Co-creating customer experiences in service ecosystems: a study in a tourist destination Open Access

The service ecosystem has emerged from service-dominant (S-D) logic as an important concept to explain the communal provision of a service by a network of actors (Vargo and Lusch, 2016; Mustak and Plé, 2020; Vargo et al., 2023). Nowadays, it is quite usual for the delivery of a service to require the participation of various actors, such as at tourist destinations, shopping malls, airports, or hospitals. According to Mustak and Plé (2020), this topic has been studied from a very optimistic conceptual point of view, which has overlooked the complexity involved in the creation of value by a network of actors: agents may seek to maximize their own interests to the detriment of others, or interactions between actors may be decoupled.

Mutual value-co-creation is one of the key service ecosystem principles postulated by S-D logic (Vargo and Lusch, 2016; Mustak and Plé, 2020; Landry and Furrer, 2023). Customer value creation is the goal of service providers and of the ecosystem. However, the complexity of the value creation process in a network that operates at micro, meso and macro levels and where individual and collective objectives may diverge means that value outcomes may vary from co-creation to co-destruction (Mustak and Plé, 2020; Barrios et al., 2023; Landry and Furrer, 2023). To generate value, each actor in the network uses its own resources as well as those provided by the context (Gardiazabal and Bianchi, 2021). Despite the importance of this perspective, multilevel co-creation value outcomes in a service ecosystem have deserved little empirical attention by researchers (Mustak and Plé, 2020; Gardiazabal and Bianchi, 2021; Sukla et al., 2023; Vargo et al., 2023; Landry and Furrer, 2023).

This paper focuses on the main outcome of value co-creation: customer experience. The literature about customer experience differentiates between service experience and memorable customer experience (MCE), because the generation of memorable experiences has become a priority for services (Moliner et al., 2023). S-D logic postulates that different service providers create their own experiences but also contribute to generating ecosystem aggregate experience. Despite the importance of this approach, to the best of the authors’ knowledge, the literature has yet to explore service experience and MCE from the perspective of a service ecosystem (Hosany et al., 2022).

The case under analysis is tourist destinations, which constitute a service ecosystem in which firms and institutions interact with tourists in a given location to co-create value. In the process of co-creation, the actors of the tourist destination contribute their own resources and activities, have institutional arrangements and take advantage of public resources and those of other actors (Gardiazabal and Bianchi, 2021; Landry and Furrer, 2023). The tourist’s experience with the destination is an outcome of the dyadic interactions that occur at the actor level throughout the consumer journey. In this paper, we study the impact on tourist destination experiences of two of the most important areas of service provision: accommodation and local food (Ye et al., 2021).

The aim of this study is to examine the co-creation of customer experiences at different levels in a service ecosystem by analyzing the case of a tourist destination. More specifically, we study three research questions:

The paper is structured as follows. In the next section, we present the theoretical background of the co-creation of customer value in service ecosystems. The hypotheses are then proposed, and the moderating variables are defined. Section 3 describes the questionnaire design, together with the size and characteristics of the sample. A survey was designed that yielded a final sample of 1,476 valid responses from Spanish tourists. In the results section, confirmatory factor analysis (CFA) is performed to analyze the dimensionality, reliability and validity of the scales and structural equation modeling (SEM) is applied to test the hypotheses. The paper ends with a discussion, findings, managerial recommendations and limitations of the study.

S-D logic defines a service ecosystem as a “relatively self-contained, self-adjusting system of resource-integrating actors connected by shared institutional arrangement and mutual value creation through service exchange” (Vargo and Lusch, 2016). S-D logic establishes four key terms that characterize service ecosystems:

1. actors (entities that integrate resources and engage in exchanges);

2. service (the process by which one actor uses its resources to generate benefits for another);

3. value (a positive or negative change in the viability of a system); and

4. institutional arrangements (set of rules, regulations, practices and values that enable collaboration between actors and enable value-co-creating actions) (Vargo et al., 2023).

Value co-creation is the main objective of the ecosystem and entails resource integration and service exchange between the actors (Vargo et al., 2023). A tourist destination is understood as a network of firms and institutions that participate in and collaborate to co-create value for the tourist in a given location. On the business side, there are the public and private service providers that participate in the consumer journey at the destination, while the institutions are the set of rules, regulations, practices and values that enable collaboration between them.

