Create account
Sign in
Home
Payments
Business operations
Financial services
Developer tools
Security
All products
Home
Payments
Business operations
Home
Payments
Business operations
Financial services
Developer tools
Support
Overview
Overview
Risk evaluation
Risk settings
Reviews
Risk insights
Uncaptured payments
Lists
Rules
Testing
Checklist
Disputes and fraud
Climate
Identity
radar
ยท
HomeBusiness operationsFraud detectionReviews

Risk insights

Understand risk factors and details about a particular payment.

Stripe Radarโ€™s adaptive machine learning system determines the risk score and risk level for a payment and uses them to decide when to block or mark payments for review. The system evaluates hundreds of signals about each payment, using data from Stripeโ€™s network across millions of businesses. The risk insights feature, available with Stripe Radar for Fraud Teams, identifies some of the highest impact signals that power Radarโ€™s machine learning system, to clarify why Radar scored a payment a particular way.

Why are certain fields not available?

If youโ€™re not seeing the customer information or locations that you expect, check that your integration follows Radarโ€™s best practices. If your integration doesnโ€™t provide important details like the cardholderโ€™s email address, IP address, or shipping address, Radar canโ€™t compute all of the data it needs to accurately evaluate each payment.

The Top Fraud Factors section notifies you with fraud signals when the payment has values that commonly indicate fraud. If you donโ€™t see any of the top fraud indicators for a payment, then fraud wasnโ€™t detected. Because Radarโ€™s machine learning detects complex patterns across hundreds of signals, itโ€™s still possible for a charge to be correctly identified as fraud, even if none of the signals appear suspicious on an individual level.

Risk insights also includes information about the customer, such as matching the cardholderโ€™s name with the provided email, and the success rate of transactions on the Stripe network associated with the email address. A low authorization rate may indicate suspicious behavior, because previous declines sometimes suggest past attempts at fraudulent transactions.

We also highlight geography-based information, including the billing, shipping, and, IP address locations associated with this payment.

Risk insights dialog

If you want to see a more complete list of Radarโ€™s highest impact signals, click the Show all insights button from the risk insights section. This opens a dialog with several of the most important signals to Radarโ€™s machine learning engine.

Understanding fraud factors

Why don't I see any fraud factors?

The data used to populate top insights and fraud factors is only populated for charges made within the last 6 months. This feature isnโ€™t supported for payments in test mode.

Some of the signals in the risk insights dialog have badges with numbers next to them. These badges show the fraud factor for a signal on this payment. A fraud factor represents the likelihood of fraud for charges with a value similar to this signal when compared to the average transaction on Stripe. A fraud factor of 3.5x means that charges with a similar value for this signal are 3.5 times more likely to be fraudulent than average. In a higher risk payment, we expect to see some fraud factors greater than 1, and in a lower risk payment we expect to see some fraud factors less than 1.

Hover over a fraud factor to see more information about the possible values for it. These factors will change over time as the data in our network changes. This data provides context for the distribution of fraud factors for a signal. This dialog also provides the network distribution of values for a signal, letting you know whether the current payment has a value thatโ€™s common or if itโ€™s rare or unique in the Stripe network.

Related payments

You can also view the network of related payments, which includes any other payments made to your business using the same email address, IP address, or card number as the payment youโ€™re currently viewing. This can help identify common fraud patterns, such as card testing (many different cards sharing a single IP address) or trial abuse (many โ€œcustomersโ€ share the same card).

Next steps

  • Review
  • Integration checklist
Was this page helpful?
Questions? Contact us.
Developer tutorials on YouTube.
You can unsubscribe at any time. Read our privacy policy.
On this page
Risk insights dialog
Related payments