education level, 61.1% of all respondents had a university education and 16.7% had a college education. Respondents’ occupations were primarily in the service and manufacturing industries, accounting for 33.4% and 33.7%, respectively. The respondents’ average number of meals eaten out in a week was five times (inclusive) or less, accounting for 58.9%, followed by 6–10 times, accounting for 27.9%.

Table 2. Demographic characteristics of the respondents (n = 365).

Percentage of Respondents			Percentage of Respondents
Gender	Male	41.4	Occupation	Student	2.3
	Female	58.6		Government sector	9.3
Age	≤20 years old	0.8		Service industry	33.4
	21–30 years old	18.9		Manufacturing industry	33.7
	31–40 years old	47.7		Financial industry	7.9
	41–50 years old	25.8		High-tech industry	9.0
	≥51 years old	6.8		Other	4.4
Education Level	High school and below	11.5	The average number of meals eaten out in a week	≤5 times	58.9
	College	16.7		6–10 times	27.9
	University	61.1		11–15 times	6.8
	Master’s degree and above	10.7		≥16 times	6.4
Marital Status	Single	45.7
	Married	54.0
	Other	0.3

presents the demographic details of the study respondents.

3.3. Reliability and Validity

First, this study based the measurement items of each construct in this study on prior literature, which was verified by a panel discussion with experts and scholars. As such, the questionnaire has good content and face validity (Hair et al., 2009). Second, this study used item analysis to measure the relevance of each questionnaire item. When the item-to-total correlation score was lower than 0.4, the item was eliminated from further analysis. The results indicate that it was unnecessary to delete any measurement items and that the research variables (i.e., service gaps, service recovery, recovery satisfaction, and repurchase intention) were appropriate. Third, this study conducted factor analysis and deleted measurement items that did not meet the criteria for factor retention. For service gaps, items SDS3 and SDS4 were removed because their absolute factor loadings were below 0.70. Item CNR3 was deleted due to a factor loading difference of less than 0.30 between its primary and secondary factors. In the service recovery construct, item SR1 was excluded because its absolute factor loading was below 0.70. For recovery satisfaction and repurchase intention, all items met the required standards and were retained. After these deletions, the factor structure aligned well with the theoretical model, and the construct validity of the scale was found to be satisfactory. Finally, this study used Cronbach’s α to measure the internal consistency of the items in each construct. Cronbach’s α for each construct ranged from 0.870 to 0.959, greater than the reliability threshold of 0.7 suggested by Nunnally (1978).

Table 3. Reliability and validity results for each construct.

Constructs		Items	Factor Loadings	Item-to-Total Correlation	Cronbach’s α
Service gap	Gap 1—the service delivery system gap	SDS1	0.865	0.438	0.870
		SDS2	0.723	0.438
	Gap 2—the customer needs and requests gap	CNR1	0.821	0.651
		CNR2	0.832	0.611
		CNR4	0.751	0.633
	Gap 3—the unprompted and unsolicited service behaviors gap	UPS1	0.813	0.777
		UPS2	0.719	0.697
		UPS3	0.870	0.802
		UPS4	0.851	0.778
Service recovery		SR2	0.874	0.834	0.957
		SR3	0.858	0.813
		SR4	0.889	0.852
		SR5	0.868	0.825
		SR6	0.868	0.825
		SR7	0.896	0.860
		SR8	0.886	0.847
		SR9	0.875	0.834
Recovery satisfaction		RS1	0.873	0.897	0.948
		RS2	0.862	0.881
		RS3	0.879	0.897
Repurchase intention		RI1	0.852	0.896	0.959
		RI2	0.896	0.928
		RI3	0.889	0.917

outlines the reliability and validity test results we performed on the final questionnaire items. The findings indicate that the measurement items on each construct met the requirements of reliability and validity.

