INTRODUCTION
The U.S. wine market is the world’s largest, valued at $58.2 billion annually, double that of the next largest country, France ($26.7b) (Statista, 2024b). It is poised for significant growth, with an expected 8 percent increase between 2022 and 2027. While some of this growth can be attributed to inflation (Mintel, 2024), it underscores the potential for expansion in the industry. It is the fourth largest producer globally, with 24.3 million hectoliters, trailing the leader, France, at 48 million (Statista, 2024d). With 11,053 wineries (Statista, 2024c), the U.S. wine market is a promising sector, accounting for 16 percent of the total U.S. alcohol market, with beer and spirits accounting for 42 percent each (Statista, 2024a).
Wine consumption is growing, as is wine tourism. Consumers want to see where wine is grown and schedule vacations around it. It is estimated that 50 countries produce wine, with 28 accounting for 85 percent of the worldwide production (World Population Review, 2024). The global market is valued at over USD 46 billion, with a compound annual growth rate (CAGR) of nearly 13 percent expected from 2024 to 2030 (Grand View Research, 2023). The U.S. growth for the same period is estimated at 14 percent. An Arizona study found that 40 percent of tourist visits were made to a winery, and 70 percent of those visitors purchased wine. The markets are segmented based on differences between domestic and international visitors in Greece. These differences include “pre-visit behaviour, primary knowledge and loyalty levels towards the winery’s brands, visitation motives, spending attitudes, income, age distribution, perception of the winery experience and future behavioural intentions” (Nella & Christou, 2021, p. 1).
This competition makes it imperative to understand consumers, especially the segments they comprise. Mass marketing will not succeed, and the industry must identify the optimal segments to focus marketing strategies on (Bruwer & Li, 2007). Thus, marketing strategy and segmentation must be grounded in competition theory (Hunt & Arnett, 2004). This study targets academics and practitioners with a theoretically grounded segmentation method that results in actionable strategies for wine marketers.
LITERATURE REVIEW
Segmentation
The theory of market segmentation began in 1956 (Smith, 1956). It enables greater strategic precision based on an understanding of consumer behavior (Dolnicar et al., 2016). Customers are assigned to homogenous and distinct groups (Wedel & Kamakura, 2000). These subgroups share similar needs, desires, and traits (Barber et al., 2007; Brunner & Siegrist, 2011). These smaller groups are more meaningful since their needs are similar. This creates resource efficiency through targeted promotional campaigns, customized products (Bruwer & Li, 2007), and attractive segments (Barber et al., 2007). Some explanatory power is lost when consumers are grouped, as segment and individual needs may differ. Recent work continues to identify heterogeneous wine segments using latent-class segmentation, including health-nutrient and origin awareness clusters (Petrontino et al., 2022).
Wine Segmentation
Socio-demographic, lifestyle, wine knowledge, and involvement level are shown to accurately reflect wine segments (Ahmad, 2003; Kelley et al., 2015; Seghieri et al., 2007). Pomarici et al. (2017) use a latent-class segmentation analysis based on specific wine attributes. They identified four segments that differ in wine preferences, demographics (i.e., education, gender, and income), and psychographics (i.e., loyalty, innovativeness, involvement, and subjective knowledge). Bruwer and Li (2017) used a latent-class model to segment wine drinkers. Five segments were identified: i) “involved, knowledge-seeking wine drinkers”; ii) “younger, relatively inexperienced wine drinkers”; iii) “enjoyment-oriented, social wine drinkers”; iv) “basic wine drinkers”; and v) “conservative, knowledgeable wine drinkers” (p. 1567-1568). A segment analysis of Croatian wine drinkers identified six distinct segments: prestigious consumers, experts, traditional consumers, hedonists, savers, and modest consumers (Kalazić et al., 2010). Portuguese consumers were segmented into experienced, less experienced, and inexperienced consumers (Ferreira et al., 2020). Age and gender were the most important socio-economic covariates in the study. Australian wine drinkers were segmented into Enthusiasts, Aspirants, and No Frills (Danner et al., 2020). Enthusiasts were more discriminating regarding wines and had more intense positive emotions when consuming wine than the No Frills segments. Aspirants were between the two extremes.
