Which website carbon calculator should you trust?

Figures last checked 30 August 2026  ·  next review November 2026

Website Carbon, Ecograder, EFWA and this site's own audit all run the same bytes-based calculation, so they broadly agree with each other, and none of them knows where in the world your visitors are. Use one of them to compare pages and track improvement, use EFWA when somebody needs a figure that was checked by an outside body, and use EcoPigs when the number has to reflect the real electricity grid your visitors are sitting on.

A green badge in a website footer is a claim, and like any claim it is worth knowing who checked it. Most are self-serve: the site owner ran a free scan, liked the result, and embedded the badge. One is a paid annual accreditation with a published threshold. They do not mean the same thing, and the buyer, funder or procurement officer who eventually asks about your website's emissions will want to know which one you have.

Part of why the figure is hard is that most of the emissions belong to somebody else. There is your hosting provider, behind them the data centre, and behind them an energy supplier selling tariffs that differ in carbon content, and nobody in that chain is obliged to tell you anything. Estimating it anyway is not new: How Bad Are Bananas? put a single Google search at one gram in 2010, a figure attributed to the book rather than to a published dataset anyone can check.

This page covers the five tools that produce a carbon figure for a web page, plus the organisation whose data and code most of them run on. It is written for someone who has to choose one, not for someone building one.

What separates one calculator from another

Five things, and the first three account for nearly all the difference between one number and another.

1. Method: counted bytes or measured energy

Four of the five estimate emissions from page weight, meaning the total size of everything the page sends to the visitor. They share one method, the Sustainable Web Design Model version 4 (SWDM v4), and most of them run the same open-source code, Green Web Foundation's CO2.js library. In plain terms it does four things:

1. Counts the bytes the page sends to the visitor.
2. Multiplies by an energy figure per byte, split across the data centre, the network and the visitor's own device.
3. Multiplies by how carbon-intensive the electricity is assumed to be.
4. Adjusts for the assumed mix of new and returning visitors, since a returning visitor downloads less.

The fifth, EcoPigs, does not estimate from bytes at all. It measures the energy the browser actually uses while rendering the page. That difference matters because bytes are a proxy for energy rather than energy itself: a page that works the processor hard can burn power the byte count never sees, and caching, where the browser keeps a copy of files it has already downloaded, means a returning visitor transfers far less than a single scan records.

2. Location: a world average, not yours

Every public badge in this set charges one global-average grid figure of 494 g CO2e per kWh, Ember's figure for the world, to the data centre, the network and the visitor's device alike (SWDM v4 methodology). The UK grid ran at 129 g in 2025, so a British site is billed for nearly four times the carbon its electricity actually carries. The same page, served and read in largely nuclear France and in a coal-heavy grid, gets the same score.

Two versions are in circulation, which is worth knowing before comparing two readings. Version 4 uses the 494 g figure above; the version 3 methodology, still running behind some tools, uses 442 g. CO2.js can accept a regional grid figure in place of either, but the public badges use the global default. If your audience is entirely in one country, the number is telling you about the page, not about your emissions.

3. Green hosting: a yes or no, not a supplier

Verified green hosting changes the data-centre part of the calculation and nothing else. What it changes depends on the version:

• Version 4 applies a green hosting factor of 1, which takes the running emissions of the data centre out of the sum. The embodied emissions, meaning the carbon spent building the hardware in the first place, stay in.
• Version 3, and CO2.js when run in its version 3 model, fixes the data-centre figure at a flat 50 g CO2 per kWh, overriding anything else you set.

Either way the switch has two positions and no dial. A host running on a direct renewable supply and a host buying certificates both read as green. Those certificates are worth knowing about: a REGO, a Renewable Energy Guarantee of Origin, lets a supplier sell electricity as renewable on the strength of a certificate bought from a generator, without having generated anything itself. The retail electricity supplier behind your host does not appear in the calculation anywhere in any case. Naming yours in your sustainability report is fine. It will not move your score by a gram.

