What all we did for the 2020 Elections

PredictWise is a digital-first audience technology company for the progressive ecosystem. At the heart of our operation is a massive database containing every adult American, and allowing us to match attitudinal with behavioral data for all records, keyed to digital identifiers and PII (ToolBox).

In this post, we’ll detail some of the many ways we helped Democrats and progressive causes up and down the ballot this last election cycle. Overall, our efforts reached over 20 Million people directly on their cellphones and provided the DNC and key swing states with fast alternatives empowering them to avoid the fall-out from the Facebook political ad shutdown. Our unique, verified data on 50 Million high-precision unregistered powered countless conversations; in just one example it was used to motivate over 90,000 unregistered women to use same-day registration in key rust belt swing states.

Below, we’ll describe our efforts in more detail:

(1) Our platform-independent data and technology, provided to the DNC and Fair Fight

(2) Delivering over 100 custom audiences to power digital and text-based advertising, with industry-leading mobile-identifiers-based advertising and machine learning-based support scores. These include bespoke audience generation powered by personalized surveys enriched with behavioral data, in close collaboration with partners.

(3) Directly running mobile-first advertising campaigns, with a variety of partners, delivering over 42M advertising impressions and spending over $500k dollars.

Of course, at PredictWise we’re not just satisfied with measuring ad spend or data delivered — we care about whether that spending was impactful in the election outcome. We’ll end this post by describing several ways we measured advertising effectiveness, including a large-scale experiment we ran in Pennsylvania.

Platform-independent data and technology

As we’ll detail below, our bread and butter is analytics on top of our massive databases with the goal of delivering digital advertisements to the best audience possible. However, organizations closer to the ground often just need trustworthy raw data to run their own operations, and we’re proud partners of leading organizations in the progressive ecosystem. Two such partnerships this past cycle were with the DNC in November, and with Fair Fight for the Georgia runoffs.

We first use our massive database to identify exactly the right individuals to target, and then we provide the best right way to reach them online: A core asset we’ve built over the past few years is a digital-first voter file that keys each individual on the voter file to their Mobile Ad ID (MAID), instead of their personal information. These MAIDs can be used to target individuals on an advertising exchange, even outside the Facebook and Google walled gardens, distributing ads inside mobile phone applications for example, without the need for unreliable data clearinghouses as middlemen. In other words, we brought the heavily analog voter file online. If you followed the advertising news ahead of this year’s election, you know that Facebook banned the upload of new political advertisements the week before November 03, hurting Democrats in particular. In advance of this ban, we provided the DNC with MAIDs tied to their own support model scores for over 79M individuals across every state. This data provided a fast, direct alternative to launching paid content campaigns, and helped avoid potentially disastrous consequences.

We also worked with Fair Fight in the run-up to the Georgia Senate run-off elections. As can be expected from a Stacy Abrams-run organization, Fair Fight had a sophisticated and effective digital operation. In particular, by the time of our partnership in December 2020, they had already amassed cell phone numbers of Georgians from many other vendors. However, we were able to provide over 345k new, previously unknown cell phone numbers to power their get-out-the-vote operations, using our partnerships with several commercial and political data vendors. As we’ll detail below, we curated and iterated on providing high-quality cell phone numbers during the November elections, during which this data yielded delivery rates over 80%.

Delivering custom audiences to power digital and text-based advertising — Signal by PredictWise

At PredictWise, our bread and butter is providing custom, digital-first audiences — powered by state-of-the-art Bayesian machine learning algorithms leveraging unprecedented amounts of data, including years of surveys and individual-level GPS and phone telemetry data (i.e., anything that is passively tracked on cell phones). We first use our massive database to identify exactly the right individuals to target, and then we provide the best right way to reach them online.

