Skip to main content
Version: Guardian v0.5.3

Risk Levels

Every verify() call returns a riskLevel and an allowed boolean. These are independent: a transaction can be allowed: true with riskLevel: high.

Levels​

low
Normal risk

Transaction looks safe. No signals flagged.

medium
Elevated risk

Some signals flagged. Consider additional verification.

high
High risk

Strong fraud signals detected. Can still be allowed depending on scenario configuration.

blocked
Blocked

Declared in the SDK's RiskLevel type. The backend does not return it.

unknown
Unknown

The SDK has no risk level from the backend. You get it when failurePolicy is allow and verify() failed. In that case allowed is true.

info

risk_level and allowed are evaluated independently by the backend. The risk level reflects the fraud signal assessment, while the allowed decision is based on your organization's scenario configuration. A high risk transaction can be allowed: true if no rejection scenario triggers.

blocked and unknown exist only in the SDK. The backend returns low, medium, or high. With failurePolicy set to allow, a failed verify() returns allowed: true, riskLevel unknown, and the error message in errors. See Configuration.

Decision Flow​

verify() returns result

allowed: true
low
medium
high
unknown

Transaction accepted. The risk level tells you how much friction to apply:

  • low: proceed normally
  • medium: proceed with caution, consider step-up verification
  • high: accepted with strong fraud signals, add friction or monitor
  • unknown: the call failed and failurePolicy is allow
allowed: false
low
medium
high

Transaction rejected by your organization's configuration. The risk level can be any of the three.

  • verify() does not return the reason for the rejection
  • The server-to-server flow returns it in result.details