By clicking “Accept All Cookies,” you agree to the storing of cookies on your device to enhance site navigation and analyze site usage.

Skip to main content

Analytics at Work: A Political Case Study

December 16, 2013

A year into President Obama’s second term, it’s easy to forget how contentious presidential elections can be. Dan Wagner has a particularly good memory of November 2012, however, as well as the year leading up to the re-election.

Over the course of two years, he built and managed a team of more than 50 analysts, data scientists, and computer scientists that helped to reinvent how campaigns and organizations think about data.  While the context was U.S. politics, Wagner essentially addressed the broader issue of how to develop an organization that bases its thinking and decision-making on science rather than gut emotions and tradition. Wagner, who now serves as CEO of Civis Analytics, described his experiences and lessons-learned at a recent MIT CDB/IDE seminar, “Analytics and Innovation – Case Study of Obama 2012.” (For more on data-driven decisionmaking see this related  blog).

Wagner’s primary task, and one of the most challenging for the neophyte campaigner, was to upend many traditional assumptions about campaign strategies. Instead of strictly adopting a strategy based on demographic segmentation, for instance, Wagner’s team focused on building data sets about voters and getting people to register. With that data in hand they could more accurately predict turnout and likely voting patterns by geography and demographics.

In addition, the campaign itself had to be reorganized. First, Wagner brought polling and analytics in-house and relied on its own frequent polling of thousands of people daily to take the pulse of voters instead of using data from outside firms. “We found that outsourcing these functions disrupts the process,” Wagner said. With immediate access to data, “we were able to pivot quickly and we trusted the data.”


Breaking Down Silos: Centralized Analytic Teams

Another successful strategy was consolidating and centralizing analytic teams from across the organization, he said. Previously, too many teams had their own analytic “nerds” who gave their bosses data they wanted to hear, not what was always accurate. The new teams were divided into two primary groups: techs who ran tests and developed algorithms for the data analysis, and problem solvers who worked with those in the field –media, finance and communications, for example—to make the data actionable for those in charge of the campaign. Overcoming cultural change can be huge, he noted.

Wagner said that all three C’s of data-based decisionmaking– what he calls, culture, computers and cerebrums—have to be aligned to unify the organization, extract truthful data and drive good insights and decisions. In the Obabma campaign case, funding had to be reallocated and messaging reworked based on the tests and experiments they were conducting. Old methods that weren’t showing results were tossed out.

Regardless of the ultimate success of the scientific tactics, it is usually tough to get buy-in from skeptics, Wagner said. “You have to convince people in stages until they see the proof and become advocates. Then, data can become empowering and trustworthy.”

And isn’t that’s what politics is all about, after all?