Google's Eric Schmidt Invests in Obama's Big Data Brains
During the 2012 campaign, Barack Obama’s reelection team had an underappreciated asset: Google’s executive chairman, Eric Schmidt. He helped recruit talent, choose technology, and coach the campaign manager, Jim Messina, on the finer points of leading a large organization. “On election night he was in our boiler room in Chicago,” says David Plouffe, then a senior White House adviser. Schmidt had a particular affinity for a group of engineers and statisticians tucked away beneath a disco ball in a darkened corner of the office known as “the Cave.” The data analytics team, led by 30-year-old Dan Wagner, is credited with producing Obama’s surprising 5 million-vote margin of victory.
Schmidt thought enough of the team that when the campaign ended, he put up several million dollars to keep its core together as a new consulting firm, Civis Analytics, run by Wagner and staffed by two dozen of his former employees. They share ownership with Schmidt, its sole investor. The plan is to bring the same Big Data expertise that guided the most expensive presidential campaign in history to companies and nonprofits. “The model of innovation in technology are these very young teams that have a brand-new idea, like each other, and work incredibly hard,” says Schmidt. “Venture capitalists are competing all over for these people. When you find them, you figure out what they want to do and you back them.” Or, as Wagner puts it, “The Cave is incorporating.”
For all its acclaim, the analytics team’s main achievement is often misunderstood as “microtargeting” or some variant on wooing voters. This reverses the relationship between campaign and voter at the heart of Wagner’s method. Recent campaigns have employed a top-down approach to identify what they thought were vital demographic groups such as “soccer moms.” Wagner’s team pursued a bottom-up strategy of unifying vast commercial and political databases to understand the proclivities of individual voters likely to support Obama or be open to his message, and then sought to persuade them through personalized contact via Facebook, e-mail, or a knock on the door. “I think of them as people scientists,’’ says Schmidt. “They apply scientific techniques to how people will behave when confronted with a choice or a question.” Obama’s rout of Mitt Romney was a lesson in how this insight can translate into political strength.
Traditional marketing has the same inherent limitation as traditional campaigning: It’s impossible to appeal to everybody, even among the groups likeliest to favor a product. “Budweiser might target football fans with an ad showing half-naked women jumping up and down with a bottle of beer,” says M. Eric Johnson, director of the Center for Digital Strategies at Dartmouth College’s Tuck School of Business. “They’re trying to build something that appeals to a huge segment of people. It may be the right segment. But they know they’re losing a bunch of people, who either won’t like the ad or won’t be persuaded by it.” Analytics offers the potential of capturing these people, too. “Big Data gives you the ability to personalize,” says Johnson, “to know me well enough to understand that, while I like beer, I don’t want to see half-naked women jumping around with a can of Bud.”
Wagner took an unlikely route to the Cave. In 2007 he was volunteering at Obama’s Chicago headquarters as an organizer of phone banks targeting Latino voters. As a hobby, he built a special calculator for caucuses that could determine how many voters a candidate would need to take from a rival to gain an additional delegate. “It was essentially a game-theory application designed for use in the Iowa caucus,” he says.
Wagner showed his creation to some colleagues and shortly after was summoned to the office of the campaign’s Illinois director, Jon Carson. “This is very interesting,” Carson recalls telling him. “How soon can you move to Iowa? Can you go in three days?” Wagner spent the remainder of the primary season traveling from state to state and led the voter targeting effort in the Midwest during the 2008 general election. He later went to work for the Democratic National Committee, where his team accurately forecast the Democrats’ blowout loss in the 2010 midterm elections.
By the time Wagner signed on as chief analytics officer for Obama’s 2012 campaign, it was clear Big Data would be central to the reelection strategy. Along with identifying potential supporters, Wagner’s team built intricate mathematical models of swing states that provided an alternative to traditional polls. “Those guys were our bible,” says Plouffe. “They consistently told us what was really going on.” During the final month, the campaign’s internal tracking polls of Ohio, a pivotal swing state, suddenly grew erratic; one night Obama would be up four, the next down by one. “There were people in the campaign completely wigging out,” Plouffe says. Wagner’s model, by contrast, showed a steady three- to four-point lead for Obama that proved accurate.
Schmidt’s introduction to Wagner didn’t seem likely to lead to a business collaboration. Each day at 4:30 p.m., to let off steam, the Cave’s inhabitants would flip off the lights, fire up the disco ball, and spend five minutes dancing to a mash-up of Psy’s Gangnam Style and a campaign robocall voiced in a dulcet baritone by Wagner’s deputy, Andrew Claster. Schmidt showed up to meet Wagner just as “Club Claster” was kicking off. No stranger to creative eccentricity, he was unfazed, and when the campaign ended, he agreed to keep the party going.
Civis Analytics, which is based in Chicago with an office in Washington, will focus primarily on companies. “Businesses are full of A/B choices—do I put my money over here or over there?” Schmidt says. “With these modern data tools, you can figure it out.” Logistics, scheduling, manufacturing, and marketing decisions would all benefit from sophisticated analysis, he says. Obama campaign veterans say that applying this rigor to their half-billion-dollar media budget made it 15 percent more efficient, saving tens of millions of dollars. “If you can change the bottom line by $100 million, you’re going to want these tools,” says Schmidt.
Dartmouth’s Johnson says nonprofits could also gain, offering the hypothetical example of how Save the Children could improve its response to an earthquake in India by using analytics to tweak the content and timing of its fundraising appeals.
The College Board, which administers the SAT, has hired Civis to figure out why high-achieving, low-income students are less likely to apply to elite schools than equally qualified students from wealthier families—and then to design a campaign to guide more of them to better colleges. “What was so impressive to anyone who was observing was the precision they brought to this fuzzy question of ‘How do you get low-income people to vote?’” says David Coleman, the College Board’s president. “We see in our work a similar need for precision and analysis so that low-income students are propelled into the opportunities they have earned.” Civis is also working with several companies and nonprofits—it declined to name them—to help enroll the millions of people newly eligible for health insurance under the Affordable Care Act.
Civis won’t be abandoning politics altogether. Wagner expects the company to work for Democratic candidates—and only Democrats—next year. That could taint the firm in the eyes of some potential clients if it comes to be viewed as an extension of the party. Wagner, however, sees his lineage as an unalloyed good.
“We came out of the campaign having helped invent this category of using Big Data to reinvent organizations,” he says. “There’s a huge commercial opportunity in that. But there’s also an opportunity to take the experience, people, and technological infrastructure from the campaign to help solve some of the world’s biggest problems.” If that sounds like a lot of pressure, Wagner swears it won’t be a grind. The disco ball will soon be hanging in the new office.