The Victory Lab: The Secret Science of Winning Campaigns

Excerpted from the Prologue from The Victory Lab: The Secret Science of Winning Campaigns by Sasha Issenberg. Copyright © 2012 by Sasha Issenberg. Published by Crown, a division of Random House, Inc. Reprinted with permission. All rights reserved. No part of this excerpt may be reproduced or printed without permission in writing from the publisher.

How to Win An Election Without Anyone Knowing

The ads aired by the two candidates in the 2010 Colorado U.S. Senate race told the story of the ideological war that defined that mid-term election. Michael Bennet, a freshman Democrat appointed to replace a man who had become one of Barack Obama’s cabinet secretaries, was deeply in hock to a liberal White House. Tea Party Republican challenger Ken Buck was— or so went the punditry and Bennet’s attacks—too conservative for the moderate, suburbanizing state.

Meanwhile, one million letters being delivered to Democratic-leaning Coloradoans in the last days of the race made no mention of either candidate, their allegiances, or the issues that separated them. They lacked any allusion to the ideological split riving the nation or reference to the policy consequences of a change in party control of the Senate. The folded pieces of laser-printed white paper were designed to be ugly, with a return address referring to a sender whose name voters were unlikely to recognize. The sender thanked the recipient by first name for having voted in 2008 and then said she looked forward to being able to express such gratitude again after the coming elections. The letter, dispassionate in tone and startlingly personal in content, might have inspired most recipients to dispatch it to a trash can with no strong feeling other than being oddly unsettled by its arrival.

It was not only in Colorado where communications in the last days before the 2010 elections seemed out of whack with such a feral season in American politics. Across the country on the Sunday night before the election, millions of Democrats received an e-mail from Obama’s seemingly dormant campaign apparatus, Organizing for America, with a gently worded reminder that they had “made a commitment to vote in this election” and that “the time has come to make good on that commitment. Think about when you’ll cast your vote and how you’ll get there.”

The voters who received either the Colorado letter or the Organizing for America message had likely never encountered anything like them before. At a moment when many candidates, admen, pundits, and organizers thought that the way to get their allies to the polls was to implore them through television ads to consider the election’s high stakes and respond in kind, these tactics, designed to go undetected by media coverage, aimed to push buttons that many voters didn’t even know they had. The people who had scripted the messages and carefully selected their recipients aimed to exploit eternal human vulnerabilities—such as the desire to fit in or not to be seen as a liar—in order to turn layabouts into voters.

The man who had sent the million letters in white envelopes did the quick math after the Colorado election from his post thousands of miles away. Hal Malchow was a middle-aged Mississippian who had spent his life conniving new ways to win elections, except for a brief detour into securities law that ended when he realized that writing the contracts to guard against complex financial schemes was less fun than trying to hatch them. Now he was playing a different angle, and calculated that the psychological influence he had exerted through his letters would improve turnout among recipients by 2.5 percent. That would mean that his language had created 25,000 new voters, most of them carefully selected to be likely votes for the incumbent. Bennet had lagged Buck for much of the year and had never approached the 50 percent threshold that many experts say is necessary in pre-election polls for an incumbent to expect victory. There was further evidence of a gap in partisan enthusiasms: at the time the polls opened, 74,000 fewer Democrats had returned their early-vote ballots than Republicans. But on election day, something was pushing Bennet even with Buck, and by the time Malchow turned in for the night the two candidates were separated by only hundreds of votes.

The next morning, he awoke to good news from the west. Bennet had pulled ahead of Buck, and was on his way to winning the race by 15,000 votes. His victory would help to keep the Senate in Democratic control. Malchow was having fun.

COLORADO WAS ONE of the rare sources of cheer for Democrats in an otherwise disastrous set of midterm elections in 2010. Party wise men were eager to mine grand lessons from the Rockies; if only they could figure out what made Colorado resist a national conservative wave, they could use Bennet’s strategy as a model for Obama’s reelection two years later. “The Bennet thing was pretty instructive,” Obama’s chief strategist, David Axelrod, told the National Journal in a postmortem. “The contrast he drew with Buck was very meaningful.”

