Facebook Says Its Ads Work, but Do They Really?

A breakdown of the platform's ongoing saga

Confusion around the efficacy of the social network abounds. Getty Images

Facebook’s ads work. But if you’ve been following the story of the biggest case in the court of public opinion right now, Facebook v. Democracy (aka, the 2016 election), you might believe the social giant’s ads, well, don’t. Which is weird. Because they totally do.

In mid-February, on the day special counsel Robert Mueller dropped an indictment on 13 Russian nationals and three Russian agencies for interfering with the 2016 election by, in part, paying Facebook more than $100,000 to run anti-Hillary Clinton ads, Facebook’s vp of ads, Rob Goldman, took to Twitter to casually explain to any who might listen:

He also linked to a Facebook blog post from 2017 about the ads that were purchased that cited stats including:

  • An estimated 10 million people in the U.S. saw the ads. 
  • 44 percent of total ad impressions … were before the U.S. election … [while] 56 percent were after the election.

Well, you know who was listening? The guy in the Oval Office, who retweeted Goldman’s tweet to his 48 million followers. Naturally, hilarity ensued. Trump supporters spiked the imaginary football, celebrating the fact that the vp of ads for Facebook said there’s nothing to see here. Clinton supporters brandished their imaginary pitchforks, yelling into the ether that Mueller needs to do his job. And media folks poked pretty large holes in Goldman’s argument. For example, total ad spend, whether before or after the election, doesn’t give a precise understanding of the kind of buys each campaign made. 

The CPM party

A week after Goldman’s tweets, Wired ran a story by a former Facebooker Antonio Garcia Martinez about how the ad machinery works. “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,” Garcia Martinez wrote.

Trump’s digital ad guy and now 2020 Trump campaign chair Brad Pascale after reading Martinez’s story and responding to Thompson’s tweet asking about a discrepancy between the Clinton campaign’s and Trump campaign’s Facebook CPMs, tweeted (naturally): 

The Verge’s Casey Newton reported that a Facebook employee said Clinton’s eCPM was 6 cents compared with Trump’s $1.06, but that smells off. He wrote: “So how much did the Clinton campaign pay? Here it gets a bit tricky. Last fall, a member of the Clinton campaign team told me that their CPMs averaged $10 to $30, which they described as typical for a targeted Facebook campaign. But that figure represented the cost only of paid impressions. As described above, ads that perform well can reach larger audiences as they receive likes, comments, and shares — so-called ‘organic reach.’ That lowers the overall cost of the ad.”

Facebook’s vp of VR and AR, Andrew Bosworth, took to Twitter on Tuesday to try and explain how digital advertising is hard, tweeting:


Well, then. A lot to unpack here, but New York Times reporter Mike Isaac succinctly summed it up in a series of tweets: “HRC’s CPM cost was much higher—dollars higher—than Trump’s because her ad engagement was lower.” 

“Trump’s team figured out how to get very low charges because their ads were inflammatory. This ultimately leads to a serious critique of the systems and its incentives: If you are charged less for making incendiary (and perhaps outright false) ads, aren’t you going to make more of those?”

Bosworth came back on Tuesday to Twitter to further clarify things, tweeting: 


The tweet also included a chart showing this. As Bloomberg’s Sarah Frier reported, there’s no mention here of the total effect of the spend because it’s only counting paid reach, not organic. (For a fun tick-tock of all this, head over to BuzzFeed.)

A side note: Facebook’s execs love taking to Twitter to defend Facebook. Interestingly, Goldman was acting as a rogue agent, not given the Zuckerberg seal of approval to be commenting publicly about this, whereas a Facebook rep told Adweek, “Boz was working with us and across FB leadership today,” even putting his tweets up on Facebook.

Brand awareness vs. direct response

One thing to focus on here, I think, is how Facebook is being used. Historically, political ads that ran on TV or in print were “brand awareness” messages—this is how you should view the candidate—and direct mailers on specific issues. However, in the Facebook age, political ads are both brand awareness and direct response. Clinton ran more video on Facebook (brand awareness), while Trump ran more static ads (direct response).

And as Adweek reported today, “Facebook’s endlessly evolving creative ad business has the same goal now as it did when it was first introduced: serving ads that are most likely to get users to click, like, share and spend more time on Facebook … where they will inevitably see more ads. According to Facebook, this is true whether the paid posts are promoting a local pizza shop or a controversial political candidate.”

Which is why Parscale says Trump was the perfect Facebook candidate.

Another interesting thing here, per the Washington Post’s Philip Bump, is that the Trump campaign used the system the way it was designed; Clinton did not. “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.”

The last two weeks have provided two truths about today’s media environment: Facebook’s ads work, and yes, Facebook is definitely a media company.  

@joshsternberg Josh Sternberg is the former media and tech editor at Adweek.