Key takeaways:
- Jump ahead to: What does the Carbon Meter measure? What does it not?
- The Carbon Meter measures activity directly related to browsing in Shift and estimates emissions based on data transfer size, not client-side processing.
- AI and cryptocurrency have high energy demands, and currently, there is no established model to accurately estimate the entirety of their emissions.
- Each of Shift’s carbon offset projects is carefully vetted to be verified, traceable, and localized.
- You can measure your digital carbon footprint in Shift by clicking the leaf icon in your browser.
Hi, my name is Mahum! I’m a software developer at Shift and the lead engineer behind the Carbon Meter, the technology that measures your digital footprint while you browse. For the past three and a half years, I’ve been immersed in the study of digital sustainability and emissions. The major research-based models I’ve studied include the OneByte model and the Sustainable Web Design Model (SWDM), which now defines how we estimate emissions at Shift.
In developing the Carbon Meter, we adapted the SWDM to fit a live browser environment, balancing what research defines as ideal and observable versus what’s actually possible to securely implement for real users.
To explain, let’s cover how the Carbon Meter works, how we translate browsing activity into estimated emissions, and what limitations we’re working to overcome—because research in this space is moving quickly and transparency is more important than ever.
How the Carbon Meter works: From bytes to CO₂e
Every click, scroll, and stream travels through physical infrastructure—data centers, networks, and personal devices—all of which consume energy and emit carbon, all of which cycle through production and disposal. As a browser, we can't meter energy at every step of that system in real-time. Instead, we use a research-based model to estimate the emissions and then offset them with verified carbon credits.
The Carbon Meter estimates, but it doesn’t directly measure energy. We’re continually iterating on its technology as science and data quality evolve.
At its core, the Carbon Meter converts browsing data into emissions estimates using the SWDM, developed by Wholegrain Digital and the Green Web Foundation.
The SWDM divides digital infrastructure into three systems:
- Data centers (where content is stored and served)
- Networks (which transmit that content)
- User devices (which receive and display it)
Each system includes both operational emissions (energy used while running) and embodied emissions (from manufacturing and disposal).
In Shift, the Carbon Meter measures the data transferred through your browsing and multiplies that by the SWDM’s average energy-per-gigabyte and the regionally specific carbon-intensity factors for electricity.
How the Carbon Meter produces an emissions estimate using the SWDM
1. Data centers
The Carbon Meter produces an aggregate estimate based on global averages for different types of hosting environments (e.g., cloud data centers vs. on-premise hosting). It does not dynamically adjust for differences in redundancy levels, hardware efficiency, or power sourcing (e.g., renewable vs. non-renewable energy use).
2. Network
The model estimates energy consumption based on data transfer size rather than specific network infrastructure. It does not differentiate between:
- Fiber optic vs. copper-based networks
- Different levels of redundancy in the networking infrastructure
- Whether a request is routed through an energy-efficient CDN vs. a high-energy-consuming routing path
Instead, it applies an average energy-per-GB-transferred value, derived from industry studies. This estimate includes contributions from core internet infrastructure (e.g., backbone networks, ISPs) and last-mile delivery (e.g., Wi-Fi, mobile networks) but does not factor in location-specific variances or ISP energy efficiency, as this kind of data is currently not possible to gather.
3. User devices
The SWDM assumes an average energy consumption rate per device type (e.g., desktop, laptop, mobile) rather than factoring in individual hardware specifications like GPU, PSU wattage, or efficiency levels.
Data size is the most practical metric for estimating emissions, but it simplifies a complex picture. A large transfer doesn’t always mean high energy use, and a small one isn’t necessarily efficient. Server-side computation, caching, and hardware differences affect the real-world impact.
Adjusting screen brightness or video quality has little measurable impact on total emissions. Most energy use occurs upstream in data centers and networks.
Factors like inefficient code (“bloatware”), frequent device turnover, and processing complexity can all contribute to emissions that data size alone doesn’t capture.
Despite its limits, data size remains the industry standard because it’s measurable, comparable, and consistent with the SWDM and broader digital-sustainability research. It’s a starting point that can evolve as models grow more precise.
What does the Carbon Meter measure?
The Carbon Meter focuses on activity directly related to browsing in Shift and estimates emissions based on data transfer size, not client-side processing.
What’s included
- Active browsing
- Loading web pages, viewing media and ads, and interacting with websites
- Using cloud apps or dashboards that refresh data
- Streaming media content
- Background processes such as preloading or auto-refreshing
What’s excluded
- Local device power draw (CPU, GPU, brightness, etc.)
