Core Web Vitals explained simply: LCP, INP and CLS, and how to actually read them

Published on 4 November 2026by Tauro Lab8 min read

In short Core Web Vitals are the three measures Google uses to judge the experience on a page: how late the main content appears (LCP, good within 2.5 seconds), how late the page reacts to a tap (INP, within 200 milliseconds) and how much it shifts on its own (CLS, within 0.1). What counts is real-user data from the last 28 days, at the 75th percentile. You can read them for free in PageSpeed Insights, in the mobile view.

You open a website on your phone, you are about to tap a button and at that moment the page shifts on its own: you tap something else. Or you tap the menu and nothing happens for a second, so you tap again. Your customers know these annoyances well, and Google measures them with three numbers called Core Web Vitals.

You do not need to be technical to read them. You need to know what each one measures, which number to look at in PageSpeed Insights (not the big one), what makes them worse and what to ask whoever looks after your site.

Core Web Vitals: the three measures and their thresholds

Measure	What it tracks	Good	Needs improvement	Poor
LCP	How long the largest element visible on opening takes to appear, usually the top photo or the heading	up to 2.5 seconds	up to 4 seconds	over 4 seconds
INP	How long the page takes to react to taps, clicks and key presses throughout the visit	up to 200 milliseconds	up to 500 milliseconds	over 500 milliseconds
CLS	How much the page shifts on its own while loading or while you use it	up to 0.1	up to 0.25	over 0.25

The thresholds come from web.dev and the Search Console Core Web Vitals report, and they are measured at the 75th percentile: at least three visits out of four must fall within the limit. A page that feels fast to you can be slow for someone on an old phone with a weak connection.

LCP stands for Largest Contentful Paint and CLS for Cumulative Layout Shift: acronyms you do not need to remember. INP, Interaction to Next Paint, replaced the old responsiveness measure in 2024. Web.dev says stable metrics change no more than once a year, and in October 2026 the three are still these: if a guide presents other 'new Core Web Vitals', check the source.

Google counts them among the page experience signals its ranking systems seek to reward. They do not replace useful content: a very fast, empty site will not rise. And to appear in AI Overviews and AI Mode, Google says no additional requirements apply: the usual rules hold, an indexed page and helpful content. Speed buys nothing special, but a page that does not load gets read by nobody.

How to read PageSpeed Insights in five minutes

Open PageSpeed Insights, paste the address of your home page and of a page that sells, a service or a product. Look at the mobile view: Google indexes mostly the mobile version of websites.
At the top is the real-user section: actual data from the last 28 days, at the 75th percentile. If it says the page passes the Core Web Vitals assessment, you are fine on this front. Other numbers appear there too, such as FCP and TTFB: they are supporting metrics, not Core Web Vitals.
If there is no real-user data, the site has few visits. Use the section below, the lab test, as a guide: it changes on every run, so run it three times and look at the trend.
The big coloured number is the lab score. It is not a Core Web Vital and it is not what Google uses for page experience. A 70 with all real-user data green is worth more than a 95 with a real LCP of 4 seconds.
Scroll down to the diagnostics: they name the LCP element and what weighs on the page. You do not have to understand all of it, you have to copy it to whoever manages the site.

If you have connected Search Console, the Core Web Vitals report groups pages with the same problem and tells you whether a defect affects one page or the whole site. Pages with little data do not appear: that is normal, not an error.

Then the hands-on test: take a phone that is not new, open the site on your everyday connection, tap the menu, scroll, press the main button. What you feel, your customers feel.

From the annoyance to the cause: what makes each measure worse

What you feel	Measure	Typical causes	What to ask
The top photo arrives after everything else	LCP	Heavy image, slow server, scripts loading before the content, deferred loading applied to the main image too	Is the main image light and loaded straight away, not deferred?
You tap the menu and nothing happens for a moment	INP	Too many scripts at once: chat, advertising pixels, social widgets, maps, scroll effects	Which scripts does the site load, and which can we remove?
You are about to tap and the page shifts on its own	CLS	Images without declared dimensions, a cookie banner pushing the content, embedded videos and maps, fonts changing once loading ends	Do we reserve the space before the content arrives?

Sliders on the home page weigh on LCP because they load several photos at once; every widget added because 'it is light anyway' weighs on INP, because they add up and a cheap phone struggles more than a computer. For image weight, format and dimensions, what we explain in how e-commerce product photos should look applies to any site.

The four questions for whoever manages your website

You do not need to know how it gets fixed. You need to know what to ask:

What are my LCP, INP and CLS on mobile, using real-user data?
Which element causes the worst value, and on which pages?
What would you change, how much work is it, and by how much do we expect the number to move?
After the change, how and when do we measure again? Real-user data needs 28 days to refresh.

If the answer is a list of technical terms without a number before and a number after, ask for it again in plain words.

Sometimes the site is slow because of how it was built, for example with many plugins or a heavy theme, and tweaking is not enough. The comparison of WordPress and custom development and the guide on when to redesign your website help you judge when starting again is worth it.

How much they matter, and when not to chase them

If all three real-user values are already in the good range, do not spend money chasing the top lab score: it is not the goal and Google does not look at it. Use the time for clearer copy and pages that answer what customers actually ask.

If one is outside the threshold, it is worth fixing regardless of Google: a person waiting in front of a page that jumps around leaves, and you have lost the customer before saying a word.

Note the three numbers today

Open PageSpeed Insights with your home page and one important page, mobile view, and note the real-user LCP, INP and CLS. If all three are within the thresholds, you are done for now. If one is not, send it to whoever runs the site together with the four questions and the diagnostics line.

Tauro Lab works on websites and SEO. If you have nobody to ask, you can write to us with the three numbers and we can look at them together.

Frequently asked questions

Does a high PageSpeed score mean I am fine?

Not necessarily. The big number at the top is a lab score from a single run. Core Web Vitals are assessed on data from real visits, which can say something different. Always look at that first, when it is there.

Why do the numbers change every time I measure?

The lab test depends on the moment, on how the server responds and on what the page loads at that instant, such as ads and external scripts. Run it several times and look at the trend. Real-user data is steadier because it sums up 28 days of visits.

Is installing a plugin enough to speed up my site?

Sometimes it helps, for example by keeping a stored copy of your pages, but it does not remove what is heavy: huge photos, too many scripts, a heavy theme. It can even add weight or break something. Find the cause in the diagnostics first, then choose the cure.

My site has few visits and I cannot see real-user data: what do I do?

That is normal: real-user data only appears once there are enough visits, and the same goes for the Search Console report. Use the lab test as a guide and try the site by hand on a phone. When visits grow, the real data arrives on its own.

Sources

Written by the Tauro Lab team, a web agency and software house in Turin. About us