Radar Earns Its Stripes Among Enterprise Orgs

Stripe’s first Radar product prevented $4 billion in attempted fraud throughout 2017. Today, April 18, the tech platform for internet businesses is announcing Radar 2.0 — a version of Radar for dedicated fraud teams at the enterprise level.

The Radar machine learning algorithms have spent two years studying hundreds of thousands of businesses’ transactions processed on the Stripe network. That learning not only informed tailored risk defense for the individual companies producing the data — it also prepared Stripe to build a bundle of advanced fraud prevention tools for organizations doing business on a larger scale.

According to a Stripe press release, Radar 2.0 adds hundreds of new signals to help distinguish legitimate customers from fraudulent ones.

For example, there are certain purchase patterns that correlate highly with fraud, which the new Radar can identify as a red flag. Another new signal is proxy detection — that is, a tool that measures the round-trip time between Stripe and the remote browser of a potential fraudster. This can help determine whether the user is leveraging a proxy or VPN to skirt the rules.

Stripe said the new Radar for Fraud Teams improves visibility and offers granular control for identifying and preventing fraud, with optimization opportunities in areas such as faster and more accurate reviews and rich analytics on fraud performance.

Users can create custom rules, risk thresholds and block-and-allow lists. An example of a custom rule would be blocking all transactions over $1,000 if the IP country doesn’t match the payment card’s country. Easier block-and-allow lists enable businesses to track and update attributes, such as card numbers, email or IP addresses that should be black- or white-listed across the board.

Radar 2.0 also reportedly updates and retrains its machine learning models on a daily basis, accounting for the specific use cases of the business implementing the tool. It does this using a cloud-based service that automatically adapts to the latest fraudster threats.