Stripe Radar’s machine learning models use many signals to distinguish between fraudulent and legitimate payments. We compute some of these signals automatically, but many of them depend on the information that your integration provides. In general, the more data your integration provides, the more successful fraud prevention can be.
If you don’t collect enough information from your customers, it can reduce the effectiveness of fraud detection. Conversely, if you collect too much information, it can negatively impact the checkout experience and result in a lower conversion rate.
Stripe Radar leverages the power of the Stripe network to effectively detect and block fraudulent transactions, regardless of how you integrate with Stripe. However, the way you integrate with Stripe to process payments can significantly impact the completeness of the fraud signals you send us. The more information you send about a payment, the better Stripe Radar is at detecting and preventing fraud. Using one of our recommended payment integrations allows you to get the most out of Radar. If you can’t use a recommended integration, consider including as much additional data as possible, as explained in our recommendations below.
|Integration type||Radar integration quality|
|Stripe Payment Links Recommended|
|Stripe Checkout Recommended|
|Stripe Elements with Customer signals Recommended|
|Direct API integration with Radar Sessions and Customer signals|
|Direct API integration with client and Customer signals|
|Direct API integration with client signals|
|Direct API integration with Customer signals|
|Direct API integration with no additional signals|
Important signals to send to Stripe
Including the following information with your payments can have a significant impact on the performance of Stripe Radar’s fraud detection models. Our recommended integrations enable you to collect this information, while direct integrations might need to explicitly include this data.
|Data||Estimated fraud model improvement |
|Advanced fraud signals||36%|
We’ve tested the following recommendations to make sure that your conversion rate remains high while maximizing the performance of our machine learning models.
The most important action you can take to guard against fraud is to collect customer payment information using one of our recommended payments integrations. Each method automatically collects important high-signal data, such as device information and IP addresses. To further improve fraud detection, collect the cardholder name, full billing address, postal code, and the card’s CVC code during checkout.
You can build a seamless checkout flow within your website or app using any of these methods, and securely pass sensitive card information directly to Stripe without passing it through your servers—greatly simplifying your PCI compliance. Determine which integration makes the most sense for your business and product goals, but any of these integration methods help optimize your integration for fraud prevention.
Using Customer objects when creating payments allows Stripe to track the payment patterns for each customer over time. This significantly increases our ability to identify irregularities in purchasing behavior. To do this, you should:
- Save cards and billing address to
Customerobjects and use them to create subsequent payments.
- Provide your customer’s email address when creating a
- Provide your customer’s name when you tokenize their card information.
- If you ship physical goods, we also recommend collecting the customer’s shipping address and saving this to their associated
Customerobject can also store multiple payment methods, so you can enhance your customer’s checkout experience by offering to save multiple cards. Stripe can continue to track payment patterns for each customer, regardless of which one they use.
- Save cards and billing address to
Include Stripe.js on every page of your site, not just the checkout page where your customer enters their payment information. By doing so, Stripe can detect anomalous behavior that may be indicative of fraud as customers browse your website—providing additional signals that increase the effectiveness of our detection.
<script async src="https://js.stripe.com/v3/"></script>