TL;DR: Michelin Green stars restaurants do, on average, run lighter and lower-carbon websites than a random control group in the 2026 dataset, but the gap is not as clean as people assume.
The headline for 2026 is this: Green Star sites average about 1.92g eCO₂ per visit versus 2.29g for the control group, and they are roughly 20 per cent lighter by page weight. However, the control group contains more “ultra-lean” outliers, while also containing the very heaviest pages.
If you are a UK restaurant, the practical takeaway is not “win a Green Star then your website is fine”. It is that digital sustainability is a solvable technical problem, and the 2026 data shows exactly where the bloat tends to hide.
Key Takeaways
- Some of the worse performing websites in the 2025 Michelin Green star study have made significant improvements; Moor Hall, Daylesford Organics, & Homestead kitchen to name a few, but there’s still work to do to match places like Chapters & Restaurant Sat Bains.
- In the 2026 dataset, Michelin Green Star restaurants average around 1.92g eCO₂ per homepage visit versus 2.29g for the control group, indicating a lower mean digital carbon footprint.
- The control group contains both more very low-carbon sites and the most extreme high-carbon sites, so medians and outliers matter as much as averages.
- ECOgrader letter grades do not track perfectly with grams of eCO₂, because grades also reflect broader performance and best-practice signals beyond page weight.
- A sub-1MB restaurant homepage at roughly 0.33g eCO₂ per visit is achievable in practice when you design for performance first.
- The industry trend is for webites to increase in size over time, “Webpages are nearly four times larger than they were in 2010.” – HTTP Archive
What are Michelin Green stars, and why does digital footprint now matter?
Michelin Green stars recognise restaurants with leading sustainability practices, but your website can quietly undermine that story if it is heavy, slow, and energy-hungry, not only is it doing additional environmental damage – but site speed is a ranking factor for SEO & SERPs.
Michelin Green stars have been a big signal to UK diners since 2020: provenance, seasonality, waste, and “how you run the place” all matter, not just what is on the plate.
Digital is now part of “how you run the place”. If your booking journey loads 30MB of imagery and scripts, the sustainability claim becomes inconsistent, even if your kitchen is doing everything right.
At QED web design, a lot of our hospitality work sits at the messy intersection of brand, performance, and real-world site maintenance. If you want the practical background on designing for lower-carbon performance, start here: sustainable web design.
Michelin Green stars 2026: what does the website carbon data show?
In the 2026 dataset, Green Star restaurants average lower eCO₂ per visit and lower page weight than the control group, but ECOgrader letter grades still skew heavily towards E.
As of the 2026 dataset, Michelin Green Star restaurants have a lower average homepage carbon footprint than a randomised control group, but most still fail modern environmental best-practice grading.
The 2026 summary looks like this:
| Metric (2026 dataset) | Michelin Green Stars | Control group (non-Green Star) |
|---|---|---|
| Sample size | 37 sites | 34 sites |
| Average eCO₂ per visit | ~1.92g | ~2.29g |
| Median eCO₂ per visit | ~1.49g | ~1.04g |
| Average page weight | ~5.22MB | ~6.50MB |
| % graded E | ~83.8% | ~73.5% |
Two quick, human-readable anchors from 2026 help you feel the range. Among Green Stars, Chapters comes in at roughly 0.37g with a sub-1MB page, while Osip sits around 5.87g with a ~15MB landing page (increasing from 14.93MB in 2025).
If you want the year-on-year context, the 2025 methodology PDF is still the reference baseline for how we are measuring websites: Study of Michelin sustainable restaurant websites 2025.
Michelin Green Star changes from 2025
How the landscape has changed, since our 2025 Michelin Green stars study was published last year, there have been some notable changes. Michelin announced seven new Green stars in 2026:
- 1887 – Torridon,
- Eight at Gazegill by Doug Crampton,
- Forest Side,
- Glebe House,
- Knepp Wilding Kitchen,
- The Free Company,
- Timberyard
Whilst deletions were restricted to just five, mainly for closures, it shows that the French Tyre company is looking closer a businesses which want to make the sustainability claim:
- Le Manoir Aux Quat Saisons,
- Crocadon,
- Tillingham,
- Silo,
- Henry Robertson – Palé Hall
Those that have made significant improvements since the 2025 release.
| Restaurant | 2025 data (page size/eCO₂ per visit) | 2026 data (page size/eCO₂ per visit) |
|---|---|---|
| Daylesford Organics | 17.18MB / 6.7g | 8.84MB / 3.44g |
| Homestead kitchen | 11.99MB / 4.67g | 7.76MB / 3.02g |
| Moor Hall | 28.7MB / 9.68g | 3.99MB / 7.34g |
| Kai | 12.55MB / 4.89g | 8MB / 3.12g |
| Hand & Flowers (from the control group) | 16.48MB / 5.5g | 2.36MB / 0.8g |
However, there are a few websites which seemed to buck the weight loss trend, significantly;
- Black Swan. They increased their already weighty website from being 38.58% heavier than the Hospitality average, to 89.68% heavier.
