Online Advertisement Allocation Under Customer Choices and Algorithmic Fairness.

This article focuses on optimizing dynamic advertisement allocation for e-commerce platforms under customer choice behavior and algorithmic fairness considerations. It introduces a novel two-stage target-debt (TTD) stochastic programming framework that first determines optimal click-through targets for ads across customer segments and then adaptively allocates ad offer-sets using a debt-weighted offer-set (DWO) algorithm to meet these targets. The authors prove that the DWO policy is asymptotically optimal as the problem size grows, achieving maximum fairness-adjusted value (FV) from advertising while ensuring smoother budget depletion compared to fluid-approximation heuristics. Numerical experiments demonstrate that the proposed approach outperforms existing fluid-based methods in efficiency, fairness, variability reduction, and computational scalability without significantly compromising advertising effectiveness.