A supplier's published emissions per kWh is an organisation-level figure, and not what any one customer buys. A supplier that generates renewably allocates that generation to its renewable tariff, so what is left on its standard and commercial tariffs runs at an intensity well above the organisation-level figure, and above the grid average as well. Two data centres on the same supplier can sit either side of that average depending on the tariff each is on. Nothing in the calculation asks which.

4. Verified or self-serve

Four of the five are self-serve: you run them, you decide whether to publish the result, and nobody checks. EFWA's accreditation is the exception, being vetted on application and re-audited quarterly. Self-serve is not the same as untrustworthy, but it does mean the badge attests to the tool's output on the day the owner liked it.

5. Cost

Website Carbon, Ecograder, EcoPigs' calculator and Green Web Foundation's checks are free. EFWA accreditation is paid annually and also requires the organisation to offset its yearly footprint, which is a commitment about the organisation rather than a measurement of the website.

The tools, one by one

Website Carbon (Wholegrain Digital)

• Free, self-serve, and the source of the embeddable footer badge most people mean when they say website carbon badge.
• Bytes-based SWDM. Scans the single page whose address you give it.
• The de facto standard, largely because it was first, free and easy to add.
• websitecarbon.com

Ecograder (Mightybytes)

• Free grader that has been running for over a decade, scoring one page out of 100.
• Combines four things in that score: performance and page weight, user experience including accessibility and findability, green hosting, and SWDM carbon.
• More useful than Website Carbon when the question is what to fix, because the score decomposes into recommendations.
• mightybytes.com

EFWA carbon-conscious accreditation (Eco-Friendly Web Alliance)

• Not a badge you can award yourself. A vetted, paid, annual accreditation with quarterly re-audits, run by a UK social enterprise.
• The threshold tightened from 0.65 g to 0.35 g CO2 per page view on 10 April 2025, averaged across the whole site, with the home page required to come in under 1 g (EFWA programme change, April 2025).
• Green hosting is expected; hosting that is not verified green adds a 10% penalty. Offsetting the organisation's annual footprint is also required (accreditation criteria).
• Averaging across the whole site is the part that catches people out. A lean home page does not rescue two hundred heavy content pages behind it.
• Young: EFWA Ltd was incorporated on 17 November 2021 (Companies House 13749776), and the scheme is still changing.
• Adopters named by EFWA include National Geographic Kids, the Welsh Government, Rewilding Britain and the British Medical Journal, alongside roughly 11,500 LinkedIn followers and a stated goal of a million carbon-conscious websites by 2030 (self-reported by EFWA, which implies current numbers a good deal smaller).

EcoPigs (OYNK)

• One small Northamptonshire agency's in-house tool, founded by Ashley John: a free calculator plus paid consultancy, rather than a badge other people embed.
• Methodologically the most advanced of the set. It measures actual browser energy through the Chrome DevTools Protocol, applies device power profiles, and uses live regional grid intensity from Electricity Maps and the IEA, time-weighted to when visits actually happen.
• Crawls every page of a site rather than scanning the single address you hand it, which is why its own totals count pages as well as sites.
• The method is documented openly in OYNK's P.E.E.R. Framework whitepaper.
• Lowest adoption of the five: roughly 430 websites and 41,000 pages audited since launch (OYNK's own figure), with the current version dating from 2024 to 2025.
• oynk.co.uk and its whitepapers

The CyberspaceRoad audit

• The carbon figure sits inside a wider website audit rather than standing alone, using the same data-size method as Website Carbon.
• Same limits as the rest of the bytes-based group: global-average grid, green hosting as a yes or no. The reading is reported with a note saying so, so a UK client can see why the figure runs high against the grid they are actually on.
• The reason it exists is context. A carbon number on its own rarely tells you what to do, whereas the same number next to the page's images, scripts, accessibility failures and performance results usually does. See what a website audit covers.

Green Web Foundation, which is not a calculator

Green Web Foundation gets no row in the table because it produces no per-page gram figure. It does two things the others depend on: it maintains CO2.js, the code most of these tools run, and it verifies green hosting. What it records is narrower than most people assume:

• A binary green or not-green flag per hosting provider, not per website and not per website owner (Green Web Dataset).
• The evidence that provider submitted: green energy certificates, renewable credits, utility bills, company reports, reverified annually (what counts as evidence).
• It does not store a retail electricity supplier as a carbon-weighted field. Retail suppliers are not web hosts, so they appear only buried inside a host's evidence documents.

