Maximising sales and fostering brand attachment: the power of great expectations in the Australian cellar door experience Open Access

The cellar door experience (CDE) belongs within the wine tourism realm and is often researched from tourism and experience perspectives. Recognition of the importance of the CDE has increased in tandem with the development of the wine tourism industry. However, the CDE is more than a tourism activity (Lease et al., 2023). For many New World winery businesses (WB), direct-to-customer retail sales via the cellar door are not only the most profitable and often the only sales stream for smaller WBs, but also provide an opportunity to develop brand awareness and more importantly, the brand attachment essential for nurturing loyalty behaviour (e.g. word-of-mouth recommendations; Khamitov et al., 2019).

Wine purchases are considered high risk in a market saturated with thousands of brands. While brand awareness reduces that risk (Johnson and Bruwer, 2004), smaller WBs might not have adequate production levels to build sufficient awareness through product placement in retail outlets and restaurants. The CDE allows customers to taste wines before purchase, thus reducing much of the perceived risk, and positive customer relationships developed during the CDE help to build strength in a brand (Nowak and Newton, 2006).

Recent research has called for compelling CDEs (Lease et al., 2023). The staff co-creating the CDE are crucial for building customer relationships and delivering enjoyable, memorable experiences (d’Ament et al., 2022a); it is the staff who draw customer visual attention during their CDE (d’Ament et al., 2023b). These co-created memorable experiences inform the content of word-of-mouth recommendations and warnings (d’Ament et al., 2023b). Elements influencing customer purchase and loyalty behaviours that contribute to brand attachment, as developed during the CDE, require further exploration (Williams, 2021). The current paper uses a Bayesian network (BN) to investigate the influence of customer states (i.e. mood and emotion), traits (e.g. personality) and schemata on purchase and loyalty behaviours.

The development of brand attachment relies upon accumulated knowledge of a product (Keller, 1993), which becomes a product-related schema (a construction of preexisting knowledge of product attributes and experiences). These schemata provide established neural pathways for encoding and retrieving memories (Brewer and Treyens, 1981) and are the foundations of expectations. Schemata are constantly evolving, with multiple sources of information contributing to the development of a schema. One source of information, word-of-mouth recommendation is considered a valuable marketing tool for WBs and its contribution to a CDE schema has been established (d’Ament et al., 2023b). Each CDE adds detail or confirms known aspects of a schema, creating cognitive shortcuts that build anticipation and assist decision-making in future events (Needham and Jacobson, 2020).

The role of schema in decision-making has been explored in tasting contexts; schema congruency of wine products found novices (lacking prior knowledge) less likely to recognise a wine to be incongruent with its category (Lanseng and Sivertsen, 2019), and Parr (2019) acknowledged the influence prior knowledge plays on the assessment of wines during a CDE. The importance of delivering a CDE that met customer CDE schemata was demonstrated by d’Ament et al. (2022a). Their analysis of participant memories and perceptions provided a deep understanding of the components of enjoyable experiences, which were the convivial connections co-created with engaging staff. Barriers to purchase and loyalty behaviours were created when CDEs did not meet these expectations. It would, therefore, follow that CDEs that were either equal to or exceeding customer expectations, based on individual CDE schema, would increase purchase and loyalty behaviours:

While sensory perception, memory and attention help build these cellar door schemata, internal characteristics such as emotion, mood and personality can influence an individual’s perception of the CDE. d‘Ament et al. (2024) found that customer state, more than trait, to be more influential but that customer personality influences the profitability of a CDE in a few critical ways. It therefore becomes essential to explore the influence of states and traits on purchase and loyalty behaviour.

As neocortical appraisals of perceptions, emotions can encompass cognitive, motivational, affective and expressive components, which are described through valence (negative/positive) and arousal (strong/weak) dimensions and each lay on a continuum of these descriptions. The same emotion can be negative or positive, strong or weak, with the perception based upon the context in which it is aroused. Due to interconnecting neural circuits with the olfactory system odours are easily associated with emotions, with odours perceived as being pleasant inducing happiness (Kontaris et al., 2020). Emotions are typically more intense than moods, shorter in duration, stimulus specific and affected by subjective perception (Lindquist et al., 2013). Furthermore, the range of emotions is considered infinite (Wang et al., 2020) encompassing more than just labels for facial expressions. Emotions can influence attitudes and/or behaviour (Shuman et al., 2017), by stimulating a response (i.e. an approach behaviour or accepting attitude) or inhibiting a response (i.e. a retreat behaviour or avoidance attitude).

