How Trump Used Facebook in 2016 and Why That Tactic Is Coming to Every Important Race This Year

In politics, winning is everything. And the political consulting industry scours the victor’s tactics so they can be sold in the next election cycle, whether they worked, were happenstance, or a mix of the two.

Exhibit A for this dynamic is the 2016 Trump campaign’s use of digital media, especially Facebook. In late February, President Trump named Brad Parscale, his digital director, as his 2020 re-election campaign manager. In the meantime, Parscale has been helping the Republican Party raise millions from small donations online.

“It begs the question, why don’t more campaigns have their digital consultant serve as their campaign manager,” wrote Peter Graves, a media consultant and ex-RNC regional political director for the trade publication, Campaigns and Elections. "After all, most federal campaigns are doing the majority of their voter contact now online. Moreover, digital is the only space on campaigns that collaborates with every department.”

But what did Parscale do for Trump in 2016 that is setting the standard for 2018?

“If I had to characterize what the Trump campaign did, they automated Facebook’s best [advertising] practices at a very large scale. I think that statewide campaigns will try to do the same kind of thing in 2018,” said Colin Delany, founder of Epolitics. “I’m wondering if any new technology is going to come out of the resistance. It’s possible. I think a lot of what you are seeing is training and capacity building.”

In other words, Democrats are playing catchup to Republicans, compared to 2016, when their presidential candidate, Hillary Clinton, and the Democratic National Committee—under a series of chairwomen—were more focused on older media platforms like TV, rather than new platforms like social media.

To begin to understand how Parscale, an online advertising specialist, went from building websites for Trump products, to running Trump’s online messaging and fundraising in 2016, to assisting the RNC with small donor outreach in 2018, to being named Trump’s 2020 campaign manager, you have to examine how Parscale used Facebook.

Consultants like Delany are quick to say they can’t tell you how the Trump campaign’s messaging specifically affected voters when they cast ballots. But they can say how Parscale used Facebook, which revolutionized how national campaigns targeted and messaged likely voters.

If you roll back the clock to before the internet, a baseline emerges showing how much has changed. When I was press secretary on the 1990 campaign that first elected Bernie Sanders to the House of Representatives, I remember picking up a half-dozen TV ads for the campaign’s final stretch. Bernie looked at the stack of 30-second videos and said his future rested on those tapes. That was a statewide race in a rural state with small media markets, but compare its scale to what Parscale and Facebook did for Trump.

It’s no secret. An October 2016 in-depth report by Joshua Green and Sasha Issenberg for Bloomberg.com laid out the basics. It started with identifying persuadable voters in swing states, and then sending messaging poking at their biases and personality types. More recent reports, such as this weekend's Guardian report featuring a Trump campaign whistleblower, added more detail showing how powerful Facebook's user profiles were in 2016.

The Trump campaign took voter files from the Republican Party, which had people’s electoral histories—age, gender, address, party registration and voting histories. They took more personalized data by Cambridge Analytica, the data-mining firm run by hedge fund billionaire Robert Mercer, which boasted it had psychological profiles of every voter. (On Friday, Facebook suspended Cambridge Analytica for not deleting data taken from the social network in 2016 to create its voter profiles. The Guardian reported Cambridge Analytica's data was culled from 50 million to 60 million Facebook user profiles it had stolen before it joined the Trump campaign.)

The campaign put all that voter profiling data into Facebook’s advertising supercomputers, which had its own deep profiles of users based on their pages, communications and networks. Facebook provided staffers to every major campaign in 2016 to facilitate this process. From that data merging and data mining, Parscale oversaw an operation that identified 13.5 million Facebook users in 16 states that it deemed were persuadable. It targeted those individuals with the kind of personality-driven content that social media emphasizes—more opinion, belief and bias than facts.

The Trump campaign didn’t just send out original content, noted Melissa Ryan, who writes the Ctrl Alt Right Delete newsletter. “The Trump campaign put out relatively few ideas of their own,” she explained on Medium.com. “Instead they chose to validate and amplify the conversation happening online. They didn’t need to confirm or deny anything, they simply gave Trump’s army their megaphone… [It] put their supporters one content front and center, taking a backseat in their own campaign.”

“They were banking that people are more likely to believe something from an unofficial source that seems true rather than to believe something from an official source that actually is true,” Ryan explained. “Once they realized there was no line to cross, they went all in using social media to bully and intimidate others, spread misinformation, and mainstream the language of hate. Those who spoke out were met with a flood of abuse, and that abuse was organized. Their targets had few places to turn as tech companies turned a blind eye, and law enforcement was of little help.”  

By late October, days before former FBI director James Comey announced the FBI was reopening its investigation into Clinton’s handling of classified emails, the Bloomberg reporting team said Parscale’s crew of subcontractors had determined “which social media ads [messages, memes, shares, etc.] were most effective. Those companies test ad variations against one another—the campaign ultimately generated 100,000 distinct pieces of creative content—and then roll out the strongest performers to broader audiences.”

