Micro-targeting on Facebook to win elections

When Donald Trump was elected US President in November last year, it ushered in more than just a new era of political turmoil. It also signalled a potentially dangerous new era in social engineering. A big part of the reason why Trump won is the incredible job done by his digital team, who used a new form of micro-targeting to push their messages, influence potential voters, and raise money. Much of this was done on social media, and in particular Facebook.

In the final weeks of the Presidential campaign, Hilary Clinton outspent Trump in TV adverts – traditionally a key factor in swaying voters. She spent $72m on TV ads, compared to $39m by Trump. However, despite raising significantly less money overall than Clinton, Trump’s team spent big on digital at the end of the campaign. While Clinton dropped about $16m on the Internet, Trump spent $29m. Trump’s Digital Director, Brad Parscale, decided to use most of that on micro-targeting. In other words, adverts that targeted a very small number of potential voters. For example, he authorised a series of Facebook ads to certain demographics in key battleground states like Michigan and Pennsylvania. These weren’t just ‘vote for us’ ads either, they were also ads to dissuade voters from voting for Clinton (a.k.a. attack ads).

To illustrate how precisely targeted his ads were, Brad Parscale sent out this tweet at the end of last year:

“CA, TX, NY = nearly 0 budget.
FL,NC,NH,OH,PA,WI,MI,CO = 95% budget. Within these states it was targeted to 14MM specific voters.”

That tweet probably referred to combined TV and digital expenditure, but the point remains the same: only 14 million voters (just over 4% of the US population) were worth advertising to for the Trump campaign. Their goal wasn’t to go wide, but to aggressively go after the vote of certain people in swing states who they thought they could influence. They completely ignored voters in states they knew they wouldn’t win, like California.

What Trump’s digital team did was a big departure from previous Presidential campaigns. Micro-targeting used to mean gathering demographic and consumer data from sources such as magazine subscriptions and club memberships. A campaign would then do a bunch of prime time TV and newspaper ads, based on which TV programs and newspapers the desired demographics favoured. But social media has changed all that. It has given us much more refined ways to determine political leanings. Namely, Facebook likes, Twitter follows, Instagram hearts, and other social media data points.

In the UK’s controversial Brexit vote last year, Facebook likes turned out to be crucial. Andy Wigmore, the communications director for Leave.EU, told The Guardian that a Facebook like was their most “potent weapon.” It told the Leave campaign “all sorts of things about that individual and how to convince them with what sort of advert.” What’s more, it gave them useful clues to what their friends liked as well. This data turned out to be gold.

Identifying which potential voters are most likely to be swayed by an advert is one thing. But micro-targeting in the social media era goes a step further: it also allows you to hone your message into the most actionable one possible. A/B testing (using several variants of an advert and seeing which one gets more clicks, likes or another measurable action) is commonly used in Facebook advertising. According to Gerrit Lansing, Chief Digital Officer of the Republican National Committee (RNC), Trump’s campaign went through between 40,000 to 60,000 ad variations per day, with 175,000 iterations on the best day of the campaign. This level of A/B testing worked because, as Gary Coby from the RNC put it, Facebook users are “conditioned to click and engage and give you feedback.” In other words, we Facebook users are no better than laboratory rats when it comes to giving campaigns like Trump’s the data they want.

Micro-targeting isn’t just a trend in politics. It’s also a big deal in business, particularly retail. I myself have tried my hand at micro-targeting, in an attempt to increase sales for my self-published science fiction novel, Presence. Since my novel was about Virtual Reality (VR), I created a Facebook ad that targeted a very specific audience: people who had ‘liked’ the author Ernest Cline, whose VR novel Ready Player One was a bestseller. Unfortunately, my micro-targeting attempt was considerably less successful than Trump’s. Probably because what I was selling was less compelling (plus I neglected to do a series of provocative tweets to complement the ad campaign).

Regardless of my own lack of success using micro-targeting, it is here to stay and is particularly powerful for political campaigns. New Zealand has a general election coming in September, and you can bet the main political parties will use social media micro-targeting to influence voters. And to be frank, we only have ourselves to blame. By using social media products – especially Facebook – multiple times every day, we are constantly revealing useful data points about our political leanings. In effect, Facebook and Twitter have conditioned us to give political parties (and businesses) the data they most want: what we ‘like’, what we willingly follow, how we respond to certain messages, and so on.

The result: organisations like the Trump and Brexit campaigns know precisely which information will provoke a desired response in us. They also know that eventually, just like laboratory rats, we will nibble their dangling cheese. Bear that in mind next time you head to a voting booth.