How consumers define brand relationships

Purpose – Understanding consumer brand relationships from the perspective of the consumer has been a research topic for years. Despite this, there are still various ways in which the construct is interpreted. This paper aims to identify the most typical interpretation of brand relationships by consumers. Design/methodology/approach – A four-study prototype analysis was conducted, in which a bottom-up approach was applied to identify lay people ’ s conceptualization of consumer brand relationships. Findings – The prototype analysis generates a comprehensive list of features of consumer brand relationships that provide a nuanced understanding of the concept. The most typical characteristics of a brand relationship according to consumers are quality, bond, value and joy. Comparing this relationship prototype with existing literature shows that there may be a gap between theory and practice regarding the concept of brand relationship. Originality/value – The prototypical conceptualization of brand relationships shows which aspects play a role in consumers ’ most common interpretation of the construct. This provides an opportunity to assess the validity of existing conceptualizations of brand relationships. Knowing which aspects are most relevant for consumers ’ brand relationships allows brands to make adjustments as needed and improve at establishing and maintaining relationships with consumers.


Introduction
Relationships between brands and consumers are a powerful mechanism for achieving organizational benefits, such as reduced marketing costs, increased brand loyalty, more purchases and customer retention (Fajer and Schouten, 1995;Khamitov et al., 2019a;Smit et al., 2007).Understanding how consumers perceive relationships with brands is therefore vital.
Although brand relationships have been intensely studied (cf.Keller, 2020), there are still many different interpretations, resulting in a complex, wide variety of concepts concerning specific types of connections between consumers and brands.The plethora of theoretical concepts does not necessarily make things more concrete.Often, concepts are defined in terms of other concepts, causing inherent overlap and confusion (Jones et al., 2018), prohibiting consensus about what consumer brand relationships are (Albert and Thomson, 2018;Fetscherin and Heinrich, 2015).For example, attachment has been named a dimension of consumer brand relationships (Esch et al., 2006;Fournier, 1998), while it has also been proposed as a separate type of relationship (Thomson et al., 2005;Whan Park et al., 2010).The same holds for love, which is both regarded as a dimension (Fournier, 1998;Robertson et al., 2022), and as a type of relationship (Batra et al., 2012;Carroll and Ahuvia, 2006;Wallace et al., 2022).
The current issue and full text archive of this journal is available on Emerald Insight at: https://www.emerald.com/insight/1061-0421.htm In examining brand relationships, research has disproportionately focused on loyal, positive, high involvement relationships (Fournier and Alvarez, 2013).In reality, most brands struggle to develop any kind of relationship with consumers that goes beyond indifference.People have limited relational capacity and simply do not have the time, interest, or emotional energy to form strong relationships with too many brands (Alvarez and Fournier, 2016;Dowling, 2002).In light of the academic-practitioner gap in branding research (Alpert et al., 2022), there is a need for a thorough review and revision of the brand relationship concept.Which concepts are most relevant, which aspects apply in most situations?The current research aims to answer these questions bottom-up, by identifying the most typical interpretation by consumers.This could clarify which aspects are involved in consumers' interpretation of these relationships and help assess the validity of existing conceptualizations of brand relationships.For brands, knowing what consumers' most typical interpretation of brand relationships is can help move currently used concepts within marketing strategies to correspond to consumers' conceptualizations, allowing their strategies to improve at establishing and maintaining relationships with consumers.

