Core Web Vitals
Core Web Vitals, measured and explained
LCP, CLS and the lab stand-in for responsiveness, each with the number, the threshold it is being judged against, and what on the page is producing it.
Three audits every 24 hours with a free account. No card, and nothing switches off.
What it measures
Checked on every run
The vitals are three specific measurements with published thresholds, which makes them unusually honest as far as ranking signals go: you can measure them, you can see whether you pass, and nobody has to guess. What they are not is a ranking strategy.
Largest Contentful Paint
When the biggest thing above the fold finishes rendering. Good is under 2.5 seconds. It is an image or a web font far more often than it is a server.
Cumulative Layout Shift
How much the page moves about while it loads. Good is under 0.1. Almost always an image or an ad slot with no reserved space, or a font swapping in late.
Total Blocking Time
How long the main thread is too busy to respond to a tap. It is the lab stand-in for Interaction to Next Paint, which needs a real visitor to measure. Good is under 200ms.
Speed Index
How quickly the page fills in visually, rather than when one particular element lands. Useful when LCP looks fine and the page still feels slow.
How it works
Three steps, about ninety seconds
Paste the address
Any public page. The engine resolves it, follows redirects and reports where it ended up.
It loads the page twice
Once at a phone viewport on a throttled connection, once on a desktop, in a real browser — not a lookup against a database of old crawls.
You get numbers, ranked
Four scores, the measurement behind each, and the problems in the order they are costing you. Results stay in your dashboard, so a re-run shows what moved.
Questions
What this does, and what it does not
- Why does this show Total Blocking Time and not INP?
- Because Interaction to Next Paint measures a real person tapping a real thing, and there is nobody in a lab run to do that. TBT is the measurement that correlates with it and can be taken from one load. If TBT is high, INP is very likely to be too.
- I pass here but fail in Search Console. Which is right?
- Both, about different things. This is one load on a simulated device; Search Console is 28 days of real visitors, including the ones on a five-year-old phone in a basement. Field data is the one that counts for ranking — but it lags by weeks, which is why you test fixes here first.
- How does the audit actually run?
- We load the page in a real browser twice — once on a simulated mid-range phone on a slow connection, once on a desktop — and record what happens. Nothing is estimated from a database of past crawls; every number comes from that load, on the day you asked for it.
- Is it free?
- Yes. Three audits every 24 hours with a free account, which exists so the engine is not run by scripts. There is no card, no trial, and nothing switches off.
Measure it now
One address, ninety seconds, and a list you can hand to whoever builds the site. If the list turns out to be longer than an afternoon, that is a conversation we are happy to have.