January 2023 Dataset

With this update we move away from our proprietary estimates of the carbon footprint of each component of the advertising supply chain and instead use open source methodology. We also switch to ads.txt and app-ads.txt as the primary source for mapping the programmatic supply chain. Finally, our overall boundary assumption is changing from "media production and distribution" to just "media distribution" to enable a more consistent comparison across media owners, publishers, and channels.

Use ads.txt and app-ads.txt as primary source for programmatic supply chain

• Use industry standards (ads.txt and sellers.json) to map the programmatic supply chain instead of our proprietary scanner.

Known issues:

• Some countries do not commonly use ads.txt files, especially Japan, Brazil, and China. Our coverage will be limited in these markets while we work with key publishers to create supply chain maps.
• We can't verify that there are not buyers and resellers that are not listed in ads.txt. While we expect that changes to ads.txt will be picked up quickly by SSPs and DSPs (and have confirmed with leading platforms that this is the case), it is possible that non-compliant reselling will continue to happen.
• The ads.txt file does not fully represent regional and format nuances. Many publishers use inline comments like #EMEA or #video to signify these, but they are not standardized (rumored to be coming in ads.txt version 1.5). We plan to support inline comments in an upcoming release and will use standard syntax when available.
• When we don't find an ads.txt file we assume the publisher is non-compliant. In practice, this may mean that the publisher doesn't have any programmatic monetization. We will provide a "no programmatic" designation in an upcoming release (likely Feb-2023).

Use open source methodology for most models and calculations

• Use average corporate footprint data from open source project instead of internal estimates.
• Use average ad tech platform footprint data from open source project instead of internal estimates.
• Use consumer device footprint data from open source project instead of simulating page load using proprietary scanner.
• Use benchmark data for ads per session instead of using proprietary scanner.
• Use "power model" from open source project for video and streaming data transfer calculations; use "conventional model" for banner and display.

Omit media production from lifecycle boundary

Philosophically we believe that most media production is done once and used many times with no marginal emissions per use other than from distribution, very similarly to creative.

This means:

• Companies that want to include media production in their emissions boundary should use Albert or other similar tools to calculate the carbon footprint.
• Publishers and media companies that primarily distribute content - for instance YouTube - will have no change.
• Publishers and media companies that produce their own content, like a newspaper, will deduct production emissions from their total. In general this is not broken out in any sustainability reports we have seen, but we suggest that it would be helpful for companies to do so.
• We will also work with DIMPACT to align with their methodology (which is similar, but doesn't try to break out news content production).