Data analytics firms make a serious play for telematics

The rise of telematics in the insurance industry has data analytics vendors' tongues wagging, Sooraj Shah investigates

The idea of paying less for something is always appealing to the consumer.

But what if those discounts are only available if you give up some your privacy? Many would still maintain that as long as they save money, they'd be happy to give up some of their data. In fact Google's - for one - business model relies on that fact.

This is where the conundrum of telematics begins; the technology used in vehicles which can track a person's driving skills and send data back to a server, which can then be analysed to allow insurers to fine-tune their insurance premiums.

Global insurer AXA's group CIO and COO Kevin Murray suggests that this will become a mainstay in the UK within five years, after making inroads in the US, despite fears of it being a little "Big Brother". He says that insurance rates should shift from being a "big underwriting pool of risk" to more targeted and strategically set prices.

But how does it all work?

A recent partnership between Japanese firm Pioneer and California-based managed cloud services provider Treasure Data unveiled a small piece of software installed into the car which collects sensor data. The data is then transmitted over cellular networks to Treasure Data's data centre and analysed.

Aldo Monteforte, CEO and co-founder of The Floow, which provides the data analytics for telematics to insurers such as the Direct Line Group, explains that his firm takes individual mobility data from a variety of devices initially.

"We collect data from smartphones where one of our apps has been downloaded, or alternatively from more conventional telematics sources like on board diagnostics devices or black boxes which have been professionally fitted into a car," he says.

"Wearable devices will probably follow suit," he adds.

Manufacturers such as BMW and Mercedes have installed black boxes as standard into cars themselves in an attempt to lure drivers back to dealers' garages for maintenance, but these too can be tapped into by insurers if they are given consent to do so.

The Floow collects data from these three different sources, and channels this into its data management engine.

"The first thing we do once we receive the data is to cleanse and standardise it; for example we know that the GPS in a smartphone won't be exactly the same as the black box, this is important because our clients want to give scores to the drivers," he says.

Then, the company takes into account contextual variables that may affect someone's driving - such as driving near a school where a lot of people are walking around or near an area where a lot of alcohol is usually consumed.

Then it shares the overall insight it has gathered with its insurance partners.
But how do insurers use the data to price their premiums?

Simon Overton, a sales manager of banking and insurance at SAS, another company that delivers data analytics, believes that customers are using their tools in an attempt to identify which characteristics of driving give the biggest indicator of risk - it may not always be the speed at which someone drives, but it could be the way they use their brakes, for example.

He suggests that customers are also using analytics to build up a picture of what normal looks like first, so that the insurance firm knows what is a risk and what is a danger, and can then price their insurance premiums accordingly.

"You can have an insurance claim where someone says they were only driving at 15mph and had a crash, but if telematics data suggests that around a particular bend drivers should only be travelling at 5mph, then 15mph would be very dangerous," Overton states.

"Some customers are looking at it from a fraud perspective, to see if they can see patterns emerging for fraudulent claims," he adds.

According to Overton, SAS is "heavily used in the pricing process" by nearly all of the general insurers in the UK.

The Floow, meanwhile, has a senior team made up of computer scientists, physicians, mathematicians, actuaries and psychologists that have built an algorithm to define whether someone is driving is safely.

"It is based on almost 30 years of scientific research and we offer these algorithms as a starting function to our clients and they can adjust and personalise these," says Monteforte.

Hannah Smalltree, director at Treasure Data, explains that although many of the data analytics tools on offer compete against one another, they can also be used together.

"It is possible for us to sit alongside a customer that already has a SAS environment. You would get advanced level analytics with SAS and with Treasure Data you would get the data collection of real-time information as well as the end-to-end storage," she says.

"But if you wanted to write specialised algorithms, you could write the basic analysis in Treasure Data and then export that into an advanced analytics environment like SAS or Microstrategy; they are different animals. We're good at crunching the data and SAS is more focused on the data analytics side of things," she adds.

Regardless of the vendor used, Gartner analyst Thilo Koslowski believes that any system is unlikely to be hacked in order to manipulate the data for a cheaper insurance premium.

"The information collected is usually based on GPS which is pretty accurate and it would take a lot of effort to mess around with the data to manipulate it, so I don't think you would see a lot of hacking to adjust this data," he says.

The drawbacks

According to AXA's Murray, a key issue with telematics is that people regard it as a "Big Brother"-style tool, as they don't necessarily want third parties to know where they are and how they're driving.

But he suggested that telematics had won over parents in the US, despite initial privacy fears, as they could install the technology in their cars to ensure their children were driving safely, and were rewarded with cheaper insurance premiums.

Although the idea for cheaper insurance premiums would be welcomed by consumers, Koslowski says that currently insurers only want high-risk consumers to sign up to use telematics.

He questions the logic of expanding telematics to every consumer.

"If [an insurer] rolled this out to everyone, would they be able to offer discounts to all of their customers? They would ultimately cut into their profits. However, if one insurer did this, then others would have to follow suit," he says.

He said that part of this is down to the hardware and data collection costs. Treasure Data customers, for example, are charged a flat-rate on a monthly pricing model based on the amount of data imported, stored and the type of processing engine used. Meaning that the more people an insurer has using a telematics device, the more it has to pay to the data analytics and hardware firms that are storing and processing it.

But Koslowski believes that the insurers could have the consumers pay an upfront fee to help to deal with these costs.

On the flipside, if this does become mainstream, then he thinks that insurance firms could change policies to be ‘pay how you drive' policies - but again drivers may not be fond of the idea from a privacy standpoint.

"Someone might get into a driving accident and [historical] data could be used against them in court, for example," Koslowski suggests.

But there are other benefits which the data analytics firms are talking up for the consumers. For example, The Floow and Treasure Data will offer consumers the ability to receive alerts such as a notification stating that the driver has a limited amount of petrol in the tank left, or that they are using too much acceleration.

This is in the hope of making mobility "safer and cheaper", according to The Floow's Monteforte.

But the insurance companies will be most concerned about their bottom line, and SAS's Overton says that the data analytics firm is in talks with many of the general insurers about how they can leverage new data sources so they can differentiate themselves against the competition.

According to Overton, telematics "will just become another data source to differentiate insurers".

"I think the real value can be added where insurance providers can offer you services on the data they've got," he says.

This could be offering a cheaper quote if a driver took a less risky route to work on a daily basis, for example.

Overall, the consensus is that there are still many more questions about telematics than answers, the biggest of which is whether any one insurer will take the plunge and offer telematics insurance to the mass market, rather than keeping it safe with high-risk drivers.