Campaign insiders and digital advertising specialists across the country spent hours Friday trying to sort out whether U.S. Sen. Cory Gardner had deliberately kept Colorado residents from seeing a digital ad touting the Republican senator's bonds with President Donald Trump — or, as the Gardner campaign swore up and down, Coloradans hadn't seen the ad because of a complex formula that determines which ads appear in front of which Facebook users.
It's hard to say what drove the bickering, sparked by a HuffPost article posted before dawn that suggested Gardner was shielding his constituents from the very sight of the pair.
The Gardner team declared the matter moot, based on partial information and a poor understanding of the campaign's digital strategy, while gleeful Democrats celebrated the discovery and asserted that their opponents' denials only proved the case.
Call it the mystery of the missing Facebook ad impressions.
The HuffPost article got plenty of attention off the bat, from national outlets, the Colorado Democrats and staffers with former Gov. John Hickenlooper, Gardner's Democratic challenger, who made hay with the incumbent's apparent attempt to banish Trump from the state's social media pages.
After the Gardner communications team had stopped laughing at the hypothesis — no one dominates the news or social media like Trump, and Democrats and their supporters are spending millions of dollars to make sure no one in Colorado thinks of Gardner without associating him with Trump — they attempted to set the record straight, but every explanation they offered only provoked scornful attacks on social media from Democratic-aligned digital media experts.
This much is beyond dispute: On Aug. 10, the Gardner campaign began running a Facebook ad featuring the senator and Trump, "asking Patriots nationwide to show that they support the pro-growth, conservative agenda that President Trump and I are fighting for."
The ad introduced a brief video set to a bouncy instrumental as a photo of Trump and Gardner dissolved into a list of the "REAL results" the campaign boasted had been delivered by the president, including tax cuts and a secure border. "Add your name," the closing graphic instructed, making clear the purpose of the ad — to build the list of the Gardner campaign's supporters.
Also uncontested was the tale told by a map of the United States that appeared on the ad's Facebook Ad Library page, which also revealed the dates the ad was active, how much the advertiser spent, and the number of impressions it had garnered, next to a tiny image of an eyeball. Each state was filled with a shade of purple, indicating the state's share of the ad's total impressions — except for Colorado, the only un-tinted state, meaning few, if any of its Facebook users had seen the ad.
The dispute raged Friday over the explanation for the ad's trajectory, with the Gardner campaign insisting Coloradans weren't seeing the ad because they had already seen it, months earlier, while the Republicans' critics argued the account made no sense and was merely a clumsy attempt to spin some news that wouldn't be that bad, if only the Gardner camp would take its lumps.
When HuffPost came across the ad and its map — political reporters make a habit of perusing Facebook ad data to see what messages politicians are dishing up — it appeared to bolster an assumption that has been hardening lately among Democrats.
According to the narrative, endangered Republican senators — Gardner is routinely listed as the most vulnerable Republican senator up for election this year — have been quietly distancing themselves from Trump as the election approaches, hoping to salvage their careers even as the president's popularity slides amid an enduring pandemic, a damaged economy and rising dissatisfaction with the county's direction.
Except that the Gardner campaign ran a nearly identical ad for about four months earlier this year, from late February to early June, and far from hiding it from Colorado voters, that ad — according to its Facebook Ad Library data page — had been seen only by Facebook users in Colorado, and by hundreds of thousands more users than saw this week's ad.
According to the Facebook data, the new ad registered around 20,000 impressions nationwide at a cost of a few hundred dollars. The earlier ad — its only difference was listing the Bureau of Land Management's headquarters' move to Grand Junction among Trump and Gardner's results — had been seen by around 850,000 people at a cost of around $10,000. (Both iterations of the ad were posted several times, a common practice in robust digital campaigns, so advertisers can track results based on multiple parameters.)
Over the last 30 days, according to Facebook data, Gardner has spent about $70,000 on Facebook advertising, slightly lagging the roughly $75,000 spent by Hickenlooper.
Like all candidates at Gardner's level, his campaign runs dozens and dozens of digital ads, targeting different audiences and tailored to produce different results. Most run for a brief period — some for just a day — and are replaced if they don't perform, while a handful stay up for a long stretch. The Trump-Gardner ad that appeared in Colorado falls in the latter category, appearing to have been one of the Gardner campaign's most successful and longstanding ads.
And that success, said Meghan Graf, the Gardner campaign's press secretary, helps explain why the version that ran this week didn't show up for Colorado Facebook users.
