Radar’s algorithms evaluate payments for suspected fraud risk and take action accordingly. Radar blocks high risk payments by default and provides additional fraud tools (if you have Radar for Fraud Teams) so that you can specify your own criteria to block suspicious payments.
Attempted payments account for all card payment requests screened by Radar, including retried payment attempts on the same purchase.
Blocked payments represents the number of attempted payments that Radar blocked. Payments are broadly blocked by Radar for two reasons:
Block rate is the percentage of attempted payments that were blocked by Radar.
Volume, blocked payments is the monetary value of attempted payments that Radar blocked. (The volume shown is in your default currency, using estimated conversion rates for payments from other currencies.)
Stripe may block a payment for other reasons not included here (e.g., payments initiated by cards on deny-lists that are globally known to be fraudulent or payments made from sanctioned countries).
Additionally, SetupIntents—which let you save customer credentials for future payments—aren’t accounted for here even though they’re screened by Radar.
The following section contains two views that help you understand changes to your block rate over the selected time period, along with the proportion of payments blocked by both Radar’s machine learning model and your block rules.
Radar — High risk score accounts for the number of blocked payments that were blocked due to high risk, their total monetary volume, and the corresponding block rate (out of all attempted payments). These are payments that received a Radar risk score greater than your high risk threshold and were consequently blocked by the default high risk block rule.
The estimated false positive rate is the estimated probability that a non-fraudulent payment was incorrectly blocked by Radar’s machine learning. This is derived using a combination of the Radar risk level of these payments and global experiments across all payments on the Stripe network.
Radar — Rules similarly, accounts for the number of blocked payments that were blocked by one of your other block rules, their total monetary volume, and the corresponding block rate (out of all attempted payments).
Depending on your business needs, your block rate or false positive rate, you may want to adjust the amount of fraud blocked by Radar’s machine learning. Radar for Fraud Teams users can adjust the risk threshold (75 by default) at which payments are blocked in their risk settings. As you increase the risk score at which you block, you’ll allow more overall payments through but you may also allow more fraud. As you decrease the risk score where you block, you’ll probably block more fraud but also block more overall payments.
Closely monitor your fraudulent dispute rate and disputes activity to understand the impact of changing risk thresholds. In general, follow Radar’s best practices to ensure your integration makes the most of Radar’s machine learning models.