Skip to main content

Identity Verification (Onboarding)

Onboarding is the default flow: take an unknown person, prove who they are, and enrol them as a verified Customer. It combines document verification, face matching, and liveness into a single session that typically completes in under a minute.

When to use it​

Opening a bank account, activating a SIM, registering for a regulated marketplace — any moment a Company needs to know a new user is a real, unique person holding a genuine identity document.

How it works​

  1. Document capture — the User photographs their identity document (both sides where applicable). Real-time guidance ensures the image is usable.
  2. Face capture & liveness — the User takes a selfie; passive or active liveness confirms a live person is present.
  3. Automated checks run in the background:
    • The document is OCR-read and checked for authenticity and validity.
    • The selfie is matched 1:1 against the document portrait.
    • Liveness, deepfake, and injection-attack checks guard against spoofing.
    • Optionally, duplicity check searches for an existing or blocked identity.
  4. Decision — the checks combine (via Trust Factors) into an outcome and a result is returned to the integrator.

Outcomes​

OutcomeMeaning
New customerTrusted and unique — a new Customer record is created.
Merged / UpdatedTrusted and matched an existing Customer — the attempt is linked to them, optionally refreshing their data.
ReviewAmbiguous — sent to the back office for an operator decision.
RejectedIdentity not proven (poor capture or fraud).
BlockedMatched the Blocklist, or rejected and added to it as fraud.
IncompleteThe User did not finish.

Duplicity check during onboarding​

When 1:N is enabled, onboarding disambiguates common situations by combining a text signal (name + date of birth) with a face signal:

SituationTextFaceTypical outcome
Stolen ID used by someone elseSameNo matchReject
Fabricated ID, same personDifferentMatchReject
Name change or twinSame DOB, different nameMatchReview
NamesakeSame name, different DOBNo matchNew customer
Returning customerSameMatchMerge to existing
Different personSame DOB, different nameNo matchNew customer

See Face Duplicity check and Document Duplicity check for how these signals are computed.

See also​