Facebook allowed Parscale to build and target what it calls “custom audiences,” as ex-Facebook executive Antonio Garcia-Martinez explained in a Wired piece, “How Trump Conquered Facebook—Without Russian Ads.” Its custom audience consisted of all the impressionable voters—Trump enthusiasts on one hand, Democrats dubious about Clinton on the other—who could be encouraged to vote or encouraged not to.

Then the campaign used another Facebook advertising feature built atop that preliminary customization called “Lookalike Audiences.” It targeted another wider circle of Facebook users in the swing constituencies in swing states. Some of this was to keep Trump’s base engaged. Some was to discourage three key columns of Clinton supporters from voting. As Bloomberg noted, “idealistic white liberal, young women and African Americans” were the targets of Trump’s three voter suppression campaigns.

Garcia-Martinez, the original product manager for "Custom Audiences,” explained how the Trump campaign targeted audiences and then fed it content, often propaganda from other online sources. This explains how Breitbart.com temporarily had more traffic than Fox News in the campaign’s final weeks. But Garcia-Martinez’s explanation also alludes to platform features that are designed to be psychologically engaging and provoke emotional responses. It starts with creating political echo chambers filled with like-minded associates. 

“We’ve all contributed to this political balkanization by self-sorting (or being sorted by Facebook) into online tribes that get morphed into filter bubbles, which are then studiously colonized by commercial memes planted and spread there by a combination of Custom and Lookalike Audiences,” he wrote. “One of the ways the Trump campaign leveraged Lookalike Audiences was through its voter suppression campaigns among likely Clinton voters. They seeded the Audiences assembly line with content about Clinton that was engaging but dispiriting. This is one of the ways that Trump won the election, by the very tools that were originally built to help companies like Bed Bath & Beyond sell you towels.”

Colin Delany, the online media consultant and author on the topic, is quick to point out that it’s impossible to trace how all this messaging affected voters’ behavior when they cast their 2016 ballots. That’s because the thread Parscale could observe—whether messaging recipients liked, commented on or shared his handiwork—was broken as voters put down their devices and picked up ballots.

But Delany said there was another metric that confirmed how powerful and engaging Facebook was before 2016’s Election Day. It helped raise a fortune for Trump.  

“Now one of the main things with Facebook is this is one of the few operations that I’ve heard of that’s been able to raise a significant amount of money by targeting people on Facebook,” he said. “Now my suspicion is that almost all the money actually got raised by email. But they would identify and recruit the people via Facebook.”

“If you can get them to respond to the first fundraising pitch and give you $5, then you are in their email program. Or you get them to sign a petition. You get them to do whatever it is that gets them the email address,” Delany said. “The small donor lists that the RNC is using to pretty good effect right now comes out of the work that Parscale did for Trump. If you want to look for results, I think you can look to social media fundraising.”

Seen from that metric, there’s no denying the political power of social media. With 10 days to go before Election Day, Bloomberg reported that Parscale’s operation raised $275 million from 2.5 million small donors. That’s more than the nearly $238 million raised by the Sanders presidential campaign from his grassroots base. Think about that: Parscale’s operation outraised Bernie’s operation—but Parscale isn’t bragging about it.

The Clinton campaign, as many political insiders and former Facebook executives have said, didn’t use Facebook as aggressively. In one intriguing series of recent tweets, these tech and political insiders said that Democrats in general rely more on TV ads than online messaging, partly because political consultants can earn more that way. They also said her campaign was relying on dated internal assessments by Facebook (from 2010) about its effects on voter participation, and that overall, its campaigns discounted—and didn’t sufficiently study—“observational political science,” or how voters responded to social media provocations in focus groups.

These cringe-worthy observations are part and parcel of political campaigns, which are like startups that rapidly pop up and then mostly disappear after Election Day. Political campaigns are perpetually in search of messaging and fundraising infrastructure they can use; just as institutions like Facebook, YouTube and Twitter have created infrastructure seeking new users. Garcia-Martinez, reached by email, said he was stunned by how little political professionals—and academics who study politics—know about advertising.

“My recent [Wired] piece tossed me into a few threads with political/election/academic Twitter, and I found it a very strange world,” he wrote. “Very punctiliously following strict experimental protocols and the scientific method, and also not knowing the first f**king thing about how to advertise on Facebook (their studies measure what they can as outsiders, which isn't much). It is a leap of faith to just assume Facebook ads work for politics.... but without question, they work for more commercial direct-response advertisers.”

His takeaway: “Parscale is hailed as this genius, but all he did is what any reasonably savvy ecommerce operation would have done, and that set him way beyond Hillary and the supposedly ‘best and brightest.’ Which is why, as Delany and Graves, campaign media consultants and columnists, said, social media will be at the center of 2018’s elections for control of Congress and constitutional offices in swing states.