Since the 2016 campaign ended, Parscale has been investing a lot of his digital marketing energy in another client: himself.
Over the weekend, that meant plugging Wired’s extensive look at how Trump’s campaign effectively used Facebook. And, in response to a question posed by Wired editor in chief Nick Thompson, a bold claim about how much more effective his campaign’s ad spending was, relative to Hillary Clinton’s.
What Parscale is saying is that Clinton’s campaign paid hundreds of times what his campaign did to run 1,000 Facebook ads. (CPM is “cost per 1,000″ — as in, how much Facebook charges an advertiser to show an ad 1,000 times.) As important, though, is Parscale’s explanation of why: Trump “was a perfect candidate for FaceBook.”
The natural response to Parscale’s claim is shock. One candidate paying hundreds of times as much for the same platform? Is that even legal? This concern, though, misses the point.
How Facebook advertising works
The Wired article was written by Antonio García Martínez, a former product manager for Facebook’s ad targeting team. Martinez himself worked on the social platform’s Custom Audiences tool, but the central focus of the Wired article is Facebook’s ad auctions.
We’ve looked at Custom Audiences before. Campaigns have lists of voters that they want to target, and Facebook allows campaigns to match voter lists to Facebook users — meaning campaigns can now target specific voter groups with ads on Facebook. For campaigns looking to wring the most value out of ad spending, targeting voters specifically (without wasting ad views on people who aren’t registered to vote, for example) is valuable.
Facebook’s targeting tools extend beyond that, of course. Martínez explains that the Lookalike Audiences tool expands a custom audience to identify people who share similar characteristics as the originally targeted user. Maybe they’re not registered to vote, but maybe they want to buy a “Make America Great Again” hat anyway.
Those ad auctions, though, are where Parscale’s argument largely comes into play. Potential advertisers on Facebook submit an ad and a price they’re willing to pay for a particular target audience — say, married women in Des Moines.
Rather than simply reward that ad position to the highest bidder, though, Facebook uses a complex model that considers both the dollar value of each bid as well as how good a piece of clickbait (or view-bait, or comment-bait) the corresponding ad is. If Facebook’s model thinks your ad is 10 times more likely to engage a user than another company’s ad, then your effective bid at auction is considered 10 times higher than a company willing to pay the same dollar amount.
That’s the central point. If Trump’s content is better clickbait, it costs less, because Facebook wants to reward engagement so that it can keep people sticking around to view more ads.
Jennifer Palmieri, director of communications for Clinton’s campaign, turned heads when she replied to Parscale’s tweet with one cryptic word.
What she meant, she explained to The Washington Post, was that she agreed with Parscale’s evaluation that Trump was a “perfect candidate” for Facebook: His content was engaging in the way that Facebook rewards, for better or for worse.
Why the issue of cost doesn’t tell the whole story
It’s hard to evaluate Parscale’s claim that his team spends hundreds of times less than Clinton’s, given that it is speaking in broad strokes about hyperspecific valuations of Facebook ads. People I’ve with whom spoken and people responding to his tweet suggested that Parscale’s estimate is inflated significantly, based on their experiences with targeting advertising on Facebook. But there are several reasons Trump’s ads might be expected to be quite a bit cheaper.
Trump and Clinton were targeting different audiences
Facebook Vice President Andrew Bosworth compared Facebook’s ad marketplace with the market for other forms of advertising.
The issue isn’t that the candidates were paying different prices for the same sort of advertisement. If it were illegal for ads to cost different prices based solely on the medium, then a television ad running in prime time in Manhattan would cost the same as an ad running at 2 in the morning in Missoula, Mont. You pay more or less to reach different audiences.
Both campaigns targeted swing-state voters, but Trump also ran fundraising ads aimed at red states, and Clinton ran ads targeting blue states. It’s much cheaper to Trump to show an ad to users in Nebraska than it is for Clinton to show an ad to people in Silicon Valley.
They were running different sorts of ad campaigns
Not only were the audiences different, but the ads themselves were, too.
“Fundamentally, we were trying to do two different things,” said Andrew Bleeker, a senior adviser on advertising to Clinton’s campaign and president of Bully Pulpit Interactive. “It’s not that one was right and one was wrong, but they were different. They shouldn’t cost the same.”
Clinton ran a lot of video ads, aiming to boost turnout or make the case for her candidacy. Trump’s ads were often single images with an enticement to click. The ads were designed to do different things.
“When I look at cost, I don’t necessarily care at all what he’s paying,” Bleeker said. “I care: What would it cost me to get this video in front of people on television or any other format? This is still a good deal.”
Trump’s campaign spent a disproportionate amount of its advertising budget on digital, outspending Clinton on digital ads as of September 2016, according to data from Pathmatics. Many of those ads were fundraising pitches aimed at doing things like encouraging people to buy campaign gear. By late October, the campaign had spent more on Trump hats than on polling.
In an interview with CBS’s “60 Minutes” last year, Parscale explained that the Trump campaign ran thousands of versions of its ads each day — up to 100,000 iterations of an ad on one day — in which language and visuals were tweaked to entice as many people as possible to click. Put another way, the Trump campaign’s use of static images that could be tweaked in infinite, subtle ways gave it an advantage for proving to Facebook’s ad auction tool that it would earn clicks and comments.
What’s more, the Trump campaign had Facebook staff embedded with their team who helped guide their ability to maximize their investment. Given that Trump’s team spent tens of millions of dollars with the company (Parscale told CBS that Facebook ate up 80 percent of his $94 million ad budget), that clearly made business sense.
Clinton’s ads may have attracted more trolls
But there’s a corollary here that’s interesting to note. The ad auction system will raise costs for advertisers whose content generates negative feedback; it’s the flip side of rewarding content that gets good engagement. And Clinton’s content attracted a lot of negative comments in the sharply polarized environment of the 2016 election. Online trolls may have cost Clinton more money to advertise. That’s true of negative comments on Trump’s ads, too, but his efforts to game clicks almost certainly made that less of a factor.
Parscale’s celebration of the discounted rates he paid for Facebook ads is, in part, a celebration of his firm’s ability to deliver for clients. In an election that came down to 78,000 voters in three states, though, he can fairly make the case that his ability to slap Trump ads in front of millions of Americans — even simple ads mostly meant to sell merchandise — might have determined the outcome of the election.
The pricing isn’t interesting. The strategy and efficiency underlying that pricing is.
Update: There’s another reason to be skeptical of Parscale’s claim. According to data released by Facebook’s Bosworth … it’s wrong.
At least broadly. It’s likely true that on specific campaigns, as above, that Trump’s costs were lower. But the Trump team’s heavy focus on fundraising and engagement means that their return on investment in terms of donations, shares and clicks may have been higher.
In a statement to The Post, Bosworth explained how different Facebook advertising campaigns have different goals.
“Direct response oriented campaigns can often afford higher CPMs than traditional brand campaigns because they know the result they are getting is worthwhile,” he wrote.
Bosworth was explaining the economics in general. But the distinction he draws is important for our purposes: Trump was running a campaign aimed at getting people to do things on Facebook, and they were successful. Clinton was trying to convince voters and get them to vote.