When literature refers to customer experiences, it distinguishes between service experience and MCE (Kim et al., 2012; Kim, 2014; Kim and Chen, 2020; Moliner et al., 2023). Service ecosystems make value propositions that consumers transform into experiences throughout their interactions along the consumer journey. Klaus and Maklan (2012) coined the term service experience to define the evaluation of the processes before, during and after the experience. In the case of a service ecosystem, it refers to the customer’s assessment of the dyadic interactions with the network of actors involved in a tourist destination. Klaus and Maklan (2012) argue that the roots of service experience lie in service quality: service quality is eminently cognitive, while service experience incorporates personal value and affects customer evaluation. These authors propose four dimensions of service experience:

1. peace of mind (emotional aspects of service related to the perceived expertise of the service providers);

2. outcome focus (reducing customers’ transaction costs);

3. moments of truth (service recovery and flexibility); and

4. product experience (customers’ perception of having choices and the ability to compare offers).

However, in reality, achieving a satisfactory service experience is not enough to build customer loyalty – it has to be memorable (Kim, 2014). According to Sthapit and Jimenez-Barreto (2018), MCE is generated through service experiences that are positively remembered and evoked (Kim et al., 2012). Following Kim and Chen (2020) and Sharma et al. (2022), MCE can be defined as a significant event that is accumulated in the memory of the customer and can be evoked later. MCE is easier to evoke and more difficult to forget, and it becomes the most important source of information influencing the customer’s future behavior (Ye et al., 2021; Sharma et al., 2022). A high level of memorability facilitates memory vividness and is a key antecedent of loyalty behaviors (Ye et al., 2021). From the service ecosystem perspective, MCE is a consequence of the customer’s evaluation of the value co-created in the ecosystem based on dyadic interactions with the service providers throughout the consumer journey.

The service ecosystem paradigm considers that on-site experiences are generated through the dyadic interactions between the customers and the actors (Stamboulis and Skayannis, 2003; Kim, 2018). From the MCE perspective, it is important to distinguish between ordinary and extraordinary experiences (Ye et al., 2021). Ordinary experiences of a destination meet the customer’s basic needs, as in the case of accommodation, food and transportation services (Qan and Wang, 2004; Ye et al., 2021). For Ye et al. (2021), extraordinary experiences, linked to multisensory elements, fantasy and emotional aspects, are what generate MCE: customers seek to satisfy their hedonic needs through experiences that go beyond simply finding solutions to everyday problems in the destination.

In service ecosystems, therefore, two types of experiences (service experience and MCE) can be generated, and two levels of relationships (dyadic and aggregated) must be considered. Regarding the relationship between service experience and MCE at the aggregate level, some studies have found a direct and positive relationship between service experience and MCE. According to Ye et al. (2021), the service experience must include extraordinary aspects for MCE to be generated, which implies that MCE is not always an outcome. Similarly, Williams et al. (2020) distinguish between memorable experiences and frictionless experiences, associated with emotional attachment and functional satisfaction, respectively. Prentice et al. (2022) consider that for a service experience to be memorable, it must be innovative and able to generate affective emotions during the cognitive process. Recently, Roggeveen and Rosengren (2022) have stressed the importance of the human experience, because if a service connects with the customer’s central ambitions, beliefs, values and/or feelings, maximum levels of experience can be reached. Therefore, the service experience is a necessary condition for MCE but is not sufficient on its own. We therefore posit a direct causal relationship between service experience and MCE at the aggregate level:

A service ecosystem is a network of actors collaborating to co-create customer experiences. Each actor generates experiences through dyadic interactions with the customer. However, collaboration with the other actors is necessary for the ecosystem to function in a coordinated manner and for the customer to have a full experience. Co-creation therefore implies interdependence, because an outcome is determined not only by one’s own actions but also by those of partners, and vice versa (Sukla et al., 2023; Le et al., 2023). For a service ecosystem to achieve satisfactory results at the aggregate level, it must pay attention to structural aspects (goal, power, rules and coordination structures) and to behavioral elements (display of team-like behaviors such as progress monitoring, information exchanging or cooperating with one another) (Le et al., 2023).

