Reach & Frequency: What Has Changed and Why it Matters
- Brian Schopper
- Bella Nguyen (Unlicensed)
- Erin Bowling
- Dylan Mabin (Unlicensed)
Historical R/F Approach - Overview
Beginning with the launch of Geopath Insights in 2019, our reach & frequency approach was built around highly aggregated, anonymous carrier data. This mobile data was used to determine average frequency and understand general home locations, which allowed us to look at the footprint of audiences in different markets.
While this was the best data resource available at the time, there was limited device-level data available for scalable validation of the approach. In January of 2020, Geopath began using observed frequency data and unit-level TRPs to estimate unit-level reach. Package-level duplication was calculated and aggregated at the market to estimate package reach – this approach replaced the surveyed data that had been used for the last several years to estimate duplication of audience.
In 2021, data releases accounted for travel pattern changes due to the Covid-19 pandemic. Reduced traffic volume, combined with existing reach & frequency methods and assumptions about trip frequency lead to outputs that were misaligned for some packages of inventory. Upon discovery, Geopath immediately began exploring options to enhance its reach & frequency methods with device-level mobile location data.
New R/F Approach - Overview
Geopath’s reach & frequency approach has been, and will continue to be, an iterative process; as new data and technology resources become available, the approach is subject to incremental enhancements. Most recently, the shift from surveyed data to observed data has allowed for more precise measurement and advanced validation methods.
Geopath’s new reach & frequency approach is centered around observed mobile location data from smartphone applications and connected cars. Observable reach metrics from mobile data provides a “ground truth” for Geopath to analyze and understand how reach builds over time for different packages of real inventory, targeting various audiences across markets. These observed reach datapoints can then be used with inventory attributes and unit level audience metrics to build and train reach and frequency models through machine learning.
This then provides us with a method that allows us to calculate package-level reach and frequency for any campaign length, aligned with the observed data with ±5% accuracy.
Please note: this new reach and frequency approach is incorporated into the 2023 data release, which will become the default dataset in January of 2023. Due to the various changes in the approach as outlined above, comparisons should only be done directionally - comparing the 2023 reach and frequency to prior datasets will not be a true “apples to apples” comparison.
The below chart covers some of the major components that have evolved from Geopath’s historical reach & frequency approach to the new approach. Along with the factors mentioned above, these components help shape how reach & frequency are compiled at the package level.
Component | What Has Changed | Why It Matters |
|---|---|---|
Audience Coverage – The geographic dispersion around a market of the home location of the exposed audience and amount of exposure in each neighborhood. | Previously, audience coverage was aggregated at larger geographies or market-levels. Now, with increased precision, looking down to the neighborhood-level coverage (census tract, within the market), we can now accurately identify the reachable and unreachable target audience population. | Audiences are not always evenly spread across a target market. Understanding the inventory which has coverage (potential reach) of local areas in which your audience lives within a larger market is key to audience targeting and efficient media planning.
|
Duplication – within a package of inventory, there are overlaps in audience delivery. Two or more media units will deliver to the some of the same audience in a market. | Duplication of reach was previously assumed to be random at the package level, within the coverage area of the market. This meant that reach and frequency estimates were not sensitive to the overlap of coverage and precise duplication. However, reach duplication is not necessarily random. Some media have significant overlap of audience and therefore will not add much incremental reach, but will add frequency. Similarly, inventory which has very little overlap will add reach, but very little additional frequency. Using mobile movement data, we can now precisely calculate the overlap and duplication of audience within a package. | At the package level, each piece of media has a unique relationship with every other piece of media in a package. Individual and collective duplication of the media units in the package is now better accounted for when estimating reach. The clustering or spread of an inventory package within a market has a significant impact on the audience that it is able to reach. If all of the inventory within a package is concentrated to one area, large portions of the target audience may never have the opportunity to see the campaign. |