The audience research tool enables users to locate & understand the highest value users to your brand by utilizing the traffic on your website, or a specific geo, to create detailed insight reports, and leverage look-a-like modeling to find new targets.

1. Define an existing audience of the brand:
   • From existing website traffic
   • OR from a specific geo that the grand has identified as high value or high performing for their product or service
2. Create a detailed insight report from the zip codes and US Census data to better understand demographics and characteristics of the existing audience
3. Through look-a-like modeling, find the zip codes that most closely match the existing audience
4. Export these zip codes for targeting across all media and channels

Create an insight report from an existing audience to research their demographic characteristics, interests they are showing across the web and their streaming behaviors. The Audience Research tool makes all of this data easily accessible for the audience selected, and using this data creates a lookalike model to create an audience extension to the zip codes most closely matching the characteristics of the existing audience.

There are three types of audiences that can be used for the ‘source’ of an ART report:

1. Pixel Report: A Pixel report will pull audience insights from the visitors to the Advertiser’s website via an IP Pixel.
2. Geo Report: The Geo report will pull Audience insights from a list of U.S. zip codes. For example, if an advertiser has a zip code, DMA or city that is a top performer, you can select this geo and run a report to create a complete profile of users in that geo, then leverage a lookalike model to find similar zip codes for prospecting.
3. Contextual Report: The Contextual report will pull audience insights from a contextual AI audience or interest category. Create an insight report to better understand the demographics of users that are showing interest in a specific topic. To create a custom interest category, see here: Custom Interest Categories

To create a ‘Pixel’ Report, the first step is to create and place Pontiac IP pixels on the Advertiser’s website. Multiple pixels can be created to track each action or page on the website, or you can simply place one pixel on all pages to capture all site traffic. Reports can then be created from each individual pixel, or from several pixels combined, allowing you to segment insights in a way that best fits your needs.

1. In the Pontiac Platform select ‘DSP’ in the top right corner.
2. Create a new Advertiser or open an existing one.
3. Select ‘New IP Conversion Pixel’, give the pixel a name, and save.
4. Allow a few minutes for the pixel to generate, then click on the name of the pixel to view the code.
5. Place the pixel code on the Advertiser’s website. This can be placed on all pages, or on a specific page of interest. For example, if you place this pixel on the ‘Thank you’ page after a transaction is completed you will be able to create a profile of users that have made a purchase. If you place this pixel on all pages, you can create a profile of all users that visit your website.

Once the pixels have been placed you will need to allow time for the sample sizes to grow as visitors are seen on the page.

Read more on Pontiac IP Pixels here: Pixels

Creating an Audience Insight Report

1. Log in to the Pontiac Platform and in the top right corner, click on the button ‘ART’.
2. Select the blue ‘New Report’ button.
3. Give the Report a name.
4. Select the target audience size from 1,000 Zip Codes, 2,500 Zip Codes, 5,000 Zip Codes, or 10,000 Zip Codes. This is the number of matched zip codes that will be returned by the lookalike model.
5. Select an Auto Refresh schedule if needed. If an audience is published from the report, this will update the report and the audience on the selected schedule. New interest data is ingested every 30 days so refreshing the audience will incorporate any new data that has been added.
6. Select the ‘Audience Type’.
   • For ‘Pixel’ Audiences, associate the pixel or pixels to build the report from. This is an IP Pixel that must first be created in the DSP section of the platform, then placed on the Advertiser’s website. Once your IP pixel has been placed and the sample size has grown over time, you can create insight reports from this data.
   • For ‘Geo’ Audiences, add the list of U.S. zip codes to build the report from.
   • For Contextual Audience, select a Contextual audience or Interest category from the list. Custom interest categories will need to be created in the DSP platform, prior to making a report. See more here: Custom Interest Categories
7. For Pixel reports – Enter the dates that you want the report to pull from. This will pull the IP addresses of visitors to your website during this date range. The maximum lookback period is 6 months.
8. Select the categories or characteristics that you would like to research for your audience. Select the demographic categories from ‘Demographics’ or ‘Focused Demos’ sections. For topics of interest, select from ‘Interest’ categories to understand the types of content that your audience is showing interest in. There is a limit of 10 categories that can be selected per report.
9. Apply filters to the lookalike audience if you would only like to receive zip codes within a specified geo or above a certain population threshold. For example, if your campaign is set to only run in New York you can limit the lookalike audience to the top zip codes in New York that resemble your existing audience.
10. ‘Submit New Report’ to generate the Report.

