Zalando Deploys AI-Powered High-Frequency Pricing System for E-Commerce.

E-commerce sales events present unique challenges for pricing, including volatile demand and the need for rapid decisions that balance short-term revenue with long-term profitability. Traditional weekly-granularity pricing systems often fall short in these dynamic environments. This paper details a new specialized forecast-then-optimize algorithmic pricing tool designed for high-frequency decisions in fashion e-commerce. The solution combines daily-resolution demand forecasting, powered by gradient-boosted trees, with a multi-objective optimization framework. This framework is designed to maximize both long-term profit and net merchandise value across over 5 million articles. The system drastically reduces pricing decision time from hours to minutes, addressing a key limitation of previous approaches. Validated through 23 A/B tests across 12 markets at Zalando, a major European online fashion retailer, the new system demonstrated approximately 6% higher profit while maintaining equivalent sales and revenue compared to their prior manual-algorithmic hybrid method. The algorithm has since been successfully deployed to production, handling the majority of sales campaign pricing decisions.