Can the same data used by brands to discover new customers and drive growth also help political candidates discover new supporters and drive voter turnout? In our series Dstillery Predicts, we intend to find out. 

In this second installment of our Dstillery Predicts series, we take a look at how a voter first discovers a candidate. ICYMI, check out Part 1 in our podcast Dstillery Predicts The 2018 Election.

It’s the first canvassing Saturday of what will be a long campaign season. All of the work up to this point has been focused on building campaign infrastructure, calling potential donors and predicting the likelihood of constituency support. But today, the candidate will finally deploy a number of loyal staffers and volunteers into their communities to increase candidate awareness and begin to spread their position on important issues. Knocking on doors and phone banking will hopefully increase web traffic and add more supporters names to their database. This process is then repeated, over and over, with increased frequency each week so that maybe, just maybe, the candidate can garner enough votes to secure the primaries and then again in the general election. But do these traditional tactics truly work? According to Berkeley and Stanford researchers Joshua Kalla and David Broockman – they don’t.


Candidates As Brands

Campaigns rely heavily on voter files as their core data source. Person by person, they build knowledge and awareness within their district and reach those people through the touchpoints outlined above. They also lean on reported behavior, like surveys and polls, to gauge political interest and predict turnout. But as we often learn working with brands: it’s not just enough to understand how a person is interacting with a campaign or company website. We’ve discovered that the amount of time a potential voter actually spends interacting with their candidate online accounts for less than 1% of their overall online behavior. It’s time for candidates to take advantage of the 99%.For our first analysis, we asked the question “what happens when a supporter first discovers their candidate and what motivations drive their digital candidate interactions?” To answer this, we mapped a typical voter journey (utilizing data up to February 27th) for the Democratic candidate Conor Lamb in the Pennsylvania 18th Congressional District. This journey highlights over 650 digital observed behaviors including the six candidate touchpoints and their catalysts shown below.

Campaign Website Visit     Visit Motivator
20-Jan      Phone Bank / Canvass
21-Jan      Phone Bank / Canvass
26-Jan      10k yard signs
31-Jan      Door to Door in Upper St Claire
14-Feb      FL Shooting
22-Feb      Donation Push/News on opponent spending


Sample Voter Journey for a Constituent from PA-18

In this journey, the supporter first visits the candidate’s website on Saturday, January 20th, perhaps as a result of a canvassing event in their area, a news report or a conversation with a neighbor. But we see that the activity on Saturday is limited to the candidate website and the Washington Post. It’s not until Sunday when this person studies up on who their candidate is by  reading a variety of new sources. Another week passes before the next visit occurs on Friday as they look to see what candidate events are being held over the weekend.

After the initial flurry of activity, we see a significant drop in the frequency of interactions. Two weeks pass without a single visit! It isn’t until a national story brings the potential supporter back to the candidate’s page. As the school shooting in Florida is picked up by every major news outlet, this supporter checks out  their candidate’s page to see how Conor Lamb is responding to the tragedy. Another eight days will pass before the supporter is motivated enough, as part of a multi-day fundraising push, to revisit the website. In the entire month of February, Conor Lamb only received two potentially meaningful interactions.


Insights Into Action

By visualizing this one voter’s journey, we can understand what online touchpoints (in comparison to offline events such as canvassing) drive supporters to engage with specific candidates.

Here are a couple ways candidates can operationalize the data:  

  • If a supporter’s online behavior reflects an interest in political current events, then speak to their intellectual curiosity and send curated emails discussing the details of the campaign, using insights to craft more targeted messages that speak to different subpopulations of voters as opposed to “one size fits all” messaging.
  • If the data shows that a supporter has visited the campaign site on a Friday, then it’s the perfect time to send a reminder with a list of that weekend’s campaign events, encouraging their involvement.


Getting to 100%

Understand voters’ journeys and see what makes their district unique. Build on the knowledge gained with each new supporter to better inform your future decisions.

Campaign events and/or calls may drive a singular visitation to the website or brief research on the candidate but in total, those actions still account for 1% of a user’s digital behavior. All of this leads to the campaign’s need to understand a voter’s interests (the other 99% of their behavior) as a way to motivate them to get out and vote.

In today’s digitally driven world, it’s never been easier for a candidate to supplement their 1% by unlocking the potential of observed behavior. In our upcoming series, we’ll take a look at how the insights we build everyday for brands can be applied to the political realm. By marrying traditional campaign tactics with digital intelligence, predictive marketing has the power to build robust stories to help candidates understand who their voters are and how to drive them to turnout on Election Day.  

Peter Lenz is a Senior Geospatial Analyst and our go-to for all things location and geography. Peter Ibarra is a Senior Analyst on our Data Science & Analytics team with a degree in Political Science. Learn more about their findings: