Ma, Shaohui and Fildes, Robert (2017) A retail store SKU promotions optimization model for category multi-period profit maximization. European Journal of Operational Research, 260 (2). pp. 680-692. ISSN 0377-2217
A_retail_store_SKU_promotions_optimization_model.pdf - Accepted Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (2MB)
Abstract
Consumer promotions are an important element of competitive dynamics in retail markets and make a significant difference in the retailer's profits. But no study has so far included all the elements that are required to meet retail business objectives. We extend the existing literatures by considering all the basic requirements for a promotional Decision Support System (DSS): reliance on operational (store-level) data only, the ability to predict sales as a function of prices and the inclusion of other promotional variables affecting the category. The new model delivers an optimizing promotional schedule at Stock-Keeping-Unit (SKU) level which maximizes multi-period category level profit under the constraints of business rules typically applied in practice. We first develop a high dimensional distributed lag demand model which integrates both cross-SKU competitive promotion information and cross-period promotional influences. We estimate the model by proposing a two stage sign constrained regularization approach to ensure realistic promotional parameters. Based on the demand model, we then build a nonlinear integer programming model to maximize the retailer's category profits over a planning horizon under constraints that model important business rules. The output of the model provides optimized prices, display and feature advertising planning together with sales and profit forecasts. Empirical tests over a number of stores and categories using supermarket data suggest that our model generates accurate sales forecasts and increases category profits by approximately 17% and that including cross-item and cross-period effects is also valuable.