ADVERTISEMENT

The Democrats’ Most Effective Election Ads Did Not Feature Trump

The Democrats’ Most Effective Election Ads Did Not Feature Trump

Four years ago, Donald Trump’s upset presidential victory shattered the Democrats’ notion that they knew what they were doing. “The mindset change that we have gone through is that, in the old world, we thought that we can understand people,” says Dan Wagner, chief executive officer of the data science firm Civis Analytics. “What 2016 really taught us is that we can’t.”

This year’s results show that they’re only beginning to figure it out. Joe Biden defeated Trump, winning more votes than any presidential candidate in U.S. history. But the Democratic Party simultaneously lost seats in the House and failed to win back the Senate, barring an unexpected double victory in Georgia’s joint Senate runoff on Jan. 5.

That disappointing outcome has fueled an intraparty rift between progressives such as Representative Alexandria Ocasio-Cortez of New York and moderates like Pennsylvania Representative Conor Lamb over whom is to blame for the party’s lackluster appeal. Privately, many Democratic strategists have developed a dour view of their party’s future. After coming up short in dozens of down-ballot races against an opposition weighted down by an unpopular president, a pandemic, and a recession, Democrats are anxious to understand what went wrong, what went right, and how they can correct course before the midterms in two years.

Civis is part of the firmament of companies and consultants whose job is to craft the Democratic message and figure out how to sell it. There may be no outfit on the left better equipped to offer insights on these questions. The company, which grew out of Barack Obama’s 2012 presidential campaign (I wrote about them in 2013), offers its data-science skills to hundreds of Democratic candidates and progressive advocacy groups. “I think of them as people scientists,” Google’s former executive chairman Eric Schmidt, who backed the company, told me at the time. “They apply scientific techniques to how people will behave when confronted with a choice or a question.”

This cycle, Civis tested 3,300 messages for everyone from the Biden campaign and Priorities USA, his main super PAC, to boutique outfits such as the Bail Project and Be a Hero, the political organization founded by the ALS patient and health-care activist Ady Barkan. As a result, Civis probably has a broader understanding of what did and did not move voters in 2020 than just about anyone.

After the election, the company and some of its clients agreed to share their findings with Bloomberg Businessweek. Rather than point out a clear path forward for Democrats, they illuminate just how difficult it’s become to move voters in a country riven by partisanship and declining social trust.

1. Making Trump the bad guy backfired

Given polling failures in 2016 and beyond, the process of testing political messages has become more expansive. “Traditional political phone polling may be dead,” Wagner says, “especially in states with heavily diverse populations.” Instead, Civis looks online to conduct thousands of interviews per week and tries to correct for the refusal by low-trust Trump voters to answer surveys by incorporating machine learning and inducing participants with in-app rewards and even video game credits.

But even a test audience more representative of the electorate has become tougher to persuade. Most people aren’t interested in ads, and attention spans are short: In this cycle, 15-second ads performed better than 30-second ads, and 30 seconds did better than 60. For all that Trump dominated the campaign, voters didn’t want to be lectured about him, so standard persuasion methods often fell flat.

Trump’s mere presence in ads frustrated Democrats’ efforts to move voters away from him. “If we wanted to say that Trump’s policies or actions were devastating, having him in darkened silhouette with ominous music playing—explaining how devastating they were—often backfired and moved people to Trump,” says Jess McIntosh, a veteran Democratic strategist who is the creative director of the group Fellow Americans. Showing the president’s face or including audio of his voice “caused white men to support him more, regardless of what you were saying or doing in the spots that we did,” McIntosh says.

Because comprehensive message testing didn’t exist until recently, it’s impossible to say whether this phenomenon is common to incumbent presidents or specific to Trump. But most strategists suspect that people made up their minds about him long ago and couldn’t be persuaded to change them.

The kind of red-meat ads that grassroots Democrats love to send virally on Twitter and Facebook were often the least effective at changing people’s minds. A study by Priorities USA conducted after the election found “no correlation” between an ad’s virality and its effect on vote choice. In June, amid the protests over police violence, Fellow Americans produced an ad that co-founder Nathaniel Lubin felt sure would score big. It lacerated Trump for marching out of the White House to hold a Bible aloft in front of St. John’s Church after police cleared protesters with horses and tear gas. The ad bombed, even with Democrats.

A rare exception to the Trump-ad backlash in Civis tests were spots that featured an animated version of the president. Perhaps because cartoons are cute and disarming, or maybe because they didn’t technically show him, these spots slipped past people’s Trump defense mechanism and thus managed to register. “It allowed us to do that tightrope walk of getting a [negative] message across,” McIntosh says. Early in the pandemic, a vaguely Seussian ad featuring an animated Trump (“The Trump Who Cried Hoax”) was one of Fellow Americans’ best performers.

2. Shaky iPhone videos inspired more trust than slick professional ads

The issues Democrats expected to dominate the presidential race mostly did not. “We started out hitting some things that we thought were going to be primary this cycle: gun violence, climate change,” Lubin says. “And then Covid happened, and everything changed.”

The sudden onset of the pandemic made health care a top concern. Liz Jaff, the president of Be a Hero, Barkan’s PAC, had already found that anti-pharma ads featuring Barkan, who uses a wheelchair and speaks with a computerized voice, tested well with voters across the political spectrum. “We were kind of shocked that those ads were moving Trumpers by seven to eight points,” she says.