Considering the influence of emotion is important for the CDE, as wine-evoked emotional responses can influence the purchase intentions of wine consumers; these emotive responses can be influenced by context (Torrico et al., 2020). Danner et al. (2016) found an increased willingness to pay for a bottle of wine in the absence of negative emotions. Further, of the five recognised concepts of consumer-brand relationships, attachment-based brand relationships are a better predictor of loyalty behaviour (Khamitov et al., 2019). Attachment is considered to be an emotion-laden target-specific bond, with product attachment being a consumer-created product-specific emotional bond (Mugge et al., 2008p. 425). The enjoyment of the CDE increases consumer loyalty (Bruwer et al., 2013). Positive experiences increase dopamine release, which strengthens memories (Duszkiewicz et al., 2019), leading to increased revisitation and enduring loyal customers (Skavronskaya et al., 2019). Hence, it is anticipated that emotion will be influential over self-rated measures of wine quality and experience quality and, in turn, impact purchase and loyalty behaviours:

Moods are general affective states which are consciously accessible and lie on a continuum between positive (e.g. euphoria) and negative (e.g. devastation), occurring without focused reference. Forming slowly through cognitive appraisal of experiences, they influence our perception of environmental stimuli, forming our judgements. Thus, a dynamic relationship is created whereby mood shapes our experience, but our experiences also shape our moods (White, 2006). However, the interconnecting neural circuity of the olfactory and emotion systems overlaps with the mood regions, and odours have been shown to modulate mood both overtly and subliminally, with unpleasant odours inducing a negative mood and pleasant odours producing calm moods (Kontaris et al., 2020). When considered with memory the mood congruency effect maintains that when a mood is congruent with the mood at the time the memory was encoded, that memory will be easier to retrieve (Bower, 1981). Further, mood interacts with emotion so that a succession of experiences evoking a positive emotion creates a positive mood, but also the valence and intensity of the evoked emotion can be moderated by the mood state (Rusting, 1999). Dopamine is associated with pleasure and plays a role in consolidating memories; it is, therefore, unsurprising that a positive mood strengthens memories, leading to stronger brand attachment and purchase intention (Orth et al., 2020):

Personality moderates an individual’s attention, processing, likes and dislikes of the abundant stimuli in our environment, as well as the affective states resulting from assessing the current experience against schemata developed through past experiences (Kritzler et al., 2020). Different personality types have unique needs (Caliskan, 2019). Saliba and d’Ament (2022) suggest that an introvert would find comfort in a quiet, reflective cellar door. Whereas the energy of the busy weekend crowds would more fulfil an extrovert. Personality has been shown to influence taste preferences with sweeter white wines preferred by individuals with higher levels of impulsiveness (Saliba et al., 2009) and affect consumer preferences for less familiar wines in out-of-home contexts (Capitello et al., 2019), perhaps influencing the effect of schema incongruency where less familiar products have less detailed or underdeveloped schema or expectations. Neuroticism and openness to experience have been positively correlated to alcohol consumption (Turiano et al., 2012), with wine drinkers tending more towards openness to experience and agreeableness (Gustavsen and Rickertsen, 2019):

Regular purchasing creates habit formation, predicting future behaviour (Wood and Neal, 2009) which, in the context of the CDE, habitual purchasing behaviour may lead to loyalty behaviour in the form of wine club membership. Most WBs with cellar doors offer a loyalty programme through wine club memberships. Such approaches to developing customer loyalty are a sound strategy for building a self-sustaining competitive advantage when selling a high-risk product in a saturated market (Dick and Basu, 1994). For example, the regular sales generated by wine clubs can carry a WB through a low wine tourism season, and Australian WBs with well-maintained databases enjoyed record sales during COVID-19 lockdowns (d’Ament et al., 2022b). Further, WBs with staff dedicated to post-visit communication, such as wine clubs, newsletters and online purchases, substantially increase post-visit purchases (Szentpeteri, 2018).

