Facebook has been in the news this week, in no small part because one of its executives took to Twitter to give his opinion about how using ads on the platform did not swing the election:
His tweet caught the attention of the President, and has been the subject of much discussion in the news. What is not up for debate, in my mind, is the efficacy of paid Facebook ads to influence people socially and politically. I know because I’ve done it.
Donald Trump has won the election and the country is reeling. My friends and I are in a state of shock. I’d told everyone who’d listen to me that Hillary would carry PA by 5 or 6 points and not to worry about tightening polls. Things are not good.
In countless postmortems of the election one thing kept coming up — Facebook ads. I am a digital marketer, and have been working with Facebook for years. I decided that I had to do something in the wake of this disaster. I got a few like-minded friends together, and we hatched a plan.
My theory was that Facebook advertising could be used to turn outrage and despair into something positive.
I was willing to put as much of my own money on the line as it took to see whether I was right. Our group discussed a few pressing topics and, as a child of educators, I chose one that was close to my heart — the nomination of Betsy DeVos to Secretary of Education.
She was unqualified and she had horrific ideas about education. We’d spoken to a few political strategists who told us that the Senate confirmation vote would come down to three Dems who were likely to waver — Joe Donnelly of Indiana, Heidi Heitkamp of North Dakota, and Joe Manchin of West Virginia.
The goal of the ad campaign was to convince people to call their Senate offices and tell them to vote No on a confirmation. I registered the domain dumpdevos.com anonymously, set up a Facebook page, and we were off. I needed content for our ads, so I wrote up a 60 second script and a 30 second version, pulled some public use images and Shutterstock art, and sent it off to a voiceover artist and video editor on Fiverr.
The total cost for my creative materials was $128.
I received the final cut of my video, and started up my ads. Despite what some con artists may tell you, I prefer a broad demographic approach to online messaging . Rather than complicate matters, I decided to target everyone in each of the three states between the ages of 28 and 44. No political affiliation sliders, no audiences, no Facebook machine learning wizardry. Every person within the age range, Republican, Democrat, or otherwise.
I set it to run during the Senators’ office hours, on the two critical days before the vote. Each ad was set to spend a whopping $50. I looked at the results when the ads stopped and the vote was SUPPOSED to be held. My jaw hit the floor.
After spending $150 to run ads, the results were in.
Since I was working solo on a shoestring budget over a holiday season, I hadn’t had the time to put together a system to track who’d dialed in from the ads. I’d created “Video Views” advertisements with a link to each Senator’s official webpage and their office phone number listed prominently in the ad copy. I did not expect much engagement with the ad itself, but hoped to get a decent number of views.
To provide some context, a good direct response Facebook video ad might see somewhere between 1% and 3% CTR (Click-Through Rate), meaning people who click on a Call to Action button, as seen here:
What I saw was dramatically higher, across all age groups.
These were West Virginians ages 28 to 44, and almost 20% of them were not only watching the clip, they were clicking through to call Joe Manchin’s office. I’ve been doing advertising for a long time, and I’ve never seen that level of engagement on paid ads. We were in viral video territory.
Also surprising was the amount of time people spent watching the videos. It’s well known that Facebook pads the video engagement numbers but I had years of my own ads for comparison. My best video ads might get an average 40% of viewers watching the entire video, but the DeVos videos averaged over 60% full watches, and went as high as 79% in one of the Indiana ads (you’re welcome, Joe Donnelly).
The vote got pushed back another week due to fears the Republicans didn’t have the votes to confirm DeVos, so I decided to double down on my success. In the interests of science, I added a new wrinkle. My next ads would target millenials. I recreated the ad sets, this time targeting the same broad demos in each state, with an 18–27 age range. The kids don’t vote, I told myself. They won’t call. It was worth another $150 to find out.
North Dakota became my outlier, with engagement closer in line with what I’d originally expected. Indiana and West Virginia, however, remained practically unchanged, with a 16–17% CTR on an ad to call their Senators about a cabinet pick. My mind was blown.
In the end, as we all know by now, Betsy DeVos became our Secretary of Education. She passed the Senate with a historic tiebreaking vote cast by Vice President Mike Pence. A tiebreaker had never been needed before in a majority controlled Senate. The targeted Democratic Senators held strong. One of them even tweeted direct evidence of how effective our ad campaign had been:
1,400 North Dakotans had called Heitkamp’s office.
That’s funny, I had some stats from the ads:
996 people had clicked the Call-to-Action link, and an unknown number had called directly from the ad itself; all calls went directly to Heitkamp’s Senate line. Even conservatively estimating the number of people who’d followed through, Facebook accounted for at least half of the calls to her office, possibly close to 75%. She’d cited those calls as the reason she voted no.
It cost $128 to cut and host a video ad. It cost $349.85 for the ad spend (no idea why Facebook didn’t want my last 15 cents).
For less than $500 I’d arguably secured at least one No vote in the Senate against Betsy DeVos, maybe as many as three. Engagement had been absolutely off the charts, and they had nothing to do with voting or an election. People were mad, and they were looking for any way to try to fight back.
Hot-button issue ads like these can be used to influence any contentious debate, and we have no shortage of those in America. This campaign was completely anonymous as far as my audience was concerned; though my site, message, and ads were consistent with my actual beliefs, they had no way of knowing whether I was a real American, an actual political organization, or a foreign entity manipulating them for more sinister reasons. It’s also worth noting that despite Facebook’s recent attempts to verify political advertisers, this sort of ad copy would not fall under their new rules, since it’s issues-based and not supporting a particular candidate.
Rob Goldman of Facebook has already apologized for his comments, and the company may be finally coming to grips with the seriousness of the problems it faces going forward. It’s clear from results like mine that Facebook is an incredibly powerful platform to influence behaviors, and it must find a way to wield this power responsibly.