3.4. Common Method Bias (CMB)

First, all measurement items of this study were reviewed and tested by experts and scholars. Second, the items were randomly allocated by separating the independent and dependent variables before they were officially distributed (Tourangeau et al., 2000). Third, before the questionnaire was tested, respondents were reminded that there were no correct answers and that they were asked to respond according to their true feelings. Fourth, the results of Harman’s single-factor test showed that the first factor explained 35.4% of the covariance obtained by converging all the items without rotating the axes. This means that the data collected in this study did not result in a single factor explaining the primary source of variance (Podsakoff et al., 2003). Based on the above assessments, we confirmed that CMB was likely not a severe threat to these research results.

4. Data Analysis and Results

To test the hypotheses, this study used hierarchical regression analyses. Since the regression equation contains one interaction term, this study standardizes all variables to reduce the potential effects of multicollinearity (Cohen et al., 2003). The technique of least squares was used with the control variables entered as a block in step 1 (gender, age, and education level), followed by the main effects in step 2 (Gap 1—the service delivery system gap; Gap 2—the customer needs and requests gap; and Gap 3—the unprompted and unsolicited service behaviors gap, service recovery, and recovery satisfaction), and the moderators in step 3. Specifically, the following regression equation was analyzed in three hierarchical steps:

Y₁ = α + β₁X₁ + β₂X₂ + β₃X₃ + β₄X₄ + β₅X₅ + β₆X₆ +β₇X₇ + β₈X₈ + β₉X₇X₈ + ε

where Y₁ = repurchase intention, α = intercept, X₁ = gender, X₂ = age, X₃ = education level, X₄ = Gap 1, X₅ = Gap 2, X₆ = Gap 3, X₇ = service recovery, X₈ = recovery satisfaction, and ε = random disturbance terms.

The results are summarized in

Table 4. Hierarchical regression results predicting repurchase intention.

		β	Beta	t	Sig.	R2	Adj. R2	ΔR2	ΔF	Sig. ΔF
Step 1	Constant	1.221		3.513	0.001	0.036	0.028	0.036	4.503	0.004 **
	Gender	−0.249	−0.123	−2.375	0.018 *
	Ages	−0.166	−0.141	−2.656	0.008 **
	Education levels	−0.109	−0.088	−1.660	0.098
Step 2	Constant	0.387		1.664	0.097	0.598	0.589	0.562	99.598	0.000 ***
	Gender	−0.093	−0.046	−1.326	0.186
	Ages	−0.093	−0.079	−2.275	0.023 *
	Education levels	0.021	0.017	0.484	0.629
	Gap 1	−0.116	−0.116	−2.837	0.005 **
	Gap 2	−0.091	−0.091	−2.289	0.023 *
	Gap 3	−0.163	−0.163	−3.564	0.000 ***
	Service recovery	0.348	0.348	4.094	0.000 ***
	Recovery satisfaction	0.376	0.376	4.430	0.000 ***
Step 3	Constant	0.208		0.910	0.364	0.623	0.613	0.025	23.424	0.000 ***
	Gender	−0.073	−0.036	−1.074	0.283
	Ages	−0.089	−0.076	−2.248	0.025 *
	Education levels	0.022	0.018	0.519	0.604
	Gap 1	−0.170	−0.170	−4.127	0.000 ***
	Gap 2	−0.090	−0.090	−2.353	0.019 *
	Gap 3	−0.190	−0.190	−4.239	0.000 ***
	Service recovery	0.335	0.335	4.061	0.000 ***
	Recovery satisfaction	0.416	0.416	5.031	0.000 ***
	Service recovery X Recovery satisfaction	0.146	0.168	4.840	0.000 ***

