Using data to shape your campaigns could leave you open to accusations of sinister scientific profiteering.
If the allegations are true, the EU referendum was as good as rigged. In a series of explosive articles for The Observer, investigative journalist Carole Cadwalladr argued that the Leave campaign conducted “military-grade psychological warfare” on voters.
She alleges Vote Leave harnessed hundreds of millions of datapoints to create micro-targeted messages. A data company called AggregateIQ supplied the numbers. Then, with pinpoint accuracy, the public mood was swayed.
Her narrative is like something out of a crime thriller. She tracks down a second data-analytics company, Cambridge Analytica, and connects all the links to a shadowy American billionaire.
Mysterious characters give her tip-offs. Ex-military officers are everywhere. One key source turns out to be the daughter of Google’s chairman. And who masterminded Trump’s presidential campaign? Only the same two firms – AggregateIQ and Cambridge Analytica.
Phew! So, did Vote Leave really use military-grade psychological warfare? And, if it did, can, and should, we all do it?
There’s no doubt Vote Leave put data at the heart of everything. Campaign director Dominic Cummings has said 98% of his budget went on data and online ads, and much of this was handled by AggregateIQ.
Cummings revealed his methods in a personal blog. He wrote: “In the official 10-week campaign, we served about one billion targeted digital adverts, mostly via Facebook, and strongly weighted them to the time period around postal voting and the last 10 days of the campaign. We ran many different versions of ads and tested them. We combined this feedback with polls (conventional and unconventional) and focus groups to get an overall sense of what was getting through.”
To manage this blitzkrieg, Vote Leave built its own tracking system called VICS, named after Victoria Woodcock, the campaign’s respected operations director. VICS tracked the performance of online ads. The best were tweaked and run again, and the worst ditched.
Cummings explained: “One of our central ideas was that the campaign had to do things in the field of data that had never been done before. This included integrating data from social media, online advertising, websites, apps, canvassing, direct mail, polls, online fundraising and activist feedback.
“We also tried a new way to do polling and we had experts in physics and machine learning use proper data science in the way only they can. We went far beyond the normal skills applied in political campaigns.”
VICS created a star-rating system to label streets and then directed the 12,000 volunteers to the most receptive areas. The focus on swing voters was relentless.
Impressive. But you may be thinking: “That’s what campaigning looks like. Where’s the military bit?” Here the story wobbles. Cadwalladr admits there’s no proof of any “special sauce”.
Industry leaders laugh off the claims of military-grade manipulation.
“If it is possible, I’d love to meet the creators,” says Sophie Wooller Dent, director of data and technology products at iProspect, part of the Dentsu Aegis Network. “Arguably, this is what politicians have always done: understand local demographics and issues, and… try to persuade [constituents] to vote for their party.”
Ed Arnall-Culliford, strategy director at media buying agency SearchStar, says the military links are “not so much paranoia as hyperbole. Just because Cambridge Analytica works with the military doesn’t necessarily mean the work it does for political lobbyists is inherently militaristic. Serco has catering contracts with the MoD and NHS, but hospital food is rarely called military-grade”.
“It’s perfectly normal these days,” says Jim Hawker, founder of PR agency Threepipe. “This is the new reality of comms and it makes complete sense to create highly personalised content.”
A more plausible explanation for Vote Leave’s success is overlooked by the conspiracy theorists. It wasn’t just that Vote Leave used data analytics. It’s that the team trusted the numbers.
Cummings’ people knew, for example, that their campaign slogan ‘Go global’ was “a total loser with the public”. For a while, politicians, unfamiliar with a numbers-driven strategy, didn’t grasp this.
Cummings is brutal about the failings of MPs, who prefer to “fail conventionally” than use data science to win. He says: “They were very happy to be on the Today programme. But they didn’t want to win that much. Not enough to work weekends. Not enough to stop having all their usual skiing holidays and winter beach holidays.” He demanded the campaign ignore MPs and admen.
Only the numbers mattered.
The rest of the success was down to old-fashioned comms and marketing – for example, getting your timing right. For Cummings, there is one constant truth: messages are more effective the closer they are to the moment of decision. So, the most popular message of all – about immigration – was held back until the last moment. It became the campaign’s “baseball bat” with which to bludgeon home the win.
By contrast, the Remain campaign was shambolic. David Cameron’s aide Daniel Korski wrote a post-mortem lamenting the “frustratingly basic” reasons for the loss. The campaign relied on old-fashioned polls rather than data analytics. It was analogue versus digital.
The killer fact is that Vote Leave found slogans that resonated. Data led to the right messages. But micro-targeting of individuals wasn’t really necessary, as the slogan ‘Take back control’ was a winner across the target demographics.
It’s clear that micro-targeting is genuinely effective, and can influence consumer behaviour. But as a method of hijacking democracy? There’s no evidence of that.
Charles Orton-Jones is an
award-winning business journalist
This article was originally published in Influence magazine, Q3 2017.
Picture credit: Farzad Nazifi