When President Barack Obama was reelected in November 2012, it was due in no small part to the team of data experts at Blue State Digital. European Media Director Rob Blackie explains how they did it.

November 2012

Pop quiz: If you had to name the most memorable moment in the history of presidential politics, what would it be? A well-chosen zinger that changed the course of a campaign? ("It's the economy, stoopid!") An ad that summed up the mood of a nation? ("It's morning in America...") Or the moment television made its mark? (Think JFK 'winning' the first debate against Nixon.)


Everybody has their favorite, but hands up if you named the failed 2004 presidential bid of Howard Dean. No? Didn't think so. And yet it was in the aftermath of this defeat that four alumni of the campaign founded an online fundraising, advocacy, and engagement outfit that today is at the forefront of a revolution. Blue State Digital is running Washington politics the Silicon Valley way, and transforming how a president connects with the people.

First hired by Obama in 2008, and again in 2012, BSD is responsible for every facet of the president's mold-breaking online campaign, from fundraising to email communications, social media marketing, design, web video, grassroots organizing, search and display advertising, staff- and user-generated content, cutting-edge analytics and optimization, and peer-to-peer mobilization.

Like Nate Silver, BSD is dragging politics into the digital age, using big data to run a personalized, responsive, and highly social campaign that reached out to an email database of 13 million people, signing up over three million individual donors and 45,000 volunteer groups. The result? Over half a billion dollars in the campaign war chest, and a historic second term for Obama.

European MD Rob Blackie took time out to tell us how they did it and the lessons they learned along the way.

Think Insights: Operationally, what was the biggest change in approach in the 2012 campaign compared to 2008?

Rob Blackie: Firstly, the databases were properly integrated. In 2008 they were synchronized occasionally but they weren't real-time, which means we missed opportunities to spot clever things we could do, or things that were going wrong. This time around it was different. Secondly, we had a very large data mining team that wasn't just looking at tactical website observations; it was looking at how messaging, donations, and events in the field all fit together, so we're spotting opportunities to make things more effective.

During the election, the number crunchers found that George Clooney had an 'almost gravitational' pull on West Coast women between the ages of 40-49, and the campaign was able to use this to pull in donations. How can big data reach insights like this and what are some of the other ways you used the analytics?

The key is to approach data as part of your overall campaign process, rather than a minor optimization thing on the side. If you use data this way - that is to say, creatively - you can spot opportunities to improve your campaign. In this case it was about who our spokespeople were, but it might be what your messages are and how those messages align to particular people.

What's important is that you pull it out of the data, which means you're pulling it out of the way people behave, rather than how they answer questions that you ask them. What market research has tended to find has mainly been driven by the questions that it asks; the advantage with big data is that you actually see how people are behaving over time so you can spot behaviors that wouldn't be intuitive to you - things you'd have never thought to question. The opportunity for organizations is to gain those insights into people's lives that get much closer to your audience than you'd traditionally be.

The opportunity for organizations is to gain those insights into people's lives that get much closer to your audience than you'd traditionally be.

What other insights came from the data?

An obvious example is that the data showed the campaign was really struggling to articulate how the economy was improving. The response, which was extremely successful, was to create a very basic infographic showing the number of jobs created under Obama. There was this curve that turned as things got worse and worse under Bush, but as soon as Obama came into office things gradually started to get better as economic measures took effect. That was an insight that came from the campaign understanding that people weren't really getting the deep economic message, and that they needed to find new ways to articulate it.

Obama's campaign manager, Jim Messina, was quoted as saying that this would be a metric-driven campaign that wouldn't rely on political instincts. Did that affect the way you targeted individuals in different demographics, like Hispanics or boomers or young people?

Demographics were targeted differently, but they were targeted on the basis of their behaviors, not on the basis of a marketing definition. In other words, if somebody shows an interest in the fact that contraception is available under Obamacare, it's much more sensible to communicate with them in the future on the basis of that interest rather than communicating with them on the basis of being a baby boomer or Generation X or Generation Y. I think those marketing definitions are not very helpful because they're artificial constructs. It's much more helpful to talk to people about what they're interested in.

In terms of the different ways in which you connected with voters, what kind of role did things like mobile, local search, and social networks play?

The obsession with data meant that all of those were used, but the campaign could really understand how useful they were. Really, the biggest single change has been smartphones. Smartphones mean that people expect to get information instantly, so that puts a very high premium on managing to react fast to events. It means that people can do things like donate money or connect with the campaign at any point because people never leave their smartphones more than a few feet from them. It means that if you need to deliver location-relevant stuff there's a way people can be found. For instance, if you want people to early vote it's very easy to send them something saying, 'Click here to find out where your nearest polling station is so you can go and vote today.' The campaign was very successful in getting people to early vote on that basis.

How important was YouTube?

It was very important - and that's only going to increase. In the month before the election, the official BarackObama.com YouTube channel got about 23 million views. Not all of those viewers were from swing states, but that's quite a high ratio of viewings to people we're trying to influence.

The videos are reasonably long - two to four minutes normally - and obviously completely on message, so they were very effective. I suspect that influence will grow a lot in the next election because as mobile phone networks develop more capacity, people will find it easier to watch mobile on video than they do now. So while video was very important for us, it's something that will grow in the next few years.

A running joke during the election was the number of emails that supporters were receiving from Obama. Why so many emails and did they work?

I know there were long-term bits of analysis done of things like email frequency, and I know it was considered to have worked very well despite the high frequency. The funny thing is, because the campaign had very good metrics they knew that the people complaining were only a very small minority. Most people weren't unhappy about it. The second thing they did, which was actually very clever, was if you did click on 'unsubscribe,' the campaign made a very determined attempt not to lose you. So some of those people would be persuaded to opt down into less frequent communications. Other people would fully unsubscribe, but the campaign would then try and find ways to get back and hook them in later and rebuild that relationship. On the whole it was very successful, despite the surprisingly high frequency.