*** p < 0.001; ** p < 0.01; * p < 0.05.

. Table 4 shows the variance of the dependent variable (repurchase intention) that could be explained by the independent variables (Gap 1–3, service recovery, and recovery satisfaction). To assess Hypotheses 1–4, this study first examined the results in step 2. In this step, Gap 1 (p < 0.01), Gap 2 (p < 0.05), Gap 3 (p < 0.001), and recovery satisfaction (p < 0.001), when entered along with service recovery, were significantly influenced customer repurchase intention. In other words, the service gaps (Gap 1–3) negatively affected customer repurchase intention. The results also indicate that recovery satisfaction positively impacted customer repurchase intention within the restaurant industry, providing initial support for Hypotheses 1–4. The inclusion of service gaps (Gap 1–3), service recovery, and recovery satisfaction in step 2 of the model explained the significant variance in customer repurchase intention beyond that explained by the control variables in step 1 (p < 0.001); the step 2 equation explained a total of R² = 0.598.

The results of Hypotheses 1–4 are robust when the entire model is specified in step 3 to assess Hypothesis 5. Specifically, the main effects of Gap 1 (p < 0.001), Gap 2 (p < 0.05), Gap 3 (p < 0.001), and recovery satisfaction (p < 0.001) and the moderators of recovery satisfaction × service recovery (p < 0.001) significantly influenced customer repurchase intention. In other words, service recovery had a moderating effect on the relationship between recovery satisfaction and customer repurchase intention in the restaurant industry. The inclusion of the moderator term (recovery satisfaction × service recovery) in step 3 explained the significant variance beyond step 2 (ΔR² = 0.025, p < 0.001). The fully specified model (i.e., including steps 1, 2, and 3) resulted in R² = 0.623 (p < 0.001). Thus, all five hypotheses were supported in the hierarchical regression analysis.

5. Discussion and Conclusions

5.1. Theoretical Implications

In the field of hospitality and tourism studies, preventing service failures and addressing them through effective recovery strategies have emerged as critical priorities. Recent research has increasingly explored the dynamics of service failure and recovery (Koc, 2019; Ogbeide et al., 2017; Shams et al., 2021). However, the extent to which consumers engage with recovery efforts following a service failure, and their satisfaction with these efforts as drivers of repeat purchasing behavior, remains insufficiently examined (Owusu et al., 2025). This highlights a research gap in understanding the interplay among service recovery, recovery satisfaction, and repurchase intentions, particularly given their pivotal role in hospitality and tourism marketing. Therefore, this study explored the effects of recovery satisfaction on repurchase intention and the moderating effect of service recovery on the relationship between recovery satisfaction and repurchase intention. Moreover, this study identified service failure incidents based on the framework established by Bitner et al. (1990, 1994) and utilized the ECT approach to construct a service gaps model tailored to the restaurant industry. This study further categorized service gaps into three types: Gap 1—the service delivery system gap; Gap 2—the customer needs and requests gap; and Gap 3—the unprompted and unsolicited service behaviors gap. It also identified strategies to bridge these gaps, which have been shown to enhance customers’ intention to revisit restaurants (N. Kim et al., 2021).

Regarding Gap 1—the service delivery system gap—restaurant operators should emphasize training staff on basic etiquette since customer evaluations are not performed solely on the outcome of a restaurant; they are evaluations of the service delivery process (Parasuraman et al., 1985). In other words, during the customer dining process, including welcoming, seating arrangements, food delivery, water delivery, etc., the service provided by staff is a significant factor in determining the customer’s perceived performance. This indicates that staff characteristics such as being uncordial, unfriendly, or uncourteous can affect the service quality of a restaurant by causing uncertainty concerning quality service delivery (Tseng, 2019). Regarding Gap 2—the customer needs and requests gap—restaurant operators should emphasize the importance of food quality, including the indicators of freshness, tastiness, healthiness, and food presentation (Konuk, 2019). For example, they should ensure ingredients are fresh and that the food is consistent with the pictures on the official website or menu. Moreover, restaurant operators should provide customers with a dining experience that represents good value for money. Regarding Gap 3—the unprompted and unsolicited service behaviors gap—this study recommends that restaurant employees be equipped with professional literacy to work more accurately and efficiently, leading to better work quality. Regarding professional literacy, this study refers to the knowledge, abilities, skills, and other characteristics that restaurant employees require to perform their work effectively (Busulwa et al., 2022). For example, employees should be familiar with the restaurant policy, menu design, checkout system, etc., to assist customers in addressing their issues. Moreover, restaurant operators should recognize that a restaurant’s employees strongly influence the service performance perceived by customers. As such, it is essential to determine the professional literacy of restaurant employees in shaping customer-perceived service performance. Therefore, professional literacy should be enhanced among restaurant employees to provide quality service to their customers.