The people who explain politics for a living—the politicians themselves, their advisers, the media who cover them—love to reach tidy conclusions like this one. Elections are decided by charismatic personalities, strategic maneuvers, the power of rhetoric, the zeitgeist of the political moment. The explainers cloak themselves in loose-fitting theories because they offer a narrative comfort, unlike the more honest acknowledgment that elections hinge on the motivations of millions of individual human beings and their messy, illogical, often unknowable psychologies. In fact, Bennet could have won the Senate seat because of major demographic changes and ideological fault lines around delicate cultural issues, or because of a single letter that exerted a subtle dose of peer pressure on its recipients— or hundreds of other factors big and small that played a part in changing people’s minds or getting them to vote.

The political craft thrives on that ambiguity. It allows just about any-one involved to take credit for good results or attribute blame for poor ones, confident of never being proven wrong. After a positive result on election night, everything a winning campaign did looks brilliant. When a campaign loses, consultants usually blame the candidate or the moment— and there rarely seem to be professional consequences for those who had set the strategy or tactics. Longevity, as well as the aura of wisdom that comes from it, is a political operative’s most valued trait.

Over a generation, helping Americans choose their leaders has grown into a $ 6-billion-per-year industry. But the new profession hums along on a mixture of tradition and inertia, unable to learn from its successes or its failures. The tools available to campaign operatives can do little to explain what makes someone vote —and few of the people toiling inside campaign war rooms seem disturbed by this gap in their knowledge. “It’s probably the only industry in the world where there’s no market research,” says Dave Carney, the top strategist on the launch of Texas governor Rick Perry’s presidential campaign. “Most things are done with only one check,” says Steve Rosenthal, a Democratic consultant who works closely on campaigns with many of the left’s top interest groups. “People’s guts.”

The unheralded arrival of the gently threatening letters in Colorado mailboxes marked the maturation of a tactical revolution against that kind of gut politics. The first stirrings had come a decade earlier, in the wake of the 2000 presidential vote, which shifted on election night from a contest between electoral strategists to a tussle among lawyers. What seemed at the time to be a low-stakes election would have a major effect on the way campaigns were waged. The narrow, almost accidental quality of George W. Bush’s victory— decided by 537 votes in Florida, or really just one on the Supreme Court—provoked a reexamination of where votes come from.

Seemingly small boosts of two or three percentage points quickly became indispensable components of a victory formula, and the intellectual hierarchy of thinking about campaigns changed accordingly. Turnout, the unsexy practice of mobilizing known supporters to vote, could no longer be dismissed by campaign leadership as little more than a logistically demanding civics project to be handled by junior staff or volunteers. Campaigns could not obsess only over changing minds through mass media. “Many strategists had been believers that ‘big things are all that matter in campaigns’—the big events, the big TV spots, the debates, the convention and the VP pick,” says Adrian Gray, who worked in the Bush White House and on both of Bush’s campaigns. “After 2000, for the first time a lot of people who shared that sentiment started to believe that there is a lot that can be done on the margins.”

The result has been an ongoing, still unsettled battle between the two parties for analytical supremacy, a fight that Bush data analyst Alex Gage likens to an “information arms race.” A new era of statistical accountability has been introduced to a trade governed largely by anecdote and lore. Each side has its own sobriquets for the intellectual rebels—Karl Rove boasted of his “propellerheads” and Rick Perry’s campaign of its “eggheads,” while those on the left were happy to call themselves “geeks.” They have made their cases in the PowerPoint palimpsest that inevitably arrives when an industry quickly learns to appreciate its own data. Suddenly, the crucial divide within the consulting class is not between Democrats and Republicans, or the establishment and outsiders, but between these new empiricists and the old guard.

The latter can be found in both parties, and it was a constellation of new-guard academics and political consultants on the left who had mastered the psychological tool used in the Colorado mailer. Six years before, one of them had first had the idea of ominously reminding citizens that whether or not they vote is a matter of public record. In the next few elections, the language and presentation had been refined through serendipitous collaboration unusual in politics, flowing effortlessly between operatives and academic researchers who previously had neither the opportunity nor inclination to work together. Functioning in a growing laboratory culture, they had jerry-rigged a research-and-design function onto an industry that long resisted it. And by the summer of 2010, they had perfected the politics of shame. It was only a matter of time before a desperate campaign, or interest group, would summon the audacity to deploy it.

