17 June 2025, updated 23 June 2025
I joined a really good workshop on Estimating digital carbon emissions by Chris Adams and Katrin Fritsch of the Green Web Foundation last week. You should join one of the workshops - it helped me understand the landscape of tools and approaches much better, and talking to participants who are on the same journey is energising. Loved it!
When I tried to work out my own company’s emissions last year it was a bit of a first-principles slog (trying to actually build it to understand it), but I learned a lot. And it felt like starting to take some action would also build momentum for more action, and I’ve done more since. So I’m interested in first steps, but also how I can encourage my customers take their first steps. I put emissions estimates into my invoices and proposals these days. Always interested to hear tactics others have tried.
It might be interesting if Green Web Foundation created follow up activities/articles that give organisations step by step playbooks for putting what they’ve learned into action. “If you’re an small organisation, start by doing X then Y then Z, but save ABC for later. And if you’re a large/multinational organisation, or operate a smelter, call us to get started” kind of thing.
All the notes and links
- Tech carbon footprint is 2-4% of global emissions?
- digital services: end user device usage, network transfer, data centre usage, production and disposal.
- Scopes
- scope 1: directly controlled emissions (fuel combustion, company vehicles, fugutive emissions). For most companies, this is small. Coffee metaphor: emissions from directly buring fuel to make hot coffee.
- scope 2: purchased energy (electricity, cooling, heat, steam). Coffee metaphor: emissions from electricty you buy, to make coffee.
- scope 1 and 2 are reported directly.
- scope 3: “upstream” purchased goods/services, travel etc, and “downstream” transportation, use of products/services, investments, disposal. Coffee metaphor: emissions in your supply chain(s), so you can have coffee.
- (These descriptions very incomplete, read the GHG protocol)
- Location-based vs market-based. My sense: location-based is concerned with the on-the-ground reality, with atoms. Market-based adds accounting adjustments: if you have emissions of 10 but buy clean energy of 6, your market based emissions = 4 but your location based emissions still = 10.
- Carbon’s not the only thing. Water, pollutants, affordable goods, overconsumption, inequality, resource scarcity, education, health, eutrophication, poverty, biodiversity, ecotoxicity and justice too.
- A simple model of digital service emissions: data transfers in GB x Annual Internet Energy per GB of 0.81 kWh/GB (how they derive this number: Sustainable Web Design Model v3 - but it looks as if this number has improved to 0.3 kWh/GB in the version 4). You can adjust it for grid intensity, % of new visitors/returning visitors/data reloaded to page (because caching reduces emissions). More on this: The Nuance of Quantifying Digital Carbon Emissions This data transfer metric is perhaps the equivalent of the quick method for freelancers: a good place to start.
- Alt/specific models for media-heavy services: DImpact, for apps/sites GreenFrame (but see other links below), for advertising Scope3.
- And for any digital service, there’s an emerging standard: Software Carbon Intensity (SCI) specification by the Standards Working Group in the Green Software Foundation - “The SCI technical specification describes how to calculate the carbon intensity of a software application”
Measuring site emissions:
Background to measurement:
AI and emissions:
Also:
OK that’s it. Good stuff, Green Web Foundation.