Fraud Detection · Documentation
User Journey
The end-to-end path through Fraud Detection, from trigger to outcome.
← Fraud Detection- 1
Booking → signal → disposition
A subset of shared bookings carry a risk signal (high-value bookings more often). Scoring computes score = min(100, 30 + reasons×16 + jitter) and names the rules that fired; a score at or above 75 auto-holds. The reviewer triages by score, and the owner sets a disposition. Re-scoring emits fraud.flagged.
- 2
Escalation → dispute → reversal
When the owner escalates a confirmed signal, a linked fraud-reversal refund case opens in #21 — pre-recommended approve, flagged as a dispute, carrying the originating signal id — and dispute.opened is emitted while the signal is marked escalated with the linked case id. A confirmed signal can also motivate a settlement hold in #34. The loop closes against the same bookings the rest of the ecosystem reads.