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.
Figures & tables
Figure 1. Example of message placement styles, urgency placement in particular.
Figure 2. Stageboost Architecture
Model
Purchase
Latency
CLE (Baseline)
0.109%
Within Budget
StageBoost
0.255%
Exceeds Budget
Lean StageBoost
0.144%
Within Budget
Table 1. Counterfactual uplift measurement for different models. CLE denotes the production baseline.
Metric
Overall
Focus Category
GMB
+0.08%±0.28%
+0.58%±0.50%⋆
ASP
–
+0.41%±0.38%⋆
Table 2. Online A/B experiment results on live traffic. ⋆ denotes statistical significance.