Literature review
Building brand relationships is a relevant and important topic already for thirty years (Blackston, 1993).In recent years, the Journal of Product & Brand Management dedicated three special issues to the topic (Fetscherin et al., 2016(Fetscherin et al., , 2019(Fetscherin et al., , 2021)), and the number of relevant publications increased significantly over the years.A text-mining analysis of 287 publications reveals 71 different constructs (Albert and Thomson, 2018).These developments and numbers confirm the great diversity of different brand relationship-related constructs.
Scholars have attempted to narrow down these constructs, often by placing them on dimensions.A frequently used dimension is the strength (Fetscherin et al., 2019) or intensity (Veloutsou and Ruiz Maf e, 2020) of the emotional bond a consumer experiences with a brand.Sometimes, this strength is linked to the concept of engagement (Fernandes and Moreira, 2019;Veloutsou, 2015), which is described as an emotional, cognitive and behavioral brand-related concept (Hollebeek et al., 2014;Vivek et al., 2012).If the engagement manifests itself in a process of repeated interactions and benefits both the consumer and brand, it can also be considered a form of (brand) co-creation (Prahalad and Ramaswamy, 2004;Vargo and Lusch, 2016).This joint creation of value can take place between the brand and an individual or group of consumers, or between consumers themselves.Brand relationships can thus also be categorized based on an individual-collective dimension (Bauer et al., 2023;McAlexander et al., 2002;Veloutsou, 2009).Brand communities can serve as an (online) platform to facilitate collective interactions and experiences (Brodie et al., 2013;Muniz and O'Guinn, 2001).Relationship quality is another widely used concept to investigate brand relationships, and there is no agreement on its dimensionality (Papista and Dimitriadis, 2019;Veloutsou, 2015).Quality has been conceptualized with elements as commitment, satisfaction, trust, interdependence, intimacy, self-connection, love, passion, partner quality and attachment (Adhikari and Panda, 2020;Fournier, 1998;Youn and Dodoo, 2021).All these elements are also presented by others as independent forms of brand relationships, which is another example of the inherently overlap and confusion described earlier.Functionality, another dimension, is based on the perceived utilitarian benefits by consumers (Fernandes and Moreira, 2019;Fetscherin and Heinrich, 2014).Finally, a dimension of valence (negative vs. positive) is used to order constructs (Fetscherin et al., 2019).Recently, more attention has been paid to negative forms of brand relationships, such as brand hate (Zarantonello et al., 2016;Zhang and Laroche, 2020).
Functional-based and emotional-based dimensions can be combined to place brand relationship concepts in four quadrants (Fetscherin and Heinrich, 2014).For example, satisfaction is an important aspect of high functional but low emotionally connected consumers (Papista and Dimitriadis, 2019).Brand love is an example of a high emotional, high functional relationship (Bairrada et al., 2018).The distinction between exchange relationships, in which benefits are provided to get something in return, versus communal relationships, characterized by prioritizing the relationship partner's needs, seems to align logically with this combination of dimensions (Aggarwal, 2004(Aggarwal, , 2009;;Herter et al., 2023).The combination of the strength and valence dimensions has also been used to classify brand relationship concepts: brand like, brand satisfaction (weak, positive); brand love, brand passion (strong, positive); brand dislike, brand avoidance (weak, negative); brand hate, brand divorce (strong, negative); and brand indifference (no strength) (Fetscherin et al., 2019;Fetscherin and Heinrich, 2014).
Although the above dimensional categorizations in principle allow for neutral or weak forms of brand relationships, most research has focused on intense, highly emotional relationships and there seems to be less interest in less intense constructs (Albert and Thomson, 2018).Many highly emotional concepts such as self-brand connections, brand attachment, brand passion, brand commitment, brand love and brand loyalty, appear in a recent meta-analysis of 392 publications (Fetscherin and Heinrich, 2015).Similarly, the five main types of brand relationships identified in a recent systematic literature all require a considerable degree of emotional bond: brand love, hate, communal relationships, friendships and addiction (Alvarez et al., 2023).Other recently researched examples of intense concepts are masstige brand relationships (Shin et al., 2022) and brand evangelism (Sashittal et al., 2023;Sharma et al., 2022).
Holistic approaches to address the complexity and multidimensionality in consumer brand relationships research did not solve the issue of idiosyncratic definitions of the concept.In this type of research one finds definitions such as, the brand person relationship is: a voluntary or imposed interdependence between a person and a brand characterized by a unique history of interactions and an anticipation of future occurrences, that is intended to facilitate socio-emotional or instrumental goals of the participants, and that involves some type of consolidating bond.(Fournier, 1994, p. 108).
Other definitions lack elements of functionality or goalorientedness, and focus, for example on the different roles consumers can have in their brand relationships, and define these as long-term commitments that people make to inanimate objects that they buy and use, as well as help make, sell and distribute (Kumar, 2007).Some definitions ignore the emotional bond and add the element of the consumer's subjective view by characterizing a consumer brand relationship as "repeated, interrelated, and nonaccidental exchanges or transactions between a consumer and a brand, wherein the brand's behavior is considered as a virtual or quasi-behavior in the subjective view of the consumer."(Fritz and Lorenz, 2010, p. 369).Another definition builds on the element of the consumer's subjective view by defining a brand relationship as the combination of a specific perception of the brand (brand image, brand personality) and a specific projection, or inference about the brand's attitude and/or brand experience (Blackston and Lebar, 2015).Apart from this specific aspect, this definition does not include any of the other previously mentioned aspects.
The wordings of these definitions confirm the disproportionate focus on intensive forms of brand relationships.Aspects of interdependence, an anticipation of future occurrences or longterm commitments lend themselves more to strong, intense brand relationships, than to neutral or weak brand relationships.At the same time, these definitions clearly focus on different, specific aspects, confirming the lack of consensus about what consumer brand relationships are.The fact that most research with a holistic orientation does not provide a clear definition of the concept is also indicative of this.
In past research, several issues stood in the way of answering to the question how consumers conceptualize their relationships with brands.A first issue concerns the brand aspect.Most articles do not explain what they consider a brand to be (see for exceptions, Dall'Olmo Riley and De Chernatony, 2000; Sweeney and Chew, 2002).In the initial phase of the current project, individual in-depth interviews with ten consumers were conducted, and these revealed that people find it hard to distinguish a brand from its product or service.By not being clear about what a brand is one obscures the question of whether certain findings actually apply to the relationship between a consumer and the brand.The current research avoids this type of confusion by including examples and an explanation of the brand concept in the survey instructions (see the online materials).
A second issue concerns the relational aspect.Most research uses the interpersonal relationship theories framework from social psychology (Fournier, 1994(Fournier, , 1998;;Miller et al., 2012).The applicability of these theories is questionable, as it is not clear to what extent consumers relate to brands in the same way as they relate to people (Alvarez and Fournier, 2016;Bengtsson, 2003;Swaminathan and Dommer, 2012).The current research overcomes this by investigating what criteria consumers themselves apply to define brand relationships.
Finally, some methodological issues complicated earlier research on how consumers conceptualize their relationships with brands.First, often respondents are only asked about specific, by the researcher selected, characteristics of their relationships with brands (Aggarwal, 2004;Miller et al., 2012;Sung and Campbell, 2009).This assumes that the provided characteristics are the relevant ones.Second, respondents are sometimes explicitly instructed to think about a brand as if it is a person (Fournier, 1994;Sung and Campbell, 2009), forcing a representation upon respondents that might not correspond to how they normally think about brands.Third, respondents are asked to select one particular brand they feel close to and then report about this brand only (Reimann et al., 2012), which may lead to them uniquely report on strong, positive brand relationships.
In summary, conceptual and methodological issues obscured the question how consumers naturally conceptualize their (potential) relationships with brands.The present research overcomes this by using a prototype analysis approach to empirically investigate what criteria consumers themselves use when thinking about relationships with a brand, without imposing any theoretical framework upon them, or constraining them in another way.This will result in a conceptualization of the most natural, typical interpretation of brand relationships by consumers.