Age
Wine consumption differs by age group. Wolf, Wolf, and Lecast (2022) found that “segmentation by generation is appropriate when creating product pricing, determining channels of distribution, and creating messaging for a specific wine brand” (p. 373). Younger generations purchased less expensive wines than older generations. Olsen et al. (2007) also found differences in the type of wine preferred among Millennials, Generation X, Baby Boomers, and Traditionalists. Younger wine drinkers tend to engage in more frequent and social drinking, whereas older drinkers are more occasional and ritualistic (Lange et al., 2010). Age groups vary in their preferences for different types of alcohol; for example, Millennials prefer liquor over beer or wine, while the Silent Generation favors wine (Statista, 2019b). Age and spending on wine are negatively correlated. Thach et al. (2020) researched Generation Z, Baby Boomers, Millennials, and Generation X. The first two groups were similar in wine consumption, while the last two were high-frequency consumers. In a comparison of those over the age of 63 and others in Denmark, quality and origin were linked for the latter group (Berni et al., 2005). Generation Y and older cohorts wine drinkers differed on wine expertise, price-quality inferences, and hedonic orientation (Kennett-Hensel et al., 2011). Generation Y are less knowledgeable and more likely to rely on price-quality inferences than older cohorts.
Education
With increased education, there was a corresponding increase in the use of wine with meals outside the home and at parties. The higher the level of education, the greater the openness to wines from new competitors (Berni et al., 2005).
Gender
Forty-three percent of women and 22 percent of men list wine as their favorite alcoholic beverage (Statista, 2019a). However, men consume more wine than women (Berni et al., 2005), and women tend to focus on the social aspects of wine (Wilsnack, Wilsnack, Gmel, et al., 2009; Wilsnack, Wilsnack, Kristjanson, et al., 2009). A segmentation study on wine found that three of the five segments were female-dominated (Bruwer & McCutcheon, 2017). Pomarici et al. (2017) found gender differences in segmentation. Men tend to be more simplistic in their assessment of wine, relying on heuristics, whereas women use more analytical approaches and conduct more information searches before making a purchase (Atkin et al., 2007; Bruwer et al., 2011; Bruwer & McCutcheon, 2017; Thach, 2012). Women are more concerned with calories (Berni et al., 2005).
Extrinsic Variables
Wine possesses both intrinsic and extrinsic cues. Intrinsic is the product itself, while extrinsic is external to the product (Schiffman & Kanuk, 2010). Intrinsic factors include, for example, alcohol content, aroma, color, flavor, organic, and texture (Boncinelli et al., 2018; Kelley et al., 2015). Since wine can rarely be tasted before purchase, extrinsic factors are often used to inform the purchase decision (Boncinelli et al., 2018). Extrinsic factors often represent quality heuristics, such as brand, labeling/packaging, place of origin, price, vintage, and winemaker (Charters & Pettigrew, 2007; Lockshin et al., 2006; Sáenz-Navajas et al., 2013; Viot, 2012). According to a Mintel study, 20 percent of buyers would opt for less expensive wines to save money (Mintel, 2024). Chinese wine drinkers were classified into three categories: intrinsic-attribute-seeking, extrinsic-attribute-seeking, and alcohol-level-attribute-seeking (Liu et al., 2014).
Involvement
Involvement measures the importance or relevance of a consumer’s decision based on their inherent interests, needs, and values (Zaichkowsky, 1985, 1987). Experiential and symbolic significance leads to more enduring involvement (Ogbeide & Bruwer, 2013); wine is an example of this. High-involvement wine consumers use more complex decision-making in choosing wines, while low-involvement consumers rely more on price and awards (Lockshin et al., 2006). Wine drinkers who are highly involved with the product consume more frequently (Hirche & Johan Bruwer, 2014). For high-involvement individuals, wine is an integral part of their lifestyle (Bruwer & Buller, 2013); they are more engaged with it. Conversely, low-involvement wine consumers put little effort into information search for the product (Bonn et al., 2016).
