Advances in Consumer Research Volume 26, 1999 Pages 352-358
RELATING CONSUMER-BASED SOURCES OF BRAND EQUITY TO MARKET OUTCOMES
Steven N. Silverman, Washington State University
David E. Sprott, Washington State University
Vincent J. Pascal, Washington State University
[The authors thank the editors and three anaonymous reviewers for their thoughtful comments.]
Among scholars and practitioners, there is some agreement on the concept of brand equity. Yet measuring this construct is less clear. This project considers how various measures of brand value relate to one another. Study 1 compares market-based outcomes (i.e., Financial World ratings, and annual sales) with consumer-based sources of brand value (i.e., familiarity, usage, and evaluation). Study 2 uses affinity analysis to examine how consumers brand associations differentiate market leaders from followers. Study 1 found a weak relationship between measures, whereas the rich results of Study 2 are encouraging for both scholars and managers.
Researchers have studied and measured "brand equity" from both the organizational (e.g., Simon and Sullivan 1993) and customer (e.g., Keller 1993; Krishnan 1996) perspectives. Recently, Keller (1998) used a customer-based definition of brand eqity to distinguish sources of brand equity held in consumer knowledge structures (i.e., customer-based brand equity), from marketplace outcomes of brand equity (i.e., market-based brand equity). [There are many "sources" of brand equity (including consumer cognitions, management creativity, corporate structure, etc.) that may ultimately impact market-based outcomes. The present paper focuses on consumer-based sources of equity--a view modeled after Keller (1993).] The relationship, however, between customer-based and market-based measures of brand equity remains unclear (although Krishnan  has studied the question).
This paper presents two empirical studies that explore this issue. Study 1 compares consumer-based measures of brand awarenessCincluding familiarity, favorability, and direct experienceCto market outcomes of brand value (annual sales, and the Financial Worlds measure of brand value). Study 2 then introduces a new method called affinity analysis to examine how brand imageCstrength, uniqueness, and favorability of brand associations (Keller 1993)Cmay differentiate between two market-leading brands.
MEASURING BRAND EQUITY
Brand equity has emerged as an important issue in marketing research (e.g., Aaker 1991; Keller 1993; Keller 1998). Much of the extant work has been aimed at defining and measuring brand equity. As Keller (1998) notes, most discussions of brand equity consider it to be "the marketing effects uniquely attributable to the brand" (p. 42). His own definition captures this distinction: "customer-based brand equity is defined as the differential effect of brand knowledge on consumer response to the marketing of the brand," (Keller 1993, p.8).
While Kellers (1993) view recognizes the marketplace impact of brand equity, he clearly places the locus of brand equity within the consumer. Alternatively, others have focused on the marketplace valuation of a brand (e.g., comparative approaches, evaluation approaches, holistic methods [see Keller 1998]). This situation has led to the development of numerous measures of brand equity (e.g., Kamakura and Russell 1993; Simon and Sullivan 1993). Some researchers suggest that brand equity measures should rely on market-based, objective, measures because consumer attitude and preference measures are inherently subjective (Simon and Sullivan 1993). Others contend that for a brand to have value it must be valued by consumers, hence their views must be included (Keller 1993). We consider both of these approaches, and explore how they are related to one another.
Consumer Sources of Brand Equity
Conceptualizations of brand equity based on consumer sources typically fall into two categories: those involving consumer behaviors (e.g., Kamakura and Russells  analyses using scanner data) and those involving consumer cognition (e.g., Keller 1993). Our focus is on the latter. This view suggests that brand equity arises from the strength and favorability of the two components of consumer-based brand knowledge structures: brand awareness and brand image. Brand awareness relates to the strength of a brand in memory, and the likelihood and ease with which the brand will be recognized or recalled under various conditions. Brand image is defined as "perceptions about the brand as reflected by the brand associations held in consumer memory" Keller (1993 pg. 3). Following an associative network model of memory, brand associations are the myriad nodes that are linked to the brand in memory (i.e., product related attributes, price, user and usage imagery, and so on).
Market Outcomes of Brand Equity
Most market-based measures of brand equity are related to some aspect of market performance. For example, valuation approaches measure brand equity from brand-related financial data (e.g., sales, assets), or changes in stock price (Simon and Sullivan 1993). Another financial method (used in the present research) was developed by Interbrand, a marketing consultng organization, and adapted by Financial World. These similar valuations first determine the earnings for a brand, and then adjust that figure with a brand-strength multiplier consisting of seven brand-related factors: leadership, stability, market environment, internationality, trend, communications support, and legal protection (Ourusoff 1994).