How to choose

If you want a number this afternoon

Website Carbon. Free, instant, and the result is directly comparable with the thousands of other sites quoting the same tool. The trade-off is that it measures one page and knows nothing about your audience's location.

If you want to know what to fix

Ecograder, or a full audit. A single gram figure gives you no instruction; a score broken into page weight, hosting, accessibility and findability does. The trade-off is a longer report to read.

If a funder or buyer wants proof somebody checked

EFWA accreditation. It is the only option in the set where an outside body measured your whole site against a published threshold and will do so again next quarter. The trade-offs are cost, an annual offsetting commitment, and a whole-site average that is genuinely hard to hit if your content library is old.

If the figure has to survive a technical challenge

EcoPigs. Measured browser energy against live regional grid data answers the objection that a bytes-based estimate cannot tell France from Poland. The trade-offs are that almost nobody else quotes it, so your figure is not comparable with the site next door, and its accuracy claim is the supplier's own.

The pattern worth naming

Adoption runs in the opposite direction to measurement quality. Website Carbon is the least precise and is everywhere; EcoPigs is the most sophisticated and has measured roughly 430 sites. What spreads is not the best measurement but the one that is free, instant and produces something you can put in a footer. Worth holding in mind whenever a sustainability claim seems to have become an industry standard remarkably quickly.

What none of them will do

None of these tools measures the emissions of your organisation. A website is a small part of a small category of a carbon footprint, and a site scoring well tells a funder nothing about the flights, the supply chain or the building. If the reason you are here is that somebody asked for your organisation's footprint, start with working out a footprint without buying software and with which scope 3 categories actually apply, then come back to the website.

Questions people ask

Does a green badge mean a website is low carbon?

It depends entirely on which badge. A Website Carbon or Ecograder badge means somebody ran a free scan and chose to display the result, with nobody checking it afterwards. An EFWA accreditation mark means an outside body measured the whole site against a published threshold and re-audits it quarterly.

Will my score improve if I switch to a renewable electricity supplier?

Not unless your web host switches. The calculation looks at whether your hosting provider is on Green Web Foundation's verified list, and that is a yes or no. Your own office electricity contract is not part of the sum, and neither is the name of your host's supplier.

Why do two calculators give my page different numbers?

Usually because they measured different bytes: a different moment, a different consent state, a different assumption about returning visitors, or one tool averaging a whole site while another scans one page. Model version matters too, since version 3 charges 442 g per kWh and version 4 charges 494 g. EcoPigs can differ by more, because it is measuring something else.

Is 0.35 g per page view achievable for an ordinary small business site?

Yes, for a site built with restraint: compressed images, few third-party scripts, self-hosted fonts, no video autoplaying on the home page. It is difficult for a site running a page builder, a marketing tag manager and a stock hero video, which is a good part of why the threshold is interesting.

What actually counts as green hosting?

Green Web Foundation records a green or not-green flag per hosting provider, based on evidence the provider submits: green energy certificates, renewable credits, utility bills or company reports, reverified annually. It is a judgement about the provider, not about your website or your own energy contract.

Do I need to pay for an accreditation, or is a free badge enough?

A free badge is enough if the audience is your own visitors and the point is to show you have thought about it. Pay for accreditation when somebody with a procurement form is going to ask who verified the claim, because that is the question a self-serve badge cannot answer.

How often should a site be re-measured?

Once a quarter is enough for a site that changes slowly, and after any redesign, new tracking script or added video, which are the three things that move the figure most. Keep the dated results, because the trend is more useful than any single reading.

More answers

• How does a small organisation work out its carbon footprint without buying software?
• What does scope 3 actually require a small organisation to measure?
• What does a net zero roadmap cost when nobody in-house does sustainability?
• Can AI build a custom tool cheaply enough for a small organisation to afford?
• How do you collect carbon data from households or members over years, not once?
• What should a website accessibility audit cover, and what does it cost?

DON'T PANIC cyberspaceroad.com