Customer loyalty can be divided into behavioural loyalty (e.g. post-visit purchasing, wine club membership) and attitudinal loyalty (e.g. intention to engage in word-of-mouth recommendation). Loyalty programmes, such as wine clubs, have been shown to enhance behavioural and attitudinal loyalty, with behavioural loyalty more resistant to decay over time (Belli et al., 2022). Interestingly, Krishen et al. (2023) found no need for an emotional connection to maintain continued wine club membership, supporting the persistence of behavioural over attitudinal loyalty. Functional behaviours such as purchasing habits (Shammout et al., 2022) are expected to be more influential in joining the wine club. Wine clubs typically offer discounted wines which, when coupled with the risk reduction provided by regularly delivered quantities of discounted known wines, appeal to customers with more practical goals. Such deal proneness has been linked to loyalty behaviours (Wolter et al., 2022), with deal-prone customers more likely to maintain club memberships and post-visit purchases when they perceive ongoing value (Esmark et al., 2016). This relationship between purchase habits and wine club membership is a crucial aspect of customer loyalty in the wine industry, particularly in the context of CDEs (Nella and Christou, 2014):

Interpretive analysis of a Bayesian Network created from 42 variables measured in a survey conducted during a CDE, capturing customer personality, mood and wine-evoked emotions while co-creating the experience with cellar door staff, will allow the discovery of the elements that influence purchase and loyalty behaviours and investigate these five hypotheses.

A survey, completed during a CDE, contained questions on consumer behaviour, including wine purchasing habits, frequency of visiting cellar doors and self-rated wine knowledge. Cellar door schemata were quantified using rating scales regarding expectations and evaluations of wine quality and experience, intentions to recommend and wine club membership. These provided insight into the influence of different variables on purchase and loyalty behaviours. The influence of individual states and traits were measured using the 44-Item Big Five Inventory Personality Scale (John et al., 1991) for personality, Brief Mood Introspection Scale (Mayer and Gaschke, 1988) for mood, openness to new wine using the Wine Neophobia Scale (Ristic et al., 2016) and the Australian Wine Evoked Emotion Lexicon (AWEEL; Danner et al., 2016) for emotions evoked by the wine tasted by customers.

Participants included customers who attended wine tastings at cellar doors in various regions of Australia, including the Canberra District, Hunter Valley, Shoalhaven, Coal River Valley, Tamar Valley, Clare Valley and Coonawarra. Customers were approached by a researcher when on-site, invited to participate by the staff member conducting the tasting, or by self-selecting via the QR code on the display poster.

The cellar door survey was completed in real-time during the participant’s cellar door experience while interacting with a staff member. Customers completed the personality and mood scale while the staff member was preparing their tasting experience. The wine evoked emotion was self-rated and recorded by the customer with each wine tasted. Collecting survey data during the CDE allowed for the mood and emotion at the time of the experience to be captured, enabling the influences of each on purchase and loyalty behaviours to be determined through Bayesian analysis. If surveys had been completed at some time following the CDE, as mood and emotions are states that change with different environmental stimuli, the data would not accurately reflect the participant’s mood and emotions during the CDE.

Surveys were accessed via a QR code on a study advertisement poster supplied to each participating cellar door. The survey contained the participant information statement, and the agreement to continue the survey was considered informed consent. Ethics approval was obtained from the Human Ethics Research Committee, and data collection ran from May 2021 to November 2022, resulting in 136 surveys being completed for analysis. The data collection was completed during the rolling lockdowns of the COVID-19 pandemic, delivering a very clean sample of Australian winery visitors, as international borders were closed. While these restrictions provided a clean data set, cellar door foot traffic was vastly reduced during this period resulting in a small sample size.

