Facebook stop Admiral from using their data to set premiums

Credit: Twin Design/Shutterstock.com

The plans were in the process of being realised by Admiral and their "firstcarquote" app, which people would've been able to download and use to prove to the insurer that they're safe drivers.

However, Facebook have told Admiral that they aren't permitted to use people's posts and data to determine premiums, since such a practice would constitute a violation of their terms and conditions.

As a result, the insurance company have been forced to delay the app's launch, which will now go live on an unspecified date with reduced functionality.

But while this reduction does indeed represent a victory for privacy and free-speech advocates, the app will still require users to share "some social data" so as to receive a reduced quote.

And even though they'll still be asked to hand over this data, Admiral have appeared to have gone back on their previous offer of a discount of up to 15% on premiums.

Organised drivers

Now, the discount will be offered to all users of the app at a flat rate, yet Admiral haven't disclosed just what this rate will be, suggesting that it could well be lower than 15%.

Neither have they responded to Choose's request to state precisely what social data they'll now be taking. However, given that all users will be automatically awarded the same discount, it's hard to imagine what importance it will have if any when it comes to determining a customer's level of responsibility and irresponsibility.

This is exactly what the app would have been doing if it'd launched with access to Facebook's data. In particular, it would've mined a customer's posts and messages for certain supposedly telltale signs of whether or not he or she were a safe driver.

For example, customers who write using short, concise sentences would've been regarded as more "organised". So too would've been customers who specified particular times and locations when arranging to meet with their friends.

Conversely, those who didn't specify particular times, as well as those who used a healthy supply of exclamation marks in their writing, would be deemed as less-than organised and wouldn't therefore receive any discount.

Admiral have stated that no one would receive a higher-than-normal premium because of their posts, yet the fact remains that they could still have received the highest possible premium they now offer.

And given that the app is for 17-24 year olds only, this could be a very high premium indeed.

In fact, at an average of £1,240 [PDF] a year, the premiums paid by 17-22 year olds, for example, are already grossly overpriced.

At least, they're overpriced in the sense that responsible 17-22 year olds have to pay an almost prohibitive amount for their insurance simply because some of their peers aren't quite as responsible.

And because the firstcarquote app would've potentially put them in the irresponsible bracket for saying the wrong things on Facebook, it would have effectively punished them with higher premiums.

Third parties

Still, the irony of this is that Facebook denied Admiral access to their data not because they dislike the idea of judging people according to the content of their posts, but because they want to reserve the right to do so for themselves.

More specifically, Facebook's business model rests largely on selling their users' personal data to interested third parties, who then target people with certain ads depending on what their data "says" about them.

As Jim Killock, the executive director of the Open Rights Foundation, said, "Social networks do not want you to feel inhibited". This is because they want people to speak freely, without fear of receiving a smaller discount on their insurance, so that advertisers will be better able to target them with "relevant" ads.

And what's particularly ironic is that Facebook enables the kind of discrimination that the likes of the Open Rights Foundation have praised them for avoiding in the case of Admiral.

For instance, it recently emerged that the social network enables advertisers to target people with specific ads, not only according to their interests or backgrounds, but according to their "ethnic affinities" (i.e. their race).

In other words, they allow companies to exclude people from receiving ads, offers and promotions according to a data-informed inference as to what kind of people they are. While this might not be as bad as making people pay the highest possible insurance premium because of an unreliable deduction as to their safeness, it operates according to a similar principle.

No strong link

And it's this discriminatory, potentially inaccurate principle that was arguably the most alarming feature of the planned use of Facebook, and not the invasion of privacy, which is already perpetrated by many employers looking to assess the suitability of applicants.

It's alarming because no link has been demonstrated to date between the content of social media posts and car insurance claims.

Searching quickly through Google Scholar, the most relevant research paper [PDF] on the use of social media in insurance mainly documents how insurers could use Twitter to conduct market research and improve customer service. It says nothing about a proven correlation between how people write and how many claims they make, and it's doubtful that any other research has demonstrated such a correlation either.

This is the main reason why the app was worrying.

It threatens to give and deny people discounts on their insurance for potentially superficial reasons, and because of this there's little likelihood that it would be as reliable as, for example, the black box insurance they and other insurers offer.

And in contrast to the black box's 25%, a maximum discount of 15% for saying the right things on Facebook doesn't seem quite as attractive.

This is especially so when the discount may turn out to be less than 15%, despite customers still having to turn over social data, for some purpose Admiral have yet been open about.