Skip to main content
ShieldLabs provides 99.9% identification accuracy and 99.9% risk signal detection accuracy. They describe different capabilities:
  • Identification: recognizing a returning device and distinguishing it from other devices.
  • Risk signal detection: recognizing conditions such as VPNs, proxies, anti-detect browsers and automation.
Neither figure means that 99.9% of flagged accounts committed fraud. A detected VPN can belong to a legitimate customer.

​
How accuracy is measured

Identification and detection require different tests. Evaluate returning-device matches and false matches separately. For risk signals, compare each detection against known conditions in your test cases; report false positives and false negatives by browser, device and network.

​
Read the evidence

Use the Risk Score together with the named detection flags, account history and the protected action. VPN (30) and Anti-Detect Browser (60), for example, produce a score of 90. The result explains the observed conditions; your account and transaction context determines the appropriate response. High-Risk Events have their own Medium or High confidence. An event’s confidence is not a percentage or an identification accuracy estimate.

​
Validate your integration

  1. Check returning-device behavior across your supported browsers, network changes and storage resets.
  2. Confirm the Identification ID reaches your backend and retrieves the expected domain and account.
  3. Test missing, incomplete and unavailable results separately from low-risk results.
  4. Review false positives with your own legitimate traffic before enforcing a new sensitive-action policy.
See Browser support, Identifiers, Risk signals and Acting on results.