A Bayesian network (BN) model was developed to determine the influence of individual traits, states, schemata, expectations and evaluation of experience quality on purchase, recommendation and loyalty decisions. A BN is a graphical representation of the joint probability distribution for a set of variables, where each is represented by a node with a dependency relationship between associated variables defined by a link (Kjaerulff and Madsen, 2013; Pearl, 1988). Using Bayes’s Theorem, BNs provide mathematical assessments of the bidirectional effects of each variable. The resulting networks compute both likely effects given specific values and likely causes of observed events. As such, the importance afforded to the significance of a variable or value does not exist as it does for frequentist analyses, as BNs provide the bidirectional influence of each variable on every other variable in the network.

The network comprises both a qualitative and a quantitative component. The former specifies the network structure and relies on dependence and independence statements among random variables, their informational precedence and their preference relationships. The researcher decides these relationships based on prior knowledge. A concept model for the hypothesised influence was developed to aid the interpretation of the data collected at cellar doors across Australia, with dependent and independent variables connected by the expected direction of influence on and between variables. Figure 1 shows the concept model for the five hypotheses of the current analysis.

The BN quantifies the strength of dependence relationships by applying probability and preference relations using utility theory (Kjaerulff and Madsen, 2013), which posits that individuals consistently rank choices dependent on preferences; therefore, decision outcomes rely on the value or utility to the individual. Inherently, BNs quantify local dependency relationships between a variable and its parent variables through links; then, all local dependency relationships are integrated based on the probability chain rule, allowing the joint distribution of interrelationships of all variables to be determined (Kjaerulff and Madsen, 2013).

Clean data sets were converted from MS Excel files to a format recognised by the software and denoted parent or child nodes with links depending on relationships determined by the researcher. Due to the large number of interrelated variables a sampling-subject-oriented approach (Xie et al., 2023) was used to create the BN in this study. For this approach, the sampling subject – the winery visitor – becomes the target variable to determine the optimal BN structure as per the Tree Augmented Naïve Bayes algorithm (Friedman et al., 1997; Norsys Software Corp., 2021), meaning that the participant IDs become the parent variable. Netica (Norsys Software Corp., 2021) was used to develop the BN from which the excerpts in Figure 2 were calculated.

One benefit of using the sampling-subject-oriented approach to creating the BN is the ability to determine the influence of specific nodes on outcome variables (Kaikkonen et al., 2021; Pollino and Henderson, 2010), which for the current study allows for a deeper examination of influences on purchase and loyalty behaviour. An example of how one variable can be examined in depth is shown in Figure 2, which illustrates the variable expectation_purchase. Each section illustrates the varying levels of influence that personality trait variables of conscientiousness and neuroticism and the mood state variable negative/relaxed, have on each level of expectation_purchase.

Responses to 136 questionnaires were analysed to create a BN containing 42 variables, from which variables that measured purchase and loyalty behaviours were selected as outcome variables. These include total_spend, bottle_no_purchased, join_club_today, recommendation_CDE and recommendation_winery. The variables total spend, and bottle numbers purchased are considered representative of purchase behaviours; join wine club, cellar door and winery recommendation variables are considered representative of loyalty behaviours.

The influence of each variable on outcome variables is contained in the sensitivity of findings report from the BN (see Table 1). Each outcome variable is reported individually, with the influence of those nodes to which the outcome variables are most sensitive in bold text. The sensitivity to these more influential variables is provided in separate tables. The joint probability analysis of a BN allows for further exploration of the influences on each of these influential variables, with these influences also reported.

The outcome variables total_spend and bottle_number_purchased are, not surprisingly, highly influential on each other, as the higher the number of bottles of wine purchased the higher the dollar value spent. A more detailed analysis of this relationship provides insight into determining the styles of wine purchased. A more expensive bottle would have a lower bottle number purchased for the same total spend compared to a less expensive wine (e.g. Shiraz has more significant production costs, hence a higher price point than Sauvignon Blanc). However, the focus of this paper is to understand broader customer behaviours, and the influence of these two variables on each other will not be discussed further.