TWO DAYS AFTER the 2010 election, while Colorado election officials were still counting ballots, Hal Malchow sat in his office in Washington, D.C., pleased to see Bennet on his way to victory. A sign hanging on the wall neatly summarized the self-satisfaction Malchow felt at moments like these: All progress in the world depends on the unreasonable man. He had spent more than three years obsessing over the technique he had used in his mailers. He had pored over the scholarly research that supported the use of what psychologists called social pressure, and he had finally persuaded the liberal group Women’s Voices Women’s Vote to overcome its fear of a backlash and send out letters to Colorado households likely to support Bennet but requiring an extra push to get to the polls.

The research had begun five years earlier. In 2005, a Michigan political consultant named Mark Grebner—whose glasses, stringy parted hair, eccentric polymathy, and relentless tinkering earned him comparisons to Ben Franklin—had written to two Yale political science professors who he knew were interested in finding new ways to motivate people to vote. The next year, they collaborated on an experiment in Michigan in which they sent voters a copy of their own public vote histories, along with their neighbors’, and a threat to deliver an updated set after the election. It was marvelously effective, increasing turnout among those who received it by 20 percent. But no candidate or group wanted to be associated with a tactic that looked a lot like bullying—and a bit like blackmail.

Prescription Drug Trials for Democracy

How to muffle such a potent weapon so that it could be used in the course of regular campaigns became an obsession of the Analyst Institute, a consortium quietly founded in 2006 by liberal groups looking to coordinate their increasingly ambitious research agendas. The Analyst Institute was a hybrid of classic Washington traits: the intellectual ambition of a think tank, the legal privacy of a for-profit consulting firm, and the hush-hush sensibility of a secret society. But its culture derived from the laboratory. The Analyst Institute was founded on a faith in the randomized-control experiment, which had migrated in the middle of the twentieth century from agriculture to medicine as a unique instrument for isolating the effects of individual fertilizers and vaccines. Social scientists later adopted field experiments, transforming research in everything from credit-card marketing to developing-world economics. Around 2000, such experiments found their way into politics, with voters as their unwitting guinea pigs. Over a decade these “prescription drug trials for democracy,” in the words of Rock the Vote president Heather Smith, have upended much of what the political world thought it knew about how voters’ minds work, and dramatically changed the way that campaigns approach, cajole, and manipulate them.

The Analyst Institute’s founding director, a psychologist named Todd Rogers, always liked to remind people that these behavioral science interventions couldn’t alter a race’s fundamental dynamics. No technique could do that; a good candidate or a bad economy would still set the conditions of an election. But experimental insights could decide close races— by nudging turnout up two points here, six points there —and none has proven as powerful and promising as Grebner’s social-pressure breakthrough.

It took three years of trial and error by academics and operatives, including Malchow, until he settled on softer, more friendly language — thanking people for having voted in the past as opposed to threatening them if they didn’t in the future —that delivered impressive results in a randomized experiment. During a test conducted during New Jersey’s 2009 gubernatorial elections, such a letter had increased turnout among voters who received it by 2.5 percent. Through other tests, Malchow had found that many political messages were most effective when delivered in understated white typed envelopes, as opposed to multicolor glossy mailers, and so he packaged the Colorado social-pressure letters in a way he hoped would resemble an urgent notice from the taxman. “People want information, they don’t want advertising,” Malchow said. “When they see our fingerprints on this stuff, they believe it less.”

The fact that Americans were tiring of political communication was, in many ways, a testament to the success of a profession Malchow had done much to develop. His métier was the direct-mail piece, the postbox- stuffing brochure so often dismissed as junk mail. That form, and Malchow’s career, had emerged in the long mid-1980s shadow of television and relegated Malchow to a second-class status in the consulting world’s star system. Direct mail is a staple of the category of campaign activity known as “voter contact,” distinguished—as compared with media advertising—by its ability to hit a preselected individual with precision. This is the way most voters interact directly with campaigns: the phone calls that interrupt dinner, the knock on the door from a young canvasser, leaflets stuffing the mailbox as election day approaches, personalized text-message blasts. Even as these voter-contact activities often go ignored by the people who write about politics, campaigns continue to spend money on these tactics, and lavishly—as much as a half-billion dollars per presidential campaign season.