The science: personalized surveys and telemetry data

Cell-phone location data to analyze movement: We used individual-level GPS data to identify individuals who stayed at home during the Covid-19 pandemic. In particular, our hypothesis was that Republicans who were staying at home more often were open to messaging about the risk of Covid-19 and the mismanagement of the crisis by the Trump administration. Leveraging 3 months of (read: terabytes of) real-time mobile GPS data for tens of millions of Americans matched to the PredictWise ToolBox, we calculated a score for each individual measuring how much they reduced (or didn’t reduce) their movement during COVID compared to other people. We then asked Republicans who scored high and low on our COVID-concern-score about their self-reported concerns. The difference in self-reported concern for COVID between Republicans scoring low and high on our behavior-based COVID concern scale was almost 30 percentage points, a clear and first-of-its-kind demonstration of the power of such behavioral data.

Figure 2: Self-reported concerns about COVID among Republicans scoring low and high on the PredictWise behavioral COVID concern score

We used this data to target almost 380k COVID-concerned Republicans in Arizona, Ohio, and South Carolina, with Covid-based advertising.

Identifying unregistered people and re-registration targets: Using the same GPS data, we further identified individuals who were no longer living in the state of their last voter registration, focusing on young people, women, and people of color. We delivered an audience of over 200k such potential voters in Pennsylvania and Michigan, targeting them with voter registration messages for their new state.

PredictWise also built the most precise database of unregistered individuals in the country, comprising over 50M unique, verified unregistered people. Our database was specially designed to counter the two most common problems plaguing such databases: that they’re filled with registered people who are simply mismatched in the voter file, and that they skew Republican because commercial data most frequently used — credit card files — tends to include wealthier, older people. We did so by leveraging our access to data such as website portal registrations and mobile phone identifiers. We then used this database to power, with our partners, voter registration and same-day registration efforts through both digital ads and text-banking, as we detail below.

Mobile Application telemetry data: We further leveraged our data on mobile phone application installs to profile individuals on their interests. One use case of our profiling was delivering an audience of 400k individuals, with details on how best to connect the audience based on insights gleaned from our database of phone application installations.

Survey-based machine learning support scores: Finally, we scored every individual on our voter file (over 260 Million Americans) on over 30 different characteristics, including which party and candidates they support as well as issue-based scores such as their support for environmental regulations, health care rights, and criminal justice reforms. This process involved both state-of-the-art Bayesian machine learning models (built on Stan and run on cloud-based graphical processors) and data, including hundreds of thousands of survey responses for multiple questions per issue over the last four years, as well as individual characteristics (such as demographics, educational attainment, voting history) and characteristics of where they live (such as historical election outcomes, demographics, density).

Of course, these scores are only useful if they’re accurate. The figure above shows the relationship between each of our scores to each other and an individual’s party — our scores both have high face validity and provide information on top of just an individual’s predicted party membership. We’re not just satisfied with face validity, however. As we detail at the end of the post, we extensively tested each part of our pipeline, and targeted advertisements based on our scores were found to outperform other audiences.

Finally, we updated our scores on a monthly basis, based on ever-rolling survey responses. These updates ensured that our custom audiences always reflect the latest public opinion. In the plot below, for example, we show how opinions about the state of the economy tanked as Covid-19 spread in the United States in March and April.

Is now a good time to search for a new job in your local community?

The delivery and impact: texting campaigns and mobile-first audiences through digital identifiers

MAID-based audiences

Our primary deliverable this cycle was custom audiences for any type of campaign, with targets delivered through MAIDs. Campaigns used our online tool, Signal (shown below), to delineate an audience by geography, demographics, issue views, and vote history.

In this manner, we served over 45 audiences to various groups, consisting of 7,329,946 million individuals and 27,781,122 MAIDs. These audiences ranged from unregistered minority voters to frequent voters who are also Union supporters, and everything in between.

Powering text-banks

PredictWise provided high-precision cell phones to organizations like OpenProgress, for our tailor-made custom audiences of voters who hold progressive views on the issues. Working closely with our partners, we identified high-quality cell phone data and ultimately provided lists with over 80% text delivery rates.

In particular, starting the weekend before the election we delivered a cell phone audience of almost 2M registered, progressive-leaning individuals across almost every swing state: GA, PA, IA, MI, MS, NC, OH, SC, MN, MT, TX, and FL.

