Case Study: Dwolla achieves a 50% drop in fraud and scales without hiring additional analysts with Sift

A payments network that can scale without worrying about fraud

Dwolla is a U.S.-focused payments network that enables fast, API-driven bank transfers. Faced with on-demand transactional fraud (with unauthorized activity sometimes not showing as chargebacks for 30–60 days), Dwolla’s rules-based in-house system was becoming hard to scale as fraud rates climbed to 0.5 basis points and analyst workload increased.

Dwolla integrated Sift’s machine-learning solution using custom fields—starting with a two-day test and a full 2–4 week part-time implementation—and hooked it into their fraud workflow via webhook. The result: fraud dropped from 0.5 to 0.2 basis points (a 50% reduction), the team avoided hiring additional analysts, average time spent on fraud investigations fell, and analysts gained faster, actionable scoring and cross-account intelligence.

Dwolla

Ryan Hodge

Financial Intelligence Unit Director