Store-Wide space planning balancing impulse and convenience

We develop a store-wide shelf space planning framework that assigns product categories to shelves with the objective of balancing impulse buying and shopping convenience. Our resulting model is a bi-objective nonlinear integer program, similar in structure to a quadratic assignment. The model parameters are estimated based on data collected from a large supermarket in the Normandy region, France. We propose a linearization scheme and utilize it to exploit the efficient frontier of Pareto-optimal solutions via the ϵ-constraint method. Our results indicate that the supermarket management can significantly increase their revenue from impulse buying, while not burdening the customer with the inconvenience of extensive walking and search. In fact, we observe that many of our Pareto-optimal layout are anticipated to perform better than the actual layout the retailer is currently using on both impulse and convenience. For example, one well-balanced layout is anticipated to increase impulse profit by 82% and decrease walking by 11% with respect to the current layout. A side contribution of our work is developing a framework for data collection and parameter estimation based on panel (customer survey) data, observations from store visits, and publicly available information.