Personalized Sales Targets with Customer Choices

Modern firms usually have personalized service/sales targets for different customer segments and products. Such sales target personalization poses additional challenges for firms to optimize their assortment policy. We propose a general modeling framework to study the assortment optimization problem in the presence of personalized sales targets. Our framework integrates both sales target optimization and personalized assortment planning as a two-stage stochastic program. Based on this modeling framework, we develop a family of simple and effective algorithms, the Debt-Weighted Assortment policy, and demonstrate their optimality for assortment planning with personalized sales targets. Our modeling framework is flexible enough to incorporate several important applications that unveil interesting insights. The first application explored in this paper is sales target personalization. In this application, we show that personalized sales targets may require a substantially higher cost to induce the necessary customer traffic. To address this issue, we propose a multi-sourcing strategy that efficiently re-balance customer types and personalized sales targets. Next, we apply our modeling framework to the classic personalized assortment optimization problem with inventory constraints. We propose a family of debt-weighted assortment algorithms that prove to be asymptotically optimal. Furthermore, the proposed algorithms achieve a higher expected revenue and a lower revenue variability than the benchmarks in the existing literature such as choice-based linear program and LP-resolving policies.