Key Audience Insights

The highest positive deviance section of the ART report will display the key characteristics that are the most defining demographics or interests for your audience:

The lowest negative deviance section of the ART report will display the key characteristics that are the least defining demographics or interests for your audience:

Site Traffic Analytics

Referring URL

The Referring URL section of the report will provide additional website analytics, displaying the URL that the user was on when the pixel load occurred.

Geographic Data

Once your report has generated, you will see a heat map displaying the geographic areas where your site traffic comes from. Below the map, there is a list of states and the percentage of website visitors that come from each state. Expand each state to see the percentage of traffic at the zip code level, their populations and number of households.

Demographic Data

ART will create a report of the selected data categories utilizing U.S. Census data and the geographic data generated by your IP pixel. The report will be broken down into sections by ‘Concept’ and each concept will contain several graphs. For example, the Concept ‘Households’ will display graphs with the number of children present, the martial status, the household income and the size of household in the zip codes that your users come from.

Each graph will display the % of users in that zip code for each demographic category, as well as compare that percentage to the national average. A zip code is considered ‘exceptional’ in a specific category if the % of users in this category is higher than the national average. In this example below, the zip code selected over-indexes for household income of $200,000 or more as 11.03% of individuals fall into this bracket, while the national average is only 8.19%.

Use the ‘View Deviance’ button to view the deviance for this zip code from the national average. The higher the deviance, the more exceptional a zip code is for a specific demographic:

Political Data

Political Data is also available for the 2020 presidential election. This data comes from Harvard Dataverse.

Interest Data

Through AI-based contextual analysis, Pontiac identifies categories of content consumed by zip code across the U.S., then matches this data to the zip codes of your existing audience for insights into the topics your website visitors are reading about. Leverage interests in content to develop cookieless contextual targeting strategies that are most relevant to your target audience. For example, in the below report the Advertiser’s audience is showing above average interest in Career Advice:

Inventory Data

In the Inventory Data section of reports you can review the specific CTV networks, publishers, channels, series, episode titles & apps that your audience is currently watching. Our CTV Bidder provides the capabilities for targeting down to the channel level, so these are all actionable insights. Build detailed CTV targeting plans to serve ads across the inventory your audience watches the most.

Here is a sample model for CTV Network viewership for Mother’s in Louisiana with high household income that are showing interest in boating:

Lookalike Modeling through Zip Code Correlation

By building a look-a-like model from your existing web traffic, the geo selected or the interest category, the report will output a list of new zip code targets that are most closely matched to your existing audience across the categories and characteristics selected. Select ‘Publish Audience’ at the bottom of the report to create a cookieless targeting solution across all channels and media types – from Display and CTV to Digital Out of Home. Reports will also include the population and number of households in each zip code to provide insight into the potential audience size.

Create your target Audience profile from the desired demographic variables to discover new zip code targets with the highest propensity match to your desired Audience profile. Leveraging the correlation between zip codes and demographics, ADT uses US Census Data to find the zip codes with the largest deviation from the US mean across the selected variables. Leveraging impression data and deep contextual analysis, interest categories can also be evaluated for each zip code to find areas that are over-indexing in content consumption for the selected topics. Audience Discovery data can be exported for targeting across all channels and media types.

To create a report in the Audience Discovery Tool, navigate to the ART section of the platform, then select ‘Audience Discovery’.

• Select ‘New Report’.
• Give the report a name.
• Select the ‘Target Audience Size’. This is the number of zip codes that will be returned in the results.
• Select an Auto Refresh schedule if needed. If an audience is published from the report, this will update the report and the audience on the selected schedule. New interest data is ingested every 30 days so refreshing the audience will incorporate any new data that has been added.
• Optionally, Add Description of this audience. This will include a few AI generated images that represent the audience you’re targeting.
• Click ‘Edit Targeting’ button to open the menu of data categories. There is a limit of 10 datasets that can be selected per report.

In the ‘Census Categories’ tab, select the demographics for your desired audience:

For example, here we are looking for households with a married couple and children present in the house.

In the ‘Contextual’ tab, select the interests of your target audience:

For example, we added interest in the topic ‘Parenting teens’. Custom interest categories can also be created using the Pontiac Contextual AI. For more information, see here: Custom Interest Categories

• If your campaign is set to run in a specified Geo, you may want to filter results by that geo to only receive the top zip codes that will match your campaign parameters.

• Zip codes can also be selected and weighted in their own area. Note that you can also create a Block List depending on your needs.

