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

The three thresholds, the gap between field and lab data, and which row to look at in Search Console and PageSpeed Insights, step by step.

· 11 min readSEO

Short answer

Core Web Vitals are three measures Google takes from real visitor data: LCP for loading, INP for response to interaction and CLS for visual stability. The good thresholds are 2.5 seconds, 200 milliseconds and 0.1. Judge them by the 75th percentile of field data in Search Console and PageSpeed Insights, not by the lab score.

You open the Core Web Vitals report in Search Console and see a red row on the mobile tab saying that CLS is more than 0.25. Your developer, meanwhile, says the PageSpeed score is 92. Both can be right, because the two screens look at different data. The useful answer to what Core Web Vitals are is less the spelling-out of three abbreviations than knowing which number to read, and where.

This article gives the order for reading LCP, INP and CLS. We covered the fixing side in how to speed up a website and the effect of speed on sales in why a fast site sells more. Here we look only at reading and diagnosis.

The three measures and their thresholds

Core Web Vitals are three measures Google uses to see how a page looks to real visitors: loading, interaction and visual stability. The table below summarises the ranges in Search Console Help and the PageSpeed Insights documentation.

Measure What it measures Good Needs improvement Poor
LCP Time from the page request until the largest element in the viewport is rendered 2.5 seconds or less Up to 4 seconds Over 4 seconds
INP Response time of the slowest click, tap or keyboard interaction during the visit (outliers are ignored) 200 ms or less Up to 500 ms Over 500 ms
CLS A unitless score for unexpected layout shifts 0.1 or less Up to 0.25 Over 0.25

Which slice of the data is read matters as much as the number. Google reports the 75th percentile of visits for each measure. So if LCP shows 2.5 seconds, up to a quarter of visits may be slower than that. A page or site with enough data for all three measures passes only when the 75th percentile of all three is “good”.

The reading order: four stops

Most confusion comes from asking the right question on the wrong screen. We suggest this order. First find the weak group in Search Console. Then read one page from that group in PageSpeed Insights using field data, look for the cause in lab data, and wait after a single change.

Reading order for Core Web Vitals: the group in Search Console, field data in PageSpeed Insights, the cause in the lab, a fix and 28 days of tracking 1 Group Weakest row in the Search Console report 2 Field 75th percentile in PageSpeed Insights 3 Lab Lighthouse shows the cause 4 Fix Make one change, then wait Confirm with 28 days of tracking in the same report Reading order for Core Web Vitals: the group in Search Console, field data in PageSpeed Insights, the cause in the lab, a fix and 28 days of tracking 1 Group Weakest row in the Search Console report 2 Field 75th percentile in PageSpeed Insights 3 Lab Lighthouse shows the cause 4 Fix Make one change, then wait Confirm with 28 days of tracking in the same report
Figure: the order for reading Core Web Vitals; the group in Search Console first, then field data, then the cause in the lab and 28 days of tracking.

1. Find the weak group in Search Console

The Core Web Vitals report is based on real-user data and shows mobile and desktop separately. Its rows are URL groups made up of similar pages. According to the help page, a group’s status is the status of its weakest measure: if CLS is poor and INP is good, the group shows as “poor”.

Keep these points in mind when reading it:

  • The report covers only indexed URLs and does not list all of them; it shows a sample of pages.
  • If a group lacks enough data for LCP and CLS, it does not appear in the report at all. Not appearing does not mean “good”.
  • Data is combined from all countries. If much of your traffic comes from a country with slow connections, the overall values fall.
  • The report is not designed to find the status of one URL; it is for seeing the site as a whole and catching problems that affect many pages.

Click a weak row and copy one of the example URLs. Pages in the same group most likely share a template and a common cause, because Google assumes the pages in a group share a similar structure.

2. Read the field data in PageSpeed Insights

Paste the copied address into PageSpeed Insights and look first at the real-user section at the top. According to Google’s documentation, this section shows the last 28 days of data and is updated daily. Each measure has a green, amber and red distribution bar. For example, if you see 11% in the amber bar for LCP, 11% of observed LCP values fall between 2500 and 4000 milliseconds. The figure above the bar is the 75th percentile, and the assessment is made on that.

Three things to watch for:

  1. URL or origin? If the page lacks enough data, PageSpeed Insights falls back to the origin level, which covers all pages of the site. web.dev also advises checking which data is showing each time; otherwise you can waste time chasing another page’s problem.
  2. The pass rule. If the 75th percentiles of all three measures are good, the page passes. If INP has too little data, it still passes when LCP and CLS are good. If LCP or CLS has no data, the page cannot be assessed.
  3. The gap with Search Console. PageSpeed Insights usually shows a single URL, while Search Console shows the group. A single URL can be an outlier in its group, so the two may not match. PageSpeed Insights also strips the parameters from the address; Search Console does not.

