Revolut Leverages SEON’s Anti-Fraud Platform With Great ROI

About

Revolut needs no introduction these days. The Digital bank juggernaut and fintech trailblazer operates in 39 countries and has over 30 currencies, allowing 45m+ customers worldwide to pay, get paid, invest, exchange currencies and manage money all in one app.

Revolut’s frictionless onboarding, immediate access to financial services, and ease of use for managing money and trading are what make it so popular with customers. Fraud is an issue faced by the whole industry and Revolut has made significant investments in its systems, processes and people to ensure that it protects its customers as it scales and grows.

Challenge

As a major player in the financial sector, Revolut faces the ever-present threat of fraud, a challenge that affects the entire industry. The company is dedicated to safeguarding its customers while expanding its global operations. To scale effectively, Revolut needed a solution that could enhance its risk models and provide better fraud prevention without compromising the customer experience.

Dmitri Lihhatsov, Fincrime Product Owner at Revolut, explains how the company fights back:

‘We have an incredible FinCrime team, continually updating and improving our strategy around keeping Revolut and our customers safe. Our data scientists are currently working on improving our global risk model because we have different configurations fo the United States, Great Britain and worldwide users. All this is part of our unique approach to fighting financial crime and fraud at Revolut.’

Dmitri Lihhatsov, Fincrime Product Owner at Revolut, and his team were looking for more data to bolster their machine-learning models and support their proactive approach to fraud prevention.

Solution

And the key to better risk models? Access to better information. This is precisely why Dmitri and his team turned to SEON, exploring how enriched data could fit into their meticulous strategy.

“We use raw SEON data alongside our custom-developed features to train our machine-learning models. For instance, in addition to social and device signals, we started looking into the CNAME data to ensure we fully leverage SEON.”

Results

The impact of SEON’s enriched data was clear. Even with Revolut’s already fine-tuned fraud prevention models, incorporating SEON’s data led to a 2% improvement in accuracy. While this may seem modest, those final increments were crucial in significantly reducing fraud cases that would have otherwise gone undetected.

“SEON improves our accuracy by an extra 2%. That doesn’t sound like a lot, but we are talking about 2% of fraud cases we would be losing otherwise. And it’s often the last few points that are so hard to win as we fine-tune our models.” 

More importantly, SEON allows Revolut to reach incredibly high rates of prevention accuracy while helping the company manage its OpEx costs.