In this paper, we study the role of point-of-sale (POS) marketing mix variables in explaining variation in brand shares (i) at different retail formats, and (ii) across national and store brands in different price tiers. Stores in different retail formats differ in their positioning, the clientele they attract, and types of shopping trips made to the store. Further, national and store brands in different price tiers differ in the quality perception, and the contracts between the retailer and the manufacturer. Together, this implies that POS variables may influence brand shares very differently in different retail formats and for different brand types, which has important implications for retailers and manufacturers. We use Nielsen store level data from 2011-2014 and analyze the top ten spending product categories spreading across four product departments – dry grocery, non-food grocery, dairy and frozen, and across four retail formats – convenience stores, drug stores, supermarkets and mass merchandisers. In order to draw generalizations across categories, rather than study a subset of brands in each category, we classify national and store brands into “premium”, “value” and “economy” brands and study the relative drivers of choices of brands in these groups. Consistent with the theoretical predictions, we find that brand shares at convenience stores are least sensitive to price changes, and most sensitive to changes in assortment (number of sizes and number of variants per size). Further, brand shares at drug stores (mass merchandisers) are more sensitive to changes in POS variables as compared to mass merchandisers (supermarkets). Interestingly, the two different assortment metrics we consider – number of sizes and number of variants per size (packaging, flavors, fragrance, etc.) – have a differential impact on brand shares across formats and on different brand types, which as we show, has important consequences for retailers reallocating national and store brand assortment at their stores.
Note: Research papers posted on SSRN, including any findings, may differ from the final version chosen for publication in academic journals.