For wine, age and involvement were found to be related (Roe & Bruwer, 2017); individuals over 44 years old had higher involvement in wine. Low-involvement consumers purchased fine wine when it reflected a higher sense of self-worth (e.g., a desire for group affiliation or a reluctance to be perceived negatively), possibly due to a lack of knowledge about wine (Roe & Bruwer, 2017). A recent study showed involvement-based differences in cue utilization for sparkling wine (Pickering & Kemp, 2024).
Income
Higher income increases consumption (Saffer & Dave, 2005). High-frequency drinkers are highly involved and have higher incomes (D’Hauteville, 2003). Income differences were found in segmentation studies (Bruwer & Li, 2017; Pomarici et al., 2017).
Theory
This study is grounded in Resource-advantage Theory (R-A Theory) and the Elaboration Likelihood Model (ELM). Our contribution is to demonstrate that involvement-driven processing (ELM) explains segment differences in consumption and provides the theory for aligning offerings with marketplace needs to create a sustainable competitive advantage. This occurs by aligning firm resources to segment needs, as posited by R-A Theory. Segments are uncovered through Latent-class Analysis (LCA). LCA probabilistically assigns respondents to subpopulation or segments based on class-specific parameters.
Segmentation aims to discover unobserved heterogeneity, that is, subpopulations that differ in needs and decision rules. Segmentation allows firms to strategically deploy the marketing mix more effectively in enticing purchase and loyalty. This is based on R-A Theory where firms align offerings and promotion to segments most receptive to it (Hunt & Arnett, 2004; Hunt & Morgan, 1995, 1996). Firms target segments where they have a competitive advantage.
Segmentation’s primary behavioral mechanism is consumer involvement, a core construct of the ELM. Involvement or perceived personal relevance or importance is based on consumers’ values, needs, and interest. With the ELM, high-involvement consumers follow a central route of persuasion characterized by extensive elaboration, while low-involvement consumers follow a peripheral route that relies on simple cues and heuristics (Petty & Cacioppo, 1986). Higher involvement leads to extensive information search, complex processing, and extensive decision making (Zaichkowsky, 1985, 1987).
Wine
Heterogeneity in price-sensitivity, involvement, internal and external cues, and demographics aid in identifying which segments to successfully pursue based on the receptiveness to the marketing mix. When choosing wine, consumers use intrinsic (e.g., aroma, flavor) and extrinsic cues (e.g., price, brand, origin, and packaging/labelling). When pre-purchase tasting is unavailable, extrinsic cues serve as quality signals in setting expectations and value. Involvement is a strong predictor of wine purchase intentions (Rinck, 2025). Hung, Tang, and Huang (2022) in a segmentation study show the importance of psychographics such as involvement over demographics. Psychological constructs such as purchase intentions and trust can weaken demographic effects (Indiani et al., 2024).
Higher-involvement wine segments place greater weight on quality attributes, production region, variety, and other cues. Low-involvement wine segments rely more on heuristics such as price and convenience (Pickering & Kemp, 2024). This is the peripheral route of persuasion, which is more reliant on heuristics. High-involvement segments for sparkling wine use multiple extrinsic cues such as brand, origin, and label in choosing wine, while low-involvement consumers rely on the heuristic of price (Pickering & Kemp, 2024). Under the ELM, higher involvement increases reliance on extrinsic cues and greater consumption frequency. Low involvement shifts processing to the peripheral route, increasing reliance on price and simple heuristics. High involvement is multi-cue, attribute processing, and low involvement is single-cue heuristic processing.
Barbierato and Alvino (2025) synthesize neuroscience studies in wine marketing and show that involvement amplifies neural responses to brand and origin cues. Sensory and extrinsic cues influence decision-making at a subconscious level. Storytelling and label design are more persuasive for premium brands with high involvement. This enhances brand equity and utilizes firm resources effectively through a central route of persuasion. This is how R-A theory can be operationalized through involvement-based segmentation.
The hypotheses are:
H1: Clusters will exist based on how often wine is consumed and the quantity per occasion.
H2: Clusters will differ by age.
H3: Clusters will differ by gender.
H4. Clusters will differ based on extrinsic factors such as brand, label/packaging, locally grown, place of origin, and price.
H5: Clusters will differ by income.
H6. Clusters will differ by level of involvement with wine.
