Detect claim anomalies
Use ClaimDS Smart to surface unusual claims — outliers and odd patterns worth a second look — computed from your own data, before they settle.
Most claims are routine, which is exactly why the unusual ones are easy to miss. Smart anomaly detection surfaces the claims that don't fit the pattern — outliers and oddities — so a mistake gets a second look before it settles.
Flagged, not judged
Like everything in Smart, the figures are computed from your own data and the AI only explains them. A flagged claim isn't an accusation — it's a prompt to check, and many turn out to be perfectly valid in context.
Find and review
Follow the numbered steps below. Smart features are part of a paid tier — if you don't see anomaly detection, check your plan.
Step-by-step
Open anomaly detection in Smart
Go to the Smart area and open anomaly detection. It scans your claims for outliers and unusual patterns — the ones that don't fit the norm and are worth a closer look.
Review what's flagged
Read the flagged claims and the plain-language explanation of why each stood out. The figures are computed from your own data; the AI only explains what makes a claim unusual, it doesn't decide it for you.
Investigate the genuine ones
Open a flagged claim and check it against its agreement and supporting documents. Some anomalies are perfectly valid once you see the context; others are the early sign of an error worth fixing.
Act through the normal flow
Where a flagged claim is wrong, correct or dispute it through the usual claims flow. Anomaly detection points you at what to look at; you act on it the same way you would any claim.
Frequently asked
Does it block claims automatically?
No. It flags claims for review — it doesn't reject anything on its own. The decision stays with you.
Why is a normal-looking claim flagged?
An anomaly just means a claim is unusual relative to your data, not that it's wrong. Many flagged claims turn out fine once you see the context; the point is to make sure they're looked at.
Still stuck?
Book a demo and we'll walk through it on your own data — or just talk to us.