Established in June 2013, Civis Analytics grew out of President Barack Obama’s 2012 re-election campaign. Its earliest roots, however, can be traced back to 2010 when Obama campaign manager Jim Messina, who would later become executive chairman of Organizing For Action, announced that “we are going to measure every single thing in this  campaign.” By this, Messina meant that he and his large analytics team—which consisted of five times as many staffers as the one which had worked on Obama’s 2008 campaign—would be putting their faith entirely in so-called “Big Data.” This term refers to “millions of data points” that are gradually collected and subsequently entered into centralized databases for analysis. The purpose of the data, in this case, was to help the Obama campaign build virtual profiles of likely voters, potential voters, and most importantly, uncommitted voters who could possibly be persuaded to support Obama.
The first order of business for Obama’s analytics team was to merge, into a single megafile, the voter information—e.g., age, sex, race, neighborhood, voting record, and purchasing habits—contained in the various databases which the 2008 campaign had already compiled. This would eliminate duplication of information, reveal gaps where more data was needed, and streamline the process by which number-crunchers could assess the proclivities of individual voters. The goal was to be able to predict which types of people could be persuaded to support Obama by certain kinds of appeals. Indeed, the call lists in the campaign’s field offices did not just contain names and phone numbers; they also ranked the names in order of their estimated persuadability. Said one senior Obama advisor: “We could [predict] people who were going to give online. We could model people who were going to give through mail. We could model volunteers.”
For the better part of two years, the Obama analytics team—stationed in a large, windowless room (dubbed “The Cave”) at the north end of the campaign’s vast headquarters in Chicago—worked tirelessly on its Big Data project. Mostly young adults who were passionately devoted to Obama’s agendas, the team members received guidance in their work from Obama supporter Eric Schmidt, executive chairman of Google, Inc. In turn, they coordinated the efforts of “thousands of staff and millions of volunteers” nationwide, who collectively made as many as 9,000 phone calls per night in order to gauge broad public opinion. Over time, they interviewed millions of respondents and gathered detailed information about them, which they continued to add to their “Big Data” files.
The combination of personal data and daily interviews enabled the Obama alanytics team to identify specifically where support was lagging and needed to be actively bolstered via some form of personalized contact such as Facebook, e-mails, phone calls, or home visits—whatever approach the analysts, based on the data they had compiled, deemed most likely to persuade each targeted individual. By CA’s calculus, these were “the most accurate voter targeting models ever used in a national campaign.”
A noteworthy example of how the Obama analytics team targeted specific categories of voters occurred in the late spring of 2012, when the backroom number-crunchers noticed that in a “Dinner with Barack” contest which had been held on the West Coast, women aged 40 to 49 were far more likely than any other demographic group to make small campaign donations in exchange for a longshot, lottery-style opportunity to dine with President Obama and film actor George Clooney. Aiming to replicate the millions of dollars generated out West by that Obama-Clooney “lottery,” the analytics team set out to identify an East Coast celebrity with similar appeal among the same, middle-aged-female demographic. This was the genesis of yet another highly publicized “Dinner with Barack” contest, where the winner was invited to dine with Obama and actress Sarah Jessica Parker at the latter’s Manhattan brownstone on June 14, 2012. Events like this not only raised a great deal of cash for the campaign, but also helped stoke enthusiasm among potential female voters.
As the Obama analytics team gathered ever-more information about voters of various descriptions, it created specific advertising and outreach campaigns designed to address their respective concerns. To test the effectiveness of these efforts, the analysts developed (and constantly refined) statistical and predictive models through which they ran more than 65,000 simulations each night to project which candidate—Obama or Romney—was leading in the various battleground states.
Ultimately, the Obama campaign’s data analysts accurately predicted the 2012 election outcomes in every battleground state within one percentage point. Highly impressed by this achievement, Eric Schmidt said: “I think of them as people scientists. They apply scientific techniques to how people will behave when confronted with a choice or a question.”
Soon after the election, Schmidt invested several million dollars in the formation of a new company, Civis Analytics (CA), that would keep the core Obama analytics team together as a unit. Not only could CA assist other Democratic candidates post-Obama, Schmidt reasoned, but it also could help nonprofit groups maximize their effectiveness. CA’s services include database integration, predictive analytics, randomized controlled experiments, and survey research.
The College Board, which administers the Scholastic Aptitude Test each year to high-school seniors, became one of CA’s first clients in 2013, hiring the company to: (a) figure out why high-achieving, low-income students were less likely to apply to elite colleges than equally qualified students from wealthier families, and (b) design a campaign that would steer more low-income students to the better schools. Several companies and nonprofits also commissioned CA to help them enroll people newly eligible for health insurance under the Affordable Care Act (“Obamacare”).
CA is currently staffed by more than 20 individuals, all with close ties to the Obama administration. The organization’s official founder and CEO is Dan Wagner, a University of Chicago graduate who served as national targeting director for Obama’s get-out-the-vote initiative in 2008, and for the Democratic National Committee two years later. In 2012 Wagner was the Obama re-election campaign’s chief analytics officer.
To view a list of CA’s other founding staffers and a brief description of each, click here.