METHODS
The survey was constructed after an extensive literature review and telephone interviews with nine wineries. The survey was developed in Qualtrics and distributed to wine drinkers on Amazon’s Mechanical Turk (MTurk). A pretest was conducted in the same manner, and minor adjustments were made to the survey based on the pretest results. Respondents were compensated for survey completion. Funding was provided by a northeastern state’s Department of Agriculture. Respondents were from nine northern states. Written informed consent was provided. Respondents were required to have purchased and consumed wine in the last three months and be at least 21 years of age. The survey ran for one week.
MTurk’s external validity has declined over time, making it less representative of the general population (Shimoni & Axelrod, 2025; Tang et al., 2022); however, MTurk is suitable for theory-driven segmentation studies focused on internal validity rather than precise population estimates or external validity. We limit claims to theoretically derived segments, not segment sizes to the U.S. wine-drinking population.
We aim to predict how often a respondent consumes wine and the amount consumed per occasion. Both variables are measured on an ordinal scale. The Spearman rank correlation between these ranked variables is -.29 (p < .000), indicating a weak association. Less than 6 percent of the variation in one variable is explained by the other, and the correlation is negative. The constructs mainly measure different things. There are six categories of how often you drink wine, ranging from every day to less than once a month. Glasses per occasion has six response categories, ranging from one to more than five.
The independent variables are age, extrinsic factors (i.e., brand, label/packaging, locally grown, place of origin, and price), gender, income, and involvement. Two of the variables are measurement scales. The first measures extrinsic factors in the decision to buy wine. It is a five-item scale. Each question is measured on a seven-point scale, anchored from “not at all important” to “extremely important.” The scale has a coefficient alpha of .77, indicating acceptable internal consistency (Fornell & Larcker, 1981). Since it is internally consistent, questions are summed to create the scale.
The second scale is a three-item scale for involvement with wine. It includes a strong interest in wine, careful selection of wine, and a love of giving wine as a gift. The scale is one-to-seven, with strongly disagree being one and strongly agree being seven. Coefficient alpha is .71. The scale is internally consistent (Fornell & Larcker, 1981).
The demographics are age, gender, and income. Actual age in years is recorded. Gender has four categories: male, female, non-binary, and prefer not to answer. Income is categorized into seven ranges, starting with less than $25,000 and ending with more than $150,000. Awards were omitted for parsimony and to avoid redundancy with brand and price. Low-involvement decision making often relies on simple heuristics such as awards and price. Awards may be part of a composite cue subsumed by brand and price perceptions. Since our focus is on involvement-driven segments, we prioritized extrinsic factors with greater impact on the Elaboration Likelihood Model and Resource-Advantage Theory.
Detail on how Latent-class Analysis was used is in the Appendix.
RESULTS
Four-hundred-and-eighty-four respondents met the screening criteria for purchasing and consuming wine in the last three months and being at least 21 years of age. Thirty-seven respondents completed only a small portion of the survey and were dropped. No respondents were dropped because of low variability. The sample size is now 447.
Mahalanobis Distance was estimated using regression to identify multivariate outliers. The dependent variable is the respondent number (one to 447), and the independent variables are the 13 Likert-type questions used in the study (Tabachnick & Fidell, 2021). The chi-squared critical value with 13 degrees of freedom and a p<.001 is 34.53. Eight respondents were identified as multivariate outliers and removed. This left 439 respondents. Fifty-three did not complete a question for the variables in the model. The LCA requires 100 percent completion. These were removed for a final sample size of 386.
The number of males (51%) and females (49%) is almost equal (Table 1). Nearly a quarter (24.6%) of respondents earned between $50,000 and $74,999. Thirty-eight percent make less than $50,000, and 17 percent earn $100,000 or more. Respondents are overwhelmingly white (85%), with 7 percent black or African American. In a separate question, 12 percent identified as Hispanic, Latino, or Spanish origin. The mean age is 40, with a range of 20 to 83.
