"I'm seeking more participation at this point because we are at a critical juncture as we define XML messages," says Bielak. "Some folks may think spending an hour talking about business requirements is not the best use of their time, but now we need people to get back to the table. What we define is going to be the standard and if it is not implementable than shame on us. We want as much participation as possible to make sure what we define can be implemented and will be used across the industry."

Because of the relative newness of telematics, getting carriers to share information is a challenge. Some major carriers have agreed to participate and are eager to hear what the working group is doing, according to Bielak. Getting those same carriers to chime in with information on their own efforts has proven to be more difficult.

"There's a problem with telematics and being vocal about it," says Bielak. "This is something we've been wrestling with from the beginning."

If you are a third-party service provider or a carrier and you take a position or ask a question about a certain point, Bielak explains you are revealing your particular interest in the subject matter. By doing that, carriers are afraid they will spill what Bielak calls their "secret sauce" recipe.

"Because telematics is new and emerging, insurance carriers particularly don't want anybody to have any notion on why they want the data, what data they want, and how they analyze the data," he says. "It's black box to the world. The service providers are following the lead of the carriers and aren't volunteering much information at this stage."

Convincing Others

Bill Jenkins, a consultant with Agile Insurance Analytics, believes top insurance carriers will be reluctant to take part in establishing telematics data standards as they are convinced their proprietary information constitutes a competitive advantage in this emerging underwriting tool.

"ACORD standards within Tier 1 carriers are not all that pervasive," he says. "The big companies basically have proprietary standards. The whole value of collecting third-party data is to collect data that will be an advantage to that particular carrier based on customer segments they have."

Jenkins points out that Progressive Insurance has seven patents that have been approved by the U.S. Patent Office concerning telematics and have taken considerable steps to defend those patents.

"Some of that evolves around data so it is pretty much proprietary at this junctiure" says Jenkins. "My guess is that the data you have around your particular customer base is going to be proprietary for that organization. That's their differentiator. They might get into a circumstance where they share things down the road, but it is too early to tell."

There are many carriers in the UK that have been focusing on telematics and Jenkins believes those insurers are more advanced compared to what is happening here in the United States.

"There are a lot of data management issues around this—what they can use, where they can store it, a lack of skill sets to analyze the data," says Jenkins.

Standards aren't just an issue for insurance carriers, points out Jenkins, many automobile manufacturers are installing different types of devices into their cars.

"There is no doubt there is a need to have a standard in place, but how do you get the auto manufacturers in line?" asks Jenkins. "In two years 90 percent of all cars will have built-in capabilities to do these types of measurement."

Some of the approaches ACORD has taken to attract cooperation include attempts to convince members of the working group to send Bielak examples—even brainstorming examples—so he can work the data to preserve anonymity when the information is presented to the working group as an example.

"I finally got three contributors who shared with me the structure of the data they are currently receiving or sending," he says. "That gave us some framework of reality within which to work. That is what we are using to initiate this XML design. At the same time, we are still struggling to get more participation."

The chair of the working group, Mark Woods from Allstate, suggested to Bielak on a recent call that if participants have questions or want to guide the discussion but are unwilling to speak on the call they should email Bielak their issues and he will try to lead an anonymous discussion so nobody is accused of spilling their company's secrets.

"One member actually had their general counsel talking to our general counsel about what we are up to because there were concerns about their people participating on the standard," he says. "I was never privy to those discussions, but that's how serious insurance companies see this. It is top secret right now."

Top Secret Challenge

John Foster, vice president, personal lines underwriting for Penn National Insurance believes the scores that companies plan to use to provide discounts—the meat and potatoes of telematics—will likely be kept confidential.

"I think carriers or third parties will tell you the most significant drivers are mileage, acceleration, maximum speed, miles driven, and hard braking," he says. "Companies will share with consumers the most important characteristics and how the driver compares to an average driver."

Companies are participating with ACORD and Foster believes the reason is to share data among various companies, but he also feels it is going to be a challenge.

"Everybody that will be playing in the UBI space is looking to create a competitive advantage," he says. "They are going to look for pieces of data—something different—outside the ACORD standard that they can use. I am suspect to see how that works out."

Penn National is an ACORD member company and Foster maintains ACORD has done great work for the industry in providing standardization.

"We support their efforts, but how that plays out in the long term with telematics is unclear to me," he says. "A company of [Penn National's] size is probably not going to be managing the data. We will select a company to do that for us, do the analytics, and create a score for us. How the data is standardized and collected is really not that important to us. It will probably be more important to the big carriers."