- Background system operations not linked to browsing (software updates)
- Cryptocurrency mining or AI workloads
- Accurate emissions data when using VPNs or privacy tools (hidden IP addresses may cause the Carbon Meter to over or underestimate due to regional renewable energy factors and data transfer distance)
Playing a video game in your browser may be CPU/GPU-intensive, but the Carbon Meter can only measure energy consumption from data transferred (loading assets, ads, or real-time multiplayer data. The same goes for AI chatbots: emissions are estimated based on the size of requests that send prompts and receive responses. AI-related processing, such as model inference cannot currently be accounted for in the Carbon Meter.
AI and cryptocurrency have high energy demands, and currently, there is no established model to accurately estimate their emissions, as this remains an emerging field.
The journey to carbon neutrality
The Carbon Meter estimates emissions from our users’ browsing and Shift offsets them by purchasing verified carbon credits. These investments fund carbon offset projects that remove or prevent CO₂ emissions, such as forest conservation, renewable-energy development, and community-based sustainability initiatives.
Carbon-neutral browsing applies specifically to activity measured through Shift. That said, we don’t have control over the servers or infrastructure used by the websites, apps, or extensions that are active in Shift—and can’t make them switch to more renewable options. But we can offset their impact. Offsets, while they aren’t as effective as reduction, do allow us to take responsibility for the emissions within our scope. We’ll continue working to expand that boundary as technology and measurement improve.
We launched our offset strategy through a partnership with Carbonzero. Each of our carbon offset purchases is carefully vetted to be verified, traceable, and localized. Carbon offsets are a bridge, but our long-term aim is to reduce emissions at the source through optimization, design efficiency, and smarter data management.
Offsets aren’t the end goal, they’re part of the journey. Awareness → Measurement → Removal → Reduction
As a Climate Action Partner with Synergy Enterprises, we’re conducting continual GHG accounting and developing a reduction strategy aligned with the highest global standards.
You can learn about our journey to carbon neutrality on our Sustainability page.
AI and the future of our digital carbon footprint
AI has changed the landscape of digital emissions. Current research suggests that one AI query produces around four grams of CO₂ (roughly twenty times more than a Google search). Estimates vary, but emerging research suggests one AI query can emit around 4 g CO₂, while full model training can range from roughly 1,400 lb to 78,000 lb CO₂, depending on scale and tuning.
These numbers are helpful for awareness but should be treated as ballpark estimates, not precise values. Emissions per query vary widely depending on model size, hardware efficiency, data-center location, and local grid mix.
These figures are evolving rapidly, and we’re tracking research closely. When a peer-reviewed methodology for estimating AI emissions becomes stable and reproducible, we plan to integrate it into the Carbon Meter.
What’s next for the Carbon Meter?
Our long-term goal is to go beyond offsets toward actual reduction. That means cutting unnecessary data transfer, optimising caching, and designing more efficient browsing experiences. We’re also:
- Exploring region-specific grid-intensity factors (Global averages make the tool usable at scale but less regionally specific. We’re exploring localisation to reflect regional grid intensity)
- Collaborating with sustainability researchers
- Publishing our learnings so others can build upon them
There is a lot of research going into product design at Shift that didn’t conclude with the release of our Carbon Meter. As technology and the carbon emissions that come with it change, we are committed to updating how we offset and reduce right alongside it.
Start measuring your impact
The internet accounts for an estimated 2–4% of global greenhouse-gas emissions and growing fast. At this rate, our digital carbon footprint could be responsible for as much as 14 per cent of global emissions by 2040.
The Carbon Meter aims to make those emissions visible—not to assign blame, but because what we can see, we can improve. Awareness is the first step to meaningful change.
While individual choices help, real progress depends on collective pressure to decarbonize the infrastructure that powers the web, including data centers, networks, and hardware supply chains.
With the Carbon Meter, Shift helps users see their footprint and encourages developers to build a more climate-conscious and sustainable digital ecosystem. Individual action matters, but collective progress across tools, companies, and infrastructure is how we move the web toward a lighter future.
To measure your digital carbon footprint in Shift:
- Open Shift, click the leaf icon (in top right corner of default layout)
- See the carbon footprint of everything you do in Shift
- View how much CO₂ Shift users have helped offset
Every measurement represents one small but meaningful step toward a cleaner, more accountable internet. You can also take our 2-minute quiz to quickly understand the carbon impact of your internet habits.
