Home
/- Largest paint
- –
- Layout shift
- –
- Blocking time
- –
- Download
- –
Performance over recent releases
// under the hood
We say we build software properly. This page is the proof for the site you're on. Every release is built, checked and measured by the same pipeline, and these are its figures for the release that's live right now, not a screenshot from a good day.
// lighthouse
Google's Lighthouse, run on its phone setting against the public site after each release ships: three runs per page, middle result kept. That is the setting most sites score worst on.
Performance over recent releases
Performance over recent releases
Performance over recent releases
Performance over recent releases
// the build
Counted from the finished build itself, the same files the server sends. Pages are plain HTML with their styles inlined, so a visit is one request before anything renders.
// every release
From the moment a change is pushed to the moment it's live, every step runs on its own. If one fails, everything after it stops, and nobody has to remember to look. The one step a person takes is deciding to release.
01
02
03
04
05
06
// security headers
Five instructions sent with every page, read from the live site when it was last measured. A release that lost any of them would have failed its checks.
strict-transport-securitynot measuredBrowsers only ever connect over HTTPS
x-content-type-optionsnot measuredFiles are only used as the type they declare
x-frame-optionsnot measuredOther sites can’t embed these pages in a frame
referrer-policynot measuredOther sites see where a visitor came from, never the full address
permissions-policynot measuredCamera, location and payments are switched off; the microphone works only for the voice demo
The same habits go into the agents and software we build for contractors: tested on every change, measured in production, and fixed by the people who built them.