Case Study: Remitly achieves 50% fewer manual reviews and catches 70% of fraud in the top 2% of transactions with Sift

How Remitly safely sends millions of dollars overseas every month

Remitly is a mobile-first remittance service that enables fast international money transfers and guarantees refunds, so preventing fraud at scale is critical. As the company expanded—especially into the Philippines—fraud activity increased and the risk team was overwhelmed by lengthy phone verifications, KYC/AML requirements, and an unwieldy rules engine (80–100 hard-coded rules), creating a pressing need to reduce manual reviews while maintaining strict controls.

Remitly integrated Sift’s API and dashboard in an afternoon, using Sift Scores, real-time alerts, and machine learning trained on Remitly’s own data to streamline workflows and retire dozens of rules. The results: manual review rate fell from 5% to 2.5% (about 50% fewer reviews), over 70% of fraud is caught in the top 2% of transactions, and new-user reviews declined—significantly cutting investigation time and phone verifications.

Remitly

Nate Spanier

Director of Operations