Follow Datanami:
June 7, 2013

Big Data Analytics Give Electoral Edge

Alex Woodie

There’s an old saying that politics is the art of the personal. Barack Obama’s 2012 presidential campaign may not have proven that saying moot, but its deep reliance on big data analytics put a fascinating new twist on it, and changed forever how major political campaigns will be fought and won going forward.

Obama’s 2012 campaign manager Jim Messina recently discussed how it used big data and social media technologies to give it a decided edge against the Romney camp in last year’s president election.

“We sent out one tweet on election day in 2008. We thought it was a silly technology that would never go anywhere,” Messina says in the video. “We had 12 people on our data team in 2008. We had 165 in 2012. We ended up using all that data to change the way we dealt with voters.”

According to Messina, who is now the chairman of a non-profit called Organization for Action, the 2012 campaign spent $1 billion amassing and using a big data-driven analytical system that was also hooked into the campaign’s social media activities. Over the course of 14 months, the campaign ran a total of 62,000 simulations of likely voter behavior based on that data.

“We ranked every single voter in America…from one to 100 about whether or not they support Barack Obama, whether you’re going to vote…and the holy grail: Whether or not you had a chance of being a ticket splitter,” Messina said. “Every single decision we made was based on the 62,000 sample every night. Data became the most important thing we did.”

The broad sampling, the richness of data, and the sophisticated analysis combined to give the 2012 campaign a very high degree of certainty about what was going on with voters, which allowed the campaign to tailor messages to individuals. “We’ve gotten really good at predicting behavior. We ended up predicting our final vote in Florida by .05 percent. We didn’t miss any state by more than .5 percent,” Messina said in the video.

Toward the end of the campaign, the team accelerated efforts to hook the big data to social media. “We built this thing called targeted sharing. It took us a year and way too much money, but it allowed us to use Facebook to persuade people,” Messina said. “We spent $1 billion to figure out a simple truth: What your friends and family and neighbors say is more important to your consumer decisions and your political decisions than anything else.”

The Facebook app allowed Obama supporters to send personalized messages to friends and family who, according to the campaign’s data, were still undecided. The app succeeded in convincing 78 percent of those undecided voters to vote for Obama, Messina said.

Relentless analysis–what Messina called “AB testing”–allowed the data team to scrutinize nearly every aspect of the campaign, from how far the front desk should be from the door of the campaign office to what font to use in campaign emails.

While Obama’s big data crunching activities have been criticized by some as eliminating the personal aspects of politicking, Messina has a different view of what his team built. “In the old days, people were treated like numbers. I wanted to treat them like individual people and what we know about them,” he said. “The way we use data is to make everything we did easier and more targeted.”

Political rules prevent the Obama campaign from selling its big data machine. By the time the 2016 campaign kicks off, backers of each political party will need to invent their own big data creations. The legacy of the 2012 campaign will live on, however, in showing future campaigns how it’s done.

Related items:

Obama Big Data Team Reforms as Civis Analytics

How Data Analytics is Shifting Politics

Civic Hacking Targets Optimizing Government

Datanami