Is the dream over? A study of the theory of customer delight in complaints against travel agencies

Based on the theory of customer delight, the article examines the effect of problem-solving and interpersonal interactions on the tourist’s delight in complaint situations. A mixed method is used in this study. First, content analysis is applied to transform the content of messages in quantifiable and pre-established categories in the literature. These data serve as input for the subsequent step, in which structural equations based on partial least squares (PLS) are used, taking two different groups. A group is formed by complaints that occurred before COVID-19, and the other is made up of complaints published during the pandemic. It is evaluated whether the factorial configuration and the structural coefficients remain unchanged from one group to another. 502 complaints against travel agencies are examined. Among the results, one of them stands out. The human element is a key factor in the formation of pleasurable experiences during the pandemic. Travel agencies need to train their employees to respond appropriately to the demands of that period. In times of contingencies, the tourist seeks more than the real solution of the case and expects to be well looked-after by frontline employees. Future tourist behaviors may depend on this performance.


Abstract
Based on the theory of customer delight, the article examines the effect of problem-solving and interpersonal interactions on the tourist's delight in complaint situations. A mixed method is used in this study. First, content analysis is applied to transform the content of messages in quantifiable and pre-established categories in the literature. These data serve as input for the subsequent step, in which structural equations based on partial least squares (PLS) are used, taking two different groups. A group is formed by complaints that occurred before COVID-19, and the other is made up of complaints published during the pandemic. It is evaluated whether the factorial configuration and the structural coefficients remain unchanged from one group to another. 502 complaints against travel agencies are examined. Among the results, one of them stands out. The human element is a key factor in the formation of pleasurable experiences during the pandemic. Travel agencies need to train their employees to respond appropriately to the demands of that period. In times of contingencies, the tourist seeks more than the real solution of the case and expects to be well looked-after by frontline employees. Future tourist behaviors may depend on this performance.

INTRODUCTION
For practitioners and academics, satisfaction is a key factor to understand consumer attitudes and behaviors. The literature shows that a visitor's spending level at the destination can be explained by their satisfaction (Disegna & Osti, 2016;Jurdana & Freta, 2017). In that line, intention to recommend to others, to repurchase, and to pay a premium price for experiences on cruise tours are also considered to be affected by tourist satisfaction (Wu, Cheng & Ai, 2018). As a result, satisfaction became popular as a useful indicator, especially as a predictor of consumer loyalty (Akhoondnejad, 2016). Many companies put customer loyalty as a strategic priority, but few actually define how to identify a loyal customer. Thus, customer satisfaction stands out in the tourism industry, becoming a guide to identify opportunities for improvements in tourist services (Aksoy, 2013). Tourist services tend to be hedonic consumer experiences that result in emotions, such as satisfaction. This is observed in unique, memorable tourist experiences in the destination (Ma, Gao, Scott, & Ding, 2013). Therefore, any service improvement actions can be guided by levels of tourist satisfaction. Companies must direct resources to the continuous monitoring of satisfaction (Arnold, Reynolds, Ponder, & Lueg, 2005;Ladeira, Costa, Santini, & Araujo, 2013).
The fierce competition for customer attention and loyalty, lead tourism businesses to promote charming experiences instead of merely satisfying visitors (Parasuraman, Ball, Aksoy, Keiningham, & Zaki, 2020). The main reason for this impulse is the positive outcome for companies (Guidice, Barnes, & Kinard, 2020). The subject is expected to judge the tourist experience so much better than expected, and in a pleasant and surprising way. Therefore, delight must be consolidated as a business strategy, particularly, since it combines the dimensions of "surprise" and "joy" dimensions (Ali, Kim, & Ryu, 2016;Ma et al., 2013). For Liu and Keh (2015), delight is more memorable than satisfaction and requires an unusual service. When receiving something unexpected and that has a certain relevance to well-being, service experiences provide a level beyond satisfaction . This euphoric stage symbolizes a profound enchantment of the tourist, for example, typical of cases in which it is possible to promote the feeling of authenticity of the destination. That said, service packages that only satisfy do not necessarily lead to customer loyalty (Barnes, Beauchamp, & Webster, 2010). The loyal consumer demands excess rewards from the service provider. As the company increases what is delivered to customers, enchanting them strongly, the more likely is it for them to have higher levels of loyalty (Barnes, Beauchamp, & Webster, 2010 funcionários para responde-rem apropriadamente as demandas desse período. Em épocas de contingências, o turista busca mais do que a real solução do caso e espera ser bem atendido pelos funcionários da linha de frente. Comportamentos futuros de turistas podem depender desse desempenho.
