The Obama campaign's director of digital analytics wants to bring a culture of testing to progressive organizations.
At the dawn of the 2012 election cycle, the digital sector was the Wild West of the campaign world and Amelia Showalter wanted in.
After honing her targeting skills at MSHC Partners and Changing Targets Media, she jumped aboard President Obama’s reelect as the campaign’s director of digital analytics—a position that had her among those leading the digital and data charge last year.
Much of her time was spent testing email. And with a list as big as the Obama campaign’s, the possibilities were practically endless. The team would often come up with a test in the morning, run it in the afternoon and get results by the evening—a change in strategy could happen as quickly as that night or early the next day.
“Everything we did with our email program was very well tested,” Showalter says. “Anything new, any sort of messaging, any gimmicks—they were all tested. And if you saw it on multiple occasions, it means we tested it and found that it worked quite well.”
Post-Obama, Showalter wants to make the same sort of rigorous testing standard practice for progressive organizations on and off Capitol Hill.
C&E: Where is your focus now that the campaign is behind you?
Showalter: I’ve re-launched my consulting business with more of a digital focus. Over the next year, I’m going to see how much I like this consulting lifestyle. What I’m really hoping to do with the consultancy is bring that testing to progressive organizations, campaigns and firms. There’s a lot of interest right now in what we did on the Obama campaign—people want to know our best practices. The best practice really is to do your own testing and figure out what works for your organization.
C&E: What approach did you take to building the Obama analytics team?
Showalter: I started with just me and I had to hire a whole team. It was harder than you’d think to find people that had every single skill set. The stats background, the database management background, the digital background—you can’t find all those things in one person. I had the political background, but not the digital. I had a stats background, but not a hardcore database management background. So I looked for really smart people who had at least one of those skill sets and the capacity to learn the others. My deputy Evan Zasoski did a lot. He had worked at the Democratic National Committee, and was a really great programmer. Almost everyone else who was on the team is someone who hadn’t worked in politics before. It was helpful to get different perspectives, particularly from people who weren’t married to the way things had always been done.
C&E: What was the effect of having just a single database in 2012?
Showalter: I think having all that information easily accessible made a real difference. In our email program, we would occasionally personalize the emails based on past behavior. If someone had signed Michelle Obama’s birthday card, and we were trying to get people to sign Barack Obama’s birthday card many months later, we could find the data set of who had taken that action and say “Hey, you signed Michelle’s birthday card. Would you like to sign Barack’s card?”
I wasn’t on the 2008 campaign so I don’t know how decisions were made. But I went back and looked at the amount of money they would ask people for and it was personalized—there were maybe four buckets. There were people who would be asked for $25, then people who would be asked for $50. This time we had a few hundred buckets. We actually had a whole algorithm. I had a very smart person—Ricky Gonzales—spend a huge amount of time on this formula. Our focus was on how much further we could push the innovations from 2008.
C&E: How much of what you did on the Obama campaign can be done in races further down the ballot?
Showalter: I think a lot of it can. It was a huge luxury to have the email list we had. It allows just a much greater breadth of testing. But every email list can be split into two. I think it’s important to have campaigns, even smaller ones, get into the mindset of testing. It can be testing two subject lines instead of 18 different variations. I think that can be applied on a pretty small level.