- Osip is 290%+ heavier than the sector average, increasing their footprint from being 285% heavier in 2025
Our client, Ancient Shepherds by Mark Poynton, has risen to a Grade B as updates and upgrades have reduced the impact. Their image heavy website has a footprint of 0.23g eCO₂ per visit from a page weighing only ~690kb (0.69MB)
The full list:
| Restaurant | Page Weight (MB) | vs Sector Average (3.875 MB) | Grams CO₂ per Visit | Ecograder Report |
|---|---|---|---|---|
| Osip | 15.13 | 290.45% heavier | 5.87g | View Report |
| Angela’s | 2.60 | 32.90% lighter | 1.01g | View Report |
| Interlude | 5.24 | 35.23% heavier | 2.04g | View Report |
| Wild Shropshire | 6.51 | 67.97% heavier | 2.20g | View Report |
| Forge | 9.97 | 157.29% heavier | 3.36g | View Report |
| The Small Holding | 2.18 | 43.77% lighter | 0.85g | View Report |
| Marle | 5.86 | 51.10% heavier | 1.98g | View Report |
| Jericho | 3.23 | 16.65% lighter | 1.09g | View Report |
| Chapters | 0.95 | 75.48% lighter | 0.37g | View Report |
| Inver | 2.19 | 43.48% lighter | 0.85g | View Report |
| The Whitebrook | 3.82 | 1.42% lighter | 1.29g | View Report |
| Where the light gets in | 3.82 | 1.42% lighter | 1.49g | View Report |
| Wilsons | 3.13 | 19.23% lighter | 1.22g | View Report |
| The Black Swan | 7.35 | 89.68% heavier | 2.48g | View Report |
| Coombeshead Farm | 7.29 | 88.13% heavier | 2.46g | View Report |
| L’Enclume | 7.63 | 96.90% heavier | 2.57g | View Report |
| St. Barts | 10.32 | 166.32% heavier | 4.02g | View Report |
| Pythouse Kitchen Garden | 9.78 | 152.39% heavier | 3.81g | View Report |
| Restaurant Sat Bains | 1.06 | 72.65% lighter | 0.41g | View Report |
| Daylesford Organic Farm | 8.84 | 128.13% heavier | 3.44g | View Report |
| Exmoor Forest Inn | 1.55 | 60.00% lighter | 0.60g | View Report |
| Apricity | 2.60 | 32.90% lighter | 0.88g | View Report |
| Homestead Kitchen | 7.76 | 100.26% heavier | 3.02g | View Report |
| Oak | 2.26 | 41.68% lighter | 0.76g | View Report |
| Petersham Nurseries Café | 4.29 | 10.71% heavier | 1.45g | View Report |
| Moor Hall | 3.99 | 2.97% heavier | 1.34g | View Report |
| ANNWN | 4.62 | 19.23% heavier | 1.80g | View Report |
| Culture | 2.79 | 28.00% lighter | 0.94g | View Report |
| Pine | 1.90 | 50.97% lighter | 0.64g | View Report |
| Kai | 8.00 | 106.45% heavier | 3.12g | View Report |
| 1887, Torridon | 6.93 | 78.84% heavier | 2.70g | View Report |
| Eight at Gazegill by Doug Crampton | 2.28 | 41.16% lighter | 0.77g | View Report |
| Forest Side | 7.82 | 101.81% heavier | 3.05g | View Report |
| Glebe House | 7.48 | 93.03% heavier | 2.92g | View Report |
| Knepp Wilding Kitchen | 6.72 | 73.42% heavier | 2.27g | View Report |
| The Free Company | 3.62 | 6.58% lighter | 1.41g | View Report |
| Timberyard | 1.72 | 55.61% lighter | 0.58g | View Report |
Are Michelin Green stars websites actually better than the control group?
On average, yes, but “better” depends on whether you care about the mean, the median, or the worst-case outliers.
This is where people get it wrong. A common misconception is that sustainability leadership in the dining room automatically translates into a low-impact website. It does not, and the data makes that painfully obvious.
In 2025, the Michelin Green Star group averaged 2.38g eCO₂ per visit versus 2.64g for the control group, but 85.7% of Green Star sites still landed a Grade E rating. In other words: “slightly better than peers” can still be “objectively poor”.
Mean vs median (why people misread the story)
The control group’s median is lower because it contains more ultra-lean sites, but its mean is higher because it also contains extreme heavyweight sites.
In 2026, the control group includes several genuinely lean restaurant sites in the 0.14g to 0.38g range, which drags the median down. It also contains the worst offenders, including pages above 30MB and above 11g eCO₂ per visit, which pushes the mean up.
That is why it is possible for Green Stars to “win on average” while still looking unimpressive when you compare best-in-class examples across the whole market.
