With a big mouth, plenty of buzz, and a yuuge media presence, does Donald Trump still need a data-driven get-out-the-vote operation to win in Iowa and New Hampshire?
The real-estate developer has disrupted the presidential campaign like few other candidates in history and has shown an amazing ability—through the sheer forces of his personality and a populist movement receptive to his message—to survive the outrage over some of his offensive statements and to dodge bullets fired at him by powerful politicians and "fair and balanced" media outlets. But now that there's a real likelihood of him winning the nomination, the Trump campaign is being forced to suffer the tedium and drudgery of the ground game, which involves hiring geeks to crunch the data and comb through voter lists.
The Trump campaign has been secretive about the extent of its get-out-the-vote operation in those two key states, but insiders believe that it’s far from the high-tech enterprises of the Cruz, Bush, and Rubio campaigns and that they're playing catch-up. True to form, the Trump campaign declined to comment or provide information on its high-tech campaigning apparatus to Fast Company.
"The silence is deafening," said one insider. The person said the community of campaign data scientists is a small one, and one usually knows who’s working for whom, and generally what they're doing—especially in presidential races. The consensus among data scientists and digital strategists is that only last October did Trump begin seriously assembling his data ground game.
Politico reported earlier this month that the Trump campaign started discussions with the voter data provider L2 in October and "ultimately entered into an agreement."
Regulations require the Trump campaign to report money spent to license any voter database. No such line item appears in any filings for the first three quarters of 2015. The most recent filing show expenses up to September 30, so we can be sure that the campaign did not formally contract with L2 or any other voter database company before October. The campaign will have to file its expenditures for the fourth quarter of 2015 by January 31, so we’ll soon know for sure if it's paid L2 for data.
Separately, the campaign acquired a voter list from the Republican National Committee (RNC), which it will combine with the L2 data. Such lists are provided free to GOP campaigns, with the agreement that the campaign adds to or corrects the data in the list, then provides the updated list back to the RNC to lend to other GOP campaigns.
Politico also reported that the Trump campaign hired a couple of low-profile former RNC data strategists—Matt Braynard and Witold Chrabaszcz—to work with the data. The two have reportedly been working to integrate the L2 and RNC data with voter data collected by the Trump campaign, with the goal of targeting get-out-the-vote communications at non-traditional or unregistered voters. But one insider says that Braynard and Chrabaszcz aren’t that well known in the small community of digital strategists and data scientists who are instrumental to making GOP campaigns more competitive using data.
Most presidential campaigns typically begin working with the voter data eight or nine months before the start of the primary season, say data strategists. The Trump campaign would have had no more than three months. And Trump's staff of two digital strategists is hardly enough for a serious data operation, which usually employ at least a dozen staffers.
Dan Pfeiffer, a former senior advisor to President Obama who worked on both the 2008 and 2012 campaigns, points out that, in recent history, a sophisticated ground game—built or borrowed—is table stakes for winning elections.
"The candidates that have won in the past have either built an extensive ground operation (Bush in 2000), or have tapped into the existing grassroots operation of the evangelical community (Huckabee in ‘08 and Santorum in ‘12)," Pfeiffer wrote in an email to Fast Company.
"Trump seems to be doing neither," he added.
The 2008 and 2012 Obama campaigns invested heavily in targeting data and analytics to drive its successful get-out-the-vote campaigns.
High-profile GOP data strategist Zac Moffatt says essentially the same thing. "Historically turning out new voters across the board consistently in multiple states without a massive field program is impossible—but this campaign has had many firsts." Moffatt, who is cofounder of the conservative ad targeting firm Targeted Victory, led the Romney campaign’s digital effort in the 2012 presidential campaign.
The data and software used in today’s political campaigns is similar that used in the advertising industry, and the goal is the same—connect with the right prospects at the right time with the right message over the right media.
One key difference is that consumers’ political opinions are far more fluid than their opinions about, say, potato chips or deodorant. As the Ben Carson campaign can tell you, a candidate can be on top one day and heading for the cellar the next. This means that the campaign must be continually polling and surveying voters.
Political campaigns typically acquire one or more datasets from a third-party provider like L2, Aristotle, or GOP Data Trust, or from their political party. Voters in these groups are often grouped in segments and scored on their party loyalty or likelihood of voting for the candidate. More resources, then, can be spent on communicating with voters with high scores.
Campaigns with any degree of sophistication usually contract with an outside targeting firm like Cambridge Analytica, Deep Root Analytics, or Targeted Victory to further analyze and segment the data, target TV ads and digital media messages, and measure the results. All of these firms use their own unique approach to the job, employing their own unique mix of software and data.
Other firms specialize in voter identification, telephonic or online polling and surveying, issue modeling, and data tools and support.
The Trump camp hasn’t said what it’s been doing with its L2 and RNC data since acquiring it. Public statements by campaign officials give some clue.
Trump’s campaign co-chair in New Hampshire, state representative Steve Stepanek, has said the campaign is focused on using phone calls to get non-traditional and unregistered voters out to vote. Voters in Iowa have also reported receiving emails and direct mail from the campaign. The phone numbers and email addresses could be coming from the RNC list Trump recently acquired.