A one-to-five-cluster latent-class solution is estimated without covariates to identify the number of segments (Table 2). Five clusters were chosen as the largest number of clusters, given the sample size of 386. If equal, six clusters would have an approximate average of 64 respondents per cluster (386/6); however, the clusters become dramatically smaller, so the sixth cluster would be significantly smaller. The BIC declines from one to two clusters and then rises. The two-cluster solution is best. The L2 is statistically significant for the one-cluster solution indicating poor fit but not statistically significant for the others. (Note: It should not be statistically significant.) The drop between the one- and two-cluster solutions is large compared to the drop-off among the others. The same is true for the log-likelihood. The bivariate residuals are below four for all clusters except the first, with the lowest at five. The misclassification rate is 9 percent with two clusters and increased to 25 percent with three clusters. The two-cluster solution is chosen.
The model was re-estimated with covariates. Cluster One is the largest, comprising 73 percent of the sample, and Cluster Two accounts for 27 percent (Table 3). The dependent variables are the frequency of wine consumption and the typical number of glasses consumed per occasion. The model explains 33 percent and 19 percent of the variability in the dependent variables, respectively. Both dependent variables were statistically significant; this supports hypothesis one. Among the covariates, extrinsic factors, income, and involvement are statistically significant at the p < .05 level. This supports hypotheses four, five, and six. Age and gender were not statistically significant. Hypotheses two and three are not supported. We reran the model adding an interaction between age and involvement; however, it was not statistically significant. The main effect of involvement was still significant.
Cluster One drinks wine more often than Cluster Two (Table 4). Twelve percent drink daily in Cluster One compared to one percent in Cluster Two. Forty-five percent of those in Cluster One drink several times a week. For Cluster Two, this is only 8 percent. When Cluster Two drinks, it is primarily one or two glasses per session (92 percent). Only eight percent in Cluster One drink one drink per occasion. Ninety-two percent have more than one drink.
Table 5 presents the three statistically significant independent variables. Extrinsic cues that are important are measured on a five-item scale anchored from not at all important (five) to extremely important (35), with 20 being moderately important. Forty-one percent in Cluster Two view the factors as unimportant, while only 17 percent do in Cluster One. Conversely, 27 percent in Cluster One view them as extremely important compared to only 4 percent in Cluster Two.
For the wine involvement scale, lower numbers mean less involvement. The three-item scale ranges from three (strongly disagree) to 21 (strongly agree), with 12 indicating neutral. For Cluster Two, 76 percent are in the bottom two groups; they strongly disagree or are neutral about wine involvement. Cluster One comprises 29 percent of these groups. Conversely, 57 percent are in the two high-involvement categories for Cluster One, and only 20 percent for Cluster Two; they agree or strongly agree that wine is a high-involvement product for them.
With Cluster One, 41 percent have incomes below $50,000 compared to only 25 percent for Cluster Two. Cluster Two has more people in higher income brackets. Forty-six percent have incomes above $74,999, while Cluster One has only 36 percent. Cluster Two has more people with incomes between $50,000 and $74,999 (23% vs. 30%). Note: Latent Gold combined the first categories into one (i.e., less than $25,000 and $25,000 to $34,999) and the last two categories.
Table 6 presents the two non-statistically significant independent variables. Among age categories, the two clusters differ by no more than 7 percent (50-83 years old or 19% vs. 26%). Cluster One is evenly split between men and women, while Cluster Two has slightly more males. (Note: Latent Gold combined ages into categories since actual age was the variable measured.)
For important extrinsic factors, we took cluster membership and ran independent samples t-tests on the scale’s five items (Table 7). The clusters are not different (statistically significant) in price but in brands, label/package, locally grown, and place of origin. Price is important for both clusters. Effect sizes were estimated. For example, Cohen’s d indicates that Cluster 1 rates place of origin approximately 1.7 standard deviations higher than Cluster 2.
We examined within cluster differences. For Cluster One, a one-sample t-test shows that brand, label/package, locally produced, and price are statistically significant (Table 8). Place of origin is not statistically significant. They were tested against the midpoint (4) of moderately important. Price is very important. Brand is important. Label/package and place of origin are slightly important.
For Cluster Two, all five variables are statistically significant, but most are below the midpoint of moderately important. Price is very important. The other variables are below important. Brand, label/package, and place of origin are slightly important. Locally grown is of little importance. This cluster buys almost exclusively on price.
Table 9 shows the hypotheses’ results.
