CanniUplift: A Holistic Framework for Mitigating Seller and Incentive Cannibalization in E-commerce Uplift Modeling

2026-07-06

AI Summary

This study addresses the challenge of biased incremental effect estimation in e-commerce platforms operating under multi-seller environments, where inter-seller expenditure shifting (seller-level cannibalization) and incentive-induced interference with organic conversions or substitution of alternative rewards (incentive-level cannibalization) distort causal inference. To tackle this, the authors propose CanniUplift, a novel framework that jointly models both forms of cannibalization for the first time. It incorporates Platform-level Global Alignment (PGA) to constrain cross-store substitution behaviors, employs Redemption-Driven Decomposition (RDD) to reduce attribution noise, and introduces a Treat-Attention mechanism to capture complex interactions between users’ historical behaviors and current incentives. Evaluated on both synthetic and industrial datasets within a full-space uplift modeling paradigm, CanniUplift significantly outperforms baseline methods, yielding a 4.08% increase in platform-wide incremental GMV upon online deployment and demonstrating markedly improved ROI in A/B tests.

Abstract

Personalized incentive allocation is vital for e-commerce, where uplift modeling is the standard for estimating Individual Treatment Effects (ITE). However, traditional models often fail in complex multi-seller environments with violations of the Stable Unit Treatment Value Assumption (SUTVA). We identify two critical challenges: Seller-level Cannibalization, where incentives shift expenditure between shops without growing the platform, and Incentive-level Cannibalization, where organic conversions or alternative rewards introduce significant noise into incrementality estimation. In this paper, we propose CanniUplift, a unified framework to mitigate these dual-source cannibalization effects. Specifically, we design Platform-level Global Alignment (PGA) to capture cross-shop substitution through global GMV consistency constraints. To tackle incentive-driven noise, we introduce Redemption-based Decomposition Denoising (RDD), which uses redemption behavior to decompose treated outcomes and reduce attribution noise within an entire-space framework. Furthermore, a Treat-Attention mechanism is designed to model intricate interactions between users' historical behaviors and current treatment options. Extensive experiments on both synthetic and large-scale industrial datasets demonstrate that CanniUplift significantly outperforms state-of-the-art baselines. Ablation studies confirm that the integration of PGA and RDD consistently improves wAUUC and wQINI. Successfully deployed online, our framework achieved a 4.08% relative increase in platform-wide incremental GMV (Delta GMV) over the production baseline and improved ROI in online A/B tests, proving effective in driving global platform growth.

Problem

• Uplift Modeling
• Cannibalization
• Individual Treatment Effect
• E-commerce
• SUTVA

Innovation

• uplift modeling
• cannibalization mitigation
• global alignment
• redemption-based denoising
• treatment effect estimation