Motor insurer Admiral announced it was going to use social media to analyse the personalities of first-time drivers. If their Facebook posts indicated they had personality traits Admiral could link to safer driving habits and fewer claims, annual discounts of up to £350 would be offered via the insurer’s new firstcarquote product.
But just before launch the insurer was forced into a U-turn. Facebook said the product would break its terms and conditions, which prohibits the use of data obtained from its service to make decisions about eligibility. “Following discussions with Facebook, the product is launching with reduced functionality,” an Admiral spokesperson announced. It will still, however, ask for access to drivers’ public Facebook profiles before it provides them with a quotation, so customers can “share some information to secure a faster, simpler and discounted quote”.
This is the new world of big data and insurance, where social media websites are vast sources of valuable information just waiting to be plundered. “Insurers have been recruiting data scientists, and I think it is very likely they are already using information from social media websites to set premiums,” says Gary Frost, managing director of 51zero, a software consultancy specialising in data gathering and analysis for financial services firms. “If they’re not, they’re missing a trick. It provides extra detail and information which can help improve their risk scoring.”
Insurers can access all the data that is posted publicly on Twitter and Facebook, Frost says. “You can’t buy access to a private social media profile, but insurers can potentially find out all the public information these sites store about you, such as who you are friends with, your public updates, public pages you have liked and posts you have publicly recommended or shared.
I think it is very likely insurers are already using information from social media to set premiums
Gary Frost, data analyst
“So if you buy a cherry-bomb sport exhaust to soup up your Vauxhall Corsa and talk about it on social media, classification and clustering technology is likely to place your profile with a community of higher-risk consumers.”
Any lifestyle information the insurer can glean about you from any source – for example, your sexual orientation, race, religion, predilections, estimated consumption of alcohol, cigarettes and even saturated fat when out socialising – could potentially be used to reassess your application and increase or decrease the cost of your premiums.
Even seemingly innocuous information can potentially be linked to a higher or lower claims ratio, and therefore affect your premium. This includes the web browser you use to buy your policy, the time of day and month of the year you do so, and whether you use a comparison site.
Each time you apply for insurance, regardless of how you pay for your policy, your insurer is likely to soft-search your credit record. “There are very significant correlations between likely claims and credit scores,” says Kevin Roberts, director at Legal & General. “Better credit scores generally result in fewer claims – we don’t know why.”
Consumers shouldn’t assume they can control the data they share. Insurers will look for any previous contact and use that to make a judgment about a new application. For example, if you visit two comparison websites in quick succession to request quotes, all the insurers which offered you quotes the first time will be able to match your profile when you visit the second site – and because they know they’ve quoted for you previously, they may take a dim view of any changes to what you disclose, such as your annual mileage. “I’d expect that to have a negative impact on your premium. It may be seen as indicative of dishonest behaviour,” says McCulloch.
Similarly, if you already hold a policy with an insurer, the answers you gave when you first applied may be analysed in relation to the new application. For example, any disclosures about high value ticket items or burglaries on your home policy may be considered in relation to a motor policy.
Insurers have also found a correlation between callouts on motor breakdown and home emergency policies and claims on motor and home insurance products, says Roberts. So if you are loyal to a breakdown service and make few callouts, you may be offered a cheaper insurance policy by any provider with access to that data. “AA members tend to be a better risk for us,” says Ian Crowder, spokesperson for the AA’s insurance arm. “We know them so can offer discounts. They tend on the whole to care for their vehicles and thus are more likely to recognise risks when it comes to motoring.”
When you apply for insurance, firms will already attempt to analyse your whereabouts and the places you travel to. “Companies can obtain lots of information about postcodes from third-party sources,” says Toni Vitale, director at law firm Addleshaw Goddard. For example, he says, if you provide your car insurer with details of your home and work postcodes, it may map out the most likely route between the two, and then take into account the crime rates of the postcodes you will be passing through each day.
Similarly, if you’re a Nectar card holder Sainsbury’
s says it will look at the size of your spend and how frequently you shop at its stores, and offer you up to 15% off its home and car insurance premiums. “Information about your individual purchases is available, but we don’t use it to make any decisions,” a spokesperson says.
Jim Killock of the Open Rights Group is sceptical about the rewards on offer in return for this data. “I think insurers are tempting people to share their data by saying they might get a discount, but in fact they’ve just priced everybody else up,” he says. “An insurer’s goal is to maximise its profits, after all, and now they are using unrelated information to make judgments about people. We are not aware precisely how far these judgments might go. For example, a machine might decide there’s a correlation between someone’s ethnic background and the claims data, and start pricing people out of insurance. In this way, using unrelated data may introduce racial biases and discriminate against people without good reason.”
In September, the Financial Conduct Authority identified that the use of big data had the potential to leave some consumers worse off, and said it was concerned that it could enhance an insurer’s ability to identify ways to charge certain customers more.
“The increasing amounts of data from a wider range of sources, alongside sophisticated analytical tools, might lead to the use of reasons other than risk and cost in pricing becoming more prevalent,” said Christopher Woolard, the FCA’s director of strategy and competition.
“To assess how different pricing factors are used, we will … look at pricing practices in a limited number of firms in the retail general insurance sector this year.”
MANAGE YOUR PROFILES
■ “Check a site’s terms and conditions,” says Toni Vitale from Addleshaw Goddard. “Know what you are signing up to and how they are going to share your information with other companies.”
■ “Don’t fill your Facebook page with fast cars or show that you’re an adrenalin junkie. Keep your profile as private as possible,” says Kevin Pratt of Moneysupermarket.com.
■ “Make sure you know exactly what is on your public Facebook profile,” says Gary Frost from software consultancy 51zero. To do this, go to your profile and click on the three dots next to “View Activity Log”. Then click “View As …” and you will be able to see what your timeline looks like to a member of the public who happens upon your profile.