Case Study: eBay achieves faster, cost-effective, and more accurate product classification and GTIN enrichment with CrowdFlower

CrowdFlower optimizes the cost and quality of a massive data project

eBay, the global online marketplace founded in 1995, faced the challenge of organizing millions of constantly changing product listings and finding missing Global Trade Item Numbers (GTINs). Because algorithms alone couldn’t handle nuanced categorization and GTIN discovery, eBay’s traditional offshore outsourcing proved costly, hard to scale, and insufficiently accurate.

Figure Eight broke eBay’s work into microtasks completed by thousands of contributors worldwide, managed workflow quality, and implemented a machine-learning classification pipeline that randomizes label order to reduce bias. They added geographic filters for native English contributors and an exhaustive, cross-checked GTIN search process. The result: faster, lower-cost processing with high accuracy—more than 15 high-value projects completed and an ongoing classification workflow in production.

eBay

Ram Rampalli

Product Management Team, The Selling and Catalogs Group, eBay