Do you really Like that?
Be careful what you like on Facebook – that’s the warning to take from research carried out by the University of Cambridge. The project used algorithms to predict religion, politics, race and sexual orientation based solely on what people chose to Like on the social network.
By correlating personality tests and the demographic information of 58,000 volunteers, the researchers were able to compare Likes with an astonishing level of accuracy. The algorithm used was 88% accurate in predicting whether someone male was gay or straight and between 65-73% accurate in guessing marital status and substance abuse for example. And it wasn’t based on simple linking – fewer than 5% of gay users clicked obvious likes such as gay marriage. Instead it used information such as likes on TV shows, films and music.
This is music to the ears of marketers (and social networks desperate to sell advertising to them). It could even help Facebook’s depressed share price perk up a little. And if you can accurately predict detailed demographic information from just one part of a person’s online footprint, imagine what you can do if you add in web browsing, search and other social network data. No wonder Google wants you to sign into its multiple services so it can collect the maximum amount of data, whatever device you are using.
From a consumer point of view there’s two ways of looking at this – most people will see it as an intrusion into their privacy and change their settings, but brands may well rationalise it as offering people exactly what they want. And as Mark Earls has pointed out in his book I’ll have what she’s having a large number of people’s decisions are herd led. So offer them an easy option that means they don’t have to think and they’ll jump at it. In many cases consumers may not even realise they are being sold to – which could be very worrying when people start being segmented on sexuality, religion or political affiliation.
So marketers need to treat this data with caution. Yes, it gives unprecedented insight but be too aggressive when using it and you’ll cause a public outcry which could damage your brand – and trigger governmental action to tighten privacy settings on the likes of Facebook.
However my own view is that we’ve been here before. Remember when store loyalty cards came in everyone predicted that we’d be laser targeted with relevant offers that drove us to up our spend? But if I get a mailing from a well-known chemists the vouchers are pretty much identical to my wife’s, with obvious male/female differences. It seems that marketers haven’t got to grips with shopping data in enough granular detail to deliver the killer offers that will drive me to automatically purchase without thinking. We may have the data, and even the technology to analyse it, but until marketers move to a digital mindset we’re unlikely to be brainwashed into buying things we don’t even know we wanted.
Related articles
Chris Measures
-
Revolutionary Measures 