An insurance giant calls on Micropole to reduce car insurance fraud
Learn how our client reduced auto insurance fraud by implementing a machine learning solution that can be easily integrated into the existing system.
Our client is an international player in the insurance and assistance sector, present in 64 countries and with nearly 107 million customers worldwide for a turnover of €24.5 billion. In France, it is :
- 6.3 million customers
- 33,000 employees
- Thousands of dealer partners
The company called on Micropole to combat car insurance fraud and evaluate the contribution of machine learning to fraud detection. The objective was to reduce car insurance fraud from dealerships without altering the customer experience.
Micropole teams used Machine Learning to identify new high-impact rules that could be easily integrated into the system and industrialize the model(s) under Spark, a unified, ultra-fast analysis engine for large-scale data processing.
In a second phase, Micropole teams built a fraud management dashboard using the TIBCO Spotfire analytical solution to monitor alerts and the results of investigations in real time.
- A conclusive experiment that allowed the detection of 30% more fraud without penalizing the number of false positives
- The number of invoices investigated has doubled
- No negative impact on the customer experience
- A new algorithm specifically developed by Micropole, generating directly industrializable rules