Ecommerce fraud is expected to cost businesses $20 billion in 2021, up 18% from 2020. Not only is fraud expensive, but fraud techniques and patterns have evolved quickly. Fraudsters are getting smarter, using more complex strategies and tools to scam businesses around the globe.
This is why we built Stripe Radar, a machine-learning-based fraud detection solution fully integrated with the Stripe platform. We recently shipped a series of improvements to Radar to help businesses better prevent fraud and more efficiently manage fraud workflows. Read on for the highlights, or log in to your Stripe Dashboard to see Radar in action.
Detect fraud more accurately with better machine learning
To help you adapt to quickly changing fraud patterns, we’ve tripled the speed at which we update Radar’s machine learning models.
As our network grows, Stripe is more likely to see critical data inputs across multiple attempted payments, making it easier to catch fraudulent usage. And with more data, our models’ recall—a standard measure of machine learning accuracy—has improved by over 20%. This means that Stripe catches more fraudulent payment attempts while minimizing the impact on legitimate transactions.
The updated machine learning models have made it even easier to manage our fraud rates—without any additional effort, we saw our fraud rates drop by half in just a few months.
By leveraging Stripe’s machine learning and custom fraud rules, we’ve seen a decrease in our fraud rates and chargeback rates. We’ve even found our customer satisfaction has improved because our team settles customer disputes faster now that they spend less time managing a manual process.
Take action from a unified fraud workspace
Between manually reviewing payments, responding to disputes, creating new rules, and setting risk controls, managing fraud can be time-consuming. To help, we’ve redesigned the Stripe Dashboard to offer a centralized fraud workspace where you can access reviews, disputes, rules, lists, and risk controls.
With a new overview chart, you now get a clear view of how Radar’s machine learning, rules, and manual reviews are working in tandem to block or allow payment attempts.
Understand how well your fraud strategy is working
Striking the right balance across fraud prevention and conversion is important to every business, but in a vacuum, it’s hard to know if you’re doing well. Radar now provides relevant benchmarks, so you can compare your fraudulent dispute rates, false positive rates, and block rates to other businesses in your region or in similar industries, right from the Dashboard. The aggregated, custom cohorts of businesses powering these benchmarks can help you understand how you may want to adjust your strategy.
Fight fraud with more fine-grained rules and controls
We’ve expanded the rule set for Radar for Fraud Teams users with dozens of additional attributes. With these new rules, you can better customize your fraud setup to your unique business needs and determine which behaviors need more attention. For example, you can now create rules based on the number of names associated with a card, the average amount in attempted authorizations on a card, and even the time since a customer email or card was first seen by your business or on Stripe.
Optimize your rules with detailed analytics
Custom rules are a powerful tool to mitigate fraud. Radar for Fraud Teams users can now get deep insights into rule performance, including how many blocked payments and estimated false positives your rules have generated. This allows you to identify which rules are the most effective.
With new performance metrics per rule, you can optimize your rule sets to block more fraud and ensure minimal impact on legitimate transactions.
It’s great to see the improvements in Radar because they save us time and lead to better performance. Stripe’s fraud solution is very thoughtful and our fraud rates have been consistently trending down.
As always, please let us know if you have any feedback—we’d love to hear from you.