Sustainable Web Design (SWD) Model — Definition and GHG Accounting Context
A web page has no exhaust pipe, yet loading it burns real energy — in the data centre that serves it, the network that carries it, and the phone or laptop that renders it. The trouble is that none of those places has a meter you can read. To put a carbon number on a website, you need a model that turns something measurable — the data it sends — into an estimate of the energy and emissions behind it.
The Sustainable Web Design model is the most widely used of those models. It estimates a website’s carbon by converting the data it transfers into energy, and then into CO₂-equivalent.
The Sustainable Web Design (SWD) model is a methodology for estimating the greenhouse-gas emissions of websites and digital products from the amount of data they transfer. It multiplies gigabytes of data by an energy intensity (kWh per GB) and then by a grid carbon intensity to give grams of CO₂e, usually per page view. Version 4 (SWDM v4) splits that energy across data centres, networks, and user devices, and separates operational from embodied energy.
Definition — Turning Data Into Carbon
The Sustainable Web Design (SWD) model is a methodology for estimating the greenhouse-gas emissions of websites, web pages, and digital products. Because the energy used to deliver a web page cannot be metered directly, the model uses the one quantity that can be measured — the volume of data transferred — as a proxy, and converts it into energy and then into carbon. It is the engine behind many popular website carbon calculators, including the one on this site.
Its output is normally expressed as grams of CO₂-equivalent per page view or per visit, which makes very different sites comparable on a common basis. The model is deliberately simple to apply — you need only the page weight and a few coefficients — which is the source of both its popularity and its limitations: it is an estimate built on internet-average assumptions, not a measurement of any specific server or user.
The full calculation and its assumptions are set out in the website page-view methodology and the SWD model reference; this page defines what the model is and the parameters it runs on. It is worth stressing at the outset what it is for: general web browsing. High-volume video streaming is better estimated with a different, per-byte model, because the SWD model’s device and network assumptions do not hold at streaming scale.
Definition at a glance
| What it is | A model estimating website carbon from data transferred |
|---|---|
| Input | Data volume (GB), plus energy and grid coefficients |
| Output | g CO₂e, usually per page view or visit |
| Current version | SWDM v4 — segmented, operational + embodied |
| Segments | Data centres, networks, user devices |
| Not for | High-volume video streaming (use a per-byte model) |
From Bytes to Carbon: How the Model Works
At its core the model is a short multiplication chain — each step converting one measurable quantity into the next:
- Data → energy. Multiply the gigabytes transferred by an energy intensity in kWh per GB, which captures how much electricity moving that data consumes across the whole delivery path.
- Energy → carbon. Multiply the energy by a grid carbon intensity in grams CO₂e per kWh — essentially an electricity emission factor — to get emissions.
- Adjust and express. Apply adjustments for green-hosted data centres and for returning visitors whose browsers have cached data, then express the result per page view or per visit.
Everything interesting about the model lives in those coefficients: how the energy intensity is derived, and how the grid factor is chosen. That is exactly what changed most between versions.
Inside SWDM v4: The Segmented Model
Rather than one energy figure for the whole internet, SWDM v4 breaks the delivery path into three segments — data centres, networks, and user devices — and, within each, separates operational energy (running the equipment) from embodied energy (manufacturing it, amortised over its life). The live coefficients, in kWh per GB:
| Segment | Operational kWh/GB | Embodied kWh/GB |
|---|---|---|
| Data centres | 0.055 | 0.012 |
| Networks | 0.059 | 0.013 |
| User devices | 0.08 | 0.081 |
| All segments (sum) | 0.194 | 0.106 |
Two things stand out. First, user devices dominate — the phones and laptops that render pages account for the largest share of energy, and almost all of the embodied energy, because there are so many of them and they are replaced often. Second, embodied energy is not a rounding error: at about 0.106 of the roughly 0.30 kWh/GB total, more than a third of a website’s footprint in this model comes from making the hardware, not running it. To convert the energy to carbon, SWDM v4 applies a default global grid intensity of 494 grams CO₂e per kWh.
What Changed From v3 to v4
The jump from version 3 to version 4 is the clearest illustration of why the model version matters when you quote a web carbon figure:
| SWDM v3 | SWDM v4 | |
|---|---|---|
| Energy figure | One blended 0.81 kWh/GB | Segmented, ≈ 0.30 kWh/GB |
| Structure | Whole system in one number | Data centre / network / device |
| Embodied energy | Bundled in | Separated and reported |
| Effect | Higher per-GB estimate | Lower, more granular estimate |
The headline energy intensity fell from 0.81 kWh/GB to around 0.30 as the model incorporated better data on network and device efficiency. That is not the internet suddenly getting cleaner — it is the model getting more accurate. It is also why a page’s carbon figure can change simply because the calculator behind it moved from v3 to v4, and why the version should always travel with the number, much as a data source should.