A prototype analysis of brand relationships
A prototype analysis is a method borrowed from psychology (Cantor and Mischel, 1977;Mervis and Rosch, 1981;Rosch, 1973), that is used to examine how people view certain constructs/concepts.Prototype analyses have been applied successfully in marketing (Batra et al., 2012;Jones et al., 2009Jones et al., , 2018)).A prototype analysis is a bottom-up approach that identifies lay people's views of a concept by having them generate features of that concept and then testing the relatedness of concept and features empirically.It results in a list of central (most prototypical) and peripheral features that form the prototypical image of the concept.A prototype analysis is complementary to (the commonly used) top-down approaches that are based on a-priory theory or small sample interviews.Prototype analyses can examine and clarify the extent to which lay conceptualizations correspond to theory.Identifying the prototype of the holistic concept of consumer brand relationships will greatly contribute to the understanding of the concept.
Prototype analyses have been frequently applied to conceptualize and better understand emotion-related constructs such as gratitude (Lambert et al., 2009), nostalgia (Hepper et al., 2012) and hope (Luo et al., 2022).It has also been applied to motivational constructs such as greed (Seuntjens et al., 2015), and relational constructs such as love (Fitness and Fletcher, 1993), commitment (Fehr, 1988) and relationship quality (Hassebrauck, 1997).In marketing, prototype analyses have provided insight into commitment in service relationships (Jones et al., 2009) and into brand love (Batra et al., 2012).Both efforts are clearly related to the aim of the current research, establishing the prototype of consumer brand relationships, but they are focused on one dimension (commitment or love) of these relationships, rather than a holistic approach to the concept of brand relationships.The current research builds upon and extends this previous work, and will lead to a prototype that is representative of a larger diversity of relationships.
Below, a four-study prototype analysis of a consumer brand relationship is reported, closely following the procedure used in other prototype analyses (Hassebrauck, 1997;Hepper et al., 2012;Luo et al., 2022;Seuntjens et al., 2015).In the remainder, this procedure is referred to as the "standard procedure".Studies 1 and 2 serve to determine which features are prototypical for consumer brand relationships.Studies 3 and 4 further validate the classification of features by examining differences in automatic information-processing of central versus peripheral features.These are all the studies conducted in this line of research; all were preregistered and all measures, exclusions (if any), criteria for data inclusion/exclusion and determining sample size are reported.All data, code, materials, and preregistrations can be found at: https://researchbox.org/ 174.

Study 1 -generating features
The objective of Study 1 was to compile a list of empirically generated features of consumer brand relationships.Participants were asked to list as many features that they could think of and these were later coded to extract the most common ones.This study was preregistered (AsPredicted #25264).