Until recently, sources say, the Trump campaign appears to have been relying almost exclusively on the "earned media" coverage generated by the candidate’s cult of personality on TV, in print, and in social media.
"He doesn’t have to buy paid media because he’s got Fox News and NBC talking about him all day," says Matt Oczkowski, director of product at the conservative ad targeting firm Cambridge Analytica.
From a campaign’s point of view, that’s the most valuable kind of exposure you can get—and it’s free.
"Some campaigns have a very hard time getting that kind of media attention," says Michael Palmer of the voter database and analytics provider i360. They’ve had to spend more time creating their own news using other media channels, like trumpeting a new endorsement on social media.
"What I tell campaigns is that if they’re weak in one of those areas they need to bolster their efforts in the others," Palmer says.
Trump’s messaging, polls show, resonates best with white, less-educated, and non-evangelical voters. It’s an angry group, and Trump seems to connect with, and give voice to, many of their frustrations, especially on hot-button issues like terrorism, immigration, and reproductive rights.
How far can all that angst carry the Trump campaign? Farther already than anyone expected, to be sure.
But will the feelings Trump inspires in his supporters spur them to action when it really counts—on caucus day? In Iowa that means getting up off the couch, getting in the car and driving through the snow and cold, then waiting for hours to caucus in some church basement or school classroom. That’s a lot harder than picking up the phone and expressing support for Trump to a pollster.
And the demographic groups Trump attracts haven’t shown good turnout records in the past. In fact, exit polls show that past caucus goers look much different than Trump supporters—they’re older, college educated, and more likely to be evangelicals.
In Iowa, around 120,000 people typically come out to GOP presidential caucuses. Some believe the Trump campaign will push that number up to around 150,000 by drawing out non-traditional caucus goers.
Whether or not those additional voters show up will mean the difference between a big night for Trump, and a close race, most likely with Senator Ted Cruz (R-Texas).
A CNN/ORC poll from last Thursday shows a 37% to 26% advantage for Trump among "likely" caucus goers. When CNN polled only people who participated in the most recent GOP caucus, Trump’s lead vanished, and Cruz takes a 30% to 28% lead.
Will Trump’s "likelys" show up? The Trump people are banking on the hope that all those people who showed up for the rallies and speeches will show up to caucus and cast a vote for the Donald.
But much of the draw for Trump’s events is the candidate’s bombast and celebrity. Many potential voters may be willing to wait in line for six hours to see Trump speak, but unwilling to caucus for him.
In contrast to the Trump campaign, the Ted Cruz campaign has perhaps the most advanced, or at least the most ambitious, data strategy, multiple sources said. Cruz has 10,000 people on the ground in Iowa, and they’re knocking on a thousand doors a day, a source with knowledge of the campaign said.
Asked about sophistication of the Cruz digital strategy, the source explained it this way: When a Cruz volunteer walks up the someone’s door in Iowa or New Hampshire, they can see a variety of information about the household in the app running on their smartphone. They know who lives in the house and can pull up the right profile for the voting-age person who answers the door. The profile will contain basic demographic information like age and gender, along with deeper data like personality type and the person’s hot-button political issues. Based on that information the app pulls up one of a number of specialized scripts. The script is designed to steer the conversation toward the issues the person cares about, and to the candidate’s positions.
The Cruz campaign hired the "behavioral science company," Cambridge Analytica — a newcomer to the U.S. political scene — to get beyond the simple demographics and polling data and develop an understanding of the personality traits that move voters to action. The campaign can then factor that data into its messaging strategy, Cambridge Analytica’s Oczkowski says.
Cambridge Analytica says it has a database of 240 million voters nationwide, with as many as 2,000 data points on each. These data points include things like corporate information, voter files, and social media data. To add in the personality classification data the company says it conducts between 30,000 to 50,000 surveys nationwide every month.
The company identifies groups with shared traits, and can match those groups with similar voter groups in a given state or precinct. A team of psychologists helps determine the types of messaging that might work best with the various groups.
How well this will actually work on the ground in Iowa and New Hampshire remains to be seen.
Cambridge Analytica's personality data are also used to decide which Cruz volunteers call which potential voters. The campaign says it's found that volunteers are more persuasive when calling people with similar personality types to their own.
The Cruz campaign's app, "Ted Cruz 2016," is also one of the most advanced of those being used by GOP campaigns. The app attempts to gamify support for the #cruzmissile, as some of his supporters call him. Points are awarded when a user retweets a Ted Cruz tweet, or when they post something for other app users, or makes a donation. Those who pile up enough points win T-shirts and bumper stickers, or even a trip to Iowa for primary day.
The app also captures the user’s contact list. It then matches the names up with names in the campaigns voter database. If the user has a friend in Des Moines, for example, the campaign may start sending emails asking the friend to attend a caucus or volunteer. To date, the app has been downloaded 37,000 times.
