Age Verification
Age verification confirms that a person meets an age threshold — without necessarily enrolling them as a Customer. It runs against a facial age estimate first (where the law allows) and falls back to a document check when the estimate is not conclusive.
When to use it
Gating access to age-restricted goods or services — alcohol, gambling, adult content, certain financial products. The person can stay anonymous; the platform only needs to answer "is this person old enough?".
How it works
The predefined age_check_v1 workflow evaluates two Trust Factor configurations in two steps:
- Face capture & liveness — the User takes a selfie.
- Age estimation — the
age_estimationTrust Factor compares the age estimated from the selfie with a threshold that includes a safety margin (default: reject below 25). If it passes, the workflow ends accepted here, with the reasonestimated_above_threshold, and no document is needed. - Document capture — otherwise the User captures an identity document (identity card, passport or driving licence).
- Age verification — the
age_verificationTrust Factor computes the age from the document's date of birth and compares it with the legal limit (default: reject below 18). A document without a readable date of birth fails this check. - Decision — the outcome is returned; a document that fails the OCR quality check after all retakes is routed to review.
Both thresholds, and whether the estimation step is used at all, are configured in the Trust Factor configurations the workflow references — default_age_estimation_trust_configuration and default_age_verification_trust_configuration in the predefined workflow. Fork the workflow to change the document types or to drop the estimation step where estimation is not legally acceptable.
Outcomes
| Outcome | Reason | Meaning |
|---|---|---|
| Accepted (estimated) | estimated_above_threshold | The facial age estimate cleared the threshold with margin. No document was captured. |
| Accepted (verified) | verified_by_DOB | The document's date of birth confirms the person is old enough. |
| Rejected | trust_reject | The document age is below the limit, or another Trust Factor (liveness, authenticity) rejected the data. |
| Review | ocr_quality_retry_exhausted | The document could not be read reliably after the allowed retakes. |
| Incomplete | — | The User did not finish. |
Notes
- The estimated path is anonymous — no Customer record is created. On the Stored tier the verified path of the predefined workflow creates a Customer from the document; remove the outcome actions in a forked workflow if you want the document path to stay anonymous too.
- Age estimation is probabilistic; the default margin of 25 years for an 18+ check reflects its error band. The legal acceptability of estimation-only decisions depends on the jurisdiction.
- The accuracy of the age estimation technology is independently benchmarked by NIST in the FRTE Age Estimation & Verification evaluation. The Innovatrics algorithm is listed as
innovatrics-003; the Challenge 25 results show how reliably it separates people under and over 25 — the same threshold the predefined workflow uses by default. - To verify age and onboard the person, use Identity Verification instead — there the
age_validationfactor also checks that the selfie age is consistent with the document.
See also
- Face detection & auto-capture
- Trust Factors & Decisioning
- Predefined Workflows
- NIST FRTE Age Estimation & Verification — independent benchmark; see algorithm
innovatrics-003