CONCLUSION
The American wine market is the largest in the world and the fourth-largest producer. From our research, it is a heterogeneous market. Predicting consumer behavior is paramount for success in an industry with myriad domestic and international competitors. Our study confirms that the latent-class model can identify segments (Bruwer & Li, 2017; Pomarici et al., 2017). Our contribution is in showing a dual-processing involvement-anchored perspective explains why traditional demographic predictors (e.g., age and gender) lose explanatory power when involvement is modeled. The ELM and RA Theory are integrated into a latent-class model. Combined they can align attributes with segment-specific processing to create actionable managerial implications for a competitive advantage. According to the ELM, segments are defined by either the central or peripheral routes of persuasion and linked to different resource allocation and positioning strategies. Using RA Theory, the firm’s resources are aligned with segment needs. We identify a stable and actionable two-segment model for decision making. Prior research usually links demographics as primary segment drivers.
Our model identified two clusters. We explained more variation in how often people drink than in how much per occasion. The clusters do differ on income (Bruwer & Li, 2017; D’Hauteville, 2003; Pomarici et al., 2017; Saffer & Dave, 2005), involvement (Hirche & Johan Bruwer, 2014; Lockshin et al., 2006; Pomarici et al., 2017), and extrinsic factors (brand, label/package, locally produced, and place of origin) (Bruwer & Li, 2017; Pomarici et al., 2017). Income influences the ability to purchase. Higher income increases consumption (Saffer & Dave, 2005). Price is not different across segments, since it is important for both.
Wine has both experiential and symbolic significance (Ogbeide & Bruwer, 2013). Involvement affects the wine decision-making process. Higher involvement leads to more complex decision-making (Lockshin et al., 2006) and greater consumption (Hirche & Johan Bruwer, 2014). Low involvement can influence a purchaser’s self-concept (Roe & Bruwer, 2017).
Extrinsic cues have been used to segment markets (Charters & Pettigrew, 2007; Lockshin et al., 2006; Sáenz-Navajas et al., 2013; Viot, 2012). Brand, label/package, locally grown, and place of origin are different across segments. As wine is rarely consumed before purchase, extrinsic variables become more important, and intrinsic variables such as aroma and taste are less discernable. Intrinsic variables such as alcohol content, color, and organic can be measured without tasting the wine.
Although prior wine segmentation studies found age and gender differences, our results show no differences on these demographics (Berni et al., 2005; Lange et al., 2010; Olsen et al., 2007; Thach et al., 2020; Wolf et al., 2022) or gender (Atkin et al., 2007; Berni et al., 2005; Bruwer & McCutcheon, 2017, 2017; Kennett-Hensel et al., 2011; Pomarici et al., 2017; Thach, 2012; Wilsnack, Wilsnack, Gmel, et al., 2009; Wilsnack, Wilsnack, Kristjanson, et al., 2009). This is consistent with the ELM where involvement-based processing can dominate demographic effects on wine decision making (Tang et al., 2022). Demographics have been shown to be subsumed by theoretically-based psychological constructs (Indiani et al., 2024). Highly-involved consumers, regardless of age or gender, process information about wines through a central route of persuasion, whereas low-involvement consumers rely on peripheral cues.
The larger segment (Cluster One) generally drinks more often and has more per occasion. Almost 60 percent drink at least several times a week. Almost half drink more than three glasses per occasion. It is almost three times the size of the smaller one.
Seventy percent of the larger segment views wine as a high-involvement product. More than four in 10 of the cluster have incomes less than $50,000, and a little more than a third have incomes above $75,000. Price is important, and brand name and place of origin moderately important. Label/package and locally grown are slightly important. They use all the extrinsic variables to a degree. Since it is high involvement, they are more willing to learn about wines. This corresponds to prior research where high-involvement wine consumers use more complex decision-making (Lockshin et al., 2006). Who manufactures the wine and where it is from are used in their decision. To a lesser extent, the label is reviewed. To value the label, the consumer must understand, for example, the unusual qualities of the wine, vineyard designation, vintage, and type of bottle. They would be approached with more information about the wine.