Method
This research uses slightly larger sample sizes for all four studies than the largest from studies that used the standard procedure for a prototype analysis.Sample sizes of previous studies 1 varied from 111 to 232.Two participants were excluded: one waited 5 min and did not list any features; the other only listed specific brand names (rather than features of brand relationships).The remaining sample consisted of 258 USAbased participants, recruited via Academic Prolific (M age ¼ 36.33,SD ¼ 13.01, range ¼ 18-72 years; 41.2% female, 0.8% did not specify gender), who received $1.00 for participation.
The survey started with an explanation of the brand concept, including examples of specific brands and their products and services.To avoid the methodological issue of enforcing a (theoretical) framework on respondents this introduction to the study did not include any information on the concept of (brand) relationships.Participants read the following instruction (cf.Hassebrauck, 1997): Please write in the spaces below all features that in your view distinguish a relationship between a consumer and a brand.In the next 5 minutes, list as many features as you can think of.There are no right or wrong answers.
It was explicitly mentioned that the survey was not about (names of) specific brands or (features of) specific products/ services.

Results
In total, participants described 2,786 features of consumer brand relationships (M ¼ 10.80, SD ¼ 5.33).First, responses that contained multiple connected statements were separated into distinct exemplars.A distinct exemplar comprises one "unit of meaning" (Joffe and Yardley, 2004).For example, one participant wrote "shipping speed and pricing".This was divided into "shipping speed" and "shipping pricing".This procedure resulted in 2,800 distinct exemplars (M ¼ 10.85, SD ¼ 5.34).Subsequently, two independent coders (the first author and a trained research assistant) categorized these distinct exemplars into larger categories.Following Hepper et al. (2012), this was achieved by grouping identical exemplars (e.g.loyalty and loyalty), grouping semantically related exemplars (e.g.loyalty and loyal), grouping meaning-related exemplars (e.g.want and desire) into categories and, finally, grouping categories of common meaning (e.g.kind and friendly).The coders met to resolve discrepancies and developed a coding scheme with 253 categories.Words that appeared opposites (e.g."necessary" and "unnecessary") fell into the same category because they were about the same concept, only assessed differently in terms of its valence.
Next, two additional trained research assistants independently applied the coding scheme to all distinct exemplars, assigning each exemplar to only one code.Interrater reliability was good (k 12 ¼ 0.71), as were the interrater reliabilities with one of the coders who developed the coding scheme (first author), who also assigned the exemplars to the coding scheme as a third rater (k 31 ¼ 0.73 and k 32 ¼ 0.75) The raters then met to resolve all discrepancies by discussion, which led to a final categorization.The categories were reduced from 253 to 225, based on cases of co-occurrence and similarity (ks go up to k 12 ¼ 0.73, k 31 ¼ 0.74 and k 32 ¼ 0.76).Based on these interrater reliabilities the ratings by the third rater were used.Sometimes a participant listed exemplars that fell into the same category (e.g. a participant writing "habit" and "routine", which would both fall into the category "habit").Because there are good arguments to include or exclude these duplicate exemplars both options were analyzed (Gregg et al., 2008).When excluding duplicates, participants described 2,533 exemplars of consumer brand relationships (M ¼ 9.82, SD ¼ 4.59).As in the standard procedure (Fehr, 1988;Hassebrauck, 1997), infrequent categories were discarded to reduce the original number of categories to a workable number.As discarding categories with an exemplar frequency (including duplicates) of eight or less would result in 98 categories (below the target of 100), this was used as a cutoff criterion.Table 1 presents these feature categories with their exemplar frequencies and representative exemplars.
Participants generated many features of consumer brand relationships.First, the valence of features (positive, neutral or negative) was analyzed.The ratio of negative versus positive exemplars per feature shows that consumers regard brand relationships as a positive phenomenon.The vast majority of features entirely consists of positive exemplars.Only two features, exploitation and manipulation, consist entirely of negative exemplars.Two other features, authenticity and balance, have a (roughly) equal amount of positive and negative exemplars and are considered neutral.Second, the attitudinal dimensions of features (affective, cognitive or behavioral) were analyzed.The results of this analysis suggest that consumers think of brand relationships more in terms of thoughts and feelings, than behavioral aspects.The data did not allow for an analysis of the object of features (brand, product/service or relationship), although this was preregistered.

Discussion
Study 1 was the first step to determine which features are prototypical for consumer brand relationships.Participants listed as many features of these relationships as they could think of, generating a large range of features that in their view distinguish a brand relationship.The goal of the next study was to determine the centrality of the 98 features generated in study 1.Study 2 was preregistered (AsPredicted #34734).

Method
In total, 224 USA-based participants were recruited via Academic Prolific and received $1.30 for their participation.Two participants that did not seem to have participated "Please rate each feature based on the degree to which you believe it is a good indicator or not a good indicator of consumer-brand relationships."(1 ¼ not a good indicator, 7 ¼ a good indicator).Each feature was displayed with some exemplars to give participants a better understanding of that feature.For example, the feature reliability was displayed together with the exemplars reliable, confidence, delivers on promises.Participants saw the features in an individual random order.

