Twitter LinkedIn Find us on youtube Find us on Facebook Find us on Google +

Telematics Brazil & LATAM 2014

24/09/2014 - 25/09/2014, Hotel Tivoli, São Paulo, Brazil

Increased Collaboration, Advanced Services and New Customer Segments Accelerate Telematics Adoption

Q&A: The Co-operative Insurance on telematics data and pricing accuracy

Grant Mitchell, chief actuary, general insurance, The Co-operative Insurance, on using telematics data to ensure pricing accuracy.

With eight million members, The Co-operative Insurance is the United Kingdom’s largest cooperative focused on personal lines insurance.

In March 2011, it began offering Smartbox, a pay-how-you-drive (PHYD) insurance telematics option for young drivers, which rewards safer drivers with discounts on premiums.

Driving performance is reviewed every 90 days and takes into consideration the time of day, average speeds on different types of roads, how fast and hard a car accelerates or brakes, and how fast corners are taken.

TU’s Susan Kuchinskas asks Mitchell about what The Co-operative Insurance has learned so far from using telematics data to price insurance and what it sees on the road ahead.

How did you develop the Smartbox offering?

We developed it in-house, in conjunction with Wunelli. At the time we launched it, in the U.K., there was a lot of concern that insurance premiums for young drivers were rising and were very high. A lot of young people potentially couldn't afford to buy car insurance. We felt that telematics offered an opportunity to identify the safest young drivers and to offer them a much more reasonable price, and, at the same time, help to educate young drivers about safety.

Do you plan to gather other data points on drivers besides acceleration and braking, cornering, average speeds and time of day?

Over time, I'm sure we will. The data we collect is very rich. We picked those four factors when we launched, and it hasn't changed since. More recently, competitors have used similar factors. I'm certain there are lots more interesting factors there, as well.

How much data typically is generated by a driver and how do you store, manage and analyze that data?

The box itself collects data every second the driver is driving. We don't tend to store all of that, unless there's an accident. If there is an event, meaning we'll need to handle a claim, we'll store all the data around that point. Otherwise, we take one-minute snapshots, at the end of each minute.

Even so, it's a lot of data. Typically, the average driver probably drives 15,000 minutes in a year.Over time, I think we will probably develop ways of storing it more efficiently and find the more relevant bits. We're still learning a lot from it and analyzing what use we might get from it.

What are you learning?

There's a wealth of things to investigate. I don't want to give away what we're finding in particular. But, clearly, the four factors we already are analyzing are highly predictive. The other obvious ones are how familiar the person is with the road they are driving on, and do they drive in congested traffic or [on] more open roads? We also want to know whether the same factors are relevant for different drivers, for example, younger or older drivers.

Do you combine telematics information with maps or GIS data, perhaps to create risk maps or identify accident black spots?

We have a database of all the speed limits on U.K. roads; we cross-reference the actual speed to the speed limit. We do look at it on maps, but not on a routine basis. We tend to do that when there's been an accident or claim or extreme speeding event. It's a manual process, which is why we don't do it frequently.

We're not doing it at the moment, but I think there is potential to identify accident black spots using public data, for example, from publicly available data, such as from police or government data sources. It's possible to then use telematics put a geofence around those black spots, set policies and rate a driver by how often they drive there. We could use the telematics data to either increase the premium or educate that driver that this is a dangerous place to drive, and you should go on a different route.

The next level of enhancement might be knowing, for example, that the accident black spot is worse at night than during the day.

Could you do that now?

Yes, the capability exists. It's really about cross-referencing multiple data sets. The limit of what you do with that is down to the imagination of coming up with different things to look at. Inevitably, there is limited human resource to do the analysis on these things, and that's probably the biggest single limitation to how fast we can progress.

What about enhancing the claims process and even reconstructing accidents?

By knowing more about the driving style immediately before the accident and also identifying a bit more about the nature of the accident, you can get a much earlier read – in theory – on whether our insured vehicle was at fault. That potentially affects the way you might choose to handle that claim. And by having real-time event data coming through, you can be proactive about contacting that customer to handle the claim, rather than waiting a few days to have them contact you.

Are you contemplating beginning to offer value-added services, for example, referring drivers to a car wash on their route?

I'm not sure coupons for a car wash is what we have in mind. But value-added services are a very interesting area. Right now, the industry is very focused on predictors of risk. There's not much focus on value-added services. In other markets like … Italy, it is a much bigger part of what telematics is all about. There is some unexploited potential in the U.K. market around emergency assistance, and perhaps fuel economy.

What else is on your radar?

One of the hot topics in the U.K. market is what kind of telematics device will win out. We use hard-wired boxes right now, but, for getting an indicator of price, a smartphone app may well be sufficient for a lot of drivers.

A lot of the value-added services are much better delivered by hard-wired boxes. The box itself has an expense, but the biggest is the installation expense, getting someone to fit it. In the U.S. market, there are self-installed devices, which work well but haven't taken off in other markets. In the United States, vehicles tend to be very standardized, and the onboard diagnostics board you plug [self-installed devices] into is accessible. In U.K. cars, it tends to be more hidden away.

Susan Kuchinskas is a regular contributor to Telematics Update.

For all the latest telematics trends, check out Insurance Telematics Europe 2014 on May 6-7 in London, Data Business for Connected Vehicles Japan 2014 on May 14-15 in Tokyo, Telematics India and South Asia 2014 on May 28-29 in Bangalore, India, Insurance Telematics Canada 2014 on May 28-29 in Toronto, Telematics Update Awards 2014 on June 3 in Novi, Michigan, Telematics Detroit 2014 on June 4-5 in Novi, Michigan, Advanced Automotive Safety USA 2014 on July 8-9 in Novi, Michigan, Insurance Telematics USA 2014 on Sept. 3-4 in Chicago, Telematics Japan 2014 in October in Tokyo and Telematics Munich 2014 on Nov. 10-11 in Munich, Germany.

For exclusive telematics business analysis and insight, check out TU’s reports: Insurance Telematics Report 2014Connected Fleet Report 2014The Automotive HMI Report 2013 and Telematics Connectivity Strategies Report 2013

Grant Mitchell, The Co-operative Insurance

blog comments powered by Disqus

Telematics Brazil & LATAM 2014

24/09/2014 - 25/09/2014, Hotel Tivoli, São Paulo, Brazil

Increased Collaboration, Advanced Services and New Customer Segments Accelerate Telematics Adoption