Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a two-stage XGBoost model designed to optimize the recommendation of signals on eBay's View-Item page, enhancing user engagement and purchase decisions. By leveraging counterfactual estimation, the model achieved a 0.08% overall lift in Gross Merchandise Bought (GMB) and a notable 0.58% increase specifically in the Parts and Accessories category. The results indicate that the model effectively boosts conversions for high-average-price items, demonstrating its practical impact on e-commerce performance.
A two-stage XGBoost model can significantly enhance eBay's sales by optimizing signal recommendations, leading to a measurable increase in high-value item conversions.
Signals are short textual or visual snippets displayed on the eBay View-Item (VI) page, providing additional, contextual information for users about the viewed item. The aim of displaying these signals is to facilitate intelligent purchase and to incentivize engagement. In this paper, we present a 2 stage xgboost based model that optimally populates the VI page with signals. This approach has shown a 0.08% lift in overall GMB (Gross Merchandise Bought) and 0.58% increase in Parts and Accessories GMB, primarily due to increase in conversion of high average price items in online experimentation.