E-Commerce Promotions Personalization via Online Multiple-Choice Knapsack with Uplift Modeling

Abstract

Promotions also affect revenue and may incur a monetary loss that is often limited by a dedicated promotional budget. We propose an Online Constrained Multiple-Choice Promotions Personalization framework, driven by causal incremental estimations achieved by uplift modeling. Our work formalizes the problem as an Online Multiple-Choice Knapsack Problem and extends the existent literature by addressing cases with negative weights and values as a result from causal estimations. Our real-time adaptive method guarantees budget constraints compliance achieving above 99.7% of the potential optimal impact on various datasets. It was deployed in a large-scale experimental study at Booking.com - one of the leading online travel platforms in the world. The application resulted in 162% improvement in sales while complying a zero-budget constraint, enabling long-term self-sponsored promotional campaigns.