The Victory Lab (43 page)

In the fall of 2008, riders who didn’t devote their commutes to their newspaper or mobile phone might have noticed Barack Obama traveling with them each day. The Democratic presidential nominee gazed triumphantly from one of the 11-by-28-inch cardboard advertisements that lined the bus’s interior overhead, accompanied by a message rendered in his campaign’s familiar sans serif typeface: “Don’t Wait. Vote Early. Our Moment Is Now.” In smaller type were a Ohio-specific phone number and website that could offer directions on early-voting procedures, which allowed state residents to cast a vote at their leisure as early as five weeks before the November election.

Obama was at that moment perhaps the most dynamic brand in the country—as omnipresent and approachable as Starbucks, as much an embodiment of the American now as Apple—but the company he kept in the interior of Akron’s buses was far less inspiring. Most of the interior ads on Akron’s buses are for the Metro system itself, or public service announcements like “Keep Your Baby Sleeping Safe” and “Schizophrenia? Accepting Research Patients Now.” In fact, it is rare these days to find
any consumer advertising in buses anywhere; those for private businesses speak to a whole different hierarchy of needs, like mental health services or personal-injury law. And yet Obama was there, and it was far from an accident.

Weeks earlier, a data analyst at Obama’s Chicago headquarters was reviewing the hundreds of individual-level variables thrown into microtargeting algorithms and realized that one—mass-transit ridership—played an outsized role in predicting which Wisconsin voters were most likely to support Obama. The analyst knew the campaign would already try to mobilize these turnout targets through mail and phone calls, but he thought his new finding pointed to yet one more medium in which it should be able to reach them where they spent time—provided it could be done as efficiently. The analyst alerted one of the campaign’s media planners, who called each of the public transit agencies in Wisconsin to see which of them allowed advertisers to target particular routes, stops, or depots instead of covering the whole system at once. Milwaukee did, and so the media planner called over someone from the campaign’s graphics department, and together they made a map showing Milwaukee precincts where individuals with high support scores were clustered, and a series of transparencies for each of the city’s bus routes. They laid the transparencies atop the support map until they found lines that intersected their target precincts, and sent an order to GMMB, the campaign’s lead advertising agency.

Danny Jester, a GMMB vice president and media director responsible for the Obama account, had never processed a request quite like this. Jester placed many of Obama’s ads, as his agency had for John Kerry’s campaign four years earlier. For a presidential campaign, this typically meant broadcast or cable television, or sometimes radio. Maybe a candidate for city council or county commission would buy bus ads, because they were easier to produce than television spots and intuitively made sense when thinking about geographically constrained electorates, but no one at this level ever proposed putting outdoor advertising on the schedule. Among those who placed political ads, progress had been treated as
effectively synonymous with the introduction of new delivery devices. The half-century-long history of refinements in media targeting were a story of technological innovation: moving from buying national ads to local ones in key markets, and then shifting from broadcast waves to cable television, where narrow audiences could be more easily pinpointed. Internet advertising, with its ability to track users’ movements through cookies and interests through search engines, was the latest breakthrough.

Obama aggressively bought ads in all of those media,
including $16 million in online advertising, among it deep reaches into mobile devices. With no hoopla, however, the campaign also bought bus ads. Milwaukee didn’t have the inventory available on the routes Jester requested, but other cities did, and Jester started writing checks. Soon Obama’s ads were rolling through select buses in ten cities nationwide, including Philadelphia, Miami, Denver, Flint, and Akron. The most technologically advanced campaign in history had so thoroughly mastered the politics of individual data and testing that it found new value in electioneering tactics many had abandoned as hopelessly last-century. “There’s all this shit we used to say no to in campaigns—bus benches, mass-transit advertising,
PennySavers
, what’s that sock they stick the newspaper in?—because we used to do it before TV got dominant,” says Larry Grisolano, who coordinated all of the campaign’s public-opinion research and media buys. “Now if I know that there are twenty-seven people I want to reach and they all cluster around this bus bench, I’ll buy that bus bench. And if I know these twenty-seven people read the
PennySaver
, I’ll buy an ad in the
PennySaver
.”

ONE OF THE FIRST THINGS
Jeremy Bird was handed when he arrived in Columbus in June to be Obama’s Ohio general-election director was a series of pages taken from the book that campaign manager David Plouffe called his “bible.” The binder had been the principal project that lead targeting analyst Michael Simon undertook with his new team, a series of simple
but profound stories outlining Obama’s path to victory for every state he was actively contesting.

Vote goals were a staple of any campaign plan, among the initial documents a general consultant or campaign manager would draft in the early days of a race. They were the basic facts on which a theory of the case could be built, but despite their essentially quantitative nature, the numbers were often plucked more or less from the air. A strategist would usually gather past election results from similar races, jigger them to reflect population changes measured by the Census, and then adjust the targets to reflect where things stood that year.

Simon thought there should be a smarter way to make the calculations that would fill the bible. His team had access to new data, from the IDs and microtargeting scores it had generated during the most engaged primary season in modern history. When he assigned his regional analysts to create vote goals for their states, however, he encouraged them to inform their estimates with human intelligence, such as interviews with county party chairmen and local pollsters. Simultaneously, he hoped that his team could approach the task with more data-driven rigor than was typical, while delivering to his pack of statistically minded political debutants a bracing reminder of the limits of data-driven rigor. They needed to understand that some parts of the process, Simon thought, were still “more art than science.”

