The largest auto insurance companies are racing to offer driver-tracking discount programs in a high-stakes effort to lure new policyholders and retain existing ones. These programs, such as Progressive’s Snapshot or Allstate’s Drivewise, require installing tracking devices into the vehicles of policyholders. The trackers monitor driving habits, including speed and distance driven, along with turning and stopping habits.
The upside for safe drivers is the companies offer discounts for good road behavior. Most recently, Allstate boasted average savings of 14% for policyholders that participated in Drivewise. Meanwhile, in 2012, Progressive announced it was able to draw conclusions about drivers’ habits that could save the company money as it alters actuarial practices to reflect the new data.
Privacy has always been a concern with these programs, even though most of insurers stress that they don’t collect GPS data. Here’s what Progressive’s Snapshot FAQ says on the matter:
“The Snapshot device doesn’t track your location or whether you’re speeding, and it doesn’t contain GPS technology … We won’t share Snapshot information unless it’s required to service your insurance policy, prevent fraud, perform research or comply with the law. We also won’t use Snapshot information to resolve a claim unless you or the registered vehicle owner permits us to do so.”
Despite those assurances, some drivers and privacy advocates still have lingering doubts about insurers’ ability to track where participants in these programs drive. And according to a team of four researchers at the University of Denver’s Colorado Research Institute for Security and Privacy, those doubts have some credibility.
Finding location without GPS
The problem, according to the researchers, is that you don’t necessarily need GPS data to determine where a car’s being driven. As the University of Denver’s Rinku Dewri explained to Online Auto Insurance News (OAIN), it’s possible with just three data points.
“Just by tracking the speed, the distance and the point of origin,” Dewri said, “we can find the exact location where the driver ended the trip more times than not.”
According to Dewri, he and his team created an algorithm using these three data points and map information to predict the destination of a driver with accuracy 60 percent of the time in their study. A total of 15 trips were used in the study.
It’s important to note that the researchers assumed insurers would already have one of the three data points: the point of origin. But the team says insurers are aware of where the insured car is parked overnight and could use this as the assumed starting point.
“The assumption of a known start-location is not unrealistic,” the team wrote in the paper. “Start-locations in subsequent trips can be obtained from the destinations of previous trips.”
Data as currency
The concern of Dewri and his fellow researchers at the University of Denver is not that insurance companies will share data, or use driving destinations to increase rates. Dewri notes that many insurance companies currently take strong stands to guard the privacy of their policyholders. Dewri, an assistant professor of computer science at the university, said his concern is that people don’t know that the data they allow insurance companies to access could result in the companies’ knowing details about their lifestyles, such as whether or not they see a therapist or attend Alcoholics Anonymous meetings.
“We want the consumers to know that location tracking can be done,” Dewri explains. “Most consumers are not educated that when they give away data they are really giving away an insight into their lives.”
“Many drivers might not care that their insurance company can figure this out,” Dewri said. “But they should know it’s possible. What this is about is transparency, and if you read any of the companies’ policies it would seem that this is not possible.”
While all the major companies have stated safeguards against sharing the information collected by the driver-tracking programs, Dewri said that doesn’t prohibit the companies from sharing the information in the future.
“We’ve seen it with other companies, such as Netflix, where massive amounts of data are collected, and shared later down the road,” Dewri said. “In this day and age, people need to know that data is money, and that it will most likely be shared with other companies or governmental groups at some point.”
An issue of transparency
At least one insurer is concerned that the University of Denver’s research could slow the growth of the driver-tracking market, according to Dewri. The researcher would not name the insurer, as the talks between the groups were private. But the insurer, Dewri said, offers a driver-tracking program and believes that emerging privacy concerns in the public could lead to a blowback against the technology.
OAIN was able to confirm that a major insurer contacted the University of Denver to inquire about the technology, although they did not make other details publicly available. Calls and emails to other insurers for comment were not returned by OAIN’s deadline.
Dewri said his team will make the algorithm public in the near future as a way to educate consumers. He also said his team is very interested in why the algorithm was wrong 40 percent of the time. In the published study, the team said the inaccuracies were most commonly attributed to long periods where the car was at rest in traffic, excessive speeding, and times when drivers did not take the shortest route to their destination.
“That could hold a key to protecting [motorists] participating in the program in the future,” Dewri said.
But ultimately, Dewri said motorists will have to decide themselves whether they want to trade privacy for potential savings on their auto insurance premiums.
“And if they do, fine,” Dewri said. “But they should know what the potential costs are. Privacy is very important, and so is transparency. And that’s really all we want, is for the insurance companies that administer these programs to be as transparent as possible.”