| Abstract: | This study uses a sample of approximately 60,000 US households to document fundamental shopping basket size patterns across a range of retail types, and examines them in relation to retailer performance metrics (unit sales and dollar revenue). Specifically, this research addresses two main questions: 1) how do shopping basket metrics (mean and median number of items, the distribution of one, two, three … n items) differ by retail type, and 2) how does the Pareto ratio (sales generated by the heaviest 20%, and lightest 80% of buyers) apply to shopping baskets. The results show that basket size patterns in retailers are predictable. Shoppers purchase more items on average in retailers that offer a greater variety of items, and the distribution of basket sizes follows the Poisson lognormal model. The results also show that the largest 20% of shopping baskets on average generate 50% of unit sales, and 40% of dollar revenue. These results set additional benchmarks of the patterns that can be expected when looking at data on a basket-level. This research offers implications to practitioners by showing the importance of small and large baskets for retailer revenue and growth, which can guide more informed decision making to better manage their stores and brands. |
