Where do advertising carbon emissions come from in a programmatic buy?
Part of Supply path transparency
The short answer
Programmatic advertising produces carbon emissions through electricity and hardware: servers that run auctions, networks that carry bid requests and creative files, and the screens that display the ad. Ad Net Zero's Global Media Sustainability Framework estimates each stage separately. Supply path length matters because more selling paths can mean more servers and more bid traffic per impression.
Key takeaways
- Ad Net Zero's framework splits a digital campaign into buying, delivery and display stages, each estimated from impressions, data volumes and electricity grid factors.
- The buying stage uses a site's ads.txt line count as a proxy for auction activity, which the framework itself calls an imperfect proxy.
- IAB Tech Lab's sustainability playbook links shorter paths and fewer duplicate bid requests to lower energy use, but says its list is not quantified in CO2 terms.
- Estimates built on default values carry high uncertainty, and changing an input can lower the estimate without lowering real emissions.
- Any environmental claim made from these estimates falls under truth in advertising rules in the US and Canada.
How does programmatic advertising produce carbon emissions?
A programmatic display or video ad moves through three physical steps. Before it is shown, servers at the publisher's ad stack, the exchanges and the buying platforms trade bid requests and responses to pick a winner. The winning creative file then travels through a content delivery network and the internet to the reader's device, which spends a few seconds of screen time and processor work showing it. Each step draws electricity, and each piece of hardware carries emissions from its manufacture.
Definition
- Advertising carbon emissions
- The greenhouse gases attributed to delivering an ad campaign, expressed in kilograms of CO2 equivalent (kg CO2e). They include use phase emissions from electricity and embodied emissions from building the servers, network equipment and devices involved.
Researchers have measured parts of this chain directly. The CarbonTag study, published in IEEE Transactions on Sustainable Computing (opens in a new tab) in 2023, states that "Both the delivery and rendering of each ad consume energy" and measured rendering energy for single ads in the browser. A 2023 study of third party tracking (opens in a new tab), using telemetry from 100,000 users and a crawl of 2.7 million websites, found that tracking increased data transmissions "upwards of 21%". That is one study's global finding, not a benchmark for a US or Canadian campaign.
Which industry frameworks measure advertising emissions?
One framework built for media campaigns is the Global Media Sustainability Framework (opens in a new tab) (GMSF), which Ad Net Zero describes as "a voluntary standard developed by Ad Net Zero and the advertising industry, to help calculate and reduce greenhouse gas emissions from media campaigns." Version 1.3 covers six media channels, and its guides and data request forms are open access.
The GMSF Digital Implementation Guide v1.3 (opens in a new tab) was refined with IAB Europe's Methodology and Framework Working Group. It divides a digital campaign into creation, distribution and consumption, with a formula for each step and a separate method for corporate overhead. Company reporting sits underneath: the GHG Protocol Scope 3 Standard (opens in a new tab) lets companies assess their entire value chain emissions, which is why a sustainability team may ask its agency for campaign figures.
| Stage (GMSF name) | What it covers | Main inputs | Default when data is missing |
|---|---|---|---|
| Ad Inventory Selection: Media Buying | Server processing and network transfer of RTB requests and responses | Impressions, country, ads.txt lines, server and call factors | 1,000 ads.txt lines when a site has no file |
| Ad Creative Delivery | Moving the creative to the device through content delivery networks and fixed or mobile networks | Impressions, payload size, connection mix, grid intensity | 0.25 MB display, 4 MB video, 6 MB in-stream, plus player overhead |
| Digital Ad Display | Device electricity and a share of device manufacturing, by view time | Impressions, view time, device split | 3 seconds display, 30 seconds video |
| Targeting | Data lifecycle and data vendors | None yet | Placeholder in v1.3 |
How does supply path length affect emissions?
In the GMSF digital guide, the buying stage is the one tied to supply paths. The guide calls it "the hardest stage to model in the lifecycle of the digital ad due to supply chain complexity." Its default method multiplies a site's number of ads.txt lines by a server factor and a call factor: for display, 1.412 activated servers and 1.464 RTB calls per line; for video, 1.316 and 1.334. A site that authorizes more sellers therefore produces a larger estimate for the same impressions.
The guide is candid about the proxy: the working group "acknowledges that the number of ads.txt lines is an imperfect proxy of programmatic supply chain." A long file can hold dormant lines, and a short one says nothing about duplicate requests from a header bidding setup. Version 1.3 also cut the default for sites with no file from 3,000 lines to 1,000, based on files from 460,720 domains.
The case for shorter paths comes from IAB Tech Lab's Sustainability Playbook (opens in a new tab): "multiple requests for the same ad slot, without clear business goals, creates unnecessary carbon emissions." It asks buyers to cut duplicate bid requests with the Global Placement ID and limit multi hop purchases, and says private marketplaces lower "the number of hops in the supply chain." It also states that its list "Is not quantified in terms of CO2 or Greenhouse Gas (GHG) emissions," so it supports the direction, not a size of saving.
What can buyers ask partners about emissions?
The IAB Europe Supply Chain Transparency Guide (opens in a new tab) added sustainability questions in 2024, noting that measuring Scope 3 emissions "requires insight into the emissions of their commercial partners." One question it gives buyers for DSPs is "Do you deduplicate bid requests?" The checklist below pairs such questions with the GMSF data levels.
Questions for a DSP, SSP or curator before a campaign emissions estimate
- The partner names the framework and version it uses, such as GMSF v1.3, and which stages its figure includes.
- The partner states its data level for each stage, from level 0 defaults to level 4 contributed data by placement and month.
- The DSP says whether it deduplicates bid requests and whether it reads the Global Placement ID to spot the same slot on several paths.