About this debate, authors begin to make efforts to distinguish satisfaction from delight. For some, customer delight is related to the concept of satisfaction, but it goes beyond this frontier. This would amount to an extreme level of satisfaction (Ali, Kim, & Ryu, 2016;. In that case, there is a higher level of abstraction in consumer judgments in relation to what they receive as a service. Others disagree with this perspective and propose several elements that delimit delight (Loureiro & Kastenholz, 2011;Ariffin & Omar, 2016). In the latter, the study by Parasuraman et al. (2020) stands out, as it proposes a more comprehensive theory for the concept of customer delight and its conclusions contribute significantly to the advancement of relationship marketing. Parasuraman et al. (2020) report that customer delight is a function resulting from a set of properties that manifest themselves collectively. However, it is known that properties are necessary, but there are strong questions about what combination of these are sufficient to promote delight. Therefore, an extensive future research agenda is proposed by Parasuraman et al. (2020), with emphasis on two points.
First, the authors mention the hegemony of two properties (problem solving and interpersonal interactions). For them, these properties lead the others, but the unfinished conclusions invite researchers to carry out further studies, especially those that evaluate specific arrangements of properties which integrate the concept of delight. It would be beneficial to the literature advances in this direction (Parasuraman et al., 2020). Thus, a better understanding of these combinations of properties can offer practitioners the possibility of creating a multitude of strategic routes to delight the customer. Second, Parasuraman et al. (2020) suggest new research that allows a better understanding of how delight manifests itself in different service contexts. Companies must manage the impacts of each property on the production of delight. It is possible that a specific combination of properties influences delight differently, from the context (Parasuraman et al., 2020).
Therefore, this article responds to the calls of Parasuraman et al. (2020) for more studies that test the validity of the theory of customer delight. Until then, these authors have limited themselves to making theoretical contributions. There is a lack of empirical studies that give the theory more robustness. Therefore, this article seeks to fill that gap. The focus is to evaluate and compare the effect of the dimensions "problem solving" and "interpersonal interactions" on customer delight in situations of service failure in tourism, before and during the COVID-19 pandemic. For the study, failures represent occasions when the consumer experiences an imperfection in the service and, in general, leads to customer complaints. Tourists who experience service failures are more willing to complain (Li, Li, Fan, & Chen, 2020;Wan, 2013). The complaint is the voice of the consumer in cases of service failures, and the phenomenon is expected to be peculiar in times of pandemic. Customer complaints are likely to make the dimensions of delight very evident. This justifies the choice of this service context. Furthermore, it is not clear whether the impacts of elements that make up delight are the same for normal and abnormal moments. Two key questions guide the evaluation of this study: (1) To what extent do "problem solving" and "interpersonal interactions" affect the level of customer delight in situations of failure in the tourist service?
(2) Are there differences between the effects of "problem solving" and "interpersonal interactions" on delight before and during the pandemic? 2 THEORETICAL FRAMEWORK 2.1 Customer delight: an initial debate In business, the semantic domain of delight was raised by Oliver, Rust and Varki (1997). Until then, no references in literature were found. In a seminal work, these authors theorized the possibility of levels that go far beyond the mere satisfaction, producing exceptional results (Barnes & Krallman, 2019;Oliver, Rust, & Varki, 1997). This phenomenon is labeled "customer delight". This leads them to offer a first step towards stripping the concept of delight and pointing out how it operates in certain service configurations. Among the results, delight is perceived as a function of joy, excitement, and surprise. Other works validate these elements as drivers of charming experiences memories (Kumar, Olshavsky, & King, 2001;Rust & Oliver, 2000). On the one hand, joy is an important ingredient to further enchant tourists (Ma, Gao, Scott, & Ding, 2013). On the other hand, surprise is seen as a strong component of delight . However, Oliver, Rust and Varki (1997) assert that delight is a mixture of these three elements. This contribution is revalidated by Finn (2005). However, this list of antecedents is not exhaustive. Other feelings are also mentioned, such as gratitude (Ball & Barnes, 2017). Thus, these discoveries reveal distinctions between the concepts of satisfaction and delight (Barnes & Krallman, 2019), something that encouraged a solid flow of research. Schneider and Bowen (1999) add that delight can come from a sense of justice with which customers expect to be treated by the company. On the contrary, they become insecure and suspicious and become "terrorists" who disseminate the bad experiences of the service to others, with emotionally charged and exaggerated reports (Schneider & Bowen, 1999). Liu and Keh (2015) also see justice as an inseparable component of delight.
So far, the academic debate seems limited to the customers' own emotional traits. There is no empirical evidence regarding the elements extrinsic to the subject as potential antecedents. This barrier begins to be overcome by the study by Arnold et al. (2005).