Similarities continue with both outcome variables most sensitive to a self-rated expectation of purchase (28.2% and 21.4%, respectively), with higher-than-expected purchases primarily associated with the highest purchase amounts (see Table 2). A deeper exploration of the sensitivity of the expectation node in the BN (see Table 3) reveals that these expectations are influenced by personality traits, specifically conscientiousness (12.6%), neuroticism (11.6%) and a relaxed mood (11.3%). Customers purchasing above expectations had lower-than-mean conscientiousness and higher-than-mean relaxed state. Lower-than-mean neuroticism scores were associated with well-above, well-below and as-expected categories. Conscientiousness is associated with self-discipline and a preference for planned behaviour, and neuroticism with overthinking and a tendency to have negative thoughts. Meaning that more spontaneous customers, with lower self-control in a relaxed state, would be at ease purchasing beyond expectations.

The outcome variables recommendation_CDE and recommendation_winery are, not surprisingly, highly influential on each other (60.2% and 57.3%), with recommendation_CDE being more influential on recommendation_winery. This indicates that a very poor CDE will have a negative impact on the overall assessment of the winery. The next most influential variable was experience quality [(16.5%), see Table 4], with most customers providing an excellent rating very likely to recommend the CDE. Most customers rated the experience quality to be well above expectations, a self-rated measure of how well the experience matched the participant’s cellar door schema. Where expectations were not met, there was very little likelihood of the customer recommending the experience (see Table 5).

After recommendation_CDE, the outcome variable recommendation_winery was sensitive to the personality trait neuroticism (13.8%) and the expectation of wine quality (13.7%). Customers with higher than mean measures of the personality trait neuroticism were progressively less likely to recommend the winery (see Table 6). Better-than-expected self-rated perceptions of wine quality were associated with a high likelihood of recommendation, however, recommendations became less likely when the wine quality failed to meet expectations (see Table 7).

The outcome variable join_wine_club was most sensitive to the winery (24.8%) and the number of bottles purchased (18.5%), as shown in Table 1. However, these variables are possibly skewed by the number of surveys completed at each winery, with two wineries showing the most robust results. The number of bottles purchased may reflect the conditions associated with club membership (i.e. the purchase of six or more bottles to join) as well as the value of continued membership. Those who either joined or were already members purchased more than those who were neither members nor joined the wine club.

The following most influential variable was the frequency of visits to the winery (17%), with many first visitors joining the wine club, and wine club members attending the winery quite regularly (see Table 8). Further analysis of the variables influencing customers were the frequency of CDEs (20.9%) and purchases at boutique retailers (16.6%), as shown in Table 9. Interestingly, customers who had never visited the winery were more likely to attend cellar doors than boutique retailers, which provides new market possibilities. Those customers attending a CDE at least twice a year were more likely to regularly purchase wine at boutique retail outlets, which might indicate an openness to experience, a desire to source specific varietals or to support smaller WBs.

The percentage of those participants who indicated they were already members but had never visited the winery, is also noteworthy. This could be due to participant error, or they may have joined the wine club online or offsite at an event, such as a wine festival in response to a tasting, a retail or restaurant purchase or on the strength of a recommendation from a trusted source.

Using a BN analysis of the cellar door survey provided the opportunity to explore elements of this complex direct-to-customer sales relationship co-created by staff and customers and determine the interaction, influence and relative importance of each in developing compelling CDEs. The interpretation of the results found at least partial support for the hypotheses, except for H2, which related to emotion and had the most decisive influence on wine club membership and attitudinal loyalty behaviours (see Table 10).

Results indicate that expectations and judgements on experience quality had the most substantial influence over purchase and loyalty behaviour outcome variables. These measures reflect participants’ self, their CDE schema and their place within that experience. Brand attachment is formed through the development of product knowledge, which contributes to a schema (Keller, 1993). As such, cellar door management and staff should aim to deliver a CDE that meets or exceeds expectations created by individual CDE schema. This delivery is achieved by prioritising engagement with each customer, to determine their needs and expectations, while using the opportunity to increase customer knowledge of the WB brand and story rather than reciting tasting notes (d’Ament et al., 2022a, 2023a).

