Tim* decided to bundle his auto insurance in with his home and life insurance. Getting it all from one carrier — in this case, a financial institution — made sense: One bill, one payment, and the prospect of savings for the bulk package.
Then the customer service representative proposed an added measure for helping Tim save on his auto premium: Let us put a device in your car that measures your acceleration and braking, and we’ll give you a discount. Tim was apoplectic. “I have enough intrusion into my life. I’m not letting someone monitor the way I drive.”
The truth is, many providers are offering just such deals: Let us monitor some key performance indicators about your driving so we can better assess your risk. Commonly known as usage-based insurance or pay-as-you-go insurance, the intrusion is balanced by the value received in reduced premiums.
And while many consumers understand the aspect of telemetry involved — the remote collection of data from a vehicle — few know what happens at the backend, when the data is received and processed.
The link between driving behaviour and lower premiums
Norman Black has a unique perspective: He not only has a connected auto insurance policy, but he’s also the insurance solutions director for analytics and artificial intelligence (AI) software giant SAS Institute.
His car is connected to his insurance company by technology loosely referred to as the Internet of Things (IoT). Essentially, all the connected devices that send measurements in real-time over the internet.
Black’s rationale for choosing to connect: He lives in London in the UK, and doesn’t drive for his commute, only for leisure. He’s coming up on 60 years of age. He has a son about to become driving age, “and I didn’t want to take out a loan from the IMF to pay for his insurance.” Having insurance based on driving performance rather than actuarial tables made sense.
Europeans are taking to IoT insurance faster than North Americans. Those in the UK are laggards, at about 5% to 6% of policies issued. Italy is a leader at 19%, Black says. IoT options are available for home insurance as well, with companies providing devices that monitor for theft, flooding, and other household threats.
It could even be extended to life and health insurance through wearable devices, Black says. “There’s no end to what you can imagine. In theory, there are benefits to both sides.”
How can analytics make driving safer?
For drivers and homeowners, rates are based on their real-time behaviour, which they can control. For insurers, it moves the needle from payment-based — shelling out when something goes wrong — to prevention, which has long been the model for commercial insurance. While motorists in Ontario who sign up for a telematics program can earn a discount for consistently exhibiting good driving behaviour, drivers who demonstrate risky driving behaviour could possibly pay more as of November 2020. Risk-management of this type on a consumer level hasn’t been possible until recently.
Because of the IoT-networked world and the power of AI, it now is. It’s anchored in the field of analytics. Growing out of the field of business intelligence, analytics falls into four categories:
- Descriptive analytics, which explains what has happened from a dataset
- Diagnostic analytics, which allows troubleshooting of processes
- Predictive analytics, which estimates results according to various changes in inputs
- Prescriptive analytics, which helps choose alternatives to produce more optimal results
AI is essentially a feedback loop to hone the models that process datasets.
The available flood of data is both a boon and an impediment, Black says.
“There’s a lot of variables, and not all of them are understood yet,” he notes. Add to that the cultural effect of an actuarial model with a 50-plus-year track record, and there will be resistance. “They understand, ‘Driver over 60, red car, no claims in the last five years.’”
It’s a small dataset compared to the exponential increase that IoT provides, and it has worked well. If you change a pricing model and get the model wrong, you go out of business.
Using real-time auto insurance to assess risk more effectively
The popular perception of AI gone wrong is that of robots turning on their masters, systems autonomically shutting down, and Will Smith defending the world against a raging mechanical onslaught. The truth of AI is more humdrum. But it can affect your credit, your eligibility for benefits, and insurance premiums.
The federal government has proposed a Framework for Automated Decision-Making, a model for determining when and how much humans must intervene in AI. Depending on the severity and persistence of possible damage, that could range from a rubber stamp to a panel of academic experts.
Black feels the real-time insurance model isn’t integrated yet. He gets feedback from his insurer in the form of an online dashboard that reflects his driving performance, and issues nuggets of driving advice. Neither his rates nor those nuggets (“Don’t brake while turning”) seem to be intimately tied to his driving performance. But he does get his discount.
North America has been slower to adopt these technologies, but as their business value becomes more evident, they will likely have an impact on car insurance in the next few years.
Keep on top of the trends and find the best rates for your auto insurance.
* -- Tim is a real person whose surname is withheld by request.