Results
The results are in Table 1.Following the standard procedure of prototype analyses, the reliability of these means, an intraclass correlation (ICC) was computed (equivalent to the average of all possible split-half correlations of the 222 subjects with regard to the 98 features).To do so, the dataset was transposed and treated the 98 features as cases and the 222 subjects as items.ICC estimates and their 95% confident intervals were calculated based on a mean-rating (k ¼ 222), consistency, twoway random-effects model and indicated excellent reliability (Koo and Li, 2016) for participants' responses (ICC ¼ 0.98, p < 0.001, CI[0.98, 0.99]).Overall, the centrality ratings were moderately, positively correlated with the frequencies of features of study 1, r ¼ 0.43 including duplicates and r ¼ 0.44 excluding duplicates (ps < 0.001), rank-order correlation ¼ 0.47 including duplicates and rank-order correlation ¼ 0.51 excluding duplicates (ps < 0.001).These results were consistent with previous prototype analyses.
Due to the large number of features (98) obtained, the 30 highest rated features were labeled as central and the 68 lowest rated features as peripheral (cf.Seuntjens et al., 2015).

Discussion
Inspection of the centrality ratings in Table 1, led to the identification of four central clusters: quality, bond, value, and joy.The quality cluster and the bond cluster are very prominent and together include all ten most central (typical) features.Table 2 shows an overview of these clusters, including all their features, within the prototype structure of consumer brand relationships.
The quality cluster includes features such as quality, customer service and satisfaction.For consumers quality plays a fundamental role in a brand relationship.Features as expectations, experience and satisfaction indicate that quality is important both before, during and after interactions with the brand.The features product and service suggest that, besides the quality of interactions with the brand, consumers also find the quality of the brand's core deliverable an important aspect for their potential relationship with the brand.The bond cluster includes features like reliability, trust, loyalty and consistency.This cluster indicates that consumers consider a reliable bond as a central aspect of brand relationships.The value cluster includes features such as value, useful(ness) and effectiveness.This suggests that consumers think of brand relationships in terms of value, something that is or should be useful.The joy cluster contains features such as joy and liking, which indicates that consumers think of consumer brand relationships as a concept that can not only make their lives better or easier, but also more enjoyable.All clusters include several peripheral features, in addition to the central features.
Also, three clusters of peripheral features were identified: A helpfulness cluster (including the central feature helpfulness and peripheral features such as support, care and understanding) indicating that consumers expect the brand to help them when needed and vice versa (see e.g. the features willing(ness) to spend more and ambassadorship).Next, a (doing) business cluster, including features like cost, economics and profitability.And finally a status cluster, including features such as popularity, style and status.

Study 3 -recall and recognition test
The purpose of Study 3 was to examine whether central features are recalled and recognized better than peripheral features (cf.Cantor andMischel, 1977, 1979).Study 3 thereby aims to further validate the classification of central and peripheral features.This study was preregistered (AsPredicted #38334).

Method
In total, 145 USA-based participants were recruited via Academic Prolific (M age ¼ 37.30, SD ¼ 13.02, range ¼ 18-71 years; 49.0% female, 1.4% did not specify gender) and received $1.25 for their participation.All 30 central features and 30 peripheral features randomly selected of the 68 from Study 2 were used.These 60 features were divided into three sets of 10 randomly selected central and 10 randomly selected peripheral features.There were no significant differences in the centrality rating of central features The survey explained the brand concept, and participants were randomly assigned to one of three sets.Following the standard procedure, features were presented in random order and presented in a sentence containing the word consumer-brand relationship to activate the concept, for example, "A consumerbrand relationship is about loyalty".Next, participants completed a 4-minute distractor task.Subsequently, participants were given 3 minutes to recall as many features of consumerbrand relationships they had seen before as possible (free recall).Next, participants were presented with a list of all 60 features (including the 40 features from the other 2 nonpresented sets) in a random order, and were asked to indicate if it had previously been presented or not.There was no time limit set for this recognition task.

Results
The number of correctly recalled central and peripheral features were compared for the three sets combined.The same comparisons were made for false recall, correct recognition and false recognition (i.e. if a respondent recognized a feature as having been presented that in fact had not been presented in the previous part of the survey).At the free recall task, participants sometimes wrote down the same feature twice; in those cases, the feature was only counted once.If participants recalled a word that did not belong to any of the three sets, it was considered neither a correct nor a false recall.
Because the data were not normally distributed, Wilcoxon signed-rank tests were used.The results in Table 3, indicated no significant difference between the number of correctly recalled central and peripheral features; significantly more falsely recalled central than peripheral features; significantly more correctly recognized central than peripheral features; and significantly more falsely recognized central than peripheral features (t-tests showed similar results).The data confirm expectations for the last three comparisons in Table 3, but not for the first, that respondents would correctly recall more central than peripheral features.There is no clear explanation for why this comparison is not significant, but it should be noted that this has been the case before in similar studies (Fehr, 1988;Hassebrauck, 1997).

