Overview
Warnings are produced both in workflow sessions and by the standalone API (POST /v3/database-validation/). All five risks are tagged with feature DATABASE_VALIDATION, appear on the report under database_validations[].warnings[], and follow the standard warning object shape (feature, risk, additional_data, log_type, short_description, long_description, node_id).
There are no hard auto-decline triggers for Database Validation. The two match-outcome risks, the superseded-document risk and the not-applicable risk map to a configurable action (DECLINE, REVIEW, or NO_ACTION); COULD_NOT_PERFORM_DATABASE_VALIDATION is the one fixed behavior — it always routes the feature to In Review.
Configurable risks
The configured action also drives the warning’s
log_type: DECLINE → error, REVIEW → warning, NO_ACTION → information.
The match outcome itself is aggregated from the per-service validation entries (any full match → full_match; else any partial → partial_match; else no_match) — see Matching methods for the full decision logic that derives 1×1 / 2×2 outcomes.
Warnings produced
COULD_NOT_PERFORM_DATABASE_VALIDATION covers two different situations, and the warning’s long_description (and, for a per-service failure, database_validation_errors[].message) now states which one applied instead of a single generic sentence: the screened data was missing or malformed before any service could be queried, or a service was queried and did not return a usable result. In the second case, when the registry reported its own diagnostic detail for that attempt, that detail is included in the message so you don’t have to contact support to learn why. The upstream service is never named in the response.When
COULD_NOT_PERFORM_DATABASE_VALIDATION or DATABASE_VALIDATION_NOT_APPLICABLE fires in a workflow session, the check is automatically re-triggered once the missing data is filled in on the session detail page in the Console — saving the field re-runs the validation, and the warning is removed on success. This also works when the not-applicable action declined the session, so a declined session can be remediated by correcting the data. For DATABASE_VALIDATION_NOT_APPLICABLE the re-run helps when the document can carry the identifier but it wasn’t read (for example, a residence card whose national ID number wasn’t extracted); document types that never carry the identifier (for example, passports without a national ID number) keep the warning. No charge is applied for skipped services; you only pay when a service returns a billable result.Example
Warning types
Each risk is assigned a severity based on your application’s configuration. Severities fall into three categories:Related
- Database Validation overview — how the feature works end-to-end and supported registries.
- Database Validation report — full response shape including per-service
validations[]. - Matching methods — how 1×1 / 2×2 outcomes (
validation_type: one_by_one/two_by_two) are derived from per-service matches. - Outcome codes — full per-result taxonomy with country-specific detail codes.
- Supported countries & services — every live registry, required fields and per-call price.
- Data models — database validation — canonical field-by-field schema.