(Image source: Porsche.)
Many insurance companies today, personal and commercial alike, have engaged in usage-based insurance underwriting. There are two ways of collecting the use information from the vehicle: One is using a telematics datalogger and the other is using a personal smart phone with an app running in the background. Both work reasonably well for collecting miles driven and time of day; however, if the underwriting includes any form of driver behavior data, the hard mounted device provides a superior data set.
Now, the real question is what data set provides the best correlation to actual underwriting risk. Assume that it was possible to generate a valid driver behavior risk score in conjunction with miles driven. Would that data set have more value in the underwriting process and is it cost effective? Naturally, the total cost of using an app on the user’s smart phone will have the lowest possible data collection cost. But what if the data cost of the valid driver behavior risk score was only incrementally higher, would it be the obvious choice?
One of the problems with the detailed driver behavior risk score is that it requires high quality data processed by capable servers. Therefore, until the vehicle can provide quality accelerometer data from the head unit we have to rely on plug-in devices to collect the data. Current plug-in devices are predominately uploading data via an internal cellular modem. The cost of the modem and the cellular transmission makes this solution relatively expensive. There are several vendors currently working on cost-optimized Bluetooth only devices where the datalogger cost is close to half of the cellular datalogger and the data will be transported by the user’s smart phone. This will enable high quality data collection at minimum cost.
For commercial vehicles, the existing telematics dataloggers can have a dual role as both GPS tracking/operational efficiency tools and as data collection for insurance underwriting purposes and hence the cost burden is acceptable. For the personal insurance market, the jury is still out what the right implementation should be. The smart phone app is a great marketing tool for the insurance companies, but the quality of the collected data is very poor. Recent articles on the subject suggest that the UBI path going forward should focus on easy to use smart phone apps; however, the challenge with that is data quality. Smart phone apps cannot collect motion related data (driver behavior characteristics) and barely get actual miles driven or time of day.
Although the most refined apps have worked out major kinks and no longer drain smartphone batteries or require policyholders to press “start/stop” to track every trip, the data quality from the smart phone is not good enough. The smart phone’s obvious inability to always be present and actively recording when the car is driving leads to significant data gaps. Any form of true driver behavior can not accurately be collected via smartphone. Therefore, one increasingly plausible option is that a combined solution of low cost Bluetooth enabled data collection device and a smart phone app will satisfy both quality data and customer intimacy for personal lines insurance.
When UBI Self-Selection Ends
Mobile UBI advocates tout that policyholders can simply download the free branded apps from the app store, virtually eliminating the upfront equipment costs for insurers. While this is true, what is not mentioned in these articles is that UBI is still in the voluntary or self-selected phase, meaning policyholders elect if they want to participate or not. People who are willing to use the smart phone app are already good drivers that would most likely volunteer for a UBI program in the first place. Once this self-selected honeymoon phase wears off, and we get past the initial pool of good drivers, you need data accuracy. We have not gotten down to the population yet that have driving issues because they don’t opt in to the program, but once they do, the poor data collection devices will face a serious challenge in proving underwriting efficacy.