For a tourist destination, two of the most important actors in service provision are food and accommodation (Qan and Wang, 2004). Both actors, along with others present at the destination, are responsible for co-creating value and generating satisfactory experiences for tourists. Several studies highlight the importance of food in destination service experiences because it has an impact on the quality tourists perceive and their satisfaction (Stone and Migacz, 2016; Stone et al., 2018; Ye et al., 2021; Pham et al., 2023). Food is one of the criteria considered when choosing a destination, especially in the case of gastro-tourists, who plan their trips, at least in part, around trying local food and having authentic experiences associated with the local or regional gastronomy (Williams et al., 2019). But even among tourists who are not especially motivated by the destination’s gastronomic offer, food influences their evaluation of the travel experience (Björk and Kauppinen-Räisänen, 2016). Eating is a basic human need that tourists must satisfy in their chosen destination, and as such, it can be regarded as an ordinary experience (Gupta and Sajnani, 2020; Ye et al., 2021).

In recent years, increased attention has been paid to local food because of the impact it has on destination sustainability. Local food can be defined in three domains (Brune et al., 2021):

1. geographical proximity between food production, distribution and consumption;

2. relational proximity between local actors; and

3. values of proximity in terms of authenticity, freshness and/or quality.

Local food is popularly understood as the use of locally sourced ingredients. Food and local food affect the destination service experience because of their ordinariness: before the trip (as part of the attractions of a destination valued by tourists), during the stay (as a necessary part of the tourist’s daily routine and a major part in their expenditure) and after the trip (local food as a souvenir) (Björk and Kauppinen-Räisänen, 2016):

Together with food and transportation, accommodation is a basic need that must be met whenever tourists travel (Ye et al., 2021). A large part of the academic literature focuses on identifying which accommodation characteristics influence tourists’ decisions (Sthapit and Jimenez-Barreto, 2018; Moliner et al., 2019). Accommodation is a basic, ordinary aspect of travel planning, representing a significant proportion of the tourist’s expenditure and meeting their basic need for somewhere to rest (Fesenmaier and Jeng, 2000; Sthapit and Jimenez-Barreto, 2018). Accommodation is therefore a crucial element in evaluating the destination service experience (Sharpley, 2000; Sohrabi et al., 2012). Perceived quality of and satisfaction with accommodation generate ordinary experiences that affect the destination service experience (Moliner et al., 2019; Ye et al., 2021):

These three hypotheses suggest that the service ecosystem experience plays a mediating role in the relationship between service providers and MCE. However, actually, customers are looking to satisfy their hedonic needs rather than simply resolving everyday problems (Ye et al., 2021). Accommodation and local food are therefore involved in both ordinary (ecosystem service experience) and extraordinary (MCE) experiences.

Various studies have examined the direct relationship between food and MCE (Lashley et al., 2004; Sutton, 2010; Stone et al., 2018). Stone et al.’s (2017) qualitative study explored the relationship between local food and MCE, concluding that local food and authentic food are memorable aspects for tourists. Part of the uniqueness of local food lies in its origins in the local culture and its history, and another part attaches to its links to the local socio-economic and environmental system (Björk and Kauppinen-Räisänen, 2016). Lee (2023) conceptualized food memories as an intangible cultural heritage: “a form of cultural expression that transcends the food’s physical properties to include the cultural symbolic properties of narratives, traditions, embodied knowledge, and memories that thread past to present” (p. 1). Local food not only provides essential sustenance to tourists but also has a hedonic, inspirational, emotional and pleasure-inducing element that affects MCE (Björk and Kauppinen-Räisänen, 2016; Sthapit, 2018; Stone et al., 2018; Gupta and Sajnani, 2020; Hosany et al., 2022; Lee, 2023; Pham et al., 2023):

While accommodation meets the tourist’s basic need for rest and generates ordinary experiences, it can also give rise to extraordinary experiences (Ye et al., 2021). In this line, Sthapit (2018) identified three dimensions that make a hotel stay memorable: a comfortable room, the friendly attitude of hotel staff and a delicious breakfast. Other studies have identified memorable elements in different types of accommodation in different types of destinations (Sipe and Testa, 2018; Mody et al., 2017; Harkison et al., 2018; Sthapit and Jimenez-Barreto, 2018; Ye et al., 2019). As with local food, accommodation has a hedonic element that can arouse emotions, surprise and pleasure for the tourist:

Figure 1 shows the model we will analyze.

One question arising from the hypotheses development and the service ecosystems approach is whether or not the drivers of MCE are the same in any type of ecosystem (sun and sand destination, rural destination and urban destination). The moderating role of destination type is attracting increasing research attention (Rasoolimanesh et al., 2017; Segota et al., 2022; Yang et al., 2022; Moliner et al., 2023). The purpose of the network of firms and institutions making up the service ecosystem is to combine available resources to create, distribute and capture customer value (Osterwalder and Pigneur, 2010). The service ecosystem perspective implies that service providers play a key role in the customer experience.

