Use of retail scanner data to gain strategic market insights: Two applications
Kumar, Amresh G.
MetadataShow full item record
The general feeling in the industry has been that grocery supermarkets have been quite successful in stimulating loyalty card usage by their customers, but not much in leveraging the true benefits of loyalty card programs in using the transactional level and consumer information data from such programs to develop value-added insights for strategic decision making. There in lies the motivation of this dissertation research. The goal of this research is to demonstrate how the typical data generated from retailers' loyalty card program can be used to gain powerful insights that are relevant not only in the context of "marketing dashboard" metrics connected to retailers' strategies, but also in terms of better understanding of retail market dynamics in academic research. Specifically, the research is organized into two essays focusing on the use of such data in the context of two distinct strategic areas for retailers: (1) price promotions; and (2) product assortment. The backdrop for the first essay is the evidence from past research that finds consumers strategically shifting their purchase decisions (timing and quantity) to take advantage of the widely prevalent within-store inter-temporal price variation in grocery markets. However, there has been hardly any systematic study to investigate the revealed effectiveness of such strategic purchase behaviors on the part of consumers in essentially "timing" the market. The insights from such a study hold considerable implications for retailers pricing and segmentation strategy. The first essay undertakes such a study, which is similar to studies for evaluating performance of "market timing strategy" in financial equity markets. We use a scanner panel data base of over 1229 households that consists of nearly half a million purchase incidences over two years and 252 store-keeping units (SKUs) across 28 product categories. We develop multiple measures to estimate consumers' revealed performance in "timing" the market and also to identify the primary determinants of such performance in terms of consumer, SKU and category specific characteristics. A major variable of interest here is the role of consumer learning on performance. In the second essay, we shift our focus on gaining strategic insights into product assortments issues for grocery retailers. In recent years, the higher costs of carrying a large assortment of SKUs as well as competition from lower cost, lower assortment-carrying discount retailers have led retailers to look for more efficient assortment strategies by eliminating low-selling SKUs. A key input to formulating such strategies obviously hinges on the ability to have a good understanding of opportunity costs of SKU reductions. The second essay develops a heuristic approach to empirically estimate such opportunity costs based on the typical retail scanner data available, and without requiring a priori data from actual reductions of SKUs under analysis. The opportunity cost estimate takes into account not only the direct cost from lost profit that is directly being generated by a SKU, but also the indirect cost from its elimination. We then analyze the nature of distribution of the estimated opportunity costs of SKU reductions across product assortments and its determinants in terms of category, store and market characteristics. We also investigate a key relevant dimension of underlying consumer behavior--heir revealed SKU choice set size--cross distinct category, store and market contexts. For our analyses, we use a scanner data set comprising all transaction data over 52 weeks for nearly a million consumers across 14 product categories in 34 stores of a regional supermarket chain.