By Clint Boulton for Wall Street Journal
Automobile insurance companies have been trying for years to convince customers to pay premiums that vary based on their driving habits, by promising to reduce premiums for better drivers. But the programs have failed to catch fire, partly because of privacy concerns, but also because some drivers worry data about their driving will be used to increase their premiums. But insurers are hoping that customers have become comfortable enough with other location-based applications such as Google Maps, which track their whereabouts, to give their programs a chance. Though these programs may not be widely used today, insurers are expending considerable resources to fine-tune the programs for the time, in what they hope is the not-too-distant-future, when they enjoy more widespread adoption.
The programs, known in the industry as usage-based driving, use devices installed in customers’ cars that relay data about their driving habits back to the insurers. Insurance providers, including Progressive Corp., The Allstate Corp., and The Travelers Companies Inc., hope the programs will help them attract new customers with better driving habits, which would have the dual benefit of increasing incremental revenue and, at least in theory, lowering pay-outs because those new customers are more likely to be safer, less accident-prone drivers. But the programs have yet to catch fire with consumers, and will continue to stall out unless the insurers can overcome concerns about privacy and fears that their premiums will increase if insurance companies know too much about them.
The technology behind these programs includes devices loaded with accelerometers and other sensors which are plugged into the motor vehicle’s diagnostic port, located under the steering column. These devices collect and transmit back to insurers data on users’ behavior, such as how many miles they drive in a given period of time, how often they drive late at night, and how often they’re forced to brake hard. The insurers analyze the data using proprietary algorithms and assign customers a score based on this analysis. The score, which is typically developed over the course of a month, is then used to calculate any applicable discount to the driver’s monthly premium. Consumers can look at their score online. “The more data we get, and the more we understand driver’s behaviors in the vehicle, the more we understand the products they need,” said Bob Otis, senior vice president of product operations at Allstate.
Companies competing in the $170 billion auto-insurance market hope that customers looking for a better deal will pick their company on the basis of this type of program. But auto insurers also said they expect to issue fewer auto accident pay-outs if they can encourage drivers to police their own driving and curtail potentially dangerous habits, such as driving late at night. Also, if insurers can collect enough data from usage-based driving, they feel they can build more accurate risk profiles of drivers, which will help them set more personalized premiums for users. Today’s insurance premiums are calculated on risk models providers have crafted using aggregates of data on driving performance, taking into account factors such as age, gender and other broad demographic characteristics. For instance, auto insurers traditionally charge 17-year-old single males higher premiums than 55-year-old, married males because they are statistically more likely to get into accidents. But these generalizations don’t account for exceptions, and the majority of good drivers ultimately pay for the sins of the relatively few bad ones, said Dave Pratt, the general manager for usage-based insurance at Progressive. He said the idea is to move toward more personalized insurance pricing.
“We’re not members of an arbitrary actuarial class—we’re individuals with our own set of driving habits, which should be reflected in the price we pay for our insurance,” Mr. Pratt said. “If we could identify better drivers and give them a better deal, it could be a great program.” He said user-based insurance can promote safe driving, lower premiums and help Progressive build better risk profiles of customers.
Sanford Bernstein analyst Joshua Stirling estimates the current market for usage-based driving at around $1 billion in annual premiums, mostly generated by Progressive. Progressive began kicking the tires on usage-based insurance in 1998 with a program called Autograph. But Gartner Inc. analyst Kimberly Harris-Ferrante said the early version of Progressive’s device was clunky – the size of a car stereo — and had to be installed by an automobile dealer. In 2010, Progressive launched its current product, Snapshot, which calculates how often drivers slam on the brakes, how many miles they drive and how often users drive between midnight and 4 a.m. (which is statistically when a lot of accidents occur). Each month, Snapshot drivers are scored and receive a discount of as much as 30% based on how safely they drove. Mr. Pratt said 1 million of Progressive’s nearly 9 million auto-insurance customers use Snapshot, which has logged over six billion miles of driving data from over one million trips per day. Though much newer to usage-based driving, Travelers and Allstate offer similar devices and programs, but will not say how many customers are currently using their programs. The companies say the data they collect is not used to track where customers go, or to increase premiums for drivers that don’t score well.
But consumer privacy concerns remain a roadblock. “Consumers get antsy about what people know about them,” said Ms. Harris-Ferrante. She also said many consumers feel that if they attach sensor-laden devices to their cars they will be penalized by the provider for speeding and other illegal driving habits. Progressive, Allstate and Travelers all said that hard-braking and other stats collected in their programs will not be used to increase premiums. And most insurers do not use GPS in their devices, which could be used to track where people drive. “We are very careful not to cross the line of privacy for consumer and customer information,” said Allstate CIO Suren Gupta.
Today, Progressive, Allstate and Travelers use fairly standard statistical tools and database software. In the future, some are planning to use more sophisticated software that will help them parse data in near real-time. Allstate is considering using Hadoop, software that allows businesses to use clusters of low-cost servers to analyze petabytes of information, to process data more quickly. Insurers are largely secretive about technologies and models they currently use – and plan to implement – because they see usage-based insurance as something of an arms race. Insurers believe that if they can collect as much driving data as possible, analyze it, and derive meaningful insights from that data, they can build new personalized models for assigning premiums to drivers before their rivals. “The more data they have on drivers, the more they can improve their formulas to improve risk selection,” Ms. Harris-Ferrante said.