A tourism destination is:

Tourism destination typologies are differentiated by their combinations of products, services, activities and experiences, and therefore by their different value proposals. Service providers (of accommodation, food, transportation, etc.), together with tourist attractions, are an essential element in value co-creation. The value proposal and the tourism experience will therefore be different in a sun and sand destination from those of an urban or a rural destination, which justifies the use of tourism destination type as a moderating variable in our explanatory model of MCE (Moliner et al., 2023).

An electronic version of the questionnaire was prepared. The research was approved by the university’s Ethics Committee (case number CD/109/2021). The questionnaire was distributed through the social media platforms Facebook and Instagram between November 2020 and March 2021. An advertisement was placed on both social media platforms rewarding participation with the opportunity to win a gift, thereby lending randomness to the sample selection. 1,476 valid responses were obtained. Responses covered three tourist destination types and up to seven different Spanish tourism destinations: 400 (27.2%) sun and sand (Peñíscola, Benidorm), 436 (29.5%), urban (Barcelona, Valencia) and 640 (43.4%) rural (Morella, La Rioja, Asturias). Participation was conditional on respondents having recently stayed at any of the chosen destinations. From an analysis of the primary data, we obtained the main characteristics of the sample for the tourist profile (Table 1).

All the scales used correspond to their theoretical definitions. The questionnaire items were scored on a five-point Likert scale, where 1 represents totally disagree and 5 represents totally agree. Table 2 summarizes the sources of the measurement scales used in the study. The scales for the four destination service experience dimensions were adapted from Klaus and Maklan’s (2012) proposal to measure service experience. The scales used to evaluate the five local food experience dimensions were adapted from the local food motivation scale developed by Kim and Eves (2012). This adaptation took into account the elements of memorable food, drink and culinary tourism experiences identified by Stone et al. (2018). The scale to measure tourists’ accommodation experiences was adapted from Nunkoo et al.’s (2017) proposal, with an item on environmental sustainability added to the existing ten aspects considered in this scale. Finally, Kim’s (2018) proposal was used to assess memorable tourist experiences.

In addition, testing the model should consider biases from exogenous variables that may alter the causal relationships between the independent variable and the dependent variables (Nielsen and Raswant, 2018). We regarded age and gender as control variables in this study, following a common trend in tourism research (Papastathopoulos et al., 2020; Zheng et al., 2022; Krey et al., 2023).

We performed CFA using SEM to refine the scales; the EQS multivariate software package (version 6.2) was used. We adopted the maximum likelihood approach to estimate the parameters.

The least relevant indicators were eliminated following the scale refinement process, based on the structures of the latent variables assumed for each construct (Steenkamp and Van Trijp, 1991; Hair et al., 2010; Estrada et al., 2020). Indicators that did not meet the strong convergence condition were removed; that is, all indicators with individual standardized coefficients (λ) below 0.6 and an average standardized factor loading below 0.7. We verified compliance with the weak convergence condition by analyzing the significance of the factor regression coefficients between indicators and their latent variables; to this end, we considered the student t-value by imposing the maximum condition (t > 2.58; p = 0.01). This process led us to remove six indicators: EXP1.3, EXP3.3, EXP4.4, ACO.3, ACO.4 and ACO.8. This elimination does not alter the essence of the service experience dimensions because at most one item of the three dimensions has been eliminated, leaving three more items that capture the essence of the dimensions. Regarding the accommodation experience, it is on a very broad scale and the elimination of three items reduces the number of details considered. However, the remaining nine items reflect the fundamental elements of this variable. Finally, as each indicator was eliminated, we monitored the evolution of the main model fit measurements.

We performed several verification tests to check whether these refinement tests had a negative effect on scale reliability (Table 3). The tests used to verify internal consistency were Cronbach’s alpha (α > 0.7), construct composite reliability (CR > 0.7) and analysis of variance extracted (AVE > 0.5) (Churchill, 1979; Nunnally, 1979; Fornell and Larcker, 1981).