Relating Consumer Sources and Market Outcomes of Brand Equity
Using Kellers (1993) model, consumer-based sources of brand equity should relate to market-based outcomes. As Keller (1993) notes: "high levels of brand awareness and positive brand image should increase the probability of brand choice, as well as produce greater consumer (and retailer) loyalty and decrease vulnerability to marketing actions," (p. 8). In other words, if consumer perceptions of brands are reflected by purchase decisions, then the measures of those perceptions should also correlate with market-based outcomes.
STUDY 1: RELATING BRAND AWARENESS TO MARKET OUTCOMES
The objective of Study 1 was to learn how brand awarenessCassessed by familiarity, usage and favorability (Keller 1993)Cis related to market-based outcomes of brand valueCmeasured by annual sales and Financial World brand ratings (Ourusoff 1994). (The Financial World and Interbrand valuations are quite similar to one another. For a detailed discussion see Keller 1998.)
Study 1 Method
Sample and Procedure. The brands (n=196) used in the study were among those included in Financial World (Ourusoff 1994) and spanned 19 product categories. Brands were randomly assigned to fourteen groups of fourteen brands each. Respondents (361 undergraduate business students) were randomly assigned one of the fourteen brand groups to evaluate. Brand names (four per page) were presented in a larger font with bold type. After the data were collected, the number of brands was reduced to only those targeted toward the population studied. Ultimately, 92 (85) brands were used for the 1993 (1995) analyses presented here. [The final set of 92 brands included frequently purchased non-durable goods of which the target market included Study 1's student subjects. This adjustment was made based on a suggestion by reviewers. Brands were included in the final set on the basis of their availability in stores directly on campus. Results changed only modestly and primarily reflected lower variance in familiarity with brands, as would be expected.]
Consumer-Based Measures. Brand familiarity (i.e., recognition) was measured by asking, "Have you ever heard of this brand name?" Brand usage was determined by asking, "Have you ever used this brand before?" Respondents answered YES, NO, or NOT SURE to each question. Two measures were then calculated for the analysis: (1) "familiarity" is the percent of the sample indicating that they had heard of the brand, and (2) "usage" is the percent of the sample that reported hearing of the brand and had used it.
Brand favorability was assessed by a three-item scale (Cox and Cox 1988). Respondents were asked, "How would you evaluate this brand regarding the following adjective pairs?" Responses were recorded on three, nine-point scales (anchored by "Bad" and "Good"; "Dislike Very Much" and "Like Very Much"; "Unpleasant" and "Pleasant"). The three-item scale was averaged to form a "favorability" measure for subjects reporting they had heard of the brand; this value was calculated for each brand (alphas ranged from 0.65 to 0.99 for individual brands; average alpha=0.93).
Market-Based Measures. The two market-based measures (for 1993 and 1995) were sales and brand valuation figures as presented by Financial World (Badenhausen 1996; Ourusoff 1994). Financial Worlds brand value formula (Keller 1998), similar to the Interbrand formula, is based on profits related to the brand adjusted for the brands strength (defined by Interbrands seven-factors). Our use of Financial Worlds brand valuation is based on the availability of thedata and the overall acceptance and use of this measure (cf. Keller 1998). In order to account for differences in brand values and sales across product categories, the data were standardized within product categories.
Study 1 Analyses and Results
Pearson correlation coefficients were calculated between all pairs of 1993 and 1995 market-based (i.e., brand value ratings and sales) and consumer-based (i.e., evaluation, familiarity, and usage) measures of brand equity. The results of the correlational analysis indicate that the focal measures of brand value are weakly correlated with one another (see Table).
Study 1 Discussion
The findings of Study 1 suggest that favorability and usage are the consumer measures most closely associated to market-outcomes of brand equity. In both cases, small, positive, and significant relationships are found. Because the market-based measures are #holistic (cf. Keller 1998), the fact that any significant correlation is found is encouraging and suggests the need for a deeper understanding of the relationships between consumer sources and market-based outcomes of brand equity.
On the face of it, it is somewhat surprising that familiarity, based on recognition, was not more highly correlated with marketplace success, especially when considering the extent to which this measure is recommended for assessing brand value (cf. Keller 1998). However, further evaluation shows that on average, 95.3 % of subjects had heard of each brand in the study. Hence, regardless of a brands sales or value it is known to the sample. Here, familiarity offers little insight. Moreover, measuring brand familiarity is not recommended when subjects are familiar with most brands in a category. In markets where this is common (e.g., airlines, gasoline, automobiles) familiarity may not be useful in relating consumer knowledge to market value.
