Triton Digital’s Podcast Metrics (PCM) Demos+ provides person-level demographic, sociographic, media behavior, and purchase intent data in the United States of America (US) and Australia (AU), to understand the profile of the audience for podcasts measured by PCM. A combination of survey plus census-level podcast download data is used to source the data.
Methodology
Triton Digital’s Demos+ serves as an important data source for the podcast industry, both at the publisher and agency level, to help understand a podcast audience and inform buying decisions. It is important to understand the methodology to provide an understanding of credibility and accuracy for the data used to generate demographic profiles.
The Survey (US)
In collaboration with Signal Hill Insights beginning in April 2021, Triton has been fielding a continuous measurement survey among a nationally representative sample of US monthly podcast listeners aged 18+. A total of 3,000 listeners are surveyed per quarter, with data from each survey added to a rolling 24-month sample. Podcast listener quotas are set by age, gender, education, region, and race (including unacculturated Hispanics) based on an initial representative national survey of more than 10,000 American adults in the general population, balanced according to the most recent data from the US Census Bureau.
The quarterly surveys are approximately seven minutes in duration and are conducted in both English and Spanish. Questions range from demographics and socio-demographics (age, gender, household income, race, education, marital status, etc.) to media consumption inclusive of print, radio, podcast, TV, streaming, etc., and purchase habits/intentions (QSR, travel, insurance, auto, etc.). To determine podcast listening, participants are asked to identify the podcasts they have listened to in the past 30 days.
The Survey (AU)
In collaboration with Signal Hill Insights beginning in April 2023, Triton has been fielding a continuous online measurement survey among a nationally representative sample of AU monthly podcast listeners aged 18+. A total of 3,000 listeners were surveyed for the first survey period, subsequent periods will have a sample of 1,500 respondents, with data from each six-month period added to a rolling 24-month sample. Podcast listener quotas are set by age, gender, and region, based on an initial representative national survey of more than 8,000 Australian adults in the general population, balanced according to the most recent data from the Australian Bureau of Statistics (ABS)
The surveys are approximately ten minutes in duration and are conducted in English. Questions range from demographics and socio-demographics (age, gender, household income, marital status, etc.) to media consumption inclusive of print, radio, podcast, TV, streaming, etc., and purchase habits/intentions (QSR, travel, insurance, auto, etc.). To determine podcast listening, participants are asked to identify the podcasts they have listened to in the past 30 days.
Why are Survey Results Paired with Census-level Data?
A podcast survey alone does not provide enough scale, as listening is spread across the long tail of podcast listening. Triton solves for this by using the vast amount of census data from the podcasts that PCM measures to determine which podcasts have listeners in common. This enables scale, as a small sample of listeners from the survey can then be expanded through the pairing of the survey data with the census data to form a “neighborhood.” A neighborhood can be defined as a grouping of podcasts that share listeners, or “look alike” podcasts.
Profiling Approaches
The podcast for which a profile is generated is called a seed podcast. A neighborhood is generated around each seed podcast using Triton PCM census data. To generate a profile, the survey results for the neighborhood are applied to the seed podcast. Various profiling methods were developed to enable profiling for all podcasts.
These levels include:
- Neighborhood: podcasts with up to 100 (US) or 150 (AU) neighborhood podcasts with at least 30 survey mentions.
- For podcasts with at least 100 survey mentions, the seed podcast is the sole neighbor.
- iTunes Category/Survey Average profiles:for smaller programs where no neighborhood profiles are produced, the average survey profile of its iTunes category is entered. When iTunes category is missing for a podcast, the average survey profile of monthly podcast listeners is then applied.
- This enables demographics to exist for every program, regardless of size, to aggregate accordingly (genre, publisher, etc.).
Procedures for Quality Control and Accuracy
Data Analysis
To check the accuracy of the neighborhood approach, a comparison of the modeled profiles and survey direct profiles for the shows with the largest survey mentions was done across the narrower demo cuts. The profiles were also segmented by genre and publisher for further analyses.
Data Caveats (US)
For podcasts of smaller size, reporting at the iTunes Category level is typically needed as adequate neighborhood mentions are often not able to be reached. Results at this level of profiling are less precise, although with having built up to a full sample the need to profile podcasts at this level has lessened. Currently 99.5% of listeners are measured using the neighborhood approach. Within the Podcast Metrics interface a dimension called "Methodology" is available to denote the profiling approach used.
New Podcasts/Publishers: On a monthly basis, new publishers and podcasts have the opportunity to be measured via Demos+. For these new publishers and/or podcasts, neighborhoods using Triton's census level data are created. The podcasts within the neighborhood are then mapped back to the latest survey available to produce demographics. This allows for profiles to be produced sooner, to have the opportunity to learn about the show's audience, versus having to wait for the current survey to end. These profiles may be subject to change on a month-to-month basis as their audience is new and developing.
Data Caveats (AU)
Survey: Additional survey results will become available in Q1 2024 and will continue to grow with additional periods for subsequent six month periods. With the survey sample growth, this will enable a more robust analysis and to expand the number of podcasts that can be profiled using neighborhood profiles. Reporting at the iTunes Category level is typically needed for smaller-scaled shows, as the demographics produced at this level are less precise. Currently 90% of listeners are measured using the neighborhood approach, based on the first survey wave. As our sample increases, the need to report at this level will decline. Within the Podcast Metrics interface a dimension called "Methodology" is available to denote the profiling approach used.