- The partner says which electricity grid factors it applies for US and Canadian impressions, and where its servers sit.
- The partner can share an organization level greenhouse gas inventory, and says whether a third party verified it.
- Any offset or neutrality claim comes with the offset type, the registry and the evidence behind it, and IAB Tech Lab's playbook advises buyers to require third party verification from partners offering climate neutral or offsetting solutions.
How reliable are advertising emissions estimates?
Reliability depends on how much of the estimate is defaults. GMSF v1.3 asks users to report an Uncertainty Level, the share of data points calculated at level 0 or level 1. In the guide's illustration, 12 of 15 data points use those levels: 80 percent, labeled high uncertainty. A figure built from default payloads, view times and ads.txt counts is a model, not a meter reading.
| Level | Method | What the buyer needs from partners |
|---|---|---|
| 0 | Default value | Nothing; used when a site has no ads.txt file |
| 1 | ads.txt line count with standard supply chain depth | The domains on the plan |
| 2 | Contributed data | Aggregate bidding volumes from publishers, SSPs and DSPs |
| 3 | Contributed data by geo | The same data split by country |
| 4 | Monthly, placement level contributed data by geo | Monthly data per placement and country |
Methods outside the framework also differ. An August 2026 scoping review (opens in a new tab) accepted at the LOCO 2026 workshop found the research "inconsistent in its terminology" and identified five measurement approaches across 15 papers, from ad blocking to traffic flow analysis, so figures built different ways are not directly comparable. IAB Europe's proposed carbon.json file for emissions intensity data had a public feedback period (opens in a new tab) from May 20 to August 19, 2026; the v1.3 guide says it is not yet in the formulae.
What rules apply to a claim about lower emission advertising?
An estimate becomes a marketing claim the moment a brand or partner advertises it. The FTC's Green Guides summary (opens in a new tab) states that "Marketers should not make broad, unqualified general environmental benefit claims like 'green' or 'eco-friendly'," that marketers "should have competent and reliable scientific evidence to support carbon offset claims," and that they should not sell the same reduction more than once. It also says marketers "should not advertise a carbon offset if the law already requires the activity that is the basis for the offset."
In Canada, Ad Standards issued its Guidance on Environmental Claims in Advertising (opens in a new tab) on April 16, 2026. It asks for "up-to-date valid, reliable and relevant scientific evidence to support all claims," and states that claims "should be limited to the specific improvement achieved (e.g., reduced emissions compared to a prior model) and should not imply a broader environmental benefit or net positive impact unless substantiated."
What does an emissions request look like on a real plan?
Fictional example
A refurbished electronics retailer asks for a campaign estimate
An unnamed retailer of refurbished laptops and phones in Alabama plans an English display and online video campaign across Birmingham, Huntsville and Mobile, on tech review, personal finance and local news sites. Its sustainability lead asks the agency for a campaign emissions estimate before approving the plan.
The planner lists each domain with its ads.txt line count and the exchanges that sell it directly. Two sites have no ads.txt file, so the planner notes that the framework's default would apply to them instead of dropping them, then sends the partner checklist to the DSP and the main SSPs.
The agency reports the estimate with its uncertainty level and the stages left out, marked as a planning figure. The retailer agrees not to use it in any consumer facing claim without evidence that meets the FTC's standard.
How can PlacementGraph show the selling paths behind a site list?
Frequently asked questions
Is video advertising more carbon intensive than display?
Under the GMSF v1.3 defaults, a video impression carries a larger creative payload and a longer default view time than a display impression: 4 MB or 6 MB in-stream against 0.25 MB, and 30 seconds against 3 seconds. Those are defaults chosen to overestimate. A campaign's real payloads and view times give a better figure.
Do walled garden platforms report emissions the same way?
Not from the same inputs. The GMSF digital guide says selection emissions on end to end platforms are harder to estimate because less information is available, so it assumes 500 activated servers and no network calls for them.
Does a private marketplace deal cut emissions?
IAB Tech Lab's playbook says private marketplaces lower the number of hops in the supply chain and the computing load of auctions, but it does not quantify the effect. A deal that still receives one impression through several exchanges keeps those duplicate requests.
Can an agency offset a campaign's emissions?
It can buy offsets, but an offset is not a reduction in the campaign's own emissions. If the purchase is advertised, the FTC calls for competent and reliable scientific evidence, accounting that does not sell the same reduction twice, and no offsets for activity the law already requires.
Sources
- Global Media Sustainability Framework v1.3 (opens in a new tab)Ad Net Zero
- GMSF Digital Implementation Guide v1.3 (opens in a new tab)Ad Net Zero
- The Sustainability Playbook: Starter Guide for a Sustainable Programmatic Marketplace (opens in a new tab)IAB Tech Lab
- Supply Chain Transparency Guide 2024 (opens in a new tab)IAB Europe
- carbon.json voluntary disclosure specification open for public comment (opens in a new tab)IAB Europe
- Corporate Value Chain (Scope 3) Standard (opens in a new tab)GHG Protocol
- CarbonTag: A Browser-Based Method for Approximating Energy Consumption of Online Ads (opens in a new tab)arXiv (IEEE Transactions on Sustainable Computing)
- Quantifying Carbon Emissions due to Online Third-Party Tracking (opens in a new tab)arXiv
- A Scoping Review of Methods to Measure the Energy and Carbon Footprint of Web Tracking and Advertising (opens in a new tab)arXiv (LOCO 2026)
- Environmental Claims: Summary of the Green Guides (opens in a new tab)Federal Trade Commission
- Guidance on Environmental Claims in Advertising (opens in a new tab)Ad Standards
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.