Matt Dover is director of campaigns at Civis Analytics. He was deputy battleground states analytics director for President Obama’s 2012 reelection campaign.

C&E: You got your start a bit differently from most political pros—tell me about how you got into the campaign game.

Matt Dover: I actually started out as a middle school teacher. I taught American History in New York City as part of the Teach for America program. After that I went back to school at the Kennedy School at Harvard for a Master’s degree in public policy. I did some quantitative work at the Kennedy School so I got a lot of experience dealing with statistics and how that interacts with policy and politics. I actually graduated in the late spring of 2011, which was just as the Obama reelection campaign was getting ready to start hiring folks. I was one of the first people to join what became the analytics department—known in the media as the cave. I was ultimately promoted to deputy director of battleground state analytics. We did a lot of the work of determining where the campaign should allocate its resources using forecast models.

C&E: Give me a sense of the sheer amount of data you had available. Larry Grisolano bragged that the campaign had detailed demographic data on every single voter you needed to persuade.

Dover: That’s true, but I think one of the big misconceptions is that Democrats had access to all this data that no one else had. People within politics and outside of politics still have this misconception. A lot of folks don’t realize that all the data we used was either publicly available data from the voter file and the census, or data that we collected directly from voters. What was new about what we did was that we carefully constructed all of these statistical models that could project the likely voting behavior of each individual person. From there we created the lists of people that we either wanted to turn out, persuade or register. But the basis for all of this was the publicly available data.

C&E: What the Obama campaign was able to do with data and analytics—does it only work at scale? How can Senate campaigns and smaller races use this?

Dover: That’s why we’re here. If you really think about the presidential campaign, it’s a series of statewide votes, as I hope my seventh grade students learned from me. So all of the work we did, whether it was our persuasion models, our support models, or our turnout models, was done at the state level. We didn’t build a big Obama support model; we built an Ohio model for Obama. We also built a North Carolina model and a Florida model. And so what we were really doing was developing tools to help us win a series of nine or twelve statewide elections in the battleground states. In that regard, it’s no different than any other statewide election might be. We did create this list of individual voters that we wanted to talk to, but that can obviously be done in other races. If you’re running a local school board race you’re going to be talking to individual voters. So if anything, this ability to microtarget is almost more important for smaller races. So it absolutely can be scaled down and that’s something we’re really focused on here.

C&E: What does that look like for down-ballot campaigns?

Dover: Anyone can access the voter file. So we really have access to the same data, more or less. OFA still has its email lists, which are proprietary. But we also have the same people who did this work on the Obama campaign, and we’re taking most of the same data and scaling down our products. In some cases, we’re essentially building the same products and improving upon those same products to serve all sorts of clients—national entities, statewide campaigns, district-level campaigns and even some local races. The mission of our political office here in D.C. is to help Democrats and progressive causes to succeed by leveraging big data and analytics to run smarter and more efficient campaigns.

C&E: How much of a learning curve is there for those running smaller races when it comes to working with data and analytics?

Dover: The types of things that we’re doing are new and they are the types of things that people are not used to. In some cases, and very understandably so, we are doing things that people don’t entirely grasp. They might be able to appreciate what we did on the Obama campaign, but they don’t really understand it. The good thing is that people do seem to have an appreciation, at all levels, for what we did. We’re not here to just sell magical products that are going to magically win races. What we’re here to do is help candidates run smarter campaigns. A lot of that involves us providing them with some products. But a more important part of that is to actually walk them through the process of creating these things. We’re showing campaigns how to implement best practices using sophisticated tools.