Overview

Revolutionizing personal payments

The Curve app, paired with a smart bank card, combines all your cards in one and works like a normal bank card anywhere in the world that accepts MasterCard. Users don’t need to open a new bank account and don’t need to wait for weeks — after a few taps and a new Curve card in the post, they’re ready to go.

With Curve, customers benefit from: simpler spending across all their cards, saving money on currency exchange, a smarter and faster way to manage expenses, instant cashback rewards at over 50 leading UK retailers, plus a host of additional security features. Curve users have so far spent £50 million in over 100 currencies, worldwide.

Challenge

Costly manual resources

With a rapidly growing customer base, thousands of good users, and an increasing spend volume to boot, Curve can also pick up the attention of fraudsters. The ability to quickly add new cards with the Curve app and then use them within moments is of particular interest to malicious users who try to circumvent Curve’s many layers of account authentication, resulting in expensive manual resources for fraud reviews and chargeback management.

Because their user base is ever-growing, Curve works hard to understand who their customers are and ensure that the card users are who they say they are. Curve has identity verification methods in place, and various teams within the company investigate new and existing users. However, with business growing at such a pace, Curve has taken to preemptively knocking out the growing threat of fraud.

Solution

A layered approach to preempt attacks

The Curve team began looking for a fraud tool, focusing on a machine learning solution to avoid having to spend precious resources on building an in-house rules engine from scratch. After reviewing several vendor options, Curve decided to go with Sift and quickly integrated the solution. Headed by Product Operations Manager Rona Ruthen, Curve first implemented the basic Sift Score API, and soon began using Sift’s findings to assess real-time transactional data.

Within weeks of training the models and learning how to use Sift effectively, Rona and her team began to trust the accuracy of Sift Scores. Although Curve’s business model was new to Sift, they found that the solution quickly learned what fraud – and not-fraud – looked like for Curve, and was able to pinpoint bad users early and efficiently. Rona and the Curve team now take a layered approach to fraud management, relying on an in-house rules engine to weed out the hard-and-fast business blacklist while utilizing Sift to spot the trickier fraudsters, ideally before they even get a Curve card.

Rona Ruthen
“Sift had a very positive effect in days. Other solutions aren’t as real-time as the Sift solution.”

Results

More time for a deeper understanding

Although business is booming for Curve, since implementing Sift, they’ve only had to add one full-time fraud resource. Instead, Sift enables streamlined fraud workflows with a single platform for automation, review, and investigation, and fraud management no longer takes a whole team. Better yet, with Sift Curve has seen their chargeback rate drop to 1/6th of what it used to be.

Now, Rona and Curve can stay ahead of bad users by looking at accounts blocked with Sift to quickly identify connected bad users. Besides the financial savings in thwarted chargebacks, automating on Sift Scores means that Curve has more time and better data to do in-depth investigations on the fraud cases they encounter.

Rona Ruthen
“The solution is doing a great job! Having some flexibility in defining the Workflows and setting them up in a way that allows us to fine-tune it to stop what we want it to stop without impacting good customers is an incredible value.”