The Geography of Brands New Insights into the Local Performance of Nationally Distributed Brands Research by Sanjay K. Dhar and Jean-Pierre Dube Sanjay K. Dhar is James H. Lorie Professor of Marketing at the University of Chicago Graduate School of Business. Jean-Pierre Dube is professor of marketing and Neubauer Family Faculty Fellow at the University of Chicago Graduate School of Business. Marketing research on the consumer packaged goods industry has traditionally focused on how brands perform over time in a small number of geographic markets. New research suggests that analyzing brand performance across markets�rather than simply over time� may be a more fruitful approach. In the consumer packaged goods (CPG) industry, the most widely used measure of brand performance is brand market share�the unit sales of a brand divided by the unit sales of the entire category. Most research about purchase behavior and the effectiveness of marketing tools for CPG brands is based on time-series studies, which typically use data covering a few weeks to a few years in a particular geographic market. Until now, there has been relatively little research exploring the geographic distribution of market shares in the CPG industry. In the study �Consumer Packaged Goods in the United States: National Brands, Local Branding,� University of Chicago Graduate School of Business professors Sanjay K. Dhar and Jean-Pierre Dube, along with Bart J. Bronnenberg of UCLA, document several striking geographic patterns in the performance of national brands using a large scanner database spanning many CPG categories and U.S. regional markets. The authors note that the geography of CPG industries is an understudied area with several important potential directions for future research. Dhar characterizes the study as descriptive research. Dube adds, �We began this study as an exploration rather than a test. Finding a relatively large geographic component of brand shares was surprising.� The authors pooled market share information across time and geographic markets for the top two leading national brands in 31 CPG categories. They find that the geographic component of these data accounts for 92 percent of the variation in market shares. Furthermore, brand shares appear to be stable over time within a market. With only a few exceptions, researchers have previously been unable to collect truly long-run information about brand performance; CPG data sources�at best�cover a couple of years in any given market. In contrast, the geographic patterns appear to be a long-run phenomenon. These data may present marketers with a rich new source of long-run marketing data with which to study the long-run effects of marketing investments. Across markets, the geographic variation in market shares, indices of brand quality perceptions, and the identity of local brand-share leaders is so significant that it calls into question the concept and relevance of a �national brand.� These findings also highlight problems in using studies focusing on performance over time to determine the effectiveness of marketing tools when the data focuses only on single markets. Since market share variation over time appears to capture a minor component of overall variation, and it is difficult to obtain databases for long time periods, inferences about the impact of marketing variables based on short time series of market shares may be imprecise. A Broad-Spectrum Approach Dhar, Dube, and Bronnenberg used scanner data from AC Nielsen spanning 31 CPG categories in the 50 largest Nielsendesignated Scantracks, each of which typically include a single metropolitan area and its suburbs. The data was sampled at four-week intervals between June 1992 and May 1995. Categories spanned food products from bread and bakery to frozen entrees/side dishes. The authors investigated properties of CPG brand share data across brands, markets, and time, as well as several marketing variables. They constructed a brand share measure based on equivalent unit sales in each category/market/month. For each of the 31 categories, the data covers two brands in 50 markets over 39 months. The most notable feature of the market share data was the high level of geographic variation. Regardless of the degree to which a product competed on the national level, the range and level of variation in a brand�s share was very similar. This suggests that the high level of variation is not simply due to brands avoiding certain markets. After breaking down shares by time, market, and brand effects, as well as the interaction between the market and the brand, the authors find that geographic variation explains considerably more of the market share variance than variation across time. Furthermore, the local market structure within a category varies considerably across geographic markets: one brand may be very strong in one market and relatively weak in another. What are the limitations of focusing on a single market when analyzing national brands? First, focusing on a single market ignores the cross-market dimension in the CPG industry, and thus does not focus on explaining the largest source of variation in the market-level performance of national CPG brands. Second, a major goal of quantitative marketing research is to determine the effects of a firm�s marketing investments. If variation in brand performance across markets is related to such investments, focusing on a single market may lead to misleading estimates of these effects. Studying variations in market shares across markets, rather than focusing on movement over time, may be more informative about the long-run effects of marketing investments, advertising, and distribution. These findings may impact strategic marketing decisions regarding the product, such as local positioning and branding. What is the Relevance of a �National Brand�? For brands that garnered the largest shares of a category�s national market, the authors observed considerable regional variation in market share, perceived quality, and share dominance. In specific Scantrack markets, the authors find a large disparity between a brand�s national share and its local share. The authors expanded the set of brands they studied to include all local share leaders across categories in order to investigate whether a national brand�defined as a brand with distribution in all 50 Scantrack markets�has inherent advantages in terms of local performance. They find that when a national brand leads in a market, it tends to lead with 12 more market share points than a local or regional brand. The authors did not observe meaningful inherent national brand effects. Can the national performance of a brand predict how the brand will perform in local markets? The authors suggest that a single geographic market provides only limited information about a brand�s overall performance. Furthermore, within local markets, there is a stronger tendency for one brand to dominate in terms of shares. Focusing on either the national market or a single geographic market may not only limit the information about a brand�s performance, it also may limit the information about the category as a whole. For example, while the national market for coffee appears to be dominated by two firms, in most local markets there is one dominant firm. In addition to scanner data, the authors used information on perceived brand quality collected by Young and Rubicam Associates for its Brand Asset Valuator Database to study geographic patterns in quality perceptions of the same brand. The quality perception data showed comparable levels of variation as the market share data. The clear lack of consistent quality perception across markets further erodes the concept of a national brand. The authors suggest that future research could benefit from a better theoretical and empirical understanding of the sources of these local share differences. Future Research �The purpose of our study is not to provide answers, but rather to suggest several potential directions for new research based on descriptive findings,� says Dhar. Dhar, Dube, and Bronnenberg write, �Insofar as geography leads to different conclusions regarding the effectiveness of marketing variables, it could suggest that we still have much to learn.� In a follow-up study, �Market Structure and the Geographic Distribution of Brand Shares in Consumer Package Goods Industries,� Dhar, Dube, and Bronnenberg explore some of the potential underlying economic forces related to branding and brand advertising that generated these striking patterns. They utilize the geographic data to document a strong relationship between market shares and advertising across geographic areas. This finding challenges the results of prior research�based on single-market time series data�that routinely provided results favoring promotional tactics, such as price-cutting, over advertising strategies. In the geographic dimension, the authors find very little correlation between promotional activity and brand market shares, suggesting that promotions only have short-term effects. Advertising appears to have longer-lasting effects, possibly through its ability to build a lasting brand �goodwill� stock and to bolster brand perceptions within a market. In two industries, ground coffee and mayonnaise, the authors also find evidence of a strong �early-mover� effect. For the top two brands in each market, controlling for the identity of the first entrant in each geographic market captures more than 50 percent of the geographic variation in market shares. This result is particularly striking in the coffee category, where brands were rolled out as early as the mid 1900s. The early-entry effect also explains a large component of the advertising variation across markets. In the second study, the authors also examine several alternative explanations for the underlying sources of the geographic patterns. In particular, they look at proximity to production facilities (i.e., potential cost advantages), relationships with large retailers, and even local parentcompany effects. However, none of these alternative sources appears to explain the geographic variation in brand shares. The authors hope that future research will focus on broader marketing databases spanning wider geographic scope and longer time horizons. Experimenting with cross-sectional geographic data and contrasting findings with single market time-series data may be a novel way to advance current knowledge about marketing effectiveness. Ultimately, establishing theories to understand the patterns documented in the study should be a fruitful new research area for quantitative marketers.