Sift Science Takes $5.5M to Wield Machine Learning Against Fraudsters
Sift Science, which helps online retailers detect fraud, has raised $5.5 million led by Union Square Ventures with Max Levchin, Chris Dixon, Marc Benioff, First Round Capital and others.
Co-founder Brandon Ballinger told AllThingsD that his company has two big strengths: First, it uses machine learning to detect user behavior patterns — one million of them already — that correlate with fraud; and second, it’s made to be easy to integrate with any existing website via free APIs.
After that, customers pay ten cents per user per month for every user beyond 5,000.
Ballinger noted that because it’s standard practice to hold online stores liable for fraud — as compared to offline, where banks often cover fraud instead — this could potentially be a very big business.
In testing, Sift Science has observed all sorts of interesting phenomena. For instance, people who register to buy something with a Yahoo email account are twice as likely to be fraudsters as the norm. Meanwhile, people who register with an AOL account are half as likely as the norm to attempt something fraudulent.
Ballinger had previously developed similar in-house systems at Google to detect malicious advertising. He and co-founder Jason Tan started Sift Science in June 2011, and participated in the Y Combinator program later that year. They now have a team of nine people in San Francisco.
Sift Science has been in testing with companies including Airbnb, Listia, Affirm (Max Levchin’s new startup) and Uber. Customers can either use the service to block transactions in real time or to flag them for review before they’re settled.