The disagreement involves the arcane details of digital advertising and the opaque algorithm Facebook uses to shuffle advertising to its users. (When it comes to the internet, the old saying goes, if you can't tell what the product is, you're the product.)
"Colorado is not suppressed," Graf told Colorado Politics. "Facebook's algorithms are just optimizing based on people who engage with similar content, which also factors in population and demographics, so obviously Texas, California and Florida are going to be amongst the top three."
At the heart of the controversy is the distinction between acquisition ads — meant to motivate social media users to add their names to a campaign's mailing list, for instance — and ads meant to encourage other types of responses, like fundraising ads or voter contact ads.
The Trump ad, Gardner's team said, was obviously built to reach Trump fans in order to collect contact information from supporters, whose names would then be added to Gardner's vast list of potential voters, donors and volunteers.
Contrary to the theory behind some of the Democrats' "aha!" claims, it wasn't a voter contact ad and would only be shown to voters expected to respond favorably to Trump who had a chance of signing up to support Gardner.
And since the ad had already spent months scouring hundreds of thousands of Facebook users in Colorado, it wouldn't make any sense to try to get the same information from the same users all over again when the campaign went national with the ad, a Republican familiar with list-acquisition strategy explained.
As is common practice, when the ad was reissued, the campaign instructed Facebook to show it to users who met certain criteria — it's a good bet that Trump fans made the cut — and, in this case, to avoid showing it to users who had already signed up as Gardner supporters, because the whole point was to acquire new emails.
In other words, politicians don't want to pay to try to get emails from people whose emails they already have, so it would be common practice to exclude the names on their list, which is probably what the Gardner campaign did.
And since they'd already spent months gathering every email they could from Colorado residents who would respond to the ad, there probably weren't many more Facebook users in Colorado to approach, effectively making it look like the ad wasn't being shown much, if at all, in the state.
Add to that Facebook's mysterious algorithm, which concentrates ads among populations that are responding to the ads — and curtails its appearance in places that aren't — and you wind up with a map that looks like Colorado was intentionally left out, when it was only the end result of a number of factors.
A leading Democratic digital strategist, however, wasn't buying it.
Washington, D.C.-based Tim Lim — he's handled digital strategy for independent expenditure groups in Colorado during previous cycles but hasn't done work in Colorado this cycle — said the Gardner campaign's explanation made "absolutely no sense" and defied basic conventions of digital advertising.
"They're just spinning," he told Colorado Politics. "Having done this for so long, I don’t know any way they could get that type of distribution of impressions without specifically making sure people in Colorado weren’t seeing that ad. If what they're saying is true, you would see lower numbers elsewhere."
He said that the notion you'd avoid reaching your existing supporters with digital advertising was "nonsensical."
"The reality is, the whole point of list-building is, the more touches a campaign can have with a grassroots supporter, the better," adding that every "touch" increases the likelihood a campaign can convert a passive name on a list to a donor or volunteer. "That's a theory that's been proven time and time again."
Lim said the Gardner campaign's insistence on refuting the HuffPost story's central point exposed the real story.
"Audience targeting, message targeting is a normal part of programming. I’ve never seen a campaign make such denials about it. That’s where it’s getting weird," he said. "it’s clear that this is a very sensitive issue. Guys, the more you deny it, the more this will have legs."
The larger story, he said, is that Senate Republican candidates are trying to distance themselves from Trump but can't be seen doing that, because Trump on Thursday threatened "people that aren't as supportive of Trump as they should be" with defeat in November.
"The ones that don't support, and I'm just talking, take a look — you have a few people that want to be cute, and those people are going to lose their elections," Trump told the Fox Business Network's Maria Bartiromo, The Hill reported.
"That’s the thing that’s driving these absurd responses," Lim said. "It’s a fear of Trump’s reprisal. That’s why they’re making these nonsensical answers to what is a fairly straightforward question."
He added: "I expect every Republican online campaign is going to be dissected to see if they’re doing what Cory Gardner is doing. I expect a couple more stories to pop up like this."
Balderdash, said a Republican familiar with Gardner's digital advertising, saying the Democrats' reading of their strategy was far off base.
Moreover, the criticism makes no sense, since the same ad ran for months in front of vastly higher numbers of Colorado voters, the source said, adding that it's standard practice in all kinds of industries to switch from lead generation to attempting to convert a lead to a sale, the retail equivalent of turning a name on a list of supporters into a donor or volunteer.