Early voting sites across North Carolina will open their doors to voters in just one week, which means political campaigns are in crunch mode to turn everyday consumer habits into vital data points that help pinpoint persuadable voters. 

While voters are currently being bombarded with text messages, phone calls, TV ads, door-to-door canvassing, and billboards, reaching them involves a long process of strategic planning and data analysis before the message makes its way to our eyes and ears.

On both sides of the aisle, political operatives rely on data vendors to analyze consumer habits and target voters with personalized messaging. An often-overlooked industry, big data vendors consist of companies selling consumer data to political campaigns and non-political advertisers, alike. These third-party vendors gather consumer information to categorize individuals into political profiles. 

In 2020, 5.5 million votes were cast out of roughly 11 million people in North Carolina. That means one out of every two people that a political candidate comes in contact with in North Carolina, one is a likely voter. While many already have their mind made up ahead of any given election, unaffiliated voters are the largest share of voters in North Carolina, and many in this group can be influenced. Big data vendors play a crucial role in supplying consumer information to help campaigns hone in on these key individuals.

“The parties are disproportionately interested in identifying voters who are persuadable,” explained Andy Taylor, a political science professor at NC State. “There’s a sophisticated knowledge here, not necessarily of the individual voter, but of the type of voter this person may be.”

Turning Decisions into Data

North Carolina is a data-rich environment where information on voters is publicly accessible without special software.

Publicly available information categorizes types of voters using details like party registration, voter history, race, age, gender, and county of residence. Demographics are key factors that play into who will receive certain campaign materials ahead of an election. Political tendencies can also be determined by factors like household dynamics and marital status, occupation and income, and religious views.

On a more sophisticated level, consumers leave a traceable path every day based on personal preferences. When an individual goes on the Internet, they leave a trail as they shop and look at websites. From online shopping habits to the type of car a consumer drives, to the auto insurance policy a driver holds, many life choices are data indicators of voter preferences. The information is not necessarily identified to the individual voter but rather gives an idea of the type of voter the person may be. 

“The campaigns and the parties know quite a lot. If they don’t know exactly a lot about me personally, they certainly know an awful lot about me and people like me – in significant detail as well,” Taylor explained. “Not just for sort of blunt instruments like which state or which county you live in, or your race, or your gender, or how much money you make. But even things like, if you drive a particular kind of car or enjoy a particular kind of music, or attend particular kinds of sporting or artistic events. This tells observers quite a lot about which party you tend to vote for.”

Using various metrics like interests and spending trends, data experts can tag people into voter profiles and ultimately target certain audiences with political messages. Groups can acquire phone numbers from purchased data with which to send text messages or robocalls. The same applies to acquired emails, though compliance rules exist for the forms of communication that limit how the accessed data can be used legally. 

But how does the message connect with a person’s specific interests and emotions? 

“It’s so easy and cheap to have high production values in advertising,” Taylor said. “You have sophisticated information about the audiences for these outlets that you can tailor them. And so you can even change sentences in an ad and effectively create two different ads based upon your audience.”

While a candidate will have an overarching narrative and approach to their campaign, targeted communications can be customized to appeal to different groups. For instance, sentences in an ad can be changed to create two different ads based on the directed audience. A pro-Second Amendment ad may only be shown to audiences that clearly favor gun rights, while anti-gun individuals will never see the candidate’s messaging around firearms. 

For the Kamala Harris presidential campaign, the overarching message has been more hopeful and positive about where things are going in the country, whereas the Trump campaign’s messaging highlights the negative direction the country is headed. The broad message controls the direction of personalized messages delivered to consumers, Taylor explained, because the general strategy is for the communications to remain consistent with the overall theme. 

From Scrolling to Voting

Data brokers collect and compile information, then sell it to others for a range of purposes.

Commercial enterprises capture the data and pass it along to data brokers, who sell it to third-party vendors. The data is ultimately purchased by political parties, consultants, and campaigns for paid advertising and targeted messaging.

While both major political parties hold massive databases on voters, partisan firms acquire their own data for individual clients. 

On the Republican side, Data Trust is a vendor that works with nationwide conservative groups like the GOP and the National Republican Congressional Committee. Public voter files, commercial consumer data, client audiences, proprietary data sources, and other data feedback are mechanisms to win elections, which the company says “starts with having the best in class, most accurate data possible.”

On the Democrat side, Catalist works for data-driven progressive organizations, such as the Democratic Congressional Campaign Committee, League of Conservation Voters, Planned Parenthood, and AFL-CIO.  The company says the best predictor of future behavior is past behavior, which involves using models, synthetic variables, and longitudinal research on behaviors and political attitudes. 

The information isn’t used to harm anyone, and personal details like Social Security numbers are never accessible. However, it does raise concerns about the extent of privacy issues. David Capen, President of Capen Consulting in North Carolina, noted that while data can provide valuable insights into individuals, he acknowledged the potential for privacy concerns.

“We have this data that is used by pretty much every consultant in any given election, and so to not use it is to a great disadvantage,” said Capen. “But it does beg the question of how much access should people be able to have to this data and what are the rules around it? And how can you maybe opt-out if you don’t want your phone number to be included on this list?”