Discussion
Study 3 found further support for the prototype structure of consumer brand relationships by examining the influence of feature centrality on the recall and recognition of features.The data were mostly in line with the expectations.Central features were better recognized than peripheral features, but not better recalled.The false recall and the false recognition data further support the classification of central and peripheral features.As predicted the activation of the prototype caused central features of the prototype to be more easily accessible in memory which led people to think they saw these features, even if they did not see these features before.Please note that an alternative explanation could be that the central features might in general be more easily accessible in people's memory than the peripheral features.An additional study on the ecological validity of the prototype that examines this alternative explanation can be found in the online materials.

Study 4 -classification and verification speed
Study 4 was needed to further validate the classification of central and peripheral features by examining the influence of centrality on classification and verification speed for the features.Previous research found that when a prototype is activated, people are quicker and better in correctly evaluating the features as central and peripheral (Fehr et al., 1982;Fehr and Russell, 1984).This study was preregistered (AsPredicted # 40512).All 30 central and 30 randomly selected peripheral features from Study 2 were selected.Additionally, a set of 60 neutral words was constructed from a list of the 5,000 most frequently used English words (Wordfrequency, 2023) and a Word Finder and Word Search website (Yougowords, 2023).The neutral words were not related to consumer-brand relationships: none of the neutral words used appeared as an exemplar in the dataset of study 1 (see the neutral words in the research box).There were no significant differences in the average number of characters of central features (M ¼ 9.37, SD ¼ 3.58), peripheral features (M ¼ 8.53, SD ¼ 2.61) and neutral words (M ¼ 8.17, SD ¼ 3.20), F(2, 117) ¼ 1.44, p ¼ 0.241, which is desirable because word length could affect the verification speed.
The survey explained the brand concept, and participants received an example and 10 practice trials.Next, the total of 120 features and neutral words were presented one by one, in an individual random order on the middle of the screen.
Participants were asked to decide as quickly as possible for each of the characteristics whether it is a characteristic of a consumer-brand relationship.For each trial the response (yes or no) and reaction time (in ms) were recorded.

Classification
The percentages of central features, peripheral features and neutral words that were judged to be a feature of consumer-brand relationships were compared (see Table 4).Because of the skewedness in the percentages, nonparametric tests were used.Central features were more often than peripheral features classified as a characteristic of consumer brand relationships, Z(195) ¼ À11.57, p < 0.001, and peripheral features were more often classified as a characteristic of consumer brand relationships than the neutral words, Z(195) ¼ À12.11, p < 0.001 (Wilcoxon's signed-rank tests; paired t-tests showed similar results).

Verification speed
Next, the average speed at which participants made their classification judgements was analyzed for "yes" responses (see Table 4).Following recommendations (Greenwald et al., 2003), extremely slow (>3,000 ms) and extremely fast (<300 ms) latencies were recoded to respectively 3,000 and 300 ms, and a logarithmic transformation was conducted (Hepper et al., 2012;Seuntjens et al., 2015).A repeated measures ANOVA with a Greenhouse-Geisser correction revealed that verification speed differed significantly between centrality levels, F(1.256, 107.991) ¼ 67.88, p < 0.001, h 2 p ¼ 0.441.Post hoc tests using Bonferroni correction revealed that participants classified central features faster than peripheral features.Participants classified peripheral features faster than control features (t-tests showed similar results).Analysis of untransformed verification speed yielded identical results.

Discussion
Study 4 found, as predicted, that participants classified central features more often and quicker as related to consumer brand relationships than peripheral features.These results support the classification of central and peripheral features in the prototype structure.In addition, participants classified peripheral features more often and quicker than neutral words, indicating that peripheral features are indeed part of the prototype, but less prototypical than central features.

General discussion
A four-study prototype analysis investigated how consumers perceive and define their relationships with brands, following a bottom-up approach, without enforcing any theoretical framework upon them.Study 1 resulted in 98 features that typify consumer brand relationships.Study 2 analyzed the centrality of each of these features, resulting in four clusters of central features: a quality, bond, value and joy cluster.Studies 3 and 4 further validated this classification.The outcome is a conceptualization and corresponding definition of consumer brand relationships in consumers' terms, which is confronted with the existing literature.Next, the implications and contributions of the findings of this research are discussed.Finally, the limitations and directions for future research are reported.

The consumer brand relationships prototype
Based on the four clusters of central features identified in this research, a consumer brand relationship is defined as: "A bond that a consumer experiences with a brand, based on positive experiences and satisfaction with the brand's quality.It consists Differences between the mean scores of central features, peripheral features and control words are all statistically significant for both the percentages of features categorized as consumer brand relationship, the response speed (in ms) and the response speed after logarithmic transformation (log10) Source: Created by authors of aspects such as reliability, trust, loyalty and consistency and makes the consumer's life better, easier or more enjoyable".Please keep in mind that the analysis provides insights into the features of the prototype, but not necessarily into the temporal or causal order of the features.

