Former digital director, President Obama’s 2012 campaign
What was the digital campaign’s key contribution to President Obama’s re-election?
It put supporters back into a primary role. Even in the 1996 and 2000 campaigns, supporters were largely taken for granted. If you happened to live within range of a field office, you might go volunteer, but the campaigns focused obsessively on undecided voters. Of course we still cared about undecided voters this time, but we realized the most important thing we could do on the digital side was to cultivate relationships with the supporters on e-mail, Facebook, Twitter. We wanted to keep them inspired, engaged and informed. If we gave people a reason to hit the retweet button every now and again, hit the share button, they could reach almost everyone in the United States more powerfully than we as a campaign operation ever could.
Digital technologies are obviously ubiquitous, but social media, at least, skews younger. Did that limit your influence?
President Obama on election day had about 34 million Facebook fans. Those people were friends with 98 percent of the U.S.-based Facebook population, which is more than the number of people who vote. People ask me how important Facebook and Twitter will be in five years, and of course I don’t know. Nobody does. But the overall direction of these technologies is clear. They’re giving ordinary people more resources, more access to information and to each other. People are so turned off to politics today. They’re turned off to political advertising. But they trust their friends. It’s the same in business. Every business is going to be increasingly relying on ordinary people to transmit their messages. Everyday people can do a lot of damage if they’re not getting a positive experience, or they can become the best marketing engine a business could hope for.
Did you deliberately craft a system of engagement across platforms?
We tried to think about our relationships in as holistic a way as possible. Political campaigns have often thought about online supporters as being a different constituency than offline supporters; volunteers as being different from donors. But most volunteers donate and most donors volunteer. So you’re able to get a much more complete picture of a person if you integrate all sources and we had some success doing that. We wound up toward the end with a unified database that pulled in donor data, online data from clicks, and volunteer data, as well as commercial data. That gave us a much more comprehensive picture of who people were, so we could serve them better.
How difficult was it to build that universal database?
There was no philosophical battle, but there was a technology battle. We were obsessed with looking at data in as granular a way as we could. If a person consistently opened e-mails about veterans issues and not about the economy, that told us something important. People are making decisions every day about what they click on and what they open and what they forward. But bringing it all together is very complicated, especially when it comes to social. We get terrific data there, but people don’t always use their full legal name and street address so you wind up making decisions about what degree of uncertainty you’re willing to tolerate.
You obviously achieved tremendous results. But is there anything you’d do differently?
Most of all, I wish we had started earlier. When you’re working toward an end date, you gain a really clear sense of the value of time.
How would you have used the extra time to your advantage?
For example, we built a saved payment information system that gave donors the option to save their payment information so they wouldn’t have to pull out their credit card for subsequent donations. That way, we no longer needed to e-mail a link to our donation page; we’d just send an e-mail that read, “Click here if you want us to charge your account $5.” The people who received that e-mail contributed to the campaign in a 3:1 ratio compared to people with identical demographics and histories. Think about that. The second that you opt in to enter that payment information, your value to the campaign is three times what it was before. Had we rolled out that product six months earlier, we could have reached 500,000 people and raised a lot more money. Experiences like that make you realize, my God, I wish we just pressed every possible angle from day one.
You mentioned skepticism toward political messaging. This is obviously also true in marketing. Can you talk about the importance of authenticity in an age of social media?
It’s simple. You can’t fake it. You cannot use a lot of banner advertising or a clever tweet to paper over the fact that your product is no good or your corporate values are awful or that you don’t treat people well. All of these changes in technology are making companies be better because they’ve got no choice. They can’t get away with not being authentic. Sure, your tweets should be clever and your banner ads should be great, but all of that stuff decreases in importance as the Internet imposes greater accountability and visibility into the business.
Do you have any advice for how companies and departments can be more agile in reaction to real-time data?
Well, the first thing I would talk about is structure. I do some consulting with large corporations that have a marketing department and a sales department and a PR department and a communications department and maybe a brand department. It’s an organizational scheme that pre-dates the Internet. That may have worked for creating a TV ad or a press release, but that’s not a process conducive to firing off a tweet out during the Super Bowl blackout because every department is fighting over the new toy. By contrast, we didn’t have to deal with legacy systems, and so there was never a question that digital would be a free-standing department co-equal to everything else. As a result, we didn’t have to have a cross-departmental battle over every last thing or argue our case to a manager who was coming at things from a completely different point of view. I guess the point is that certain structures and incentive systems prevent agility. It’s better to not have to contend with that.
How did you use predictive analytics?
We predicted where people stood, how likely they were to vote, and even the likelihood that our contact would have an effect. Our analytics team not only predicted how people were going to vote but also how people were going to vote at various times of day. In a way, the entire campaign was predicated on a couple of big predictive models. A campaign has a finite number of door knocks or phone calls that they can make or commercials that they can run. So the whole ballgame is making sure you use those resources as efficiently as possible. You never, ever want to send a volunteer to go knock on a door where the person is not going to vote for President Obama because that’s one fewer door knock you can make. There were certain situations where we determined that a token knock actually had a counterproductive effect. So we weren’t only looking for undecided voters; we were looking for persuadable voters likely to be moved in the right direction. That’s all predictive modeling.