The Independent View: The Use of Data in a Presidential Election Campaign

It was two weeks before Election Day when Mitt Romney’s political director signed a memo that all but ridiculed the notion that the Republican presidential nominee, with his “better ground game,” could lose the key state of Ohio or the election.

But the claims proved wildly off the mark, a fact embarrassingly underscored when the high-tech voter turnout system that Romney himself called “state of the art” crashed at the worst moment, on Election Day.

To this day, Romney’s aides wonder how it all went so wrong.

But a reconstruction of how the campaign unfolded shows that Romney’s problems went deeper than is widely understood. His campaign made a series of costly financial, strategic, and political mistakes that, in retrospect, all but assured the candidate’s defeat, given the revolutionary turnout tactics and tactical smarts of President Obama’s operation.

These failures are now the subject of scrutiny by national GOP officials who say they plan to “reverse engineer” the Romney effort to understand what went wrong. A number of Romney’s top aides stressed in interviews that, while they remain proud of their work, they feel an obligation to acknowledge their numerous mistakes so lessons are learned.

Democrats said they followed the trail blazed in 2004 by the Bush campaign which used an array of databases to “microtarget” voters and a sophisticated field organization to turn them out. Obama won in part by updating the GOP’s innovation.

President Obama’s national field director, Jeremy Bird, drew his inspiration from the time around 2001 when he witnessed, as a young Harvard Divinity student, a group of African-American students in a Roxbury church, pressing their case for school funding with members of the Boston City Council. It was a model, in miniature, of grass-roots engagement that would shape Bird’s career in politics and attract him to Obama, who had himself been a community organizer. Bird was confident that Obama would commit massive resources to building an organization that zeroed in on individual voters. It would be like that Roxbury church encounter, multiplied a thousand times.

A first-class ground operation in 2012 required leading-edge technology, and here also an early gap opened between Obama and Romney.

The goal was to create the political equivalent of a Facebook or Twitter, a platform that would change the way presidential campaigns are run. And Obama’s team found just the man for the job: a 34-year-old programming whiz named Harper Reed, who got his start as an 11-year-old pecking on an Apple II and had never held a top job in a political campaign. As Reed assembled his team, he insisted on being given leeway to hire some of the best techies in the country, from Facebook, Craigslist, Twitter. Moreover, he insisted the team be largely internal, rather than have the enterprise be divided up among outside consultants.

The group was haunted by the failure of a similar venture in Obama’s 2008 campaign, when a get-out-the-vote computer program called Houdini crashed and could have cost the election if the race had been closer. This time, Reed and his team created a successor that they named Gordon, after the person who punched Houdini in the stomach shortly before the magician died.

Separately, the Obama team created a system called Narwahl, named after an Arctic whale, which linked disparate computer programs together. Narwahl and Gordon would be tested repeatedly in exercises that Obama’s team called “game day.” Every imaginable failure would be thrown at the systems — hacker attacks, database meltdowns, Internet failures — and the team would be challenged to write up a manual for how to deal with each disaster. It was, they said, more fun than the fantasy war game Dungeons & Dragons.

Zac Moffatt, Romney’s digital director, did not have the luxury of Reed’s time or resources. Moffatt came from the world of politics, had worked at the Republican National Committee and had long believed Romney would be the best GOP candidate for president. Moffatt played catch-up from the start. He had 14 people working for him in the primaries and then, around May 1, he submitted a general election plan that required at least 110 people and would eventually have 160. Obama was far ahead. Moffatt recalled his assignment in daunting terms: “Can we do 80 percent of what the Obama campaign is doing, in 20 percent of the time, at 10 percent of the cost?”

Moffatt’s team nonetheless managed to create big projects on short notice. For example, one of the highest priorities was a Facebook app that would enable the Romney campaign to locate voters who otherwise could not be found by telephone. By some estimates, half of younger voters do not have a landline or cannot be reached by cellphone. Three weeks before Election Day, the app was unveiled by the campaign and downloaded by 40,000 Romney supporters.

