The 2020 election cycle posed unprecedented challenges for campaigns, journalists, and political scientists examining voting trends across the United States. In response to a raging pandemic that upended all aspects of public life and threatened to torpedo voter participation, state and local officials unleashed a wave of new laws and regulations that left the voting environment unrecognizable. Expanded early voting and vote-by-mail programs in all fifty states and DC fueled a massive surge in votes cast before November 3 – over 100 million voters returned a ballot prior to Election Day in 2020, more than doubling the number of early votes cast just four years prior. And as it became clear that voter turnout would exceed all expectations, two key questions emerged: just how much would this surge of early voters alter the overall electorate, and what segments of the electorate stood to make the greatest overall impact?
This rapidly changing landscape left us at TargetSmart with a unique opportunity to position ourselves as the leading provider of richly detailed and timely early voting data to the progressive ecosphere, the media, and the general public. Armed with the nation’s most comprehensive and accurate voter file, a best-in-the-industry data products team rapidly collecting and cleaning early vote data around the clock, and investment in platforms that allow for lightning-fast processing of large national databases, the TargetSmart team built TargetEarly, the nation’s only publicly available real-time demographic analysis tool for early voting. A first in the political market, TargetEarly allowed election observers around the world an opportunity to study daily pattern shifts in voting behavior, as they were happening, at the click of a mouse.
How did we build TargetEarly?
We separated the development process into (1) data aggregation and (2) data visualization and web development.
Comparing demographic and partisan data across three election cycles posed significant complications, not the least of which was a need for extreme agility in our processes after an unprecedented wave of early voting ballooned the size of our build files. We were able to address this problem and rebuild our aggregation flows quickly and on the fly with minimal disruption because of agile and beautiful workflows that looked like this:
TargetSmart has leaned heavily into leveraging the variety of tools offered by Amazon Web Services (AWS), the cloud computing service that powers businesses across all industries worldwide. TargetEarly data is stored as a series of CSV files stored in AWS S3 buckets, which were picked up by our front-end to calculate summaries and draw visualizations using code written in HTML, CSS, and JavaScript. We leveraged D3.js, a top-of-the-line JavaScript data visualization library, to build the graphs. The site is served through AWS CloudFront, a low-latency Content Delivery Network (CDN). These technologies enabled TargetEarly to withstand an unexpectedly large wave of public interest, scaling to hundreds of thousands of users securely and without significant latency concerns.
The TargetEarly core team consisted of folks from TargetSmart’s Client Tech and Strategic Consulting & Analytics teams. The cross-team collaboration was instrumental to getting TargetEarly out and ensuring the success of the project. We genuinely enjoyed building a project that was so important to the movement.
We have to say, after all the work that went in to TargetEaly, the hardest part of the entire process was choosing the colors. Here’s just a few of the iterations.
How did TargetEarly impact the election?
TargetEarly’s greatest success became evident as media coverage of the 2020 election reached a fever pitch. In the days leading up to November 3, TargetEarly allowed journalists across the globe to analyze early voting patterns at levels of specificity and granularity never before seen in a presidential election. While a wide array of national and international news outlets used the site to generate overall toplines, a full host of local outlets were also using the site to paint a more nuanced picture of what early voting patterns meant for their coverage areas.
The very existence of TargetEarly had democratized the political data landscape – suddenly small newspapers and local broadcast news programs had up to date access to the same aggregate data that in prior cycles had only been available to network news and high-profile national dailies. And with expanding access to accurate data comes a new ability to publicly and transparently combat misinformation. In Georgia, where a Republican-led effort threatens to seriously restrict voting rights to millions of voters, activists are leaning on data from TargetEarly to illustrate the clear racial bias at the heart of these anti-democratic measures.
TargetEarly was a vital resource in the Democratic ecosystem in 2020/2021, and we plan to make TargetEarly even bigger and better in elections to come.