The president’s failure to contain the coronavirus opened up a fresh avenue of persuasion. As the pandemic worsened, voters became even more skeptical of slickly produced ads designed to manipulate their sentiments—but responded positively to cheap, low-fi content suddenly made necessary by the disruptions Covid imposed on the ad-making process. (That’s one reason some campaigns turned to animation.)

“Some of the ads we had that were high definition and very well set—they just did nothing in comparison to slapped-together iPhone videos of nurses running out of the hospital, making mistakes as they talked into their phone about what they were experiencing,” Jaff says.

Other Democrats found the same thing. “The brighter and shinier an ad was, the less it persuaded people,” Lubin says. “Bad lighting, for example, was a predictor of better performance.” That may reflect changing expectations due to the flood of user-made videos on social media.

Jaff says one of Be a Hero’s best-performing ads was just such a DIY affair, shot by a Covid nurse speaking to her phone camera just after finishing her shift. “Nurses across the country are voting for Joe Biden,” Dominique Hamilton, the exhausted nurse, says, “because we need someone who will listen to the science and bring us back to a functioning society.” Civis testing found that the gritty ad increased support for Biden across the political spectrum, but most heavily with voters often at different ends of the barbell: those under 35 who lean heavily Democratic and those who identify as “strong conservatives.”

3. Unconventional ads resonated in hard-to-reach communities

For all their experience, even veteran political strategists can’t predict how an ad will fare in the real world. “We tested 492 ads this cycle, up from 101 ads in 2016,” says Nick Ahamed, analytics director for Priorities USA. “The cumulative takeaway is that it’s impossible to predict which ad will be best.”

November’s election results laid bare the Democratic Party’s struggles to appeal to voters outside its urban/suburban core: It lost seats in suburban House districts it had flipped in 2018 and made no headway in rural areas where Biden was thought to hold outsize appeal. What saved Biden is that Black and Latino voters in battleground states turned out in sufficient numbers to secure his victory. (While their vote share appears to have shifted slightly to Republicans this year, both groups still strongly favored Democrats.)

One problem Democratic strategists recognized early on was that what they presumed would appeal to Black, Latino, and other voter groups often did not. Like many professional Democrats of their generation, McIntosh and Lubin are both white and live in a big coastal city on the Acela corridor (in their case, New York). To surmount this limitation, Fellow Americans outsourced idea generation for ads to more than 50 “creators” in targeted communities around the country, asking them to produce ads they thought would resonate.

Many outperformed the ads from the Washington pros. “They were filmmakers, documentarians, artists,” McIntosh says. “So we had a very diverse set of people coming up with ideas that didn’t necessarily appeal to us, but we let them go forward because they came from someone within the community we were trying to talk to.”

Although Biden retook the Upper Midwest states Democrats have long relied on to win the presidency, the changing nature of the party’s electoral coalition means future victories will depend on winning more diverse states, such as Georgia and Arizona, both of which Biden narrowly carried. Developing better methods to activate minority voters is therefore critical to the party’s fortunes in the years ahead, and it’s one reason Wagner and his clients are bullish on this unorthodox approach.

“We had a project around trying to turn out African-American audiences,” Lubin recalls. “Many of the top-performing spots were made by African Americans who understood the community way better than we would. Many times in our internal conversations reviewing pitches, we’d say, ‘Well, I don’t think this will work, [but] we should probably try it anyway.’ And often it did work.”

4. Collapsing social trust threatens Democrats, who don’t have many answers

Collectively, these insights could help Democrats in future races. Jaff’s group, Be a Hero, is currently using them to target moderate suburban women in Georgia’s Senate races with an ad featuring a poorly lit nurse from Augusta talking straight into her phone.

But they’re better thought of as tactical advances than groundbreaking strategies that will free Democrats from their geographic prison. For all the excitement over Biden’s victory, the major issues in the race—Trump’s fitness for office and the coronavirus—were both sui generis and all but certain to disappear before the midterms. Democratic strategists had hoped the bevy of tests and data would nevertheless yield clear lessons that could guide the party toward a brighter future; for the most part, they have not.

“We were constantly looking and asking, ‘What causes an ad to do best?’” Ahamed says. “We measured 50 different attributes of ads—and the only one that was statistically significant was the ad being about coronavirus.”

An even bigger concern for Democrats lies in the deep distrust that’s spread among so many voters. It makes changing their minds much harder. “I think we did see some of the challenges of Democratic persuasion this cycle,” says Civis’s Wagner. “We consistently saw a backlash against negative Trump ads, and the effect of Covid ads tended to diminish over time (especially as it soaked the news cycle). Trump maintained a consistent favorability rate all the way to the end.”

Opinion surveys show that rising distrust has already weakened Americans’ faith in the results of the election. It also threatens to erode support for government programs aimed at the broad public good whose endurance depends on a national sense of mutual obligation and goodwill. “Most progressive policy involves a social safety net and understanding that your neighbors have needs just as you do,” McIntosh says. “If that goes away, it becomes a whole lot easier for that hyper-individualistic fear-mongering that the other side peddles to take hold.”

In the meantime, Biden’s win bought Democrats time to figure out how to broaden the party’s appeal. Wagner’s goal of arming Democratic candidates and issue groups with data to quickly figure out what messages connect and how to exploit them was, in hindsight, a bigger deal than most people had anticipated, if only because the election turned out to be much closer than the polls indicated. Looking ahead to the midterms and beyond, distributed data testing may help Democrats hone their pitch. But it will be up to Biden and the Democrats to figure out what they’re selling.

“I don’t know what the hell’s going to happen in 2022,” Jaff says. “All we do know is that in 2020, depressed people on their beds worked.”

©2021 Bloomberg L.P.