Promotion Optimization for Multiple Items in Supermarkets

Promotions are a critical decision for supermarket managers, who must decide the price promotions large number of items. Retailers often use to boost sales different items by leveraging cross-item effects. We formulate promotion optimization problem multiple as nonlinear integer program. Our formulation includes several business rules constraints. demand models can be estimated from data and capture postpromotion dip effect effects (substitution complementarity). Because functions typically nonlinear, exact is intractable. To address this issue, we propose general class programming approximations. For with additive effects, prove that it sufficient account unilateral pairwise contributions derive parametric bounds on performance approximation. also show unconstrained solved efficiently via linear program when substitutable set has two values. more cases, develop efficient rounding schemes obtain an solution. conclude testing our method realistic instances convey potential practical impact retailers. This paper was accepted Yinyu Ye, optimization.