AXA UK claim handlers are piloting a machine-learning tool that will support them in making quicker and more accurate decisions regarding motor claims.
RoRI, for Repair or Replace Intelligence, captures the relevant data from the phone conversation and, using machine-learning tools, assesses how much it would cost to fix the car and how much it would cost to write it off; the idea is to take the right decision as soon as the claim is notified.
This comparative approach differs widely from the previous one, which took into account the age of the vehicle and the extent of the damage. In that decision matrix, only half of the total losses were identified immediately, at First Notice of Loss (FNOL); the other half were identified once the vehicle was already at a garage.
This led to delays for customers, which had to be tackled to improve service. From an insurer perspective, it also generated unnecessary storage and credit hire costs.
AXA UK decided to smooth out the motor claims process by providing better support to claims handlers. The focus was on capturing more relevant data but without making calls longer or adding any complexity. It turned out that the phone conversations with customers were much richer than the previous matrix assumed.
RoRI collects and uses more relevant information and, based on an ensemble of predictive models, it compares the cost of repair with the cost of total loss. The former is affected by the type of damage, the parts needed, the labour required, and whether the work is done by an approved repairer. The latter is usually the replacement cost minus salvage.
Those are simplified descriptions; the system is more sophisticated and, with machine learning, models are constantly improved.
The pilot starts on 10 February in the Ipswich office. Afterwards, the plan is to integrate RoRI into the Guidewire software used by AXA Claims later this year.