Manual review

A human looking at the cases the system could not decide.

When the automated checks are unsure, the case goes to a person who looks at the evidence and decides whether to approve or decline.

Analyst adjudication of queued cases with full evidence and audit capture. Adjudication means making a final decision on a case, and audit capture means recording what evidence the analyst saw and what they decided.

A manual review typically works like this:

1. The automated system cannot confidently approve or decline a case and places it in a queue.
2. An analyst opens the case and sees the evidence, such as document images, selfie results, and risk signals.
3. The analyst decides to approve, decline, or request more information.
4. The system records the decision and the evidence for audit.

Cases usually reach manual review for a few reasons:

• Borderline scores, where a model or rule is unsure.
• Conflicting signals, where some checks pass and others fail.
• Unsupported or unclear documents, which the system cannot fully read.
• High-risk customers, where policy requires a human decision.

Manual review sits next to terms that are easy to mix up:

• Case management is the workspace where analysts handle queued cases.
• Straight-through processing is the share of cases decided with no human at all.
• False positive is a flagged honest customer, and review can rescue them.