🤖 AI Summary
This paper addresses the problem of user segmentation and policy assignment under feasibility constraints—specifically, achieving near-optimal profit via coarse-grained personalization with a limited number of policies. The proposed method introduces a two-stage joint optimization framework: Stage I employs machine learning–based causal models to estimate conditional average treatment effects (CATE), capturing individual-level heterogeneous responses; Stage II formulates segmentation and policy assignment as a discrete-constrained optimal transport problem—marking the first integration of heterogeneous causal inference with optimal transport. Evaluated on a large-scale promotional field experiment, the approach achieves over 99.5% of the incremental profit attainable under full personalization using only five policies. It significantly outperforms conventional feature-based clustering and grid search. The framework delivers a scalable, interpretable, and high-return solution for precision marketing under resource constraints.
📝 Abstract
Advances in estimating heterogeneous treatment effects enable firms to personalize marketing mix elements and target individuals at an unmatched level of granularity, but feasibility constraints limit such personalization. In practice, firms choose which unique treatments to offer and which individuals to offer these treatments with the goal of maximizing profits: we call this the coarse personalization problem. We propose a two-step solution that makes segmentation and targeting decisions in concert. First, the firm personalizes by estimating conditional average treatment effects. Second, the firm discretizes by utilizing treatment effects to choose which unique treatments to offer and who to assign to these treatments. We show that a combination of available machine learning tools for estimating heterogeneous treatment effects and a novel application of optimal transport methods provides a viable and efficient solution. With data from a large-scale field experiment for promotions management, we find that our methodology outperforms extant approaches that segment on consumer characteristics or preferences and those that only search over a prespecified grid. Using our procedure, the firm recoups over 99.5% of its expected incremental profits under fully granular personalization while offering only five unique treatments. We conclude by discussing how coarse personalization arises in other domains. https://arxiv.org/abs/2204.05793