When proposing a series of elements associated with charming experiences, Arnold et al. (2005) highlight the role of employees. In addition, they address an item already confirmed by literature (surprise). Their narratives explain delight from two dimensions: interpersonal and non-interpersonal. The first refers to situations in which the source of the delight experience is attributed to the actions of a salesperson. Examples are the effort, engagement, problem solving, distance (impartiality), and time commitment shown by front office staff. This dimension is supported by more studies (Barnes, Ponder, & Dugar, 2011;Guidice, Barnes, & Kinard, 2020;Jiang, 2020;Magnini, Crotts, & Zehrer, 2011;Torres & Kline, 2013;Torres, Zhang, & Ronzoni, 2020). The second comprises situations in which the source of delight comes from the surprise that the purchase of products or value acquisition provides at the time of purchase. Unforeseen acquisitions and bargain values are included in this set (Arnold et al., 2005).
In fact, there is a consistent literature that explores company-customer interactions as an explanatory variable of delight (Ali, Kim, Li, & Jeon, 2018;Barnes, Collier, Howe, & Hoffman, 2016). However, Ball and Barnes (2017) believe that the value of delight needs to be carefully evaluated and managed based on the market. Some specifics of the service can be taken into account. An element is likely to generate customer pleasure in one sector, but not in another. This is evident in the study by Bartl et al. (2013) when discussing the "delight" phenomenon in online environments. The findings point to very peculiar predictors in this context. For them, promoting charming experiences is a task that involves the sum of perceived utility, entertainment value of the site, and surprise (Bartl, Gouthier, & Lenker, 2013).
Ultimately, previous research deals with customer delight in a very limited way. This breads new life into the study by Parasuraman and collaborators (2020), allowing the projection of a theory of customer delight. This theory reveals a more holistic view, with the presence of six elements capable of explaining the delight that customers feel when purchasing. Figure 1 presents a summary of the studies that served as basis for identifying the main dimensions related to delight. The concept of delight involves (...) Context Oliver, Rust, & Varki (1997) joy, excitement, and surprise. amusement park and musical event Schneider & Bowen (1999) justice It is not specified Rust & Oliver (2000) positive emotions and surprise It is not specified Kumar, Olshavsky, & King (2001) joy and surprise It is not specified Finn (2005) joy, excitement, and surprise online retail Arnold et al. (2005) interpersonal factor: effort, engagement, problem solving, detachment, time commitment. non-interpersonal factor: unforeseen purchase and value. Torres & Kline (2013) problem solving, personalized service, outstanding performance, and aspects of the employee (charisma, knowledge, and skills. Hotel Bartl, Gouthier, & Lenker (2013) surprise, utility, and entertainment. online retail Ma et al. (2013) joy tourism Liu & Keh (2015) justice, excitement, and surprise. restaurant Barnes et al. (2016) employee effort and experience, surprise, and joy.
grocery store Ball & Barnes (2017) joy, surprise, and gratitude. musical event Ali et al. (2018) physical environment (comfort, cleanliness, layout), interactions with the team (professionalism and knowledge) and with other customers (appropriate behavior).

Theory of customer delight by Parasuraman et al. (2020)
More recently, Parasuraman and colleagues design a new and broader approach to customer delight design. For them, empirical evidence robustly supports six dimensions of the phenomenon ( Figure 2). First, they cite emotions of positive value as an antecedent of delight. In this set we find joy, surprise, admiration, satisfaction, peace, relief, and others. Reports point out that feelings that precede the experiences can interfere in the perception of the service encounter. The more positive the subject's emotional state, the more pleasant the service experiences will be. A happy visitor tends to better evaluate the services received at the destination. On the contrary, a moody tourist usually has negative perceptions of the place visited (Parasuraman et al., 2020).
Second, time is seen as a variable that affects delight. Temporary assessments are constantly reviewed throughout the customer's journey (Parasuraman et al., 2020). Tourists can evaluate the same restaurant differently, depending on the trip purpose. That said, a tourist can feel delight when a restaurant attendant devotes more time to explain, in detail, the consumption items and the respective value. This can be common in leisure-oriented trips where tourists have more time to enjoy the events. The same is not true for business travelers who seek to optimize time. In general, they do not allow as much time for service completion.
Third, Parasuraman et al. (2020) cite customer role in service experiences as a predictor. It is possible that customer co-creation of service leads to delight, as it seems to provide the individual a state of justice and control over what is received. For example, guests can feel extremely delighted when they find the room well decorated, clean, pleasantly scented, and with a sparkling wine to celebrate a marriage, even though they know that preliminary information was highlighted by them when booking the hotel. The delivery of the service in strict accordance with what is desired is related to a good performance of the tourist during the service encounter.