A Worked Example
Take a fairly typical web page weighing 2 MB, estimated with SWDM v4’s live coefficients and default grid intensity (snapshot 2026-07-27):
- Data: 2 MB = 0.00195 GB
- Energy: 0.00195 GB × 0.30 kWh/GB = 0.000586 kWh
- Carbon: 0.000586 kWh × 494 g/kWh ≈ 0.29 g CO₂e per page view
A tenth of a gram here, a third of a gram there — trivial for one view, but a site with a million monthly views is emitting on the order of hundreds of kilograms of CO₂e a year just to be delivered. That is the scale the model is built to expose. The website carbon calculator runs this same chain on a real page weight and lets you vary the grid and green-hosting assumptions.
Scope, Green Hosting, and Limits
The SWD model is powerful precisely because it is simple — but that simplicity sets its boundaries, and using it well means respecting them.
- Green hosting reduces the data-centre segment: if a site is hosted on verified renewable energy, that segment’s operational emissions are cut. It does not touch the network or the user’s device, which the site owner does not control.
- Returning visitors transfer less data because their browsers cache assets, so the model assumes a mix of new and returning visits rather than a full download every time.
- Data transfer is a proxy, not a measurement. Two pages of equal weight can do very different amounts of computing; the model averages over that.
- It is a web-browsing model. For video streaming, a per-byte model (such as the Carbon Trust / DIMPACT approach) is more appropriate, because streaming’s network and device energy scale differently from page loads.
Read that way, an SWD figure is best understood as a well-founded estimate for comparison and for tracking improvement — is this page lighter than that one, is the redesign better than the old site — rather than a precise, auditable measurement of a specific server’s electricity bill.
Common Confusions
- Quoting a figure without its model version. SWDM v3 and v4 give materially different numbers for the same page. The version is part of the result.
- Treating the estimate as a measurement. The model converts data transfer into carbon using internet-average coefficients — it is an estimate for comparison, not a metered value.
- Using it for video streaming. The SWD model is built for web browsing; high-volume streaming needs a per-byte model with different assumptions.
- Assuming green hosting fixes everything. It only reduces the data-centre segment. Networks and user devices — often the larger share — are unaffected.
- Ignoring embodied energy. In SWDM v4, making the hardware is more than a third of the footprint. A model or figure that counts only operational energy understates the total.
Frequently Asked Questions
The Sustainable Web Design (SWD) model is a methodology for estimating the greenhouse-gas emissions of websites and digital products from the amount of data they transfer. Because the energy behind a web page cannot be metered directly, the model uses data volume as a proxy: it multiplies gigabytes transferred by an energy intensity (kWh per GB), then by a grid carbon intensity (grams CO₂e per kWh), to estimate emissions — usually expressed per page view or per visit. Its current version, SWDM v4, splits that energy across data centres, networks, and user devices and separates operational from embodied energy. It is the engine behind many website carbon calculators.
It follows a short chain. First it converts data to energy: gigabytes transferred multiplied by an energy intensity in kWh per GB. Then it converts energy to carbon: that energy multiplied by a grid carbon intensity in grams CO₂e per kWh, which is essentially an electricity emission factor. Finally it applies adjustments — a reduction for green-hosted data centres, and an allowance for returning visitors whose browsers have cached assets — and expresses the result per page view or per visit. In SWDM v4 the energy step is done segment by segment (data centres, networks, user devices, each with operational and embodied energy), using a default global grid intensity of 494 grams CO₂e per kWh. The full method is set out in the website page-view methodology.
Version 3 used a single blended energy figure of 0.81 kWh per GB for the whole internet. Version 4 replaced it with a segmented model that splits energy across data centres, networks, and user devices, and separates operational energy (running the equipment) from embodied energy (manufacturing it). The segmented total comes to roughly 0.30 kWh per GB — lower than v3, not because the internet suddenly became cleaner, but because the model incorporated better data on network and device efficiency and made its assumptions explicit. The practical consequence is that the same page can return different carbon figures under v3 and v4, so the model version should always be stated alongside the number.
Not really — it is designed for general web browsing. The SWD model treats energy as broadly proportional to data transferred, an assumption that holds reasonably for page loads but overstates emissions at the scale of high-volume video streaming, where much of the network and device energy is largely fixed rather than scaling with each additional gigabyte. For streaming, a per-byte model such as the Carbon Trust / DIMPACT approach is more appropriate. Using the SWD model for streaming is one of the most common ways digital carbon estimates go wrong, typically by a wide margin.
It is an estimate, not a measurement, and is best treated as such. The model converts data transfer into carbon using internet-average coefficients — secondary data — rather than metering any specific server, network, or device, and data transfer is itself a proxy for the underlying computing. That makes it well suited to comparison and tracking (is this page lighter than that one, did the redesign help) and less suited to precise, auditable reporting of an exact electricity figure. Its accuracy also depends on the inputs you give it: the real page weight, a representative grid intensity, and whether the hosting is genuinely on renewable energy. Used within those limits, it is a sound and widely accepted basis for understanding and reducing web carbon.