This site, measured honestly
I tell people to instrument before they optimise. It would be a poor look not to do it here. These are Core Web Vitals collected from real visits to this site — not a lab score from an ideal machine on an ideal network.
p75 across recent visits
0 samples · refreshing every 5sThis visit, measured in your browser
- TTFB
- measuring…
- FCP
- measuring…
- LCP
- measuring…
- CLS
- measuring…
- INP
- interact with the page…
Build
- Commit
- 026f1c1
- Branch
- main
- Context
- production
- Built
- 2026-09-16T05:46:23.482Z
- Rendered
- 2026-10-18T03:41:18.310Z
What this is, and what it is not
Every visitor's browser reports LCP, INP, CLS, TTFB and FCP to a small endpoint on this site. The figures above are the 75th percentile of those samples — the percentile Core Web Vitals is actually assessed on, because an average hides the slow tail that people actually complain about.
Samples are stored in Redis, so they survive a deploy and are shared across every server instance — this figure is the real p75 across recent visits, not just whichever instance happened to answer your request.
Nothing identifying is collected: a metric name, a number, and the path. No cookies, no fingerprint, no third party.
The submission endpoint is rate-limited per IP, so one script cannot dominate the sample the percentile below is computed from.
Let's build something worth maintaining
Have a platform problem, a team that needs a hand, or just want to argue about API design? My inbox is open.