Fourth, sensory experiences influence delight assessments (Parasuraman et al., 2020). The authors argue that images, sounds, flavors, textures, and odors can impact the way customers experience services. One example is the field of tourism that is moving in this direction. There are many tourism service packages that value the senses. In Argentina, the commercialization of tango classes has become common. The tourist is invited to learn the typical steps of this dance and not just passively watch a tango show. Therefore, visitors are increasingly exposed to spatial elements that evoke rich and strong memories of a destination.
Fifth, interactions with companies influence service experiences (Parasuraman et al., 2020). Customer behavior depends on the functional quality (Grönroos, 1984) represented by the way customers receive the technical result of the service. In this regard, employee performance plays an important role. It is beneficial that customers receive cordial and attentive service, not least because interpersonal interactions are regularly evaluated due to the intangibility of services. The more positive the interpersonal experiences are, the better evaluated the services will be. Courteous treatment by the receptionist can contribute to good evaluations of hotel services. In general, sincerity, authenticity, and attention are other elements that lead to delight in tourist experiences.
Sixth and last, the authors point to problem solving as a potential element to explain consumption assessments (Parasuraman et al., 2020). Customers experiencing service failures can experience delight from the way companies receive, register, handle, and respond to complaints. For example, when complaining about the temperature of their food, a visitor may feel extremely delighted when the chef promptly forwards a new order, replacing the previous one, but with the addition of new favorite ingredients to the dish and at no additional cost. In fact, the solution to the problem was considered pleasant.
In short, the theory of customer delight by Parasuraman et al. (2020) is based on the following assumptions: (a) there are six dimensions associated with delight; (b) these properties are manifested in arrangements; and (c) two properties lead the others. For them, problem solving and interpersonal interactions are most frequently remembered by customers when referring to delightful experiences (Parasuraman et al., 2020). This justifies the choice of this study to work only with these two properties that will be best used in the next sections. However, it is assumed that offering delight to the customer may require a greater (or less) effort from the company depending on the degree of normality of the service context. In emergency situations, customers are probably less demanding (Zeithaml, Berry & Parasuraman, 1993). When there are contingencies in service performance (for example, the COVID-19 pandemic), customers recognize that this eventuality is not the company's fault and end up accepting lower service levels. Thus, enchanting them strongly seems to be a less arduous task. In the face of abnormal conditions, it is likely that customer delight will be more powerful.
In the case of failures, this is seen when a successful service recovery occurs. That said, the above discussion suggests that the background effects on customers' delight differ from one context to another. Zeithaml, Berry and Parasuraman's (1993) ideas support this direction. Therefore, the study proposes that: Hypothesis 1 (H1): The effect of the dimension "problem solving" on customer delight is greater in times of pandemic (COVID-19) than in other contexts. Hypothesis 2 (H2): The effect of the dimension "interpersonal interactions" on customer delight is greater in times of pandemic (COVID-19) than in other contexts.

"Problem solving" dimension
Problem solving is a potential path for customer delight, however, this dimension has been investigated less frequently in the literature (Parasuraman et al., 2020). In services, problems arise when customers receive something below expectations. It is what researchers call service failures. Failures are inevitable and genuine consequences in service provision (Reynolds & Harris, 2005) and in the hospitality industry, for example, monitoring them and the quick solution are crucial for the good evaluations of tourists (Mostafa, Lages, Shabbir, & Thwaites, 2015). Therefore, managers must pay attention to how customers respond to service failures.
From then on, the focus is on the identification of criteria that allow companies to understand how customers judge organizational performance in the face of failures. It is imperative that managers understand what elements comprise a successful service recovery. After failure, if the company does not implement an adequate and well-structured service recovery procedure, there is a risk of losing customers (Cheng, Gan, Imrie, & Mansori, 2018;Mostafa, Lages, & Sääksjärvi, 2014).
In the field of business, a heated debate about the triggers of problem solving is going on. Numerous factors are cited in the literature, such as compensation (Liu, Jayawardhena, Dibb, & Ranaweera, 2019;Mostafa, Lages, & Sääksjärvi, 2014;Reynolds & Harris, 2005;Wirtz & Mattila, 2004). The organization can offer gifts in response to the inconvenience caused by failures. A retail store can give discount coupons for next purchases due to the delay in delivering an order, just as it is possible for a restaurant to offer an additional and free dish to the customer after a long wait for a table.
The apology is also an important element for problems solving (Mostafa, Lages, & Sääksjärvi, 2014;Reynolds & Harris, 2005;Wirtz & Mattila, 2004). This portrays psychological compensation for stress, inconvenience, and damage caused to customers (Abisuga, Wang, & Sunindijo, 2020). It is seen as a basic requirement after the occurrence of negative events, as it demonstrates the company's willingness to deal with the problem, and reflects empathy with the situation experienced by the customer (Reynolds & Harris, 2005).