Finally, one of the more interesting findings is the comparable performance of brand value and sales. Based on our study, it appears that Financial World measures offer minimal advantages over raw sales figures. This result suggests sales may, at some level, be a useful way of capturing brand value.
STUDY 2: RELATING BRAND IMAGE TO MARKET OUTCOMES
In Study 2 we examine the relationship between category position (a market-based distinction on whether the brand is first or second in a particular category) and brand image (the other component in Kellers  brand knowledge model, assessed by the favorability, uniqueness, and strength of brand associations).
Research and marketplace examples demonstrate the importance of being a market leader. Coca-Cola, Nike, and Tylenol, first in their product category, hold a unique position in the marketplace, and presumably in the minds of consumers. Many benefits, including long-term financial strength (Collins and Porras 1994), accrue to brand leaders. Yet as Collins and Porras (1994) point out, secondary brands are not failures by almost any measure. Nevertheless, they are often unable to overtake the primary brands that they follow. Studying brand image among category leaders may help explain why they are so formidable. For present purposes, market position (either first or second in a category) can be viewed as a dichotomous, market-based measure of brand equity. When viewed from this perspective, assuming a correspondence between this market-based measure and consumer knowledge structures, we would expect a high correspondence between consumers brand associations and market leadership.
Study 2 Method
The sample (n=119) included the same undergraduate business students from Study 1; however, Study 2 data were collected two months later. The focal brands for Study 2 included the first and second brands for soft drinks (i.e., Coca-Cola and Pepsi) and over-the-counter medications (i.e., Tylenol and Advil) as identified by Financial World rankings (Ourusoff 1994, Badenhausen 1996).
Brand associations were obtained via a thought listing task asking participants to write down "all of the thoughts that come to mind when you think of the following brand." Subjects then rated the valence of each thought on a 7-point scale (-3 to 3).
RESULTS OF STUDY 1 -- PEARSON CORRELATION COEFFICIENTS AMONG MEASURES OF BRAND VALUE
Study 2 Analysis
Each brand was considered by 30 subjects (except one for which there were 29 subjects). A range of 1 to 19 associations were provided by each subject (median=5). A mean of 143 brand associations were generated for each brand. These associations were then analyzed by several coders using affinity analysis, a process which generates an "affinity diagram" (see Figures 1 and 2). [It is important to note that these diagrams are neither mental maps (Zaltman, LeMasters, and Heffring 1982) nor node-arc associative representations. They are cluster representations based upon jointly shared perceptions of affinity among the items presented. Their visual organization is otherwise arbitrary.] The affinity methodCoriginally developed to study facts, opinions, or ideas of a complex nature using verbal data (Futami 1986, Silverman and Silverman 1994)Callows one to form groupings (i.e., base clusters) of verbal data based on the mutual affinity among the items. Affinity analysis is conceptually akin to a statistically-based cluster analysis or factor analysis, and is functionally similar to card-sorting techniques. Unlike card-sorting, however, the diagram is created using groups of people (external coders or subjects). Because the result must be acceptable to all who are coding the data, this method can generate additional insight that does not emerge from an individual level analysis.
More specifically, the affinity diagram process proceeded as follows. First, ideas associated with a given brand were transferred to Post-it7 notes and placed on the wall of the research lab. Next, the Post-it7 notes were arranged by coding teams ranging from three to six members. The groups worked in silence (a rule of this method) to create base clusters of participant responses based on the affinity of the ideas on the Post-it7 notes (represented by unshaded boxes in Figures 1 and 2). Following silent organization of ideas, coding disagreements were discussed and resolved. Finally, similar base clusters were grouped together to form mid-level clusters (represented by shaded boxes in Figures 1 and 2), to which descriptive terms were then assigned. Although each brand in this study was analyzed separately by the researchers and several assistants, similar mid-level clusters emerged for individual brands. Thus, similar labels were then applied to mid-level clusters (where appropriate) to aid in cross-brand comparisons. Each brands affinity diagram took approximately one hour to create (three to six human hours). Each brand affinity diagram was later reviewed for completeness, clarity, and consistency by at least one additional coder.
In our adaptation of the affinity diagram process, we have added a feature to allow for the assessment of brand favorability, strength, and uniqueness (Keller 1998). Using the brand associations and favorability ratings, we calculated two measures for each of the base clusters (unshaded boxes). First, we calculated the mean valence for all comments in a cluster. This calculation is designed to capture the relative "favorability" associated with the cluster. Next, we counted the number of responses in a given base cluster. We interpret this value as an indication of the "strength" with which the associations in the cluster are present across the sample. A larger number of associations in a given clusterCand the underlying cognitive structure it presumably reflectsCsuggests greater strength. These values representing favorability and strength are reported in the unshaded boxes in Figures 1 and 2 (strength figures are in parentheses). Where meaningful, the strength vlues in these unshaded boxes been have been aggregated and labeled (e.g., "ad themes" in the Coca-Cola and Pepsi diagrams).
