AIGP: An LLM-Based Framework for Long-Term Value Alignment in E-Commerce Pricing

2026-06-25

AI Summary

This study addresses critical limitations in traditional e-commerce dynamic pricing models, which often suffer from poor interpretability, an inability to incorporate unstructured information, and misalignment with long-term business objectives such as cumulative GMV, ROI, and milestone achievement. To overcome these challenges, the authors propose the AIGP framework, which innovatively integrates large language models with domain knowledge, structured data, and textual context to generate interpretable pricing decisions. Furthermore, a long-horizon value estimator based on offline reinforcement learning, combined with Direct Preference Optimization (DPO), steers the policy toward alignment with strategic commercial goals. Large-scale online A/B tests on Taobao Factory demonstrate significant improvements: over 14 days, GMV increased by 13.21%, ROI by 7.59%, and milestone attainment rate by 8.20%, while providing transparent and traceable rationales for pricing actions.

Abstract

Traditional dynamic pricing models in large-scale e-commerce suffer from limited interpretability, poor utilization of unstructured information, and misalignment with long-term business objectives such as cumulative Gross Merchandise Value (GMV), Return on Investment (ROI) and milestone achievement. We propose AIGP, a novel framework that leverages a Large Language Model (LLM) prompted with domain knowledge, structured data and textual context to make interpretable, knowledge-aware pricing decisions. For efficient deployment while maintaining high-quality outputs, we employ supervised fine-tuning for knowledge distillation. Central to AIGP is the Long-Term Value Estimator (LTVE), trained via offline reinforcement learning on historical data, which serves as a reward model to score candidate pricing actions and select preference pairs for Direct Preference Optimization (DPO), thereby aligning the pricing policy with long-term business objectives. Extensive offline evaluations and large-scale online A/B tests on Tao Factory demonstrate that AIGP achieves significant improvements: +13.21% in GMV, +7.59% in ROI, and +8.20% in milestone achievement rate over 14 days compared to the production baseline, while simultaneously providing interpretable and transparent pricing rationales.

Problem

• dynamic pricing
• long-term value alignment
• e-commerce
• interpretability
• unstructured information

Innovation

• Large Language Model
• Long-Term Value Alignment
• Direct Preference Optimization
• Offline Reinforcement Learning
• Interpretable Pricing