Sift science logo

Sift Science Could Be The Key To Ending Credit Card Fraud

Sift science logo

Sifting through false positives with science

Sift Science, a credit card fraud prevention company, has managed to secure $18 million in Series B funding and it’s still relatively unknown.

Sift Science is setting out to change how credit card fraud is detected, reducing the cost and hassle surrounding it – while also helping tackle the issue head on. Instead of going by the industry standard of 50 – 80 per cent of false positives, Sift uses a smart UI and machine learning to detect fraudulent charges as they’re happening. This means that Sift is able to see fraudulent activity across a whole network of merchants and thus reduce the percentage of false alarms down to around seven per cent or less.

“Most fraud detection systems are based around rules,” said founder Jason Tan speaking to Tech Crunch. “If the charge is more than $5,000, and from a Nigerian IP address, and the card has been rejected three times before, then the company won’t process the transaction. But we say ‘no rules, all data.’”

This means that Sift Science can actively find and stop true fraudulent transactions, rather than relying on a set of rules that stop genuine transactions. I, for one, think this is a wonderful idea after having had my card blocked for fraudulent activity a few times in the past, largely because I travel around the UK with reasonable frequency.

For merchants, Sift also seems like a rather reasonable purchase, with no fee for the first 10,000 transactions per month, for every client. After that, it’s around a one per cent transaction fee to make use of Sift’s services.

It’ll be interesting to see if Sift’s method of fraud observation takes off, because if it does, it could seriously bring down the impact of card fraud and its cost to merchants. However, it is relatively new technology, so perhaps some of the bigger fraud prevention firms could catch up.

Having reached out to Tan for comment about how Sift Science’s method has only really become possible recently, he said “machine learning is a relatively new technology that has only really matured to commercialization in the last decade. Google and Amazon have pioneered a lot of this work, but, for the most part, it is still very cutting-edge technology that’s difficult to build. You need to have the right type of software engineers and data scientists to build this stuff in-hourse and most of the good ones tend to already be employed by Google and the like.”

Now it’s just the sticky issue of moving the US over to something more secure than magnetic strips for credit card payments.

Leave a Comment

Current ye@r *