Discussion and theoretical contributions
What does this bottom-up prototype analysis have to offer?Not only does the resulting conceptualization provide a complete and balanced view of the common consumer interpretation of the construct.It also provides guidance in determining what the most relevant forms of brand relationships are.In this way, the insights from this study can help researchers to assess the validity of existing conceptualizations and make (or modify) choices for their research strategy.
The results clearly show that the way consumers conceptualize their brand relationships differs from the way it has been done in previous research.The main difference is that consumers experience brand relationships as more functional and less emotionally intense than one would expect on the basis of the current academic literature.A comparison with functionally and emotionally based dimensions, shows that a typical brand relationship is at the "high" end of the functional dimension (Fernandes and Moreira, 2019;Fetscherin and Heinrich, 2014).The importance of the quality and value features reveals that consumers view brand relationships primarily as instrumental.A consumer brand relationship can (or should) make people's lives better or easier.With regard to the emotional bond, the prototype shows that this is an important condition for consumers to experience a brand relationship, but this bond does not need to be very intense.Results show that, according to consumers themselves, the typical brand relationship is not a love affair, but a relationship of trust.This is at odds with the aforementioned focus on intense, highly emotional brand relationships (Albert and Thomson, 2018;Alvarez et al., 2023;Fetscherin and Heinrich, 2015).Existing concepts such as brand trust and brand loyalty, correspond to the most common type of emotional bond but do not include an aspect of functionality and joy (Safeer and Liu, 2023).As a result, these concepts do not cover the full scope of the holistic concept of brand relationship as typified by consumers.
In addition, the prototype analysis also shows that consumers generally view their brand relationships as something positive rather than negative.The core feature "joy" indicates that consumers think of brand relationships as a concept that makes their lives more enjoyable.The dominance of positive features does not exclude negative forms of brand relationships, but indicates that these are the exception, not the rule.The recent increased focus on negative brand relationships such as brand hate (Zarantonello et al., 2016;Zhang and Laroche, 2020) does not seem to be a proper reflection of consumers' everyday experience of brand relationships.
The current results further suggest that with respect to engagement aspects (Hollebeek et al., 2014;Vivek et al., 2012), consumers think of brand relationships more in terms of thoughts and feelings, than behavioral aspects.Finally, the prototype analysis reveals that consumers typically experience a brand relationship on an individual level.Although interactions may take place at the collective, community level and brand experiences may be created and shared collectively in brand communities (Brodie et al., 2013;Muniz and O'Guinn, 2001), the experience of the relationship with the brand seems to be primarily a personal one.
What is more, consumers include in their construal of brand relationship features that theoretically could be seen antecedents (the central feature quality) or consequences (the central feature value) (Batra et al., 2012;Prahalad and Ramaswamy, 2004).According to consumers, these aspects are inherent to the concept of brand relationship itself and thus should be included in brand relationship research.
Do consumers relate to brands in the same way as they relate to people?Results of this study show that the answer is twofold.On the one hand, the same principles quality, bond, and functionality/value, are described in both contexts (Fournier, 1994(Fournier, , 1998;;Miller et al., 2012).On the other hand, brand relationships are a specific type of relationships, with different interpretations of relational principles as compared to (most) interpersonal relationships.For example, in the context of consumer brand relationships the bond is typically based on self-interest or, put differently, consumers typically view brand relationships more as an exchange relationship than a communal one (Aggarwal, 2004(Aggarwal, , 2009)).Whereas people define interpersonal relationship quality more in terms of communal sharing (Hassebrauck, 1997).

