Software · 2026
The Dev Lab site
The site you are on. We built it under the same rules we apply to client work, and measured everything we promised.
- Client
- Dev Lab
- Year
- 2026
- Discipline
- Websites
- Timeline
- ongoing
The problem
A development studio's own site is its main case study. It has to look expensive and open fast, and those two usually pull in opposite directions: impressions are normally paid for in page weight. We wanted both, in three languages, with real SEO underneath.
What we did
We fixed a performance budget before starting and treated it as a requirement rather than an aspiration. The immersive scroll runs on native CSS scroll-driven animations instead of a library, so eight narrative sections added under a kilobyte of JavaScript. The WebGL hero loads as a separate chunk after first paint and never starts where it is not wanted: on phones, under reduced motion, and without WebGL support, a still frame of the same scene takes its place.
The outcome
Twice during the build a feature did not fit the budget. Both times we rewrote the feature rather than raising the limit: the animation library gave way to an IntersectionObserver and CSS transitions, and the scroll library turned out not to be needed at all. Every number below was measured on a production build, not estimated.
Stack
- Next.js 16
- React 19
- TypeScript
- Tailwind v4
- three.js
In numbers
- JavaScript on first loadbudget was 180 KB
- 172 KB
- LCP with the 3D scene on the pagetarget was under 2.0 s
- 1.2 s
- WCAG 2.1 AA violationschecked with axe-core
- 0
- languages, statically generated
- 3