Understanding “Cherry Pickers:” How Retail Customers Split Their Shopping Baskets

Recent studies of store choice assume that shoppers develop shopping lists a priori, then visit the store that allows them to buy needed items at the lowest expected cost. These studies implicitly assume that shoppers purchase their entire list at a single store. However, many households shop at multiple stores without any intervening consumption—a behavior that retail managers commonly term "cherry picking." To explain this behavior, we develop a mathematical model that extends the standard cost minimization framework in two important ways: First, we allow shoppers to split their basket purchases across multiple stores; and second, we introduce price uncertainty into their decision-making process. The latter is necessary since one cannot assume all prices are known when the purchase decision is made. Because deferring purchases to a subsequent store visit is essentially a gamble constructed by the shopper, the buy or defer purchase decision will depend on her risk attitude. Despite the shopper-specific nature of this attitude, we discover a rather general result: Unless the shopper is perfectly risk neutral, the probability of an item being “cherry-picked” depends on its expected price difference divided by the variance of that difference. This leads to interesting implications for retailers regarding the timing and depth of retail promotions. Our model also predicts that the amount a cherrypicking shopper purchases at a given retailer depends on the order in which the retailers are visited. We develop a discrete mixture model that incorporates various aspects of householdlevel heterogeneity, including differences in loyalty, risk attitudes, and price knowledge. We apply this model in the analysis of a panel dataset of consumer purchases made by cherrypicking shoppers and find uniform support for our model’s predictions. In addition, we find empirical evidence that most cherry pickers are risk-averse in their purchase decisions, and that they study retailer ads.