This study underscores the importance of understanding and implementing effective service recovery strategies to enhance recovery satisfaction and repurchase intentions. Building on justice theory, the present study examined whether service recovery moderates the relationship between recovery satisfaction and repurchase intention. The findings confirm that fair, well-executed recovery efforts strengthen the positive effect of recovery satisfaction on customers’ repurchase intention. In particular, when customers perceive the recovery process as just and considerate, their likelihood of returning increases. Accordingly, this study encourages restaurant operators to communicate courteously, resolve problems promptly, and adopt the customer’s perspective by providing appropriate compensation for any inconvenience or loss (Ding & Lii, 2016; Ha & Jang, 2009; Ozkan-Tektas & Basgoze, 2017). Hazée et al. (2017) further demonstrated that co-created service recovery—wherein customers actively participate in shaping or customizing the recovery experience through collaborative engagement with the provider—significantly enhances both recovery satisfaction and repurchase intention. Given the frequent and close interactions between customers and service personnel in the restaurant industry, service failures are often inevitable. However, involving customers in the formulation of recovery plans not only conveys respect but also increases their satisfaction with the recovery outcome and strengthens their intention to return. Moreover, service recovery strategies should be context-sensitive and thoughtfully tailored. As noted by Kuo et al. (2011), frontline employees may adopt different recovery approaches, even in response to similar service failures. Therefore, staff should be trained to diagnose the root cause of the failure and select the most appropriate recovery strategy depending on the intended goal—whether to restore satisfaction or promote repeat patronage. A carefully designed and strategically implemented recovery process can thus play a critical role in improving customer satisfaction and encouraging long-term loyalty.

5.2. Practical Implications

This study investigated the effects of service gaps and recovery satisfaction on customer repurchase intention, as well as the moderating role of service recovery in the relationship between recovery satisfaction and repurchase intention within the restaurant industry. The results indicated that all three service gaps had a significant negative impact on customer repurchase intention, while service recovery positively moderated the relationship between recovery satisfaction and repurchase intention.

The results of this study provided strong support for the proposed model hypotheses. As shown in Table 4, all three service gaps (Gap 1–Gap 3) were found to have a significant negative effect on customers’ repurchase intention. This indicates that narrowing these gaps is associated with a higher likelihood of customer repurchase. Accordingly, restaurant operators are advised to proactively mitigate discrepancies between customers’ expectations and the actual service performance to enhance repurchase intention. Among the three gaps, Gap 3—the unprompted and unsolicited service behaviors—had the strongest negative impact on repurchase intention (β = −0.163), followed by Gap 1—service delivery system failures (β = −0.116)—and Gap 2—failures to meet customer needs and requests (β = −0.091). These results underscore the particularly detrimental role of Gap 3 in influencing repurchase decisions. This insight is especially relevant in the context of recent challenges faced by restaurant operators in Taiwan. In the aftermath of the COVID-19 pandemic, many restaurants experienced a sharp increase in workforce demand, prompting them to rely heavily on part-time or inexperienced staff. In many cases, inadequate training contributed to frequent service mistakes—even in routine tasks—leading to negative customer evaluations and reduced willingness to return. Given that service failures typically occur during interactions between frontline staff and customers (Parasuraman et al., 1985), this study recommends several targeted strategies for addressing each service gap:

