Which Visits Lead to Purchases? Dynamic Conversion Behavior at e-Commerce Sites

This paper develops a model of conversion behavior (i.e., converting store visits into purchases) that predicts each customer’s buying probability based on an observed history of visits and purchases. We offer an individual-level probability model that allows for cross-customer heterogeneity in a very flexible manner. We also allow visits to have a variety of effects in the purchasing process. For example, some visits are motivated by planned purchases while others are simply browsing visits -requiring a model that would accommodate different visit-topurchase relationships. Finally, consumers’ shopping behavior may evolve over time as a function of past experiences, and we capture these effects as well. Our proposed Conversion Model decomposes an individual’s purchasing conversion behavior into separate visit effects and purchasing threshold effects. Each component varies across households as well as over time. We apply this model to the problem of “managing” visitor traffic at a leading e-commerce site (Amazon.com). By predicting purchasing probabilities for a given visit, the Conversion Model can identify those visits that are likely to result in purchases. These premium visits may be redirected to a server that will provide a better shopping experience while those visits that are less likely to result in a purchase may be considered as targets for promotions.