If the Trump campaign finds itself in a close and protracted race with Cruz in the primary, its lack of data could hurt it in other ways.
Many of the presidential candidates are using their data and tools to manage long-term strategies that go well beyond the borders of single primary states. These strategies dictate which specific voter enclaves can be targeted in each state to yield the delegates the campaign needs to win.
"It’s a race for delegate votes in states where it’s not winner take all," Cambridge Analytica's Oczkowski says. Oczkowski was chief digital officer for Wisconsin governor Scott Walker’s presidential bid.
The end game of the primaries is to collect enough state delegates to earn the nomination. In some primary states—like Florida (99 delegates)- the candidate who wins the most votes takes all the delegate votes, which will be added up at the end of the primary process to land on the party’s nominee to run in the presidential race. Other states, like Iowa (30 delegates), award delegates proportionately to the number of votes the candidate wins in the state.
The Cruz, Rubio, and Bush campaigns each will take a deep look at the delegate picture in every primary state that awards delegates proportionately. They’ll take a hard look at their voter data to pinpoint the voter enclaves in which they have a solid chance of grabbing delegates. This overall strategy can make a huge difference, especially in close races.
"Candidates who aren't taking a scientific look at which pockets of delegates they can realistically compete for are naive," Oczkowski says. "You start to see the difference when you look at the delegate allocations as we go past the first four states; there is a real science in how to do that."
The modern age of data-driven campaigning began in 2004 with the Howard Dean and George W. Bush campaigns for president. The Dean campaign was the first to use digital channels, mainly email, to reach voters. The Bush campaign was the first to bring in numerous demographics and political data, and use it at scale in a national campaign.
"The GOP would conduct mega surveys—tens of thousands of respondents in a particular state (whereas a typical statewide poll in your average industrial state might have 500-800 respondents)," says Josh Zeitz of the ad tech powerhouse AppNexus. "They then overlaid consumer, credit, organizational membership, and financial data (widely available for purchase) to develop much more sophisticated voter models."
Zeitz was senior policy adviser to former Governor Jon Corzine of New Jersey. He’s also lectured on history at Harvard, Cambridge, and Brown.
The Obama campaigns in 2008 and 2012 took it to a new level, both in content and scope, Zeitz says. The campaign integrated more types of data and used modeling software to determine the messaging and medium for communicating with individual voters.
"When I started working on campaigns in the mid-'90s, you could, at best, target specific households within a precinct," Zeitz says. "Now, you can be somewhat certain which people will respond to which messages."
In 2016, campaigns are analyzing and modeling voter types. They’re targeting and testing ads on smartphones. Some are working with cable services to target TV household by household. They have much better tools for tracking whether or not voters are reacting to, or interacting with, various forms of advertisement.
"Instead of blanketing the airwaves or stuffing mailboxes, campaigns can be surgical about their outreach, engagement, and fundraising and build the coalition that gets them to 50+1 [a majority of votes]," Zeitz says. "It’s moneyball for politics."
The Trump campaign’s disinterest in data could ultimately hurt the Republican party (even more than it already is). Any insider will tell you that the GOP has fallen well behind the Democrats when it comes to voter data.
The Democratic party decided long ago that all the party campaigns would contribute the data they acquire or collect to a central pool, so that all state and federal candidates can benefit.
The GOP announced that it intended to do the same thing, but competition and lack of integration among the major database providers (L2, i360, et al.) has hindered those plans.
With Trump’s celebrity and appeal to new voter groups, his campaign could be gathering massive amounts of demographic and outreach data that could help other state- and federal-level GOP campaigns in the future. But he’s not.
"We’ve come so far, we’ve been chasing our tails for the past 10 years, but we’re just starting to catch up on data and now we have a front-runner who doesn't care about any of it," says GOP data and social media guru Vincent Harris of Harris Media.
Harris, too, believes Trump’s earned media appeal and effectiveness on social media may not create enough of a bond between candidate and voter to bring about the wide margin of victory suggested by the polls today.
"At the end of the day a lot of his support is wishy-washy and back and forth, based on the some comment he made," Harris said. "I think his poll numbers are inflated, and it all comes down to what will actually turn people out to vote."
Harris points out that the Trump campaign doesn’t seem to want voters’ data, and it doesn’t want their money either. This is important, Harris explains, because donations often serve two purposes for a campaign—they pay for the cost of campaigning, but they also form a sort of handshake between voter and campaign. Harris points to studies showing that Iowans who have donated to a political campaign are likely to show up to caucus.
Trump has volunteers knocking on doors in Iowa and New Hampshire like everyone else does, but for many voters the bond to the candidate, however compelling, might still be impersonal, consisting of viewing the candidate on television or from afar at a campaign rally.
Of course nobody knows exactly how angry or committed Trump’s supporters are, and we’ll have to wait until February 1st to find out. If they come out in droves, the campaign’s data operation (or lack thereof) may be a moot point.
But there’s reason to doubt that Trump can ride his cult of personality all the way to the White House, and if that proves to be the case, he may not have much of a get-out-the-vote strategy to fall back on.