In addition to these attributes, there are references to the effective solution of the problem and ease of complaint (Mostafa, Lages, & Sääksjärvi, 2014;Reynolds & Harris, 2005). The first refers to the immediate action to correct what went wrong (Mostafa et al., 2015). The company's proactive behavior is what governs this course of action. For example, an airline that cancels a flight should offer the customer alternative and fair arrangements. The second addresses the procedures and tools that the organization has, to assist customer complaints. This indicates the company's willingness to listen to the customer.
Justice is another item that can measure adequate problem solving (Giampaoli, Ciambotti, & Bontis, 2017;Mostafa, Lages, & Sääksjärvi, 2014;Wirtz & Mattila, 2004). Justice symbolizes how appropriate is the recovery of service that customers receive (Wirtz & Mattila, 2004). This arises from judgment between the costs employed during the failure and the benefits perceived with the solution (Mostafa et al., 2015). For example, a rude conduct of flight attendants in the face of a complaint can cause a passenger to feel wronged and, consequently, to no longer relate to the airline. In general, customers evaluate justice from three perspectives: distributive justice (what they receive), procedural justice (applied procedures), and interactional justice (how the customer is treated). In fact, the company must understand what constitutes a fair result so that they can develop an effective service recovery plan .
Historically, there are other forays into the literature that proclaim the explanation, monitoring, speed of response, courtesy, and effort as components of problem solving (Liu et al., 2019;Matos, Rossi, Veiga & Vieira, 2009;Mostafa, Lages, & Sääksjärvi, 2014). By publishing an explanatory note, the company spontaneously acknowledges the problem, takes responsibility for the failure, and sets out the measures that are being taken to solve the case (Abisuga, Wang, & Sunindijo, 2020). This can make the customer feel comfortable because he or she knows that the company is committed to solving inconsistencies in the service. In turn, monitoring is the organization's systematic procedure for verifying whether the company has actually solved the customer's problem. It is an important strategy for managers to deal with failures, especially in the service sector, in which customers evaluate performance, and not tangible products (Matos et al., 2009).
Furthermore, the last three are attributes inherent to human resources. Speed of response is the readiness of service providers to record and address failures. This responsiveness can prevent the spread of negative comments about the service (Abisuga, Wang, & Sunindijo, 2020). For this reason, successful companies seek to make decisions faster than competitors (Giampaoli, Ciambotti, & Bontis, 2017). Courtesy and effort cover, respectively, the polite way in which employees deal with imperfection in the service and the energy invested by them to correct the failure (Mostafa et al., 2015). From the above, it is suggested that better problem solving generates a more intense feeling of customer delight. Therefore, the following hypothesis is proposed: Hypothesis 3 (H3): "Problem solving" has a positive influence on customer delight.

"Interpersonal interactions" dimension
Reports show the influence of employees on customer behavior (Bowen, 2016;Cheng et al., 2018;Delcourtet, Gremler, Van Riel, & Van Birgelen, 2016). This seems more intense in services, as the customer journey is marked by multiple points of contact between company and consumer. Interpersonal interactions can make customer service more enjoyable (Parasuraman et al., 2020), as some skills and attributes of frontline employees can greatly enchant them. In addition, these contacts can promote "emotional contagion" (Barnes, Ponder, & Hopkins, 2015;Fox, Deitz, Royne, & Fox, 2018). Customers tend to imitate and synchronize their movements with employees' facial expressions, voices, and behavior. It is assumed that the subject may feel depressed when talking to a depressed person, while conversations with confident individuals are likely to increase consumer self-confidence (Fox et al., 2018).
This debate has led studies to show elements that can make service experiences very enjoyable. Creativity and emotional competence are examples. Human capital is a source of innovation, in which good ideas are put into practice (Bowen, 2016). This boosts the customers' feeling of surprise and joy during their service experiences. In the hotel industry, for example, maintenance personnel can adapt the room with an audible signal to identify the water temperature after knowing that the guest is visually impaired. In turn, emotional competence involves the employee's ability to identify, understand, and regulate consumer sentiment. During service experiences, customers may be overwhelmed by one feeling or another. Suppose a tourist is angry during a boat trip, he or she will feel good when employees know what is their emotional state, what reasons led to that fury, and how to act to calm them (Delcourtet et al., 2016).
The nature of delight experiences extends to other employee attributes. It is possible to do a destination memorable due to staff's courtesy, empathy, attention, effort, communication, and level of knowledge (Arnold et al., 2005;Berman, 2005;Mostafa, Lages, & Sääksjärvi, 2014;Parasuraman et al., 2020). These intangible elements can explain delight. When experiencing a terrible and tiresome situation during the flight, the tourist can be enchanted with the service provided and the friendly and attentive way in which the hotel staff welcomes you. This makes the stay comfortable and reveals one of the paths that managers can follow to make the moments in the hotel more pleasant (Lee & Park, 2019).
