My new adventure: The Target Labs
Depending on how you got here, you may or may not know that I graduated from the Berkeley-Haas MBA program a couple of weeks ago. So school's out forever(?), and it's back to real life.
I'm here to tell you what I'm doing next-- it's exciting! If you're wondering how you can help (bless you!), that information is down at the bottom of this post.
I'm founding a company, called The Target Labs, which is bringing advanced data analytics and predictive modeling techniques to state and local level Democratic and progressive campaigns.
In 2008, when I helped lead field operations for the Obama campaign in Wisconsin, we did something that had never been done before. Like a number of other states, Wisconsin doesn't track which political parties its voters belong to, complicating the task of figuring out which voters will support a given candidate. Statewide Democratic campaigns had traditionally used the following strategy to mobilize their voters: they opened up one field office in the heavily African American section of Milwaukee, which is the most highly segregated city in the country, and another in the student wards around the University of Wisconsin at Madison. Their GOTV strategy was basically getting all of the people in those two spots to the polls.
In 2008, we did things differently. We opened up more than 60 field offices all around the state, including in rural towns and heavily Republican suburbs. Many people in these towns had never seen a Democratic field office before, and they said things to us like, “Wow, this Obama guy is really different. He is so post-partisan; he really wants to reach out to Republicans and independents.” And while that may well have been true, that’s not why we opened up those offices. The reason we were reaching out to voters outside of the Democratic strongholds in Madison and Milwaukee is because we were using data analytics to build models that used all the information we had about voters, including demographics, geography, and past voting history, to make extremely accurate predictions about how likely they were to support us. These models made it possible to find “our voters” on every block, in every region of the state. In 2004, John Kerry won Wisconsin by 10,000 votes. In 2008, our margin of victory was 400,000. Although I’d worked on many campaigns before, this was my first experience using predictive modeling, and it convinced me completely of the potential of these techniques, adopted at scale, to change the way campaigns are run.
In my MBA program at Haas over the last two years, I learned how to build the type of predictive models we used in the Obama campaign myself using statistical software. Last fall, I applied these techniques to Eric Mar's campaign for Supervisor in San Francisco's Richmond District. Eric Mar and his opponent David Lee were both Chinese American, both men around the same age, and both Democrats. These demographic similarities made it very difficult to predict which voters were likely to support one candidate or another without using advanced statistical techniques. We used these techniques to build a model for Eric's campaign that was able to predict support much more accurately than they would have been able to do themselves, and the campaign used the model to direct their voter contact efforts. In the end, Eric Mar was outspent by a factor of four-to-one, but using our model, his campaign achieved a 13 point margin of victory.
Our successful experience with the Mar campaign convinced me of the potentially huge value of modeling in smaller races. It made me determined to bring advanced data analytics to more state and local-level Democratic and progressive campaigns, democratizing and scaling the techniques for which the 2012 Obama campaign got so much (well-deserved!) press.
So, with a wonderful friend of mine, Seth Walker, who also happens to be a brilliant computer engineer, I am co-founding a company, The Target Labs, to do just that.
What we're offering in 2013:
Consulting services. We'll use individual level polling data and voter file data provided to us by campaigns to build models that allow us to make highly accurate predictions about how likely each individual voter in a given district is to turnout to vote and/or to support the candidate or issue the campaign is working on. We'll then support campaigns to use these predictions (expressed as voter scores) to guide their voter targeting decisions. We're currently building our roster of customers for 2013 elections-- let us know if you or others you know might be interested in working with us!
What we're building for 2014:
A web platform that campaign managers and consultants can use themselves. There aren't enough human analysts to do the work of building models for all the campaigns that need them. Some campaigns that could benefit from modeling can't afford to pay for a consultant to do it for them. For this reason, we're building a platform that campaign managers and consultants with minimal to no background in statistics can use to build voter models themselves. Users will upload their data into the web-based platform, and answer some simple questions about their data. Then, our tool will run the data through an algorithm, which incorporates our experience building other models, plus machine learning, to build models and make predictions about support and turnout for the campaign. This will enable us to serve vastly more customers than we (and all the other analytics consultants out there combined) could do as consultants.
1. Introduce us to local and state level campaign consultants, Democratic caucus organization staffers, labor folks and other folks who run field IEs, and anyone else you think might be interested in working with us on state and local races. They can be anywhere in the country (or, hey, the world, what the heck!).
2. Connect us with smart, reliable students studying stats, econometrics, computer science, etc who might be interested in being (paid) interns for us in the fall.
3. Tell us what you think. We value this community as thought partners-- we'd love to hear your questions and ideas as we develop our business. Hit us up!
Thank you all so much in advance for supporting us as we grow this thing!
