Circa 70% of the car insurance customers to the UKs biggest price comparison site now have more accurate premiums.
The Situation
A colleague was moaning one day about the cost of his car insurance. I asked for his details, replicated his answers and started playing with his answers to see what I could do. I asked him about his 15K annual mileage. He was adamant this was correct as he was about his car purchase date. Using a few known open data sets that expose historic mileages and a commercial data set that exposes purchase dates, I discovered quite quickly that Mr Volvo was wrong on both and would pay £150 too much. Nothing about his driving circumstances had changed. This got me curious; surely this wasn’t unique to Mr Volvo 240. I started to dig a little further as I could sense a gold nugget.
The Task
My first step was to get my hands on a little data for analysis. So I sourced over 100 million MOT records; a similar number of purchase date records and performed a few simple checks on what customers input and what the facts said. The results of the analysis were pretty conclusive in terms of my hunch that there was a gold nugget or two to be had. It would appear that circa 70% of UK car drivers do not provide an accurate annual mileage and circa 40% do not put in an accurate purchase date. The premiums for these customers would almost certainly have been impacted as could the validity of the policy. I came up with a solid business case where I said I could definitely help with the accuracy of data provided for millions of customers. I also suggested that if conversion metrics mirrored those 30% who we found to get their mileage accurate, there could be a multi-million pound increase in revenue, but I caveated this massively!
The Action / Approach
I decided that I wanted to build a data product that could initially answer the annual mileage and car purchase date. The data product would be wrapped in a Google Cloud Platform API and called during the web and mobile app customer journeys. Data was sourced from various Open and commercial data API/sets (real-time). Necessary rules and calculations were performed and results presented back to the customer. The customer would be informed how many miles they had previously done and purchase date and given the option to amend the details. Both features were tested together using Optimizely (an experimentation platform) on the live Car Insurance channel to a percentage of customers. The results from the test supported rolling the changes out to 100% of traffic.
The Result
Circa 70% of the car insurance customers to the UKs biggest price comparison site now have more accurate premiums. In terms of doing the right thing for customers, this has been a great result. The revenue aspirations are still to materialise. The product change to car insurance remains live and will be copied by competitors and Insurers alike. A really basic and simple piece of data innovation, benefiting many parties and I am proud of the achievement. Sometimes small things and being a little curious, can make a big difference.