Often, the style of CDE delivered is dictated by the at times limited on-the-job training, which relies on the sharing of knowledge and advice between employees. The lure of immediate sales at the cellar door tends to drive a traditional “push” approach, which risks diminishing the customer experience. The more valuable outcome might be to foster the customer experience. However, this requires specialist skills, for example, well-developed emotional intelligence to recognise when a customer may not be engaging due to dissatisfaction rather than their enjoyable but serious contemplation of wine characteristics. Many cellar doors use a predominantly casual workforce, or during busy periods rely on assistance from highly knowledgeable staff employed elsewhere in the WB. For WBs to effectively adopt these suggested changes to enhance staff-customer interactions, it is likely that further training will be required.

Trait and state measures influenced purchase expectations and attitudinal loyalty behaviour (d’Ament et al., 2024; Nayeem, 2014). A relaxed mood state indirectly affects purchase behaviours, further supporting the necessity of delivering a positive experience for each customer by tailoring individual experiences to co-create a relaxed mood state, rather than focusing on levels of wine involvement (Bruwer et al., 2018) or developing standardised experiences (Lee et al., 2021). This relaxed state may stem from reducing the cognitive dissonance associated with experiences that do not meet expectations (d’Ament et al., 2022a). Personality exerted a slightly greater influence on expectations in a moderating capacity. Conscientiousness has been shown to have a negative relationship with impulse buying (Tarka et al., 2022). As such, self-discipline containing purchases at or below expectations is not unexpected. Neuroticism influences attitudinal loyalty, which further supports the idea of tailoring experiences to the individual. Attitudinal loyalty behaviours, such as word-of-mouth recommendations, have been incorporated into the knowledge base, contributing to CDE schemas (d’Ament et al., 2023a). Neuroticism is associated with negative affect and overthinking situations. Hence, individuals may require different interaction styles with cellar door staff to sufficiently relax their mood and co-create the connection and engagement essential to meet the expectations necessary for increasing purchase behaviour.

Such adaptability in cellar door engagement styles becomes essential when considering the dynamic mediation between personality and trait affect (Kritzler et al., 2020). Situations, such as a CDE, and state affect (i.e. mood) mediate the relationship between personality trait and trait effect. As such, the negative affect associated with the emotional instability of neuroticism, which was found to influence purchase behaviours, may be tempered by the mood created during the CDE.

As it is widely held across the literature that an enjoyable CDE is necessary, an unexpected result was the influence of the quality of the experience ratings. These were moderately influential for the recommendation of the CDE (16.5%), and to a lesser extent the winery (11.1%), but exerted far less influence on other outcome variables [i.e. total spend (5%), bottle no. purchased (4.24%) and join wine club (1.64%); see Table 1]. When considered in conjunction with other results from the BN, a deeper understanding can be applied to the findings from previous studies that enjoyable CDEs are essential to the profitability of the winery business. Lease et al. (2023) correctly suggest that the predominant function of the cellar door is to generate sales, and delivering a compelling CDE is critical to achieving these sales. However, the most influential variable on the total_spend outcome variable was purchase expectations (28.2%). Such expectations were perhaps formed from word-of-mouth recommendations or based on CDE schemata. d’Ament et al. (2023a) found that the co-created human relationship at the centre of the CDE was most treasured and recalled in greater detail than the views or built environment. When considered in conjunction with these findings, the current study not only supports the power of word-of-mouth communication but also indicates that a non-sales-focused approach from staff may increase profits. It is posited here that while direct-to-customer sales are essential for future or ongoing WB profitability, delivering CDEs that exceed expectations and ongoing development of positive CDE schemata are more beneficial focuses than sales.

Our findings suggest that WBs need to take a dynamic approach to engaging with customers and to recognise individual differences and micro behaviours; this is at the cutting edge of consumer behaviour research and is challenging for WBs to implement. The notion of recognising an individual’s preferred style has theoretical roots in research on proxy raters, where, often in a medical context, an external assessor is asked to respond on behalf of a patient who cannot. Two issues present: first, the reliability of the proxy rater is yet to be understood (e.g. Northfield et al., 2023) and second, whether accurate assessments can be established. Further research is needed before WBs can implement the style of dynamic approach that we have found can lead to higher value customer outcomes. Future research could also address a limitation of the current study, which only evaluated the customer experience, by surveying the staff conducting the CDE. Research expanding the use of proxy ratings beyond a physical or mental health application will provide further evidence of the importance of training staff to recognise the influence of individual differences on experience delivery and success.

