Fraudulent claims are a major cost for the insurance industry but detecting them can be hit and miss. Agents are typically given a list of terms that could indicate fraud; when they hear one, they are required to flag the case on the system.
Our insurance client wanted to test the effectiveness of an automated approach to fraud detection. We did this by using the same list of terms that the client provided to its frontline agents, we applied speech analytics to thousands of recordings of claim calls. The technology identified four times as many suspect calls as the frontline agents had.
Starting with the same list of terms given to agents, plus detailed call recordings, we built the speech analytics engine. It then reviewed thousands more calls. In the first set, the automated approach identified four times as many potentially fraudulent claims.
Of course, not every alert results in a real fraud case: our client has deliberately cast a broad net to identify a large number of potential fraud cases. The automated solution brings greater consistency to that process – and has even identified real fraud cases that had been missed with the manual process.
That meant the client had to increase the size of its fraud team, but with a strong return on investment, that’s money well spent based on initial results, the approach will lead to annualised savings of £1.6m.
For more details of our analytics services or to discuss how we can help you generate richer insight from your customer interactions, contact:
Director of Insights & Analytics
With a team of more than 550 customer service advisors, Vitality manages over one million calls every…
The main driver for the project was to ‘Balance’ efficiency with a high quality customer experience, a…
When you have a wealth of customer interaction data at your fingertips, it pays to use it.…