This debate has a well-defined theoretical direction. Better interactions between employees and customers result in higher levels of customer delight. Therefore, it is proposed that: Hypothesis 4 (H4): "Interpersonal interactions" have a positive influence on customer delight. Finally, Figure 3 summarizes the model proposed in this study, exposing latent variables and the relationships between them. It is increasingly common to purchase tourist services and products over the internet. Thus, the tourism industry value comments from tourists that are posted on websites and forums (Su & Teng, 2018). For this reason, content analysis is employed to assess customer reports about service failures in travel agencies. In this study, cases of complaints are examined through an intentional sample of textual extracts that provide rich experimental content. However, a hybrid approach was chosen. The data extracted from the encodings of tourist comments serve as inputs for the use of structural equations. This technique makes it possible to assess the impact of "problem solving" and "interpersonal interactions" on customer delight in situations of tourism-service failure, as well as examining whether the effects vary from one group to another (before and after the pandemic).

Data collection
For this study, the complaints posted on the ReclameAqui website are used as a data source and which, at the same time, have been evaluated by consumers. Customer reviews are used as a proxy to represent their degree of delight with the solution of the service failure. This degree of delight is captured by a unique score that ranges from 0 (very bad) to 10 (excellent). The choice for ReclameAqui is based on the popularity of the site. It is one of the main Brazilian channels for consumers to share their unsuccessful experiences with others. The number of companies and consumers registered is significant, reaching 120 thousand and 15 million, respectively. As a cut, only complaints against three travel agencies in e-commerce were investigated, provided they were published three months before and after the first case of COVID-19 in Brazil. The option for this service category is based on the intense economic impact that the current pandemic has on the tourism industry (Karim, Haque, Anis, & Ulfy, 2020).
In turn, the focus on complaints from tourists has several reasons. First, complaints symbolize the consumer's voice in the face of imperfections in the service , and should be welcomed and encouraged by organizations (Reynolds & Harris, 2005). These testimonials are rich sources of information about a brand and provide fertile ground for investigating the delight phenomenon, given that complaints, in addition to wide adoption, also influence other consumers' decision-making (Fox et al., 2018;Torres & Kline, 2013). Second, Barnes and Krallman (2019) propose that experiences loaded with emotion (for example, vacations and pandemic situations) have a stronger level of delight than in more utilitarian experiences (for example, routine shopping). This brings a breath of fresh air to research on this path. A total of 502 complaints were assessed.

Content analysis
Content analysis was performed based on the guidelines of Su and Teng (2018). This technique assists in the task of quantifying and analyzing words, concepts, and themes in a text (Yin, 2016). It is possible to extract what the subjects think from their own words (McAlister & Erffmeyer, 2003), verifying that elements of delight are highlighted in situations of service failure. In the light of this, the researcher makes a constant comparison between textual evidence (complaints) and dimensions (items) originated by the literature (Figure 4). To eliminate bias, the texts extracted from ReclameAqui are codified by two evaluators who act as judges. This stage is recommended by Torres and Kline (2013). Each judge, independently, assesses complaints based on a set of literature codes (item diary). Their task is to assess whether each code is present (or absent) in the textual fragment under analysis. Thus, the dimensions of delight are captured by nominal measures (dummies) where 0 indicates absence and 1 points to the presence of the attribute in the text. The same complaint is read at least twice by the evaluator and may include one category or more. Then, differences in the codification process are discussed among the judges who seek consensus to resolve the inconsistency. The average agreement rate was 93%. The reason for the high agreement rate can be attributed to the fact that the texts issued by the claimants are brief, clear, and direct. Final evaluation: grade given by the tourist (0 to 10) Proposed by the authors

Structural equations from clusters
Structural equation modeling based on partial least squares (PLS) is used in two different groups, using the coding data as input. A conglomerate is labeled "before the pandemic" and has 246 complaints situations. And another group, called "during the pandemic", has 256 cases. Thus, this technique allows the assessment of causal relationships between constructs and tests the existence of differences between the structural parameters of one group and another (Hair et al., 2005). This is useful because the COVID-19 period is supposed to change the structure and the impact of the dimensions "problem solving" and "interpersonal interactions" on tourist delight.

Initial adjustments
When applying the PLS algorithm) on the theoretical model (Figure 3), using the Path Method, it was found that some manifest variables had factor loads below 0.40. This occurs in seven variables (PS1, PS2, PS3, PS4, PS6, IP3, and IP4). Under this magnitude, these items were removed from the analyses in line with notes by Hair and collaborators (2005).

