Skip to content
search-engine-optimizedby Netsolution
  1. Home
  2. Technical and performance
  3. Core Web Vitals Checker

Core Web Vitals Checker

The figures Google actually uses, measured on your visitors rather than simulated.

Data from Google’s CrUX API, aggregated over twenty-eight rolling days from real Chrome visitors who accepted statistics sharing. History available across twenty-five weeks. Free and unlimited.

What this tool is for

There are two ways of measuring a site’s speed, and only one counts for ranking. Lab simulation is useful for diagnosing and for testing a fix. But what Google uses is the measurements taken from real Chrome users.

That data cannot be simulated and cannot be gamed. It comes from real people, on their real devices, with their real connections — including those looking at your site on an entry-level phone in a poorly covered area.

That is what explains the disconcerting gap between a flattering lab score and an unfavourable Google verdict. The test machine had a perfect connection; your visitors did not.

The tie-breaking rule: when the lab and the field diverge, believe the field. It is what Google looks at, and it is what your visitors live through.

How to read the results

LCP measures the delay before the largest visible element paints. Good below 2.5 seconds, needs work up to 4, poor beyond. The usual causes are an image too heavy, a server slow to answer, or a font blocking the render.

INP measures the page’s responsiveness to interactions. Good below 200 milliseconds, poor past 500. The cause is always JavaScript monopolising the processor: consent banner, tracking scripts, third-party widgets. It is the metric most often failing since it replaced FID.

CLS measures layout shifts during loading. Good below 0.1, poor past 0.25. The cause is nearly always an image with no declared dimensions.

The distribution is worth more than the single value. A comfortable LCP at the 75th percentile coexisting with a fifth of visits well over the threshold signals part of your audience being poorly served, invisible in the headline figure.

The twenty-five-week history lets you tie a degradation to an event: a rebuild, adding a tracking tool, a change of hosting. That is often where the diagnosis is made.

Common mistakes

Fixing and checking the next day. The window is twenty-eight rolling days: your fix is diluted in four weeks of earlier data. Wait a week before concluding anything.

Looking only at desktop. Google has indexed from its mobile version since 2019. Mobile data is what counts, and it is almost always worse.

Confusing page data with domain data. When a page lacks visits, the tool falls back on the whole domain. That is useful, but it is no longer the measurement of your page: always check which source is shown.

Neglecting INP believing FID is enough. Plenty of sites were green on FID and went red on INP without changing anything. If you have not rechecked since March 2024, do it.

Optimising without measuring the real effect. The lab will say your fix works; only the field will say whether it changes anything for your visitors. Come back here after a few weeks.

Going further

Frequently asked questions

How does this differ from the speed test?

The speed test simulates a load on a test machine and gives you recommendations for fixing a page. This tool simulates nothing: it reports what your real visitors experienced over twenty-eight days, with the history across twenty-five weeks. One serves to diagnose and test, the other to know where you actually stand.

Why do I have no data?

Because your page does not receive enough Chrome visits for Google to build a statistically reliable sample. That is neither a penalty nor a fault. In that case the tool falls back on whole-domain data where it exists, otherwise you have to rely on lab measurements.

What does the 75th percentile mean?

That 75 % of your visitors had an experience at least as good as the value shown. Google deliberately set aside the mean, which a handful of very slow visits is enough to distort, and the median, too forgiving. The 75th percentile guarantees a clear majority of your visitors are well served.

How long before a fix shows up?

Count on up to twenty-eight days for the full effect, since the rolling window gradually replaces the older visits. An improvement starts to show after about a week. It is frustrating, but that is the price of real rather than simulated data.

Do Core Web Vitals really influence ranking?

Yes, Google has confirmed it, with moderate weight. Between two pages of equivalent relevance, the faster one goes ahead. But no technical optimisation will lift a page that does not answer the question asked. These metrics separate, they do not qualify.

Has FID gone?

Yes, it was replaced by INP in March 2024. FID only measured the delay before the first interaction, which artificially flattered slow but barely interactive sites. INP takes every interaction of the visit into account: it is markedly stricter and far more honest.