• Under the ‘Population’ tab, you can filter results to zip codes that meet a specific threshold of population or # of households. Often times, zip codes with very small populations will index highly across the selected variables, but it may not be valuable to the Advertiser to consider zip codes below a certain population threshold. These filters will remove these zip codes from results.

• Traffic Filtering will be applied to Contextual categories. There are defaults in place here to ensure that there is enough data in each zip code and interest category to evaluate and determine the most exceptional zip codes:
  • Zip traffic filters for zip codes that have contextual data over the last month for more impressions than the number specified
  • Graph traffic filters for zip codes that have contextual records in aggregate for the targeted graphs which exceed the number specified. Graphs are the upper level categories such as Health & Fitness and Home & Garden.
  • Category traffic filters for zip codes that have contextual records which exceed the number specified for the targeted category. Targeted categories are the most granular level such as Gardening or Exercise.
• Under the ‘Indexing’ tab, check the box if you would like each zip code to over-index on ALL demographic and interest categories selected. If this box is unchecked, the report will use a composite score on all variables and return the most exceptional zip codes overall, but each zip code may not over-index on all categories.
• Once you have selected all the desired variables to build your target audience profile, ‘Close’ and ‘Submit Report’.

The report is broken apart into different sections: audience insights, audience interests, inventory and published audiences. Follow the Extreme Sports Fans Audience Discovery Report to understand each section of the report. Note that the Audience Discovery Report setup is as follows:

• Name: Extreme Sports Fans
• Target Audience Size: 1000 Zip Codes
• Audience Description: Fans of BMX
• Targeting > Interests > Sports: Mountain biking, Surfing/body-boarding, Bicycling, Snowboarding, Skateboarding, Power & motorcycles
• Filters > Population: Population greater than 1000, Households greater than 400
• Filters > Traffic Filtering: Zip Traffic greater than 1000, Graph Traffic greater than 20, Category Traffic greater than 2

Audience Insights

If you added a description when generating the audience discovery report, four AI generated images of your target audience will be displayed. These images help paint the picture of who you’re targeting. In our example, the audience description is “Fans of BMX” and the bellow figures were output.

A heat map displays the geographic areas across the United States with the highest variance from the U.S. average on the selected variables. Below the map, the zip codes will be ranked according to their propensity match to your desired demographics and interests, with #1 being the closest match.

The ‘Exceptionalism’ column will display the average percentage difference between the displayed zip code and the U.S. average across all selected variables. A higher difference will indicate a higher likelihood for the zip code population to meet the criterion selected. Use the download button to export the list of zip codes. This zip code list can then be used for targeting on Pontiac campaigns across all channels and media types.

In this table, the report will also display the population and number of households for each zip code to provide insight into the potential audience size and reach of your campaign targeting these zip codes.

Audience Interest

In the Audience Interest expandable section, you will see graphs of your selected targeting. Displayed is the distribution of the the population in the returned zip codes across each of the selected demographic and interest variables. Based on our example, we selected a variety of sports. Hover over the graph to see the name and corresponding percentage, deviation from normal, and US average. The category with the highest deviation from normal means that it is more likely to appear in this target audience. In our example, our audience is interested in Bicycling, Surfing/body-boarding, and Football the most.

Note that in the top right corner, users can toggle between Deviance and Percentage views depending on needs and interpretability. Users can also select to show all data points, top 10%, top 20%, or top 50%.

Inventory

In the Inventory section of reports you can review the specific CTV Title, Series, Genres, Channels, Publishers, Networks, and Apps that your audience is currently watching. Our CTV Bidder provides the capabilities for targeting down to the title level, so these are all actionable insights that can be applied to your Campaigns. Build detailed CTV targeting plans to serve ads across the inventory your audience watches the most.

Here is a sample model for CTV Genres viewership for extreme sports fans:

Looking at the graph in the percentage view, the light blue columns on the graph display the national average for each demographic or interest category (Sitcom, Crime drama, Thriller, etc.). A zip code is considered exceptional in a category if the % of users in that category for the zip code is higher than the national average. This is calculated in the ‘Dev from Normal’ column.

Switch the graph type using the ‘View Deviance’ button to view how a zip code or the complete set of zip codes differs from the national average on each category.

Published Audiences

To create a targetable audience from the report data, click the ‘Publish Audience’ button. This will give you the option to select the threshold of match you want to include in the export. Filter results to select ‘all results’ or the ‘top 10%’, ‘top 20%’, ‘top 50%’ of matched zip codes depending on how closely matched you would like the zip codes to be. This audience will then be readily available for targeting in any Line Item under Custom ART Audiences.