Managerial contributions
For brands, this research offers insights that help recognize and create associations consumers have with the most common form of brand relationships.From any gap between what lay people typically understand by the term brand relationship and how a brand applies the concept within current marketing strategies, the insights from the prototype analysis can help a brand make the right adjustments.In this way, brands can improve their strategies for establishing and maintaining relationships with consumers.For example, many brands seem to strive for a love or attachment relationship and therefore base their marketing strategy on these types of concepts.The current results offer brands the insight that these concepts do not reflect the most typical form of brand relationships for consumers.Perhaps love and attachment forms of brand relationships are not feasible for most brands.People may have limited relational capacity and therefor simply do not have the time, interest or emotional energy to form strong relationships with a wide variety of brands (Alvarez and Fournier, 2016;Dowling, 2002).
It is important to note that the prototype represents the most typical form of brand relationships.This interpretation can therefore provide a useful starting point for brands.Both for functional and emotional brands, both for product and services brands, it is good to be aware that, for example, quality is an important aspect of a brand relationship for consumers.Be it the quality of the functionality of the product or the quality of the (hedonic) experience of the service.The same applies with respect to the other central features trust, value and joy.The research makes clear that establishing and maintaining relationships with consumers are not only responsibilities for the marketing, branding and customer service departments.By building products and/or providing services in a reliable, quality manner, the product and/or service department also contribute to the relationships with consumers (by filling in the central features of quality and trust).It is then up to the marketing and customer service departments to communicate these aspects to consumers in the right way.Or by letting consumers themselves communicate the quality and reliability of the products and service through brand channels, or stimulate conversations about them in brand communities (Mardumyan and Siret, 2023;Penttinen, 2023).The central features value and joy underline the importance here of being able to formulate a good value proposition and having the right tone of voice towards consumers.The target group must recognize the value of the brand (relationship) in an enjoyable way.By mutually starting from the consumers' conceptualization of a typical brand relationship, this study can ultimately increase congruence between science and practice.
Limitations and directions for future research Three avenues for further research may overcome limitations of the current analysis.First, consumers' perceptions on brand relationship concepts may vary across different types or categories of brands (Fetscherin et al., 2014;Khamitov et al., 2019a) and within consumer typologies (Lin, 2010;Monga, 2002).This suggests that the consumer brand relationship prototype could have a different structure for relationships with different types or categories of brands and for different typologies of consumers.Analyzing whether features are more typical for relationships with utilitarian brands rather than hedonic brands, or for brands building products rather than brands providing services, may increase understanding of consumer brand relationships and potentially allow brands to focus on specific relationship features based on their offering.In a similar vein, it can be interesting to investigate whether features are more typical for relationships on an individual level than on a collective level, or for relationships in a traditional or virtual brand community.Or, to test the extent to which the consumer brand relationship prototype is related to the concept of brand co-creation.Future research could also advance insights as to whether the prototype structure is valid across different categories of gender, age, personal traits, etc.
Second, this research recruited only USA-based participants.Culture and cultural background may affect the way people perceive a consumer brand relationship (Khamitov et al., 2019b).To further extend its applicability the prototype structure should be validated in samples of different cultures.The suggestions for future research described above could be conducted using methods other than prototype analysis, including qualitative approaches.
A third and final avenue for future research concerns the development of an instrument (questionnaire) to measure the extent to which consumers consider their associations with a specific brand to be a relationship.In this research the most relevant (prototypical) features of a brand relationship were identified.Future research could analyze whether the presence of these featuresas perceived by a consumercan serve as a valid indicator of the presence of a brand relationship from the consumer's perspective.If so, consumers who were previously not identified as having a relationship with the brandfor example because they do not meet the "high standard" of an intense emotional bondbut who in reality do experience a brand relationship, could then be identified as such.In other words, the risk of missing out on consumer brand relationships could be reduced.Conversely, consumers who are now labeled as a relationship in the CRM systemfor example based on behavioral indicatorsbut who in reality do not experience a brand relationship, could be stripped of this label.In this way, the measurement instrument could potentially help organizations to communicate the proper message to the proper target groups for initiating or maintaining a relationship between their brand and consumers.

Table 1
Features of consumer brand relationships prototype, sample exemplars, frequencies (Study 1) and centrality ratings (Study 2) Created by authors seriously were excluded: one indicated "not remember" on the attention check question "Which concept were you asked to rate features for?" and another responded uniformly to all 98 features.There were 222 respondents (M age ¼ 36.95,SD¼13.37, range ¼ 18-74 years; 55.9% female, 1.4% did not specify gender).The survey again first explained the brand concept.Next, participants rated each of the 98 features, following this instruction (cf.Hassebrauck, 1997, p. 169): used a scale from 1 (not a good indicator of consumer brand relationship) to 7 (a good indicator of consumer brand relationships).The 30 highest rated features are considered central to consumer brand relationships and the 68 lowest rated features as peripheral to consumer brand relationships Source:

Table 3
Recalled and recognized features in Study 3 (N ¼ 145) Notes: Participants were shown 10 central and 10 peripheral features.There were 20 nonpresented central features and 20 nonpresented peripheral features Source: Created by authors

Table 2
Clusters and features of the consumer brand relationships prototype Cluster Features Method A sample of 200 USA-based participants was recruited via Academic Prolific and received $1.40 for their participation.The preregistered exclusion criteria did not result in exclusions, but four respondents for whom time registration malfunctioned were excluded.The final sample has 196 respondents (M age ¼ 37.50, SD ¼ 12.82, range ¼ 19-73 years; 57.7% female, 0.5% did not specify gender).

Table 4
Percentages and speed in classification of central and peripheral features of consumer brand relationships in Study 4 (N ¼ 196) Ms and SDs of the response speed (in ms) and the response speed after logarithmic transformation (log10) are based only on the yes responses.