Disrupting the Auto Insurance Market

​

In the race towards digitisation, an auto insurance start-up in Thailand engaged PulseMetrics in 2017 to assist with its new product development. Traditionally, the prices of auto insurance products are determined by three factors—the car make, the car model, and the age of the car—nothing about the driver and the risk-level. This led to a problematic phenomenon of drivers gaming the system by switching among insurance companies to maintain their insurance risk rating and claims allowance. Faced with this loophole, our client wanted to be the first to market by providing personalised auto insurance products that incorporated a holistic customer profile. 

 

The only way was to make it data-driven, predicting customers’ chances of claim through advanced analytics. To this end, our consultants were tasked to create more features that can determine customers’ insurance risk so as to offer better prices for drivers of varying risk types. One of which is based on the customers’ place of residence. Given that the business would be done on a digital platform, our consultants devised a strategy to extract useful insights from the address that customers provide when filling in their particulars.

​

Measuring Risk through Geolocation Analysis

​

Using Google Maps Application Programme Interfaces (APIs), we decoded the address and determined the exact location of the customers’ place of residence. By taking the mass transit stations as points of reference, regression analyses were drawn to determine the relationship between the accessibility to public transport and claim rates. The team successfully revealed that the claim rates were higher for customers who were staying further away from mass transit stations in central Bangkok, for they require driving more than peers who had better access to public transport. 

​

In order to assign a justified measure of insurance risk to customers, we then designed a comprehensive scoring model that accounted for the average claim per year, the average period where no claims were made, the average tenure, the number of claims and the claim amount.

​

Driving Home with the Optimal Auto Insurance Product

​

Because of these innovative algorithms, our client could attract its target customers with a smarter pricing and product design. It was able to offer personalised discounts immediately after receiving the customers’ addresses and history of claims. Using external data, our consultants enriched the profiling of customers’ risk-level, thus making it more appealing to target customers with lower insurance to purchase these personalised auto insurance products. With the acquisition of the right customers, our client can minimise the cost of claims, thereby successfully hacking the start point of the customer journey.