Distraction is Now Measurable and Thus Improvable Thanks to Telematics

Although the risk of texting, and more generally, hand-held phone motion while driving is well documented, recent telematics-based research indicates that both handheld and hands-free calls while driving also correlate to losses.

In fact, many telematics factors are causative – and all are correlated – with risk. Additionally, most are user-controllable, and all are non-discriminatory.

Actuaries and data scientists vet new risk factors for insurance pricing by conducting univariate and multivariate analyses on losses. By measuring billions of miles driven each year they are able to evaluate the predictive power of distraction rating factors on future insurance losses.

Phone distraction risk factors are built from the ability to detect distraction events precisely and then extract risk features. These factors include the event duration as well as its intensity, which is a factor of the type of distraction. Events can be a time while the phone is being moved around in the hands of the driver, if the driver is interacting with the phone such as texting, or if phone calls are incoming or outgoing.

From the data collected, the distraction risk factors are developed using machine learning algorithms and their predictive power is evaluated in partnership with insurance carriers who provide external expert validation using claims data.

Using only a subset of phone-distraction risk factors, Cambridge Mobile Telematics – the global leader in smartphone telematics – demonstrated the potential lift on a frequency model. The company’s actuarial team observed that the average claims frequency increases from best to worst drivers and the 10% of most-distracted drivers have a loss frequency that is 2.2 times the 10% of least-distracted drivers.

As the insurance industry strives to become more transparent and fair, new scoring factors bring solutions that satisfy both the product and the actuarial teams. Many telematics factors, especially phone distraction, are not only predictive, but they are also causative, not merely correlated with crashes. Additionally, they are controllable by the user to improve their score, unlike factors such as age or location. By surfacing phone distraction rating factors in a telematics-based insurance program, insurers can communicate effectively around safety, actively coach drivers by rewarding improvement with higher discounts or lower premiums, and, by the same token, reduce claims frequency and severity.