Fraud Detection in Home and Motor Insurance Claims
The Claims and Loss Adjusting Director receives thousands of claims each month and needs to automatically detect which show signs of fraud or overcharging in order to send an adjuster in person, while paying legitimate ones instantly.
Where the money comes from
What to do now
- 1Label claims with confirmed fraud from the last 3 years.
- 2Train the model and validate the top alerts with adjusters.
- 3Immediate payment for low-risk claims and an adjuster for high-risk ones.
For your committee
One page with the decision, the impact, the risks, the steps and the KPI to track.
The print dialog opens: choose "Save as PDF".
Watch out: more volume is not always more profit
Investigating too many claims delays payments to honest customers, who leave at renewal. Investigate where the expected saving exceeds the cost of the adjuster and the risk of losing the customer.
KPI for the committee
Loss ratio and average payment time.
Detail for your data team
Proposed technique: Random Forest
Why: Random Forest combines hundreds of rules on timings, amounts and repair shops to detect claim anomalies with great robustness and a low rate of false accusations.
Cost of getting it wrong: Paying fake claims drives up the claims cost; investigating everyone delays payments and raises the operating costs of loss adjusting.
Data needed: Claims Management: Policy start date vs. claim date, amount claimed and description of the damage.; Policyholder and Repair Shop History: Frequency of previous claims, repair shops involved and witnesses.