Generative Optimization for Incentivized Advertising with Global Level Constraints

2026-08-05

AI Summary

This work addresses the challenge of optimizing continuous incentive allocations in rewarded advertising under strict global return-on-investment (ROI) constraints, while accounting for non-Markovian dynamics such as high-frequency user interactions, delayed feedback, and user fatigue. The authors propose GOAL, a novel framework that formulates incentive allocation as a conditional sequence generation task. It employs a hierarchical causal state encoder to capture both local behavioral dynamics and long-range dependencies in user responses. To handle diverse ROI constraints without retraining, they introduce Safe Constrained Policy Optimization (SCPO), a single policy learning algorithm that generalizes across varying constraint thresholds. Experiments on large-scale real-world data and synthetic environments demonstrate that GOAL significantly improves long-term revenue and user retention while substantially reducing constraint violations.

Abstract

Incentivized advertising allocates monetary or virtual rewards to drive user engagement, where a key challenge is optimizing continuous incentive magnitudes under strict global constraints. This problem is complicated by high-frequency interactions, delayed feedback, and non-Markovian user dynamics such as fatigue, which limit the effectiveness of existing uplift modeling and constrained reinforcement learning approaches. To address these challenges, we propose GOAL, a constraint-aware generative framework that formulates incentive allocation as a conditional sequence generation problem. GOAL directly generates incentive magnitudes conditioned on user histories and system-level global pressure, and integrates a hierarchical causal state encoder to capture both local behavioral dynamics and long-range dependencies. To enable flexible constraint control, we introduce \textbf{S}afe \textbf{C}onstrained \textbf{P}olicy \textbf{O}ptimization (SCPO), which learns a single generative policy that generalizes across a spectrum of ROI constraints without retraining. Experiments on large-scale real-world data and a synthetic fatigue-aware environment show that GOAL improves long-term revenue and user retention while substantially reducing ROI violation rates compared to strong baselines.

Problem

• incentivized advertising
• global constraints
• incentive optimization
• non-Markovian dynamics
• delayed feedback

Innovation

• Generative Optimization
• Incentivized Advertising
• Global Constraints
• Constrained Reinforcement Learning
• Conditional Sequence Generation