This is like giving every seller on eBay a smart assistant that can (1) tell them what a fair price is for their item based on millions of similar listings, and (2) instantly show shoppers other items that are most similar to what they’re viewing or searching for.
Reduces guesswork and inconsistency in item pricing while improving product discovery by automatically suggesting similar items to buyers. This helps standardize prices, increase conversion, and keep shoppers on the platform even if the exact item isn’t ideal or available.
Large proprietary marketplace data (historic listings, transactions, user interactions) powering price estimation and similarity models, deeply embedded into core eBay workflows (listing flows, search results, product pages).
Hybrid
Vector Search
High (Custom Models/Infra)
Feature computation and embedding generation at eBay scale (tens/hundreds of millions of items), plus low-latency serving of similarity search and price predictions under heavy query load.
Early Majority
Combines price prediction and similar-item search tightly around marketplace dynamics, likely leveraging very rich listing and transaction history rather than generic catalog data, and integrated into core user journeys like listing creation and search results.