Multicategory purchase behavior: basket choice, shopping frequency, and promotional analysis

This research introduces a new tool for analyzing both what customers buy and how often they shop. Unlike traditional models that focus only on in-store purchases, the MVL-Poisson Model captures shopping frequency, basket composition, and consumer response to prices and promotions. It segments customers by preferences and visit-frequency, reveals cross-category demand relationships, and highlights how promotions influence not just purchases but also store visits. It is computationally practical and can be implemented with standard retail data and analytics software. In an application to convenience store data, the model had high predictive accuracy and generated insights aligned with managerial intuition. We found that shoppers with similar preferences may visit at very different rates—a critical finding for targeting promotions effectively. Focusing only on in-store behavior underestimates the impact of promotions, as promotions also drive store traffic. Using insights on consumer preferences and cross-category relationships, we show how our model can be used to create optimal bundle promotions customized to particular segments.