All insights

Customer risk scoring without black-box theatre

Nov 14, 2026

Scores must be governable: factors, overrides, and review cadence — not mystery numbers.

What a useful score does

A customer risk score should help prioritize work and drive proportionate controls. It should not pretend to be a verdict, a moral judgment, or a substitute for investigation.

Useful scores answer operational questions: who deserves attention first, which customers need enhanced due diligence, and where monitoring intensity should rise or fall.

Factors, not folklore

Documented factors — geography, product, channel, behavior, adverse media signals, relationship complexity — beat opaque model mystique for regulated ops. Stakeholders need to know what moved a score and whether that movement aligns with policy.

Overrides are inevitable. They need owners, reasons, and expiry. An override without governance is just an informal exception factory.

Keep scoring next to cases

When scoring lives beside screening and investigation, analysts can challenge a score with evidence instead of fighting a separate system. That adjacency also helps model owners see where scores create noise or miss risk.

Scoring that cannot be inspected in the same console as the case becomes politics.

Review cadence matters

Scores decay. Customer behavior changes. Products launch. Sanctions landscapes shift. A score without a review rhythm becomes a stale label that teams stop trusting.

Define who reviews factor logic, how often, and what evidence triggers an out-of-cycle change.

Fatcon’s posture

Fatcon treats scoring as part of the decisioning workspace — visible, challengeable, and connected to cases — not a black box bolted on for marketing. The goal is governable prioritization for banks and regulated fintechs.