• To minimize the service delivery system gap (Gap 1): Restaurants can implement standardized operating procedures and ensure their consistent execution across all service staff. This includes comprehensive onboarding and continuous training on service protocols, food delivery timeliness, hygiene standards, and order accuracy.
• To reduce the customer needs and requests gap (Gap 2): A more personalized and customer-centric approach is essential. Staff may be trained to actively listen, clarify special requirements, and communicate clearly with guests regarding preferences such as food allergies, portion sizes, or seating needs.
• To bridge the unprompted and unsolicited service behavior gap (Gap 3): Employees can be empowered to take initiative and offer proactive hospitality without prompting. This includes checking on customer satisfaction during the meal, offering refills, or providing service recovery gestures when issues arise.

Effectively addressing these service gaps can not only improve customers’ immediate experiences but also foster stronger loyalty and contribute to long-term business sustainability in the competitive restaurant industry (Matsuoka, 2022; Yan et al., 2025).

This study further revealed that recovery satisfaction positively influences customer repurchase intention. Additionally, service recovery was found to have a significant moderating effect on the relationship between recovery satisfaction and repurchase intention in the restaurant industry. These findings suggest that effective service recovery can amplify the positive impact of recovery satisfaction on customers’ future behavioral intentions. According to the service recovery paradox, customer satisfaction following a service recovery can exceed the level of satisfaction prior to the service failure, particularly when the recovery performance is perceived as high (Gohary et al., 2016; Lunardo et al., 2023). In other words, a well-executed service recovery may lead to even greater customer satisfaction than if no failure had occurred. Satisfaction, in turn, serves as a critical predictor of repurchase intention (Liao et al., 2017; Wei et al., 2023). Given these insights, this study recommends that when service failures occur, restaurant operators may respond promptly with a service recovery strategy that meets or exceeds customer expectations. In terms of distributive justice, restaurants may provide fair and appropriate compensation for any inconvenience or losses experienced. Regarding interactional justice, it is important that restaurant staff communicate respectfully, empathetically, and sincerely with affected customers. As for procedural justice, restaurants may adopt flexible, transparent, and efficient processes for resolving customer complaints. By enhancing these dimensions of service recovery, customer satisfaction with the recovery process is likely to increase, ultimately fostering greater customer loyalty and repurchase intention.

5.3. Limitations and Future Research

Drawing on the ECT and the existing literature on service failures, a service gaps model was developed and categorized into three sub-constructs: Gap 1—the service delivery system gap; Gap 2—the customer needs and requests gap; and Gap 3—the unprompted and unsolicited service behaviors gap. This study has further provided a convenient way to review these service gaps and identified effective recovery methods to enhance the impact of recovery satisfaction on repurchase intention. Despite these significant findings, several limitations should be acknowledged, and future research is needed to ensure the generalizability and robustness of the results. First, the measurement items for the three service gaps were designed based on the service failure classification framework proposed by Bitner et al. (1990, 1994). Therefore, future studies are encouraged to refine and validate the service gaps scale further to strengthen its psychometric properties. Second, this study employed a purposive sampling method, inviting respondents with prior unsatisfactory dining experiences to participate in the survey, resulting in 365 valid responses. Given the relatively limited sample size, the generalizability of the findings may be constrained. Future research should consider adopting broader sampling strategies and collecting larger, more representative samples. Third, as the data were collected from respondents in Taiwan, the cultural context may have influenced customers’ service expectations, failure perceptions, and recovery evaluations. Future studies should explore cross-cultural comparisons by conducting research in countries with different cultural backgrounds to enhance the generalizability of the findings. Additionally, this study encourages future research to further investigate the roles of customer participation and the illusion of control in shaping customer perceptions of service quality, service gaps, and service recovery, particularly within the tourism and hospitality context. Exploring these psychological mechanisms may enrich the theoretical foundation of service recovery models and offer deeper insights into customer behavioral responses.