Political mail has been perhaps the least glamorous of all the voter-contact tools. At a young age, however, Malchow was drawn to the fact that brochures, unlike broadcast ads or rally coverage on the nightly news, could be unexpectedly personal. Working with mail gave him a distinctive perspective on the electorate —which he saw as an array of individuals rather than a puzzle of blocs and zones—and the ambition to measure the effect of his work on a similar scale. As a result, Malchow had ended up playing a key role in the two most radical innovations in political communication: the use of field experiments to measure cause and effect, and the so-called microtargeting that allows campaigns to confidently address individual voter instead of the broader public.

But even as Malchow found a growing circle of allies in academia and liberal interest groups, partisan campaigns remained skeptical of ideas that would radically disrupt the way they thought about how votes are won. For years, when Malchow couldn’t convince campaigns to use the microtargeting techniques he said would help them locate otherwise unidentifiable pockets of persuadable voters, he paid for them himself, at a total loss of around eight hundred thousand dollars. The challenge of innovation came to excite him more than the predictable terms of partisan conflict. In fact, Malchow’s giddiness—perceptible as his eyes open wide behind his glasses, and his words break into a gallop — emerged most readily not when he was plotting how to win a specific race for a candidate but when he was figuring out a way to run all campaigns more intelligently.

Even amid the low points of 2009 and 2010 for Democrats, Malchow was heartened to see that the party’s few successful campaigns were ones that had some of the most creative analytics on their side. In Nevada, Senator Harry Reid’s pollster, Mark Mellman, tested the campaign’s messages through a continuous cycle of randomized online experiments, allowing him to see which people were actually moved by specific arguments, not only those who told a survey taker that they might be. In the same race, one independent group backing Reid used data on how neighborhoods voted on ballot initiatives (which show voter opinions on controversial issues like marijuana, taxes, and eminent domain) to define the political ideology of election precincts with a nuance impossible to gauge in partisan vote totals. Along Lake Tahoe, which has become home to wealthy California refugees, Reid’s allies defined pockets of rich libertarians they thought were winnable. So they downplayed Reid’s statewide message portraying Republican challenger Sharron Angle as an antigovernment extremis intent on dismantling Social Security—a stance the Tahoe targets could in fact find appealing—and instead played up her conservative views on social issues. Reid won the state by five points, boosted by expanded margins in upscale redoubts like lakefront Incline Village.

Data-driven methods were carrying the day in parts of the political process where Malchow hadn’t even imagined they would have a use. When Al Franken’s lawyers began the 2008 Minnesota U.S. Senate recount by pondering which of ten thousand challenged absentee ballots they should work to have counted, they brought one of the campaign’s microtargeting experts into the strategy sessions. Andy Bechhoefer ran each of the disputed voters through the campaign’s database, which used a complex mix of personal and demographic information, along with polling, to give each voter a score of 1 to 100, predicting his or her likelihood of supporting Franken over his opponent, Norm Coleman. Armed with these scores, Bechhoefer was able not only to point lawyers to the unopened envelopes most likely to yield Franken votes but also to identify which of the secretary of state’s categories for excluding votes had put them in a rejected pile. With that knowledge, Franken’s attorneys drafted expansive legal arguments that covered entire categories of problems instead of merely contesting individual ballots in a piecemeal fashion.

Bechhoefer recounted this experience to one of the regular lunch sessions hosted by the Analyst Institute, each detail in the scheme transfixing Malchow. Step by step, Bechhoefer illustrated how lawyers were primed to defend absentee ballots that had been challenged for change-of-address discrepancies (which leaned Democratic) while hoping that those with witness-signature problems (tilting Republican) remained uncounted. At times, Franken’s lawyers watched their adversary challenge ballots they knew were almost certain to be votes for Coleman, only because the Republicans had not used such sophisticated methods to model them. Over an eight-month recount, Franken gradually turned a 477-vote deficit on election day into a 312-vote lead when Coleman’s last court challenge was exhausted in the summer of 2009, giving Democrats their sixtieth senator.