We tested convergent validity by returning to the CFA performed at the start of the process and by confirming the high estimated value and significance of the correlations between the dimensions in the scales. Table 4 shows the discriminant validity of the constructs considered as assessed by AVE (Fornell and Larcker, 1981) and by confident interval tests (Anderson and Gerbing, 1988). Regarding the first of these tests, discriminant validity is confirmed when the square root of the AVE between each pair of factors is higher than the estimated correlation between them; this was the case in our study. The purpose of the confidence interval test is to verify that the value of 1 is not present within the confidence interval calculated for each pair of latent factors, taking into account the covariance of ±2 standard errors around the estimated value yielded by the final CFA. Once again, the results confirm the discriminant validity of our model.

Several other tests were also performed. First, the variance inflation factor among the latent variables in our proposed model revealed no signs of multicollinearity. The results showed values between 1.677 and 4.793 (well below the threshold of 10), suggesting multicollinearity was not a concern in this study (Diamantopoulos and Siguaw, 2006; Kock, 2015). Second, we performed a t-test of independent means on the dimensions of the model variables using the first 50 and last 50 respondents (Armstrong and Overton, 1977). No significant differences were found between these respondents at the 0.05 level, thus confirming the absence of non-response bias. Third, we assessed the possibility of common method variance bias using Harman’s test (Harman, 1976); this test assumes that, if this bias exists, in a factor analysis, a single factor should accumulate most of the covariance of independent and dependent variables. We performed a factorial analysis on the indicators yielded by the process using principal component analysis, in which we examined the unrotated factor solution (Podsakoff et al., 2003; Friedrich et al., 2009; MacKenzie and Podsakoff, 2012). The results of the factorial analysis showed that several factors had eigenvalues greater than 1. These factors explain 77.322% of the variance among the 49 items, with the first of the factors accumulating 33.141%. Hence, because several factors are identified and most of the variance is not accumulated by the first factor, a substantial part of the common method variance bias seems to be absent.

Table 5 displays the covariance matrix resulting from the scale refinement process described in the previous section.

Based on this data, the hypotheses are tested using structural equation models to explore a series of dependence relationships simultaneously (Hair et al., 2010). Figure 2 presents the step diagram of the resulting relationship model after its specification and identification.

In the next step, the hypotheses are tested with the module Lavaan de R. The results (Figure 3) show that accommodation experience and local food experience have a positive influence on MCE, both directly and indirectly, through their effect on the destination service experience. The direct influence of local food experience emerges as a major determining factor of MCE (H4: λ = 0.533, t = 18.474**) when compared with the antecedent effect of accommodation experience, which is lower (H5: λ = 0.208, t = 4.934**). Turning to the indirect influence, however, the effect of accommodation experience on destination service experience has a greater weight (H3: λ = 0.781, t = 23.565**) than that associated with local food experience (H2: λ = 0.074, t = 2.722*). These indirect influences are a result of the positive effect of destination experience on MCE (H1: λ = 0.222, t = 5.377**).

Moreover, when the total effects derived from the proposed effects model were considered, these results were reinforced, highlighting the direct and indirect antecedent roles of accommodation experience and local food experience on MCE (Table 6). Table 6 also displays the effect of the control variables. Age has a positive effect on destination service experience, showing that the older the person, the greater the perception of destination service experience. Gender has no influence on the results of the model.

Multigroup analysis is applied to examine the moderating effect of destination type; this analysis is appropriate when the moderating variable is categorical (MacCallum et al., 2002). Seven tourism destinations corresponding to three typologies were selected: sun and sand, urban and rural.

The multigroup analysis first estimated three models (Table 7):

1. Model 1 (configurational model): the same model is imposed on the three groups;

2. Model 2 (weak invariance): the same model, but with factor loadings constrained to be equal across groups; and

3. Model 3 (strong invariance): the same model, but with intercepts and regressions constrained to be equal across groups.

All three models showed good fit indices. To uncover any between-group differences, we compared the Chi-squares of the configurational model (Model 1) with the Chi-square for the models in which the paths were constrained in the groups (Table 8). We ran two ANOVA tests: the first test compared Model 1 with Model 2, and the second compared Model 1 with Model 3. The two p-values were significant, showing that this data set supports the configurational model (Model 1).

Therefore, the model with the best fit to the data is the one in which the paths were estimated separately and freely in the three groups; the regressions and the intercepts are not the same in the groups.

To identify the significant differences between the unconstrained model and the fully constrained model, we set partial constraints variable by variable and compared the three groups (Table 9). Significant differences were only detected in the relationship between local food and MCE, which showed a significantly lower value for the urban destination type than for the other two destination types (p = 0.0457*). This result implies that there are no significant differences in the rest of the relationships proposed in the model.