Next, more aggregate indices of brand strength and favorability were calculated. Specifically, consistent with basic multiattribute attitude models, consumer beliefs and evaluations of those beliefs were used to calculate a strength x favorability product for each of the base clusters (i.e., the unshaded boxes). These values were then summed to form "mid-level cluster values" (reported in the arrows emanating from the center of each diagram). These mid-level cluster values were then aggregated to create an "overall cluster value" (reported in the center of each diagram). Using this approach, strong associations are represented by larger numbers, and positive or negative values indicate associated valence.
Finally, we turn to see how brand uniqueness is reflected in the affinity diagrams. Because base clusters emerge from the data, differences between brands can emerge in two ways. First, certain base clusters and mid-level clusters may appear within one brand and not another, and thus identify an element of uniqueness (which can be further interpreted using the cluster values to assess strength and favorability). Second, when similar base clusters or mid-level clusters emerge for different brands one could compare the associated values to assess which brand has a relative, possibly unique, strength.
Study 2 Results
Brand affinity diagrams were used to generate the assessments of brand image strength and uniqueness of association within each brand. Each brand was analyzed and interpreted by several researchers. Due to space limitations, we will present "highlights" of the analyses. The results considered here are based on Figures 1 and 2.
Soft Drinks. Coca-Cola and Pepsi clearly have different brand market values for 1993 as reported by Financial WorldC$35.95 billion and $4.94 billion, respectively. Overall cluster values for Coca-Cola (199.50) and Pepsi(83.52) are consistent with the overall market performance of those brands, and are significantly different from one another, t=4.60, p<.01. Affinity analysis also shows that the sheer number of associations is not very different between the two brands, suggesting similar strength. Yet, the additional specificity provided by the affinity analysis offers insight not available from a global favorability measure. Specifically, additional analyses taken from Study 1 show that global brand favorability measures are not different between Coke (x=7.63, n=25) and Pepsi(x=7.27 n=26), t=.60, p>.55.
A comparison for mid-level clusters is instructive and shows that the strength x favorability calculation gives Coca-Cola dominance on every dimension except for that labeled "competitors." A closer examination of the "competitors" mid-level cluster finds that the Pepsi associations are actually positive statements about Coke. Hence, this dimension also reflects a relative strength and preference for Coke.
Uniqueness in each brand can be assessed using the two perspectives described earlier. First we look to identify associations that are held in one brand and not another. Starting with the mid-level clusters, we find that in this pair of brands, there are almost no differences. Only a "miscellaneous" category in the Pepsi affinity diagram distinguishes the two. In this sense there are no obvious points of uniqueness between the brands. This might be expected given the head-to-head strategies used by these competitors.
Next we look for uniqueness in the extent to which a given brand "clearly dominates" the other. First, the strongest indication of uniqueness is found in the "consumer-related" mid-level cluster where Coke receives a 32.99 mid-level cluster value compared with a -9.80 for Pepsi; a 42.79 point spread. This difference is significant (t=4.07, p<.01). This category represents thoughts that describe the relationship between the brnd and consumers own life and experience. Here, Coca-Cola clearly has uniqueness. Base clusters appearing on the Coke diagram and not on the Pepsi diagram included "fulfillment," "my feelings," "my behavior," "family and others," and "good times and fun." By contrast, Pepsis diagram contained mentions of "negative personal effects," "non-usage," and "loyalty." The nature of these ideas is quite different. While Coca-Cola has many personal associations, Pepsi appears not to have any such relationship. These associations might also be important with respect to the development of person-brand relationships, as suggested by Fournier (1998). Again, the presence of uniqueness with the leading brand is consistent with Kellers (1993) model.
To the extent that one has a great strength on a dimension where another has none, this dimension can be considered a point of "uniqueness." Using this definition, Pepsi has the greatest strength in the advertising and promotion mid-level cluster. The subcomponents of this dimension include "advertising themes," "endorsers and spokespeople," "promotions," "ad themes," and "commercials." Yet, while uniquely strong, the mid-level cluster value is higher for Coke. One interpretation of these data is that while Pepsi has built its image on a strong sense of advertising, this strong association is not helpful to the brand relative to Coke on the same dimension.