Measurement model
The measurement models are evaluated based on validity (convergent and discriminant) and reliability. Convergent validity is obtained by checking the average variances extracted (AVE) and represents how much the manifest variables can explain the latent variable to which were proposed. For Hair et al. (2005), if the value of the average variances is equal to or greater than 0.50, it is possible to admit that the model converges to a satisfactory result. This is identified in interpersonal interactions for both groups: "before" and "during" the pandemic. Values range from 0.500 to 0.570. The same is not true for problem solving (AVE between 0.36 and 0.40). However, the researchers opted for a parsimonious stance and did not exclude this dimension from the analysis. The reasons are based on two axes: the values are relatively close to the parameter established by Hair et al. (2005), and there is theoretical support to maintain it (Parasuraman et al., 2020).
The discriminant validity portrays the level of uniqueness of one latent variable in relation to others. It symbolizes how different the dimension is from the others. This analysis is based in the exam, simultaneous, of cross loadings and the criteria of Fornell and Larcker. As for the first criterion, it appears that the highest factor loads were found in the respective latent variable. There is no evidence of high factor loads in two or more constructs. The second criterion compares the variance of the construct (AVE square root) and the shared variance (correlations between constructs). It is expected that there will be a higher value for the first in relation to the other (Hair et al., 2005). The results are shown in Table 1. The square root of AVE scores (in italics) for the latent variables InI (0.751; 0.717), PS (0.599; 0.576) and Dlg (1.000; 1.000) are mostly higher than the correlation scores between the construct and the others. The exception is the latent problem solving (PS) variable, in which this rule is not satisfied for both groups. Faced with this statistical anomaly, the researchers maintained the construct PS. This procedure was justified by the absence of cross loadings in this dimension, as well as in the theoretical base that supports the correlation between the three constructs under analysis. Submerged in these arguments, the findings indicate admissible discriminant validity. As for reliability, the composite reliability that imprisons the level of internal consistency of the scales is evaluated. The values varied between 0.616 and 0.727. These levels reveal acceptable levels, according to Hair et al. (2005).

Structural model
The bootstrapping approach (with subsamples of 500) was used to estimate the structural model. The results were presented in Figure 5, which include explained variance (adjusted R² value), path coefficients (significant paths are marked with asterisks) and their respective p-values. The indices of the "before the pandemic" group are highlighted in bold, while those of the "during the pandemic" group are in italics.
Problem solving and interpersonal interactions explain 53.3% of the total variation in subject's delight, according to records of complaints made in the period before the pandemic. This level is slightly reduced, when it comes to cases of complaints that occurred during the pandemic (45.10%). These indices reinforce the thesis of Parasuraman et al. (2020) in which these dimensions can well portray experiences of delight. It is noted that these explained portions are very expressive when referring to the social sciences (Cohen, 1988). Therefore, the data indicate that problem solving and interpersonal interactions are substantial properties of delight. An expressive portion of delight is explained by the two dimensions, and in times of pandemic it is likely that subjects value even more other elements in addition these. This justifies the difference in the percentage between the periods. As indicated by the path loads, problem solving has a significant, direct, and positive effect on delight. This impact undergoes changes in magnitude between the periods before (β = 0.631 and p <0.01) and during the pandemic (β = 0.460 and p <0.01), as well as maintaining statistical significance. There is a reduction in the effect from the first moment to the next. A similar scenario is observed in the dimension of interpersonal interactions, but in the opposite direction. The effect is amplified, jumping from 0.170 (p <0.01; before) to 0.353 (p <0.01; during). In times of pandemic, it is possible that the subjects attach greater importance to the attendance of employees and less value to the effective solution of the problem. At such times, the employees play an important role in offering delight to consumers.

Discussion
The article sheds light on the theory of consumer delight of Parasuraman et al. (2020), adding an empirical effort to examine its validity. It is proposed that problem solving and interpersonal interactions form a structure capable of satisfactorily representing tourist delight in case of complaints against travel agencies. This configuration is supported in the seminal theoretical essay by Parasuraman et al. (2020). It is hoped that this arrangement will appropriately capture the delight. However, this feeling that can vary from time to time, especially in situations where there are contingencies (Zeithaml, Berry, & Parasuraman, 1993). In those situations, customers are more understanding and inclined not to blame the company for any service failures. Thus, it is possible that there are differences between the effects of the dimensions of delight, before and during the COVID-19 pandemic. For this, four hypotheses are tested.
The main conclusions are summarized in three points. First, delight seems to be well symbolized by problem solving and interpersonal interactions (explained variance greater than 45%) either before or after the pandemic. This finding is consistent with the thesis of Parasuraman et al. (2020) which puts these two dimensions as key elements of delight. In their work, Park and Park (2016)  Regardless of the moment, the contributions of these elements to explain delight are similar. Apparently, the atypical moment does not bring radical changes in the configuration of the factors that portray delight. For the two periods, other dimensions that are omitted in the analyses seem to have less explanatory power than the two. Furthermore, the results indicate that other dimensions can circumscribe delight. Problem solving and interpersonal interactions are necessary ingredients to explain charming experiences, but they are not the only ones. The aforementioned percentages show this.