“Everybody in the Analyst Institute was grinning ear to ear—what a triumph,” says Malchow. “I was like a Cheshire cat. I thought this was the coolest thing I ever heard.”

THE POLITICAL BOOKSHELF is filled with works that have heralded epochal change: 1972 had The New Style in Election Campaigns, while 1981 brought The New Kingmakers, which profile the first generation of political consultants. Every time a fresh communication technology has become available, those who practice politics have been quick to announce that elections would be remade in its image “A campaign rally is three people around a television set,” Democratic media consultant Bob Shrum, who made his money from TV ads, boasted in 1986. A decade and a half later, Dick Morris predicted that television-focused media consultants like Shrum were about to be eclipsed by an emerging cadre who communicated online. “The current crop are like silent film stars—their skills will no longer be valuable in the Internet era,” Morris told the Washington Post in 2000.“They’re good at condensation, at the 30-second spot. The new environment of the Internet calls for elaboration, for expansiveness.”

Neither consultant’s prediction has been entirely right, ignoring less flashy but more influential shifts in how campaigns win votes. The scientific revolution in American electoral politics has relied on lots of technology, particularly to assemble and sift through large databases, but its most lasting impact may be a resurgence in lo-fi tactics. The genius alchemists behind microtargeting spend their days deciding where candidates should send postcards. A gubernatorial campaign in Texas conducted meticulous experiments to learn which was a more worthwhile use of its time and budget: sending the candidate to meet with a newspaper editorial board or to a barbecue restaurant filled with one hundred supporters. Within the headquarters of a presidential campaign widely heralded as the most technologically advanced in history, some strategists think their most impressive accomplishment wasn’t their iPhone app but the time when a staffer figured out how to buy ads on a bus whose route he was certain was used by voters the campaign was trying to reach.

By 2012, it has become impossible to correctly interpret campaign strategy without understanding the revolution in tactics. Some of the early decisions that shaped how the presidential race would be run were built on technical innovations invisible to the outside world. Texas governor Rick Perry considered withdrawing from select primary-season debates in part because the social scientists he had invited to run large-scale randomized- control experiments in an earlier race concluded that the candidate could have his biggest impact not through media appearances but through localized travel to targeted states. (In retrospect, no social scientist could have calculated how atrocious a debater Perry turned out to be.) Former Massachusetts governor Mitt Romney, who a decade earlier had been the first candidate in the country to use microtargeting formulas, knew he could hold back on committing resources to identify voters in Iowa because he had algorithms that would instantly tell him how every caucus-goer would be likely to vote even before they had made up their minds. And in Chicago, Barack Obama’s aides looking to expand the playing field against their Republican opponent thought they might be able to use psychological tricks—which had significantly reduced the cost of registering new voters—to remake the electorate in certain states in a way that could permanently confuse red and blue.

Electoral politics has quietly entered the twenty-first century by undoing its greatest excesses of the late twentieth. Just as architects have atoned for the vainglories of their field’s high-modernist period by pummeling its concrete superblocks and putting sleeping porches and Main Street–style shopping strips in their place, some electioneers are starting to conclude that political campaigns lost something when they became warped by broadcast waves. The campaign world’s most sophisticated new thinking about who votes and why, informed by an intuitive understanding of the political brain, has naturally turned attention to the individual as the fundamental unit of our politics. The revolutionaries are taking a politics distended by television’s long reach and restoring it to a human scale — even delivering, at times, a perfectly disarming touch of intimacy.

Our campaigns have not grown more humanistic because our candidates are more benevolent or their policy concerns more salient. In fact, over the last decade, public confidence in institutions—big business, the church, media, government—has declined dramatically. The political conversation has privileged the nasty and trivial. Yet during that period, election seasons have awakened with a new culture of volunteer activity. This cannot be credited to a politics inspiring people to hand over their time but rather to campaigns, newly alert to the irreplaceable value of a human touch, seeking it out. Finally campaigns are learning to quantify the ineffable —the value of a neighbor’s knock, of a stranger’s call, the delicate condition of being undecided—and isolate the moment where a behavior can be changed, or a heart won. Campaigns have started treating voters like people again.