Multi-channel Uplift Policy Learning

2026-07-30

AI Summary

This work addresses the challenge of optimally allocating a fixed marketing budget across multiple channels on e-commerce platforms to maximize utility, a task hindered by observational confounding and poor out-of-sample generalization in conventional predict-then-optimize approaches. The authors propose ReAlloc, a novel framework that formulates budget reallocation as a gain-based decision problem under simplex constraints. ReAlloc innovatively integrates causal inference with structured knowledge distillation: an orthogonal teacher model extracts unbiased local gradients from short-term observational logs, which are then distilled by an explanation-guided student into a long-horizon structured marginal field. This enables conservative yet effective capture of cross-channel substitution effects. Large-scale online A/B tests on Taobao demonstrate that ReAlloc significantly improves both paid order volume and revenue.

Abstract

E-commerce platforms must allocate fixed marketing budgets across multiple channels to maximize business utility. However, standard predict-then-optimize (PTO) paradigms fail in this compositional space due to observational confounding and severe extrapolation. We formulate this challenge as a simplex-constrained uplift decision problem and propose ReAlloc, a fast-slow causal framework. Specifically, an agile Orthogonal Teacher extracts unbiased local gradients from short-term logs, while an Explanation-Guided Student distills them into a structured marginal field over long-term horizons. This design enables support-aware, conservative decisions that capture cross-channel substitutions. Extensive simulations and large-scale online A/B tests on Taobao platform demonstrate that ReAlloc achieves simultaneous lifts in both pay order and income.

Problem

• multi-channel
• uplift
• budget allocation
• observational confounding
• extrapolation

Innovation

• Uplift Modeling
• Causal Inference
• Multi-channel Allocation
• Orthogonal Learning
• Explanation Distillation