Case Study: Purse reduces fraud and speeds transactions with Sift

How Sift helped Purse build a trustworthy bitcoin marketplace

Purse is a San Francisco–based bitcoin marketplace that matches bitcoin-holders who want to buy on Amazon with users who cash out Amazon gift cards, running 5,000–10,000+ orders per month for a global base of 150,000+ users. Because bitcoin transactions are final and Purse offers a “Purse Guarantee” of up to $10,000, the company faced costly fraud from malicious actors using hacked Amazon accounts; their initial manual review process required three full-time staff investigating hundreds of cases daily and was unscalable.

Purse integrated Sift’s machine‑learning solution via webhook in two weeks, using Sift Score, network visualizations, and device ID data to auto‑ban high‑risk users and surface genuine transactions for faster processing. The automation reduced manual review, cut fraud, increased transaction throughput, and improved customer trust—delivering measurable ROI in lower fraud losses and a better, more scalable customer experience.

Purse

Steven McKie

Head of Business Development & Product Content