Simon’s targeters came back with an estimate of the total number of votes they expected to be cast in each state in November. He sent them back to mine public and private polls and the campaign’s own IDs to tally the number of votes Obama could already count on. Then they had to work out Obama’s path to 52 percent, far enough over the halfway mark to offer a safe buffer. “What do we need to do here in order to win?” Simon asked them. “What percentage of this victory is going to be boosting turnout and what percentage of it is going to be convincing independents or other people to come our way?” Simon’s team split their state goals into three different categories—new registration targets, persuasion targets, and turnout
targets—and developed spreadsheets splitting each of them further into geographical and demographic subgroups. As they put the numbers into spreadsheets, it became clear that the path to victory was distinct in each state. Obama could carry Indiana only if he succeeded in persuading Republicans and independents. In Pennsylvania, he could get there just by successfully mobilizing his turnout targets. Nevada had enough new residents in the state that registering them had to be part of the formula.

The Ohio spreadsheet was the handiwork of Dan Wagner, who had been assigned to the Great Lakes/Ohio River Valley pod, which was charged with monitoring the densest concentration of high-stakes battlegrounds—Pennsylvania, Michigan, Indiana, and Ohio—informally known as the “all the states he lost in the primaries” region. This was the toughest assignment of all; it fell to Wagner because he was, in Simon’s words, the “purest hybrid of quant and translator” among the numbers geeks. This was a mix of skills he had been forced to develop as a management consultant who had to present his economic forecasts to corporate decision makers. Nothing Wagner produced might be as closely studied by the Obama campaign leadership as his Ohio vote goals. The midwestern state had voted for the winner of every presidential election since 1960, decided by barely 100,000 votes in 2004. Neither party had a reliable strategy for winning the White House in 2008 without it. Wagner’s analysis confirmed what many Democrats suspected about their path to victory in the state. Ohio’s Democratic base had been ruthlessly raked over by America Coming Together and other independent groups in 2004, leaving little room for the potential registration gains available in other states. Obama would need a set of tactics tilted almost equally toward persuasion and turnout.

Wagner carried the air of the dispassionate outsider, confronting campaign practices for the first time with fresh eyes and an empiricist’s skepticism. After February 5, Wagner had been assigned to Ohio and Indiana, where he managed the voter file, and Puerto Rico, where he aided Obama’s media-buying team just because he knew enough Spanish to call a radio station and ask about their ad rates. Along the way, he learned how
field organizers did their jobs—and the elaborate multilevel structures of state and county leaders presiding over local organizers and the volunteers who talked to voters—and realized that the campaign needed to do a better job packaging its microtargeting scores for these end users. “It was like I was a floor manager at Walmart and I could see how people shop,” says Wagner. “Do they look at the top shelf first or the bottom shelf?”

The gap between a campaign headquarters, with its claustrophobic landscape of wall-to-wall carpeting under a drop-ceiling sky, and the world of voters had never been so far apart. Obama’s campaign was run from the eleventh floor of a skyscraper abutting Michigan Avenue, buffered from the electorate by building security. Even a decision to open ubiquitous storefront offices was identified in corporate jargon, as “the Starbucks model.” When the campaign’s Web team wanted to see how normal people responded to their latest innovations, they set out like anthropologists looking to study how natives would handle unfamiliar tools, descending to their building’s basement food court with a laptop equipped with Silverback software that could videotape the user’s face and track keyboard hand movements.

In the world of campaigns, all politics conducted outside the stale air of headquarters—or the virtual space of broadcast airwaves and the Internet—was referred to as “field.” The primary job of a field organizer was cutting turf, as terrain for canvassing was known. By 2008, it made more sense than ever that the world of interacting personally with voters had become suffused with the imagery of agriculture. Both were recalled as pre-industrial practices that seemed to be on their way out of American life just a decade or two earlier, and now enjoyed a resurgence. Like the fad for small-scale urban farming, the notion of people talking to people carried an almost rebellious quaintness, a moral riposte to the tyranny of mechanized mass communication.

As Ohio field director, Bird brought to his tactics a humanistic approach that belied his young age. No one represented the evangelical side of Obama’s advocacy as perfectly as Bird,
who had grown up in a Missouri trailer park to conservative Southern Baptist parents. As a student at Harvard Divinity School, he had taken a class called Organizing: People, Power, and Change, which he credited with awakening him to the value of politics.
The professor, Marshall Ganz, had dropped out of Harvard to register black voters in Mississippi during the civil rights era and went on to serve as an organizer for César Chávez’s United Farm Workers and Robert F. Kennedy’s presidential campaign in California. Ganz became known for convening meetings in which he would have activists tell the stories of how they came to the movement, a mobilization device Ganz prized as a counterpoint to what he derided as the transactional “
politics as marketing” of modern elections. But as television ads and direct mail started to dominate the attention of campaign managers, Ganz withdrew from electoral politics and eventually settled at Harvard, where he was prized as a curio testifying to a lost trade, like a blacksmith continuing to forge tools at a colonial village. Ganz’s “personal narratives” were often likened to the stuff of Alcoholics Anonymous gatherings and tent revivals, and the latter fit naturally with the candidacy of a onetime community organizer who had already written a memoir before first seeking office.

A loyal Ganz disciple, Bird preached his mentor’s ideas of “motivational organizing.” But he also had a modern appreciation for how data could help build new-style structures to support it, a relationship he described to his field staffers as “the yin and yang of organizing.” Bird had earned Chicago’s notice by directing Obama’s shocking primary victory in South Carolina, where Bird controversially ignored the state’s culture of mercenary political professionals and built his own network around African-American community hubs like barbershops and beauty salons. Within Obama’s campaign, that tactic (which was repeated elsewhere in the country) was lauded for organizing in unconventional locales, but Bird believed that one of the underappreciated keys to the program’s success was in the analytics.