New Times,
New Thinking.

  1. Science & Tech
10 May 2017updated 01 Jul 2021 12:13pm

Will “dark ads“ on Facebook really swing the 2017 general election?

Micro-targeting and personality profiling are set to be a key part of this year’s election – but are claims about their effectiveness exaggerated? 

By Amelia Tait

Since the inauguration of Donald Trump, it has been widely understood that Facebook swayed the outcome of both the United States presidential election and the EU Referendum. Micro-targeted advertising on the site – which is used by political parties to serve highly specific adverts to highly specific groups – has revolutionised politics as we know it.

During the 2015 election, the Conservatives spent £1.2m on Facebook advertising, and this year, Labour plan to spend almost the same. Like broadsheets, billboards, and broadcasts before them, Facebook adverts are a powerful tool for spreading a message to the masses. Unlike broadsheets, billboards, and broadcasts, however, Facebook advertising is at present almost entirely unregulated, leading to wide speculation about what exactly is going on.

“It’s difficult to work out exactly what messages and facts campaigns and political parties are using to gain support,” says Louis Knight-Webb, a founder at “Who Targets Me”, a new plug-in that tracks how targeted adverts end up in your feed. Knight-Webb explains that Vote Leave claim to have spent 98 per cent of their advertising budget on digital adverts, but only those targeted can see the ads and therefore know where the money went. “It’s clear that more research needs to be done.”

Who Targets Me hopes to shed light on who political parties are targeting and how. Amid multiple attention-grabbing headlines about the dangers of Facebook micro-targeting, The Guardian are asking their readers to submit examples of political adverts they see on Facebook, in order to better understand the phenomenon. In the meantime, grand claims are surfacing about the power of the social network and what are now known as “dark ads”.

“You can say to Facebook, I would like to make sure that I can micro-target that fisherman in certain parts of the UK so that they are specifically hearing that if you vote to leave that you will be able to change the way that the regulations are set for the fishing industry,” Gerry Gunster, a political campaigner at Leave.EU, told BBC Panorama on Monday night. The statement has now been extensively quoted in the media as proof of Facebook’s sinister micro-targeting and its role in securing Brexit.

Anyone who as ever used Facebook Ads Manager – the site’s service for creating and targeting adverts – knows that you can choose very specific, custom audiences for your posts. You can target people by location, age, gender, languages, job titles, and their interests. The first five are pretty fool proof – we tell Facebook all of this personal information willingly – the last relies on which Facebook pages you’ve “Liked”.

Only 48,296 people worldwide have bothered to list “Fisherman” as their job title on Facebook, and an additional 17,245,300 people have “expressed an interest in or like Pages related to Fisherman”. If I try to target an advert to both of these groups in, for example, Cornwall, Facebook puts the number of people who will see my advert at less than 1,000. It then warns me: “Your audience is too specific for your adverts to be shown. Try making it broader.” I can continue to post my advert. I doubt, however, that it will affect any referendums.

Give a gift subscription to the New Statesman this Christmas from just £49

There is, in fact, a very strong chance you are one of these 17 million people who have expressed an interest in “Fisherman” – even if you are a receptionist/banker/doctor/person who is deathly allergic to fish. This is because despite all the hype, Facebook targeting can be exceptionally flawed. By clicking on your ad preferences, you will be able to see a list of things Facebook thinks that you like, which it then uses to target adverts towards you. Facebook’s knowledge can be eerie (it knows I like the colour red) but more often than not, it is hilariously wrong. The site thinks I like “pig”, “prohibition”, the sport “curling”, and the concept of “law enforcement”. It is also easily confused. A Twitter user told me the site thinks he likes the film The Sentinel simply because he worked at a local paper of the same name.  

By itself, then, Facebook targeting may not really be the spooky, conspiratorial, election-fixer that many headlines have make it out to be. Yet since January, it has emerged that political parties use other tools alongside the social network. Digital marketing companies such as Cambridge Analytica (henceforth CA) and Aggregate IQ have been working with politicians to better target ads. The former can mine your digital data (things like your online purchases, search history, and even your battery life) to profile you and then target you based on this profile. Simply put, even if you didn’t tell Facebook you’re a fisherman, CA can allegedly figure this out.

In January, Motherboard reported that CA were playing with psychographics by measuring people’s psychological states and personalities and then using this information to micro-target political adverts. The implication was that the company could predict anyone’s political leaning and therefore sway their votes – something many claimed lead to the election of Donald Trump. Yet CA now claim they never used psychographics for the Trump campaign and former employees have spoken out to say the company’s claims are exaggerated.

“There are a lot of big claims made by firms trying to win contracts,” says Who Targets Me founder Knight-Webb. Many have taken CA on its word about what it is capable of, leading to great publicity for the firm. “There is absolutely not enough transparency around how the big data firms helping political parties are using data. I would love to see the real level of detail big data firms, like Cambridge Analytica and Deep Index, have on citizens and just how effective they are at using that data in targeted marketing strategies.”

The kind of profiling done by CA is actually already a widespread practice. That isn’t to say it isn’t creepy, but there is as yet little evidence that it is as effective as CA claim. As Frederike Kaltheuner writes for Privacy International: “It’s one thing to profile people, and another to say that because of that profiling you are able to effectively change behaviour on a mass scale. Cambridge Analytica clearly does the former, but only claims (!) to succeed in the latter.”

As Kaltheuner points out, basically every party is using profiling and micro-targeting, meaning that it’s incredibly easy to retroactively shout “It works!” based on whoever wins. (Note: CA originally worked for Trump competitor Ted Cruz.) As it stands, it feels bold to claim that this type of campaigning will completely sway the 2017 general election in any party’s favour. Yes, Facebook will be a key battleground in this election, and yes, Facebook advertising will have a big part to play. But is it as grand as mind-control conspiracies make it seem? Can it really change an election which already seems so cut and dry (not to mention dull)?

Just because micro-targeting and profiling isn’t that grand, however, doesn’t mean it shouldn’t be reformed before it gets better. The Guardian revealed yesterday that Facebook have hired ex-political aides to help create political campaigns on the site, meaning that without regulation the site’s influence on elections will only increase. Greater transparency is needed from Facebook, data companies, and political parties in order for the Electoral Commission to affectively scrutinise campaigns.

It seem likely, however, that it is money – rather than pseudo-psychological profiles – that will ultimately affect the election’s outcome. The party that pours the most money into Facebook will see the greatest results, simply because more money equals more people reached. In Philip Cowley and Dennis Kavanagh’s comprehensive book The British General Election of 2015, they reveal how budget differences meant Labour reached 16 million people on Facebook in their best month, compared to the Conservatives reaching 17 million each week. Ultimately, Knight-Webb hopes Who Targets Me will provide enough information to validate or debunk the widely-circulated claims about Facebook targeting. Until then, dark ads remain in the dark. 

Content from our partners
Building Britain’s water security
How to solve the teaching crisis
Pitching in to support grassroots football

Topics in this article :