Details for the curious
Open any section for the full detail — figures, design decisions and how it fits your business.
The challenge
Fraud on online loan applications was being caught too late — after approval or funding. The lender needed to stop it before money moved, without blocking genuine customers.
How it works
Every application gets a 0 to 1 risk score and an explanation.
Under the hood: about 100 raw fields expanded to 650+ behavioural signals, a supervised model and an anomaly detector. Drift is monitored, and a retrained model is promoted only if it beats the previous one. Nobody is auto-declined.
What it is worth
Industry research (LexisNexis True Cost of Fraud, 2025) puts the total cost to US lenders at $5.38 for every $1 of fraud lost. Per $1M of direct fraud loss at the 48% baseline:
Client-reported results on 394 online applications, shown without the customer name. The multiplier is a US figure; we validate it for your market.
