We’re Bringing Advanced Data Infrastructure to Thousands of Midterm Campaigns

Nellwyn Thomas
DNC Tech Team
Published in
6 min readMar 23, 2022

This fall, thousands of Democrats will run for office in elections up and down the ballot. Most campaigns will have less than nine months to build an organization, recruit volunteers, draft messaging, buy advertising, and get information in the hands of voters. Many will operate with only a handful of staff! These scrappy campaigns will need every ounce of help they can to reach voters and win votes — and that includes having the data and tools they need to power their work. On the DNC Tech team, we’re laser focused on making sure we’re turning the progress we’ve made over the last four years into tangible improvements for campaigns of every size in 2022 — and for years to come.

The basics: the data and tech campaigns need to win

At the DNC, our goal is to help Democrats get elected, at every level of office and in every state. On the DNC Tech team, we focus on providing key components of the technology infrastructure campaigns need to run a successful campaign.

Successful campaigns must use limited resources (time, money, volunteer efforts) to efficiently net enough votes for their candidate to win. This means focusing on the people most likely to be moved by the campaigns’ programs, spending money in the right places, and (just as importantly) not spending money on programs that don’t yield results. Making smart decisions around resource allocation relies on having reliable, fast, and high quality data and data infrastructure to inform strategies around voter contact.

Some campaigns are well positioned to do this, thanks to their size and funding. Case in point: presidential campaigns are the best-funded campaigns in American politics. This means they have the most resources to spend on building up the infrastructure needed to make efficient calculations and power voter outreach programs. This also means that presidential campaign technology and data teams tend to be bigger and better funded than most any other campaign — allowing those teams to build and innovate, and rely less on external vendors.

However, the short-lived and ephemeral nature of campaigns means that the progress made by large campaign teams can often fall by the wayside, and other campaigns cannot benefit from the progress.¹

Since 2016, the DNC has been focused on rebuilding the core data and voter protection infrastructure used by Democratic campaigns, both in the lead-up to the 2020 presidential campaign, and over the year since, to lay the strong foundation required to maintain and iterate on new tools piloted by the Biden campaign. The work the DNC has done over the past year, and into this year, means we will bring the sophistication of a presidential cycle to thousands of down-ballot campaigns this midterm cycle.

Visualizing the cyclical investments in campaigns — and how DNC Tech can help maintain and iterate on technologies between cyclical peaks

Ten ways we’re bringing better technologies to down ballot campaigns

We’ve overhauled our data infrastructure and our approach to data acquisitions:

  • Campaigns who use advanced targeting or modeling can now work directly in the party’s cloud-based data warehouse built on top of Google BigQuery, called Phoenix, and leverage robust data workflow automation software designed for technical data staffers, for free. The multi-million dollar investments mean that individual campaigns or committees do not have to spend money or time to build their own data warehouses and analytics tooling. In addition to the Biden-Harris campaign, more than 50 campaigns leveraged this advanced analytics warehouse in 2020.
  • At the DNC, we like to think of ourselves as “data omnivores.” We test datasets widely, and buy the highest quality data we can, at the best price. Case in point: we’ve more than quadrupled the number of vendors we’re buying data from over the last four years.

We’ve made data about potential voters more accurate:

  • More voters are now reachable on their cell phones, by call or text or digital ad matching — DNC investments in cell phone purchases, QA, matching, and ingestion have increased the number of voters for whom we have a cell phone number from 30% in 2016 to over 70% in 2022.
  • More not-yet-registered voters are also reachable! Compared to years past, campaigns have access to more high quality lists of unregistered voters, in addition to registered voters, who they can focus on getting to register and mobilizing to vote. In 2022, that includes more than 10M potential new voters!
  • Campaigns benefit from continued and ongoing investments ($3 million this year alone!) in models predicting voters’ ethnicity, education level, religion, likelihood to support a Democratic candidate, likelihood to turnout in midterms, likelihood to be persuaded by political messaging, among other traits and behaviors. Over the past years, we’ve worked hard to increase the accuracy and coverage of these models. In 2020, our rebuilt party support score had a 10% improvement in ranking accuracy compared to the prior version. A 10% improvement in accuracy means more accurate targeting, fewer wasted resources, and more time talking to the right voters.
  • Compared to years past, our data is updated more frequently so campaigns have the freshest data possible. We’ve increased the frequency of voter file refreshes by three times — now reimbursing our state party partners for 12 refreshes a year, up from four in prior cycles — meaning that campaigns have fresher data about who has recently registered (or been purged from the voter rolls)!
  • Along with more frequent voter file updates, we’ve increased the frequency of adding data about new folks who appear on the voter rolls, and decreased our reliance on slow-to-update vendor systems. Four years ago, many campaigns would have waited up to three months to see predictive models about a newly registered voter — those scores now refresh in our system up to daily to incorporate data as it is collected by volunteers talking to voters.
  • In 2019 we launched a record-linkage algorithm that redefined our ability to connect individual state files into one whole national system. This change allowed us to identify voters who moved between states, and carry through their data from one state to another. In 2021, 42 million voters moved between states — thanks to our investment in this algorithm, we are now able to identify those movers’ prior party membership. A few years ago, many of those movers would have been treated as completely new registrants with no track record of voting behaviors.

We’ve invested in the community using and building these tools:

  • Campaigns can now work with a dedicated team of community analysts at the DNC Tech team, whose job it is to support state party, sister committee, and campaign staff in all 50 states.
  • Our community of users can now rely on a DNC Tech team that is larger and more stable than ever. Historically, this team has expanded and contracted along with the cyclical funding of the DNC. In 2021, our team size remained stable following the presidential cycle, and we retained two-thirds of our staff from 2020 — meaning that we also retained critical domain expertise and momentum.

Why make these data & tech investments from within the DNC?

Building a strong foundation of technology and data for Democrats is a huge task, and thankfully, there are many fantastic and innovative organizations in the political tech and data ecosystem. We are proud and grateful to partner with many of these groups.

To ensure that we are making the most of what we, uniquely, can do, we try to stay focused on where we believe we can add specific value as the national party. The DNC has two superpowers: 1) unlike a campaign, we are an enduring organizational entity, and 2) we have unique campaign finance powers to be able to provide tools and data, for free, to other organizations in the ecosystem.

This year — and every year — the DNC Tech team is making the most of these superpowers to building enduring infrastructure and offer it at no cost to our users. We are laser focused on the investments that will save other Democratic committees (like the DSCC, DCCC, DLCC, DAGA, and DGA), state parties, and campaigns time and money, meaning they can spend less time building infrastructure and evaluating new data to buy — and more time talking directly to voters.

[1] Shout out to all those campaign staffers who do their best to share their products, learnings and tools for the benefit of future campaigns and the wider political tech ecosystem. Recent examples include a detailed write-up from the Jon Ossoff for Senate runoff campaign and the open-sourced tools of the Warren for President campaign.

--

--