An Integrated Model of Grocery Store Shopping Path and Purchase Behavior

We develop an individual-level integrated Bayesian model for store shopping trips to capture the relationship between a consumer’s shopping path through the store and her purchasing behavior. Using dynamic latent “attractions” of different zones and product categories in the store as fundamental constructs, our model jointly captures three key aspects of a consumer’s within-store behavior: which zones she visits, how long she stays in each zone, and what purchases, if any, she makes within that zone. We calibrate our model using a novel dataset from the PathTracker system, an electronic shopping cart monitoring system developed by Sorensen Associates. Our study also contributes to in-store research by modeling the degree of forward-looking behavior that consumers may exhibit, an area of significant recent study in marketing and experimental economics. By incorporating the total process of in-store shopping (movement and purchases) in our model, retailers may obtain a more comprehensive understanding of consumers’ grocery shopping behavior. As an illustration of the model’s potential usefulness, we conduct a hypothetical “policy experiment” in which two categories swap locations. We examine the impact of this change on the affected categories as well as other aspects of the shopping trip as a whole.