Foster has heard horror stories from companies that have tried to collect telematics data themselves, analyze the data, and create their own score.

"When you are transmitting data every couple of seconds from these devices, it takes a lot of expertise to determine how you are going to collect it, organize it, compress the data, and ultimately how you are going to use it," he says. "If ACORD is able to standardize things, I see some value if they can take a leadership role and create standards. It would be easier for medium-sized companies to collect the data and analyze it themselves, but I don't see that happening in the next couple of years."

There are many different ways of representing the data and Bielak believes the examples that were contributed certainly reveal that. Bielak comes from a scientific background—oil and mining—and has dealt with big data in those industries and seen standard ways of representing big data files.

"Seismic surveys have a specific format and it allows the data to be portable from one system to another," he says. "I don't see why telematics data should be any different. I think that past experience gives me an eye to focus on just the data. I do think we are on to some good approaches that de-couple the whole issue of proprietary information from the data itself. By being smart and coming up with a good definition for the data, it ought to provide flexibility to allow different providers to send different types of data to consumers. I think we have the bases fairly well covered right now for a prototype."

Commercial vs. Personal Lines

Bielak explains the original plan was to separate commercial from personal lines—with personal lines coming first—as a way of following what Polaris, the European standards organization, did in Great Britain.

"I thought that was a good idea, but the members on the call were not worried about the differences between personal and commercial lines," says Bielak. "Eighty percent of the data we want to deal with is pretty much the same—GPS, accelerator data, ODB 2, speed. When you get into the commercial lines there will be wrinkles you don't have in personal lines. We need to focus on the 80 percent that is common and then we can focus on any whistles and bells we might need for commercial lines."

The working group's scope is to define a mechanism to communicate from the data provider to the data consumer. The data provider could be a technical service bulletin or some other third party that dumps the data to the data consumer—in this case the insurer, explains Bielak.

"Telematics goes all over the place, but all we are dealing with is data provider to data consumer, so that helps limit the scope," he says. "It's an insurance standard so we are serving the purpose of the insurance industry."

According to Bielak, three things the working group has defined and are beginning to explore in more detail are"

• Recognition that vehicles generate events. Events are one category of data that the working group agrees needs to be in the standard.
• Streaming data is the bread-crumb trail of GPS, according to Bielak, and speed at different points in time that the vehicle is moving across the landscape. The group envisions this potentially to be a large data set depending on how many columns of data carriers are going to report. This is streaming data with an observation every second, 10 seconds or in the case of an accident, down to milliseconds of data.
• Report or summary data. You can generate report or summary data from the streaming data, but some insurers have explained that they are not interested in second-by-second observations. Instead, they want a summary on how many minutes the vehicle was being driven during certain hours of the day to build a histogram of the number of minutes. There also is the number of seconds the car was accelerating between 0Gs or 5Gs or decelerating; and the number of minutes the car was being driven between 60 and 70 MPH. These are abstractions from the big data to generate a summary on the behavior of the vehicle.

"We are almost done talking about events, but we have to come up with lists of codes and how much granularity we are going to have," says Bielak. "My business architect and I are investigating the right way to do the streaming data. In fact, the data might actually be outside the XML message itself and a referral to a file might be needed."

Bielak believes that from a pure data perspective, ACORD is doing things the right way.

"We are not taking any old baggage with us," he says. "[Our work] could easily be used in Australian standards, South Africa, or London. We are just focused on the data itself and wrap it in an XML message that is domestic United States basically, but you could use the same mechanism for data reporting in all our other global standards. Hopefully this will gain traction globally because there are only so many ways you can slice it. It's data, for heaven's sake."

Bielak would like to have the base of the telematics—vehicle telematics notifying message—defined and submitted by this spring, or at the very least, have events and streaming data in this message.

"We can expand the message with summary reports, but I believe we need to fish or cut bait," he says. "Let's get the thing out there. I'm shooting that by June or July we will have the specification to communicate telematics data from a provider to a consumer. If we miss anything then that will get non-participants to join the group. We recognize other needs, but that's the problem: to get them to come to the table to reveal what they want without revealing what they really need."

Bielak has been a member of ACORD since 2003, but he's been an employee for just two and a half years, so it is difficult for him to compare the job he faces today with earlier battles over standards that ACORD has survived.

"When I look at the standards, a policy is a policy is a policy," says Bielak. "There's not a whole lot that is proprietary about it. I do know there are plenty of different approaches that insurance carriers have to the way they structure their data and how those carriers use the standards. The people who consume the data have to be flexible enough to manage all the different variations."