Core Web Vitals monitoring
LCP, INP and CLS, measured per template rather than per site, and tracked over time so a regression has a date and a cause. Start with one page, no account.
The free run measures response time for the URL you enter. Vitals need a real browser, so they start on a monitored site.
What these three numbers actually mean
Vitals are three separate complaints wearing one name: the page took too long to appear, it moved while I was reading, it ignored my tap. Each has a different cause and a different owner, so they are reported separately and per template — a slow product page is not a slow site.
- LCP · Largest paint
Main content visible. Good under 2.5s, poor over 4s.
- INP · Interaction
Response to a tap. Good under 200ms, poor over 500ms.
- CLS · Layout shift
Movement while loading. Good under 0.1.
- Per template
Home, category, product and article measured as separate things.
What a result looks like
Was 2.1s before the release on the 14th.
Comfortably good. Nothing to do here.
An ad slot with no reserved height.
Each metric names the element responsible — the image that is the LCP, the node that shifted, the handler that blocked. A single performance score would average a healthy checkout with a slow blog and hide both.
A measurement tells you about today. Monitoring tells you at 04:00.
Vitals degrade in increments no one notices: a font swap, a third-party tag, an image uploaded at full resolution. By the time a quarterly report shows the decline, a dozen releases could be responsible. Continuous measurement makes it one release, named, on the day it shipped.
Common questions
- Is this the same engine as the paid monitoring?
- Yes — the same checks, run the same way. The difference is that monitoring repeats them on a schedule and keeps the history, which is what turns a number into a trend and a trend into an alert.
- Does testing my site slow it down or change anything?
- No. Everything is read from the outside, the way a visitor arrives — no agent, no script on your pages, nothing written. Our requests identify themselves so you can recognise them in your own logs.
- Is this lab data or what real users experience?
- This is lab data — the same test, the same conditions, every time, which is what makes a regression provable. Field data is better for judging your true audience and worse for pinpointing a change, so we treat them as different jobs.
- Why does this differ from PageSpeed Insights?
- Different conditions and a different question. We hold the test conditions fixed and compare against your own history, rather than scoring you against a general benchmark on whatever connection the run happened to get.
Related tests
A free test is a snapshot. Monitoring is the product.