Skip to main content

«  View All Posts

Tunnl | Products | Audiences

How Does Tunnl Build Audiences?

September 16th, 2022 | 6 min. read

Brent Seaborn

Brent is the Co-founder and Chief Data Science Officer at Tunnl.

Print/Save as PDF

When organizations are looking to improve their outreach, they turn to audiences to increase both their effectiveness and efficiency. Whether it’s for an advocacy campaign, political campaign, brand efforts, or consumer marketing, audiences allow organizations to get more petitions signed, more consumers on their side, more buyers in the funnel, and more voters to pick their candidate.

To help advertisers make the most of their outreach efforts, Tunnl builds audience solutions, including custom audience subscriptions, the U.S. Policy Opinion Makers audience, and hundreds of prebuilt audiences.

But you may be wondering, How does Tunnl build audiences?

The Tunnl audience-building process usually takes one of two paths: survey-based and lookalike-based. By the end of this piece, you will have a better understanding of how Tunnl builds audiences and be able to answer the following questions:

  • What is survey-based audience building?
  • What is lookalike audience building?

Let’s get started.

Tunnl's 3 Steps to Building Survey-Based Audiences

Survey-based audience building gives Tunnl the opportunity to identify and define niche audiences.

The majority of Tunnl audiences are built using a survey-based approach that is commonly referred to as the microtargeting process. If you’re familiar with survey research and polling, it’s similar. Tunnl is in the field surveying at least 5,000 people through a combination of cell phone, landline, and online surveys.

We can survey the entire country, individual states, or another custom geography based on our clients’ location and needs.


Step #1: Survey Americans on How They Feel About Pressing Issues

In our large-sample microtargeting surveys, Tunnl asks various questions to determine how Americans feel about today’s most pressing issues and causes. We might ask questions as simple as whether or not someone believes the country is currently headed in the right direction, or as complicated as asking someone to align their support or opposition to the expansion of nuclear energy production. 

These questions allow us to see fault lines in policy issues and complicated positions, and are the genesis of the audience-building process.

Step #2: Build a Model

After we collect responses on a survey, we pull out observations and use them as inputs to  model around. The modeling process is a combination of art and science, wherein our analysts hand-build the models and review for quality and accuracy output.

Simply put, we use thousands of data points to inform and predict how and why a set of people behave the way that they do. 

For instance, if 30% of the country is extremely concerned about inflation, we might see more obvious predictors in our model, like income, partisanship, and geography. We could also see other, less obvious predictors, like Twitter use, dissatisfaction with Medicare plans, and being highly likely to get the COVID-19 vaccine.

Step #3: Scale the Model to Build an Audience

Once we’re satisfied with the model, we scale the model up to reality. 

The model is the smaller version of the prediction. We use a process called “scoring” to assign everyone in the geography we’re working in a score between 0 and 1. The higher your score is on the model, and the closer your score is to the number 1, the more likely you are to be aligned with whatever the model is predicting. 

Every American adult gets a score, which means we can rank-order everyone’s preferences from most likely to least likely to behave a certain way. 

Tunnl then uses various combinations of the precise scores to define our target audiences. It’s at this point that our audiences are ready for activation by our clients. 

If you want to see what an audience looks like on our platform, request a demo with our expert team.

Tunnl's 3 Steps to Building Lookalike-Based Audiences

While our survey-based audience building process has been highly refined, there are other ways to build audiences that produce equally accurate results. They simply use a different context to get us there.

Many of our clients come to us after they have collected their own set of data. These might be current customers who recently purchased a product or service, or potential industry advocates who responded to a call to action on behalf of an industry issue. 

Once you have a group of people who have taken an action you care about, you’re going to want to find more people who are similar to them that you can also appeal to. 

Tunnl uses our lookalike audience-building process to do exactly that: find you more people who are similar or identical to the ones you were already successful with. 

Step #1: Match Your Data Into Our Database

Tunnl starts by ingesting your data and matching it into our database. Linking external data from your organization to our internal data is done using a proprietary methodology that finds both exact matches and nearly exact matches. 

The more complete your dataset, the easier it is to connect it with other datasets. 

Once Tunnl has the matches in our database, we begin our analysis of the dataset.

Step #2: Analyze the Data

Although you sent one cohesive dataset over to us, you may have more than one consumer or advocate profile within the data. Simply put, there might be a few groups within your data that appear similar.

For instance, if you send us a list of people who went to Starbucks, that group can probably be further broken down into classic-coffee drinkers and specialized-coffee drinkers. We want to pull those groups apart before we find new people for you to reach out to, because that will enhance how you target them — amplifying your ability to appeal to what matters to them.

When we know how many types of people we are looking for, Tunnl will begin the lookalike modeling process.

Step #3: Build a Model & Create Audiences

We use thousands of variables to predict the model of your lookalikes. Unlike when we build survey-based audiences, the prediction isn’t based on survey responses; rather, it’s based on an action that brought this person into your dataset. 

Again, we use the scoring process to assign everyone in the geography we’re working in a score between 0 and 1. The higher your score is on the model, and the closer your score is to the number 1, the more similar you are to the original person. 

Depending on the original dataset, the modeling process, and the client’s end goals, Tunnl will deliver a set of new lookalike audiences for the client to connect and communicate with. 

These lookalikes can be refined annually, or more frequently if the client has a more robust marketing plan.

Become an Expert on Audiences

Now that you have a better understanding of how Tunnl builds audiences, you may want to delve deeper into our audience solutions.

How do you use Tunnl audiences?

Advertisers use Tunnl audiences to inform their outreach strategies and ensure their target markets see and engage with their messaging. Learn more about how brands and agencies use Tunnl audiences to guide their outreach strategy in the article linked below.

What is an audience refresh?

Once audiences are built, they need to be routinely refreshed to keep the data up-to-date and useful for advertisers. Learn more about how Tunnl refreshes our audiences and why it’s so important below.

Is your ideal audience available in Tunnl's prebuilt audience library?

Tunnl has identified and built hundreds of prebuilt audiences that are available inside the Tunnl platform to use with your campaigns right now