There was only one problem. Months earlier, Obama’s campaign had developed a similar app, which had been downloaded by 1 million people.

In campaign postmortems, Republicans have been criticized for spending too much on advertising compared with Democrats, even as some reports said Obama was able to book television spots more cheaply, run more ads in key states, and reach key voters more effectively.

The Obama campaign used a program called “the optimizer” that linked data from its voter databases, focus groups, and television ratings to determine how to reach people who do not typically see campaign ads. As a result, Obama purchased ads on channels such as TV Land and Hallmark that were watched by voters who rarely saw news programs where ads often appear.

A key difference was the depth of voter contact. Romney took comfort in polls that showed voters had been contacted equally by both campaigns. But the polls were misleading, perhaps equating a recorded robocall on the phone with a house call by a worker.

As dawn broke on Election Day, 800 Romney volunteers filled the floor of TD Garden in Boston. This was the centerpiece of the campaign’s turnout operation, code named ORCA, that was supposed to swallow Obama’s Narwhal program. But the Romney team was so determined to keep ORCA secret that it had never run a test at TD Garden; it had only gone through some lesser runs in a different building.

The ORCA workers were supposed to be in contact with more than 30,000 volunteers stationed at polling places across the country. Those volunteers were told to bring a smartphone and go to a secure Web page on which they could report the names of everyone who voted. In this way, the Romney campaign could determine if supporters had failed to show up and urge them to vote. But as volunteers on Election Day began tapping in the names of voters, it became clear something was wrong. The system was so overloaded with incoming data from volunteers that it exceeded capacity and crashed.

The Obama campaign, which had suffered a similar meltdown in 2008 and had been zealous about testing its systems this time around, had no glitches. Tens of thousands of Obama volunteers across the country sent real-time data from polling places, enabling workers at Chicago headquarters to ensure that expected vote totals were on track. More importantly, the field organization put in place by Jeremy Bird hit its goals, turning out the needed number of voters to reelect the president.

In the coming months, Romney, ever the data-driven analyst, plans to contemplate how his political life came to an end, and what the party should do next, according to his son Tagg. The fight for the ideological soul of the party will play out for months. But recommendations are already pouring in for the party to create a ground-game infrastructure long before a nominee is selected, to catch up to the Democratic advantage in high-tech turnout operations, and to find ways to make the party more inclusive for minorities and women.

So today, the Republican data firms i360 and the Data Trust announced an agreement to share each other’s voter files, providing a possible boost to GOP candidates on the 2014 ballot. The Data Trust serves as the primary voter data warehouse for the Republican National Committee’s overhauled digital voter turnout program. The sharing agreement with i360, one of the top GOP-friendly data outfits, means that candidates that have been relying on RNC data to target voters will now have access to i360’s voter file, and vice versa.

The agreement augments the volume and sophistication of the data on America’s 190 million registered voters available to Republican campaigns. It also helps ensure that all Republicans have access to the same data so that they can avoid strategic missteps.

This has been among the advantages held by Democrats on the data front, as they all primarily use the same firm, NGP VAN.

“For the first time, the Data Trust and i360 will work together to reduce duplication and make right-of-center voter contact efforts more efficient, resulting in our partners having access to more and better data,” John DeStefano, president of the Data Trust, said in a statement.

“i360 is excited that through this data sharing partnership, free market organizations will be able to more effectively use data in their voter outreach efforts,” Michael Palmer, president of i360, added.

Here is how the press release from the Data Trust described the agreement:

“Through this partnership, voter contact information gathered by clients of either The Data Trust or i360 can improve the data shared with all clients. For example, if a client of either company conducting voter outreach identifies a voter attribute or preference, clients of the other organization will benefit from that information. As a result, conservative groups and campaigns will have more information about voters at their disposal for their own activities than ever before.”