Second, the impacts of problem solving and interpersonal interactions on delight are positive and significant. These findings support hypotheses H3 and H4. The faster, fairer, and more effective the solution to the problem and the better service experienced, the greater delight the consumer will feel. This discussion takes up the ideas of Grönroos (1984). For the author, the client can be enchanted with the level of service that the company provides in terms of solving the case (technical quality) or how the employees handle complaints (functional quality). Not only does the real solution of the problem lead to delight, but the attention and courtesy of employees can awaken that feeling. The works of Fatima and Razzaque (2014), Roberts-Lombard and Petzer (2018) support this relationship. These authors recognize that employee skills influence customers' delight experiences. In this line, other works reinforce the thesis. The human element is a central attribute for the delivery of successful experiences (Coelho, Meira, & Gosling, 2018). For example, delight experiences can be classified as moments when the airline goes beyond standard procedures and demonstrates full customer support in situations where technical problems with aircraft occur (Vasconcelos, Machado, Almeida, Arruda, & Matos, 2015).
Third, there are differences in effects between the periods. The impact of interpersonal interactions on delights increases during the pandemic. In atypical times, the human element is likely to gain greater prominence. Complainants seek much more than the solution of the case itself. Difficult times, with social distance, lead consumers to attribute greater value to subjective elements, for example, welcoming, empathy, cordiality, and attention. Items that Ladeira et al. (2013) label "social environment". It seems that good employee behavior can alleviate the negative feelings (distress, anger, and fear) that the consumer feels at the initial moment of the complaint and that are heightened by the current health risks (Jiang & Wen, 2020). It is important that employees treat customers in an equally friendly manner (Jerger & Wirtz, 2017;Ladeira et al., 2013). That said, hypothesis H2 is supported by the data. In the opposite direction, the effect of solving problems on delight experiences is greater in normal moments. In part, the arguments of Zeithaml, Berry and Parasuraman (1993) can justify this fact. They cite that subjects are more tolerant of failures in emergency situations. Customers may not have strong expectations regarding the effective resolution of the problem, as they recognize that service failures during the pandemic may not be the responsibility of the company. Repairing the problem is less representative to explain delight in times of pandemic. Thus, H1 is rejected.

CONCLUDING REMARKS
Finally, the study reveals theoretical and practical contributions. Advances are important to the theory of delight by Parasuraman et al. (2020). Because it is recent, the theory still lacks empirical studies that allow the framework to be robust. In this direction, this study discusses the effects of problem solving and interpersonal interactions on the ability to provide customer delight before and during the COVID-19 pandemic. These contributions add value to the field of service management, as the debate expands around aspects of consumer behavior. As for the practical implications, the findings provide tourism agencies with insights into the ways delight can be achieved, as well as expand the debate about delightful experiences in case of complaints. Based on this, the findings help companies to improve services and improve their performance in service recovery. Managers will have a foundation for directing efforts and resources on important dimensions of delight. As a result, it is possible to have adequate organizational performance in handling complaints and making a difference. The way they receive, treat, and respond to cases can became the experiences even more successful, regain the claimant's trust, and improve the corporate image (Park & Park, 2016;Vasconcelos et al., 2015). Being able to offer delight to tourists is synonymous of future repurchases (Roberts-Lombard & Petzer, 2018), and consequent leverage of post-pandemic businesses, an outcome that is desired by the tourism industry, so affected by COVID-19 (Jiang & Wen, 2020).
In addition, the work has some limitations. First, the study does not include all dimensions of the theory by Parasuraman et al. (2020). These authors see emotions of positive value, time, the role of the consumer, and sensory experiences as ingredients of delight experiences, however they were not included in this article. For this reason, it is recommended that future researches fill this gap and assess the effect of the mentioned dimensions on tourist delight. Second, a longitudinal monitoring of the effect of each dimension of delight can clarify points of inflection: how long will the impact of problem solving and interpersonal interactions on tourist delight return to the magnitudes of a normal time? Understanding the pandemic cycle in tourist service management is healthy for managers and for a better recovery of the sector. Consequently, this research flow can be explored by researchers as it helps professionals working in the field of tourism services. Third, both the use of nominal scales and the use of non-probability, convenience sampling can promote statistical bias. The results of this study cannot be generalized to broader populations. Therefore, further research is needed in order to overcome the limitations of this work. It is recommended that other service configurations and cultures be evaluated so that the results can be compared and generalizations are possible. These replications will allow to validate the model proposed here.