Over the Counter Medicines. An analysis similar to that just presented was conducted for Tylenol and Advil. Given space limitations, only a few results are spotlighted. Again, the overall cluster values for Tylenol and Advil (148.73 and 72.30, respectively) were consistent with market-values ($1.98 billion and $0.805 billion, respectively), and significantly different, t=3.12, p<.01. As with the soft drink analysis, global analysis of brand favorability did not show differences between Tylenol (x=7.13, n=21) and Advil (x=7.83, n=22), t=-1.44, p<.16. Strength and favorability domination for Tylenol was found on all but the "product-related" mid-level cluster. [The "product usage" category is a special case as it represents inherently negative ailments, more of which are associated with Advil.] Like Coke, the greatest Tylenol differential appears to be found on the consumer-related mid-level cluster; however the difference is not significant, t=-.22, p<.81. The only point of uniqueness for Tylenol may be price, though response numbers are too small for analysis. Alternatively, the large number of "product-related" comments for Advil (76) suggests that this mid-level cluster may be a point of uniqueness for that brand.
Study 2 Discussion
Affinity analysis was used to study Kellers (1993) concept of brand image and relate it to a brand position measure of market performance. The affinity diagrams themselves offer a vivid description of how a brand is held in consumers minds based on associations they have with the brand. The analyses presented here show that Kellers concept of brand image is consistent with the first and second positions of brands in two categories. Evidence of relative strength, favorability, and uniqueness are also found to be consistent with brand positions suggesting the potential value of these measures as assessments of brand performance.
This paper has examined several different measures related to brand equity, some of which have not been presented elsewhere before. Generally, the findings of these studies support Kellers conceptualization of brand equity and its components. The results provide a first step toward understanding (1) how consumer-based sources are related to market-based outcomes of brand equity, and (2) which elements of consumer-based knowledge structures may be most closely associated with market-based outcomes.
The overall implications for consumer research suggest that measures of consumer-based bran perceptions are, at least to some extent, accurate reflections of brand performance in the marketplace, particularly with the affinity analysis. Moreover, global favorability measures did not show differences between brand favorability, whereas the affinity process did. Further work on this issue will help to better inform researchers as well as practitioners about the relationship between market- and consumer-based measures of brand equity.
AFFINITY DIAGRAMS: COCA-COLA AND PEPSI
AFFINITY DIAGRAMS: TYLENOL AND ADVIL
Aaker, David (1991), Managing Brand Equity. New York: The Free Press.
Badenhausen, Kurt (1996), "Blind Faith," Financial World, 165 (July 8), 50-65.
Collins, James C. and Jerry I. Porras (1994), Built To Last-Successful Habits of Visionary Companies. New York: Harper Collins Publishers, Inc.
Cox, Dena Saliagas and Anthony D, Cox (1988), "What Does Familiarity Breed? Complexity as a Moderator of Repetition Effects in Advertisement Evaluations," Journal of Consumer Research, 15 (June), 111-16.
Fournier, Susan (1998), "Consumers and Their Brands: Developing Relationship Theory in Consumer Research," Journal of Consumer Research, 24 (March), 343-373.
Futami, Ryoji (1989), An Overview of Seven Management Tools for QC, Tokyo, Japan: Union of Japanese Scientists and Engineers.
Kamakura, Wagner and Gary Russell (1993), "Measuring Brand Value With Scanner Data," International Journal of Research in Marketing, 10 (March), 9-22.
Keller, Kevin L. (1993), "Conceptualizing, Measuring, and Managing Customer-Based Brand Equity," Journal of Marketing, 57 (January), 1-22.
Keller, Kevin L. (1998), Strategic Brand Management-Building, Measuring, and Managing Brand Equity. Englewood Cliffs, NJ: Prentice Hall.
Krishnan, H.S. (1996), "Characteristics of Memory Associations: A Consumer-based Brand Equity Perspective," International Journal of Research in Marketing, 13, 389-405.
Ourusoff, Alexandria (1994), "Brands: Whats Hot. Whats Not," Financial World, 163 (August 2), 40-56.
Silverman, Steven N. and Lori L. Silverman (1994), "Using Total Quality Tools for Marketing Research: A Qualitative Approach for Collecting, Organizing, and Analyzing Verbal Response Data," in Advanced Research Techniques Forum, Chicago, IL: American Marketing Association.
Simon, Carol J. and Mary Sullivan (1993), "The Measurement and Determinants of Brand Equity: A Financial Approach," Marketing Science, 12 (Winter), 28-52.
Zaltman, Gerald, Karen LeMasters, and Michael Heffring (1982), Theory Construction in Marketing, New York: John Wiley and Sons.