In states with same-day registration, we further helped text over 140k unregistered, progressive-leaning individuals with directions to their nearest polling place, in: NC, MI, MN, IA, ME, MT. While we are still waiting on updated voter files from most of these states, we know that 6,361 individuals out of the 40k that we targeted in North Carolina eventually ended up registering (and presumably voting) on election day.

Running mobile-first advertising campaigns

We didn’t just create audiences — we also directly delivered advertisements, serving over 42M impressions to 16.8M unique individuals and spending over $500k on just ads. Especially key for this cycle, PredictWise specializes in serving mobile ads directly inside applications, in concert with our application partners — bypassing Facebook and Google walled gardens. We’re especially proud of helping run the following three campaigns.

Predictwise partnered up with the Future Majority to run a same-day voter registration campaign targeting 400k unregistered women with progressive views in three key states: Minnesota, Wisconsin, and Michigan. We first closely worked with the Future Majority team to ensure we targeted the right people. Once the targets were confirmed, PredictWise built a mobile-first static creative that went hand in hand with their Jennifer Hudson video. In the end, the campaign served over 5M impressions. Out of these 400k unregistered women we targeted, over 92k ultimately registered: 19k in Wisconsin, 52k in Michigan, and 21k in Minnesota.

We further worked hand in hand with the Ohio Democratic Party to build our specialized custom audiences for women and BIPOC communities, for turnout, registration, and persuasion. We successfully helped Ohio reach larger audiences through programmatic channels, increased video views, and hyper-targeted messages, serving over 20M impressions.

Finally, the Florida Democratic Party contacted Predictwise to support their ad program when last-minute issues with Facebook blocked them from uploading new campaigns and creatives. We worked closely with their team to build a media plan to target key voters in Florida battleground congressional districts. Overall we served over 15 million impressions in less than 4 days.

Validating data and measuring impact

We believe that measuring impact and validating effectiveness is a key part of the process. We’ve made a methods primer available online, and here will briefly overview some further experiments we ran this cycle. These efforts together demonstrate the effectiveness of every part of our technical pipeline, from data curation to targeting to delivery.

We take a multi-tiered approach to validate our methods and measure impact. In off-cycle years, we build and validate our core methodologies. This cycle, meant building out the MAID-first voter file and survey infrastructure and then testing it in midterm elections, including for the Katie Porter campaign as described in our methods primer. Then, during the cycle, we always closely monitor ad performance and iterate on spending across ad campaigns. Finally, for a select few of our ad campaigns, we run true randomized control trials.

In our primary randomized control trial, we partnered with TruthNotLies to measure both their campaign ad effectiveness and our targeting and delivery performance. Full details forthcoming, but for now: ads targeted to a treatment group of 400k individuals of 2016 Trump voters in Pennsylvania significantly decreased Trump support over the control group, as measured by surveys sent to each group after the election.

In a second experiment, we tested whether our support model scores truly capture opinion above and beyond partisanship, as well as our ability to deliver targeted ads through MAIDs and mobile applications. We created a specialized audience of people who score high on our Environmental support score, as well as otherwise equally progressive individuals who score low on the Environmental score. We then randomized each audience to be shown either a generic ad or an ad for an environmental protection group and found that the Environmental group indeed performed best relatively (over both other ads and other audiences) when shown the pro-environment ad.

What’s next

PredictWise has always believed that maximum impact is not about frantically spending resources near election time, but is rather achieved through continuous dialogue with voters through targeted, policy-driven messaging, aligned with each individual’s values and narrative frames, and through both organic and paid content. We believe that a single backend, streamlining and processing streams of both survey and alternative/telemetry data, can achieve such a purpose. Our goal has always been to feed into such a structure or build this structure ourselves. The end goal for PredictWise is sharing better, more accurate, deeper data in service of building an always-on messaging campaign, leading to more meaningful interactions with everybody from habitual voters to unlisted. We believe that every progressive organization, from national campaigns to the school board, should be able to access such a repository without any hurdle.

As PredictWise is repositioning its offerings, we are exploring ways to make this proposition a reality. More to come!