Why ECOgrader grades can disagree with grams
Grams of eCO₂ correlate strongly with page weight, but letter grades also reflect broader technical signals, not just carbon output.
In your 2026 data, Green Stars have lower average grams, yet a higher share of Grade E than the control group. That looks contradictory until you remember that tools like ECOgrader assess a bundle of factors including performance patterns, third-party behaviour, and best-practice hygiene, not simply “how many grams”.
Practical implication: you can reduce grams by slimming images, but still score badly if the page is script-heavy, blocking, or built in a way that creates unnecessary computing. In restaurant websites, that often comes from booking widgets, marketing tags, and theme bloat.
How was the digital carbon footprint measured, and what are the limitations?
The study uses consistent desktop homepage testing with ECOgrader-style outputs, benchmarked against a hospitality-sector average page weight.
The 2025 methodology has been reused to give the consistency that Michelin inspectors demand, this standardises the testing protocol: desktop checks, homepage or landing page only, and assessments run in a tight time window to reduce day-to-day variation.
Benchmarks matter: We anchor to a hospitality sector “average landing page weight” of 3.875MB and a global “average eCO₂ per page view” of 0.8g That gives readers a reference point that is not just “restaurant A vs restaurant B”.
Here’s a reader-friendly explanation of what “eco website design” actually means in practical terms: eco website design.
Why Michelin Green stars websites still get heavy (and how to fix it)
The biggest drivers are oversized photography, unnecessary scripts, and video-first design choices that prioritise aesthetics over efficient delivery.
Restaurant sites are unusually prone to weight gain because the default brief is “make it feel premium”. In practice, that often means 4K hero images, autoplay video, layered animation, and a pile of marketing tags on top.
If your homepage is above 5MB, you are paying for every extra megabyte on every visit, in carbon, speed, and lost conversions.” It is not moralising, it is just physics and bandwidth. Google have also, recently announced a new ceiling on their crawl limits – this might only affect websites with huge amounts of HTML, but is it really worth a risk?
The fixes are boring, and that is the point:
- Image discipline: convert to WebP or AVIF, enforce sensible dimensions, and stop shipping full-resolution photography to mobile.
- Third-party audit: remove anything that is not pulling its weight (live chat, heatmaps, unused pixels, redundant font loaders).
- Theme and builder control: cut template bloat, avoid “one page does everything” feature stacks, and ship only what the page needs.
- Video restraint: use posters, lazy load, or move video off the critical path.
Another misconception worth correcting: “Dark mode websites are always greener.” That is not reliably true in the real world, because device display technology, brightness, and user behaviour dominate. Do not treat ‘dark mode’ as a silver bullet when page weight and scripts are the real drivers.
For a practical view of how this fits into a build or rebuild workflow, link your process page as a supporting post: our web design process.
What should a UK restaurant do next?
Treat digital sustainability like any other operational sustainability project: measure, prioritise the biggest waste, then lock the gains into process.
If you are running a UK restaurant, you do not need a perfect website to improve. You need a baseline, a short list of fixes, and the discipline not to re-break performance every time someone uploads new photography.
A simple, high-return sequence (and it works whether you are a Michelin Green Star or not):
- Measure your homepage weight and grams, then measure one booking or menu journey.
- Ship image fixes first, because that is usually the biggest chunk of waste in hospitality.
- Cut third-party scripts down to what you can defend commercially.
- Re-test quarterly, especially if you are running seasonal menus, events, or PR spikes.
As a grounded example from your existing published work: your methodology PDF includes a QED-built restaurant website achieving approximately 0.33g eCO₂ per visit and ~988KB page weight, which is stronger than every Green Star website in the 2025 cohort.:contentReference[oaicite:7]{index=7} The point is not bragging rights, it is proof that the benchmarks are achievable with normal tools and a performance-first brief.
Conclusion
Michelin Green stars correlate with better average website carbon performance in 2026, but the majority of sites still have avoidable digital waste.
The honest read of the 2026 dataset is not “Green Stars are failing”. It is that hospitality as a whole is still shipping heavyweight websites by default, even when the brand values are explicitly sustainability-led.
If you want your website to match your sustainability story, pick one measurable target (page weight, grams, or the heaviest user journey), fix it, and make it part of how you run the business. That is how you stop this becoming a once-a-year audit exercise.
Sources
- Michelin Guide, Michelin Green Star awards.
- Wholegrain Digital, Website Carbon Calculator methodology and global average page emissions.
- HTTP Archive, web page weight and performance datasets used for benchmarking (ongoing project, updated regularly).
- ECOgrader, environmental website assessment approach and outputs .
- GTmetrix, performance testing and page weight reporting.
- WebPageTest, performance testing methodology and metrics
- QED Web Design, “Study of Michelin Sustainable restaurant websites 2025” methodology PDF.
If you want a UK-based team to cut digital waste and lower your website’s carbon footprint.