Consumption for the small segment (Cluster Two) is much lower: 9 percent drink several times a week and almost all have less than two drinks per occasion. Approximately three-quarters of respondents view wine as a low-involvement product. Nearly half have incomes above $75,000. Price is important but not different in importance than larger segment. Brand is slightly important. Label/package, locally grown, and place of origin are between little and slightly important. Price largely dominates choice. This corresponds with prior research indicating that price and awards influence low-involvement consumers more (Lockshin et al., 2006).
Retailers can focus on both segments. These segments are like prior studies examining experienced and inexperienced consumers (Danner et al., 2020; Kalazić et al., 2010). The larger cluster is high-involvement and needs more information. The brand name is moderately important and should reflect sufficient brand equity for quality. Place of origin must signify quality, although it was not statistically significant for our results. The label must provide sufficient and persuasive information for a high-involvement consumer. Locally produced also garners between slight and moderate importance.
For the high-involvement segment, managers should emphasize storytelling, origin, and label information (QR codes for vineyard narratives and certifications) to leverage central-route processing and reinforce premium positioning. This information leads to persuasion and enhances perceived brand equity and credibility. This is supported by Barbierato and Alvino’s (2025) finding that rich cues amplify neural responses for premium brands. Cobb-Walgren, Donthu, and Pilling (2022) focused on in-store decision making and recommended that high-involvement consumers be targeted with richer in-store information because consumers are motivated to process and use that information.
For the smaller cluster, it is a low-involvement product bought mainly on price. A basic affordable product will suffice here. They put little effort into information search (Bonn et al., 2016). Labels should be simplified and retail availability enhanced so that peripheral cues—such as price promotions, simplified displays, awards, and influencer endorsements—guide choice with minimal cognitive effort (Cobb-Walgren et al., 2022). These heuristics are successful in a peripheral route to persuasion. Companies with multiple brands should have a price laddering to distinguish perceived quality.
LIMITATIONS AND FUTURE RESEARCH
The sample was drawn from MTurk and from one U.S. region, which limits its generalizability to the US wine-drink population. Sample representation is a risk with Internet surveys (Couper, 2011). Future research should include a more geographically diverse sampling method to improve representativeness.
Responses were self-reported, making them susceptible to social desirability. Anonymity was guaranteed to mitigate social desirability bias (Tourangeau & Yan, 2007). Online surveys have been shown to have less bias than face-to-face surveys (Krumpal, 2013). A randomized response technique could be used in future research to reduce social desirability more (Volicer & Volicer, 1982).
The study was cross-sectional, rather than longitudinal. Data was collected over one week. Seasonality effects could have occurred. Intrinsic cues were not included in the model because of the overlap between intrinsic and extrinsic and parsimony. We did not focus on specific wines these segments enjoy. We did not test for interactions. For example, age and involvement were found to be related in prior research (Roe & Bruwer, 2017).
Future research should focus on three extensions: i) heuristic cues and signal strength; ii) involvement moderators; and iii) longitudinal consumption and involvement. The effect of awards as extrinsic cue could be studied to disentangle its effect with brand and price perceptions. This is particularly relevant for low-involvement consumers where awards aid in reducing decision-making risk. Different types of awards could be examined: regional vs. international.
Future research could incorporate contextual factors that are likely to interact with involvement. Consumption location (e.g., bars, friend’s home, home, and restaurants), occasion (e.g., anniversary or parties), purchase reason (e.g., gift or personal) (Boncinelli et al., 2019), and type of wine consumed (Berni et al., 2005) were not included.
Consumers behave differently depending on whether wine is purchased as a gift or personal consumption (Boncinelli et al., 2019). Involvement has been shown to influence purchase depending on the occasion (Roe & Bruwer, 2017). Gifts are influenced more by extrinsic factors. Segments could be examined for when buying gifts or personal consumption. Occasion and involvement can be manipulated to see the effect. For purchase intentions, loyalty is the strongest indicator (Rinck, 2025). Additionally, the extrinsic factors that are most important within the categories tested could be studied. For example, what place of origin is perceived as most important?
Finally, temporal dynamics could be explored. Examining longitudinal data could examine seasonal and lifecycle patterns. Involvement appears to increase with age or experience. A longitudinal study can examine consumption across and within seasons.