FoxStudio
WorksLabStudioTeamJournalContact
FR·EN·IT

FoxStudio

Subsidiary of FoxCase.

Navigate

  • Works
  • Lab
  • Studio
  • Team
  • Journal

Contact

  • hello@foxstudio.fr
  • Cannes, FR
  • Ombrys

Legal

  • Mentions
  • Privacy
  • Footprint

Cannes, FR · GMT+1MEASURING…v0.1.0
03▸Footprint≤ 0,20 g CO₂ / view

What this page costs.

We announce a carbon budget, measure it, and put it on the page. No greenwashing, no yearly average — the real footprint, per view.

01▸Methodology

We use the Sustainable Web Design (SWD) v3 simplified model: 0.81 kWh per gigabyte transferred, times the global average grid intensity (442 gCO₂e / kWh). That works out to ≈ 358 gCO₂e per gigabyte, or 3.33 × 10⁻⁷ gCO₂e per byte.

We measure server-side at build time: the sum of gzip chunks actually loaded by each page. It's conservative (we assume everything is served cold, no browser cache).

02▸Current measurements

Measured in your browser right now — fetching each route + linked assets, summing bytes, converting via the SWD v3 coefficient (3.33 × 10⁻⁷ gCO₂e / byte).

Home /MEASURING…
Works /worksMEASURING…
Project /works/[slug]MEASURING…
Lab /labMEASURING…
Studio /studioMEASURING…
Admin CMS /admin735 kB · reserved for the editorial team, excluded from the public budget.
03▸Budget

Our target is ≤ 0.20 gCO₂e per view (specs §7.2). Real values sit in the section above — measured in your browser, not estimated by a third party.

For context, the 2024 web page median transfers ≈ 2.5 MB (HTTP Archive), roughly 0.83 gCO₂e per view — about four times our target.

The budget is enforced in CI: on every PR, the `Performance` workflow runs `scripts/measure-carbon.ts`, which gzips the home chunks and exits 1 over 0.20 g. See `.github/workflows/perf.yml`.

04▸Hosting
ApplicationVercel — cdg1 region (Paris). Not certified low-carbon, offset by an aggressive transfer budget and global edge cache.
MediaCloudflare R2 — certified low-carbon network, zero egress fees (therefore zero incentive to push load elsewhere).
DatabaseVercel Postgres or Neon, eu-central-1 (Frankfurt). On-demand wake-up (serverless cold start ~ 200 ms).
05▸Beyond transfer

The SWD model only covers the use phase (network + device + datacenter). We don't measure — for now — manufacturing of visitor devices, nor server hardware amortisation. Our estimate is therefore a floor.

To go further: assets cached aggressively, dynamic imports systematic, dark mode by default (OLED savings), zero video autoplay. It's all documented in specs §7.