Providing an enjoyable CDE is widely recognised as essential for WBs, wine tourism and the wine industry. The current study has adopted a unique approach to gain a deeper understanding of several elements contributing to co-creating enjoyable CDEs and makes notable theoretical and practical contributions to understanding customer purchasing decisions and loyalty behaviours in cellar door experiences. Through a Bayesian network analysis of 42 variables across 136 customer responses from seven Australian wine regions, this research offers new insights into how expectations, individual states and traits interact to influence key business outcomes.

Ratings of the CDE quality, which measures the subjective enjoyment of the CDE, consistently upheld in much of the CDE literature as essential, was most influential for attitudinal loyalty, being word-of-mouth recommendations of the experience. Delivering CDEs that consistently meet expectations requires two important components; first, adequate training and education of cellar door staff, which empowers their delivery of wine-focused information that importantly, is relevant to the wines they are selling and the wine business they are representing. A global knowledge of wine is only useful if staff apply this knowledge to the wines being offered during the CDE. As such, cellar door staff need not be sommeliers, but open to learning and talking about the wines they are selling. Secondly, the delivery of wine-focused information needs to be delivered with awareness of the well-established importance of developing rapport with customers. As shown by the BN a relaxed mood influences a customer’s expectation of purchase, which is the most influential variable on purchase and loyalty behaviours.

However, expectations were shown to exert influence over and shape commercial outcomes. Our research advancing schema congruity theory by demonstrating that exceeding customer expectations – more than merely meeting them – is a primary driver of purchase behaviour. Schema variables were shown to have a strong influence over purchases, as well as attitudinal and behavioural loyalty. The finding that purchase expectation exerted the most decisive influence on total spend (28.2%) and bottle purchases (21.4%) provides empirical validation of the crucial role of cognitive schema in consumer decision-making.

The study advances the literature on individual differences and consumer behaviour by highlighting the influence of individual states and traits on purchasing situations. Our identification of conscientiousness and neuroticism as key moderators of purchase expectations, along with the impact of relaxed mood states, offers a detailed understanding of how individual differences affect consumer responses. Notably, the finding that less conscientious customers in relaxed states tend to purchase above expectations provides new insights into impulsive buying behaviours. Methodologically, the research introduces Bayesian network analysis as a powerful approach to understand complex, interconnected relationships in experiential retail settings. Unlike traditional statistical approaches, examining isolated variables, the BN model captures bidirectional influences among the 42 variables simultaneously, offering a holistic understanding of the CDE. Real-time data collection during experiences represents an advancement over retrospective studies, capturing authentic moods and emotional states as they influence purchase decisions.

Our findings fundamentally challenge traditional sales-focused approaches during CDEs, demonstrating staff should prioritise creating experiences that exceed customer expectations rather than focusing primarily on immediate sales conversations. As per d’Ament et al. (2022a) who recommend CDEs be tailored to customer needs, to meet a minimum standard rather than being standardised. Such a paradigm shift has profound implication for staff training, suggesting that developing emotional intelligence and adaptive interaction styles yields greater commercial returns than extensive wine knowledge alone. By recognising that personality and mood moderators offer practical insights for personalising CDEs, staff can adopt different engagement styles and interaction strategies, to foster attitudinal loyalty. Further, this study provides empirical proof supporting strategic investment in experience quality over traditional marketing methods for smaller wineries. Also, that visit frequency and purchasing patterns have greater influence over wine club membership than emotional responses challenges conventional relationship marketing beliefs.

In conclusion, our research fundamentally enhances understanding of customer behaviour during the CDE by highlighting the importance of expectation management over traditional sales focused strategies. The innovative use of Bayesian network analysis in wine business research, combined with practical insights for personalising customer experiences, makes this study a noteworthy contribution to both academic knowledge and industry practice.