3. Use lab data to find the cause

The lower section of PageSpeed Insights is Lighthouse’s lab test. According to Google, Lighthouse simulates a mid-tier device on a mobile network for mobile, and a wired connection for desktop. A score of 90 or above is good, 50 to 89 needs improvement and below 50 is poor. Two runs on the same page can differ; local network and device load are among the reasons.

Google’s documentation says lab data is useful for debugging and field data for seeing the real experience. Good lab data does not mean real visitors are having a good experience.

The lab behaves differently for each of the three measures:

  • LCP: the lab measures one load on one device, while the field collects real visits across many devices and connections. If the field value is much worse, we suggest thinking about visitors on slower devices and connections.
  • CLS: field data covers the whole life of the page, while the lab usually covers only the first load. web.dev notes that content loaded as you scroll can cause shifts after loading that the lab does not show.
  • INP: a test that only watches loading produces no interactions, so it may not show INP. web.dev advises starting, ideally, with field data. Where there is none, it lists following common user flows in the lab and trying interactions while the page is still loading among the ways to find a slow interaction.

4. What moves each measure?

Once you know which part is swollen, the order of fixes is clear too. The table below summarises the parts and common causes in web.dev’s optimisation guides.

Measure Parts or common causes Reading tip (our suggestion)
LCP Four parts: time to first byte, resource load delay, resource load duration, element render delay. On a well-optimised page the shares are about 40%, under 10%, about 40% and under 10%. See which part eats the time. The two “delay” parts should be close to zero.
INP Three parts: input delay, processing duration, presentation delay. Find the slow interaction in the field, then repeat the same interaction in the lab.
CLS Images without dimensions, ads, embeds and iframes without dimensions, content injected later, web fonts. If field CLS is much larger than lab CLS, the shift may start after loading.

web.dev also notes that the LCP shares are guidelines and not strict rules: if LCP is consistently within 2.5 seconds on your pages, the proportions matter little. You will find the fixes step by step in our site speed guide.

A hypothetical reading

The situation and the figures in this section are invented; they do not belong to a real site or client and are here only to show the reading order. Suppose Search Console shows CLS above 0.25 on mobile for a category template. Suppose too that you enter the example URL in PageSpeed Insights and field CLS comes out at 0.3 and lab CLS at 0.02. The gap is a reminder that the lab sees only the first load. web.dev’s CLS guide says differences like this are likely to come from post-load shifts.

That gives the first question: is something loading when the page is scrolled or after a delay? It could be product cards, a late banner or an embed without dimensions. The lab report will not show this on its own, so you need to scroll the page by hand in mobile view and watch. The report then narrows where to look instead of leaving you to guess.

5. Make one change, confirm after 28 days

Field data covers the last 28 days, so the result of a fix does not show at once. In Search Console you can start tracking from the issue’s detail page (“Start Tracking” in the English interface). Google counts an issue as fixed if it is not seen in any URL during the 28-day window. If it appears in even one URL, the issue stays unfixed. Tracking does not trigger re-indexing.

If you make several changes to a template in the same period, you cannot tell which one worked. Note the three values before the fix and wait after a single change.

Common mistakes

  • Making the lab score the target. A score of 92 does not show that real visitors have a good experience.
  • Looking at the average. The assessment is at the 75th percentile; a quarter of visits may be worse.
  • Deciding from one screen. Search Console shows the group and PageSpeed Insights one URL; do not conclude before comparing the two.
  • Skipping mobile. Mobile and desktop are assessed separately; one can be good and the other poor.
  • Counting “No data available” as “good”. Groups without enough data do not appear in the report.
  • Expecting the result within a week. The window is 28 days.

Common questions

What happens if one of the three measures is bad?

The page does not pass the assessment. In Search Console a group’s status follows its weakest measure; for example, if CLS is poor and INP is good, the group shows as “poor”. Find the weak measure first, then look at its parts.

My PageSpeed Insights score is high, so why does Search Console say “poor”?

The score is a lab test Lighthouse runs on a single device. Search Console shows the last 28 days of real-user data at group level. They measure different things, and Google says the two kinds of data can sometimes disagree.

Why does Search Console say “No data available”?

You see this message if the property is new or if the Chrome user experience data has too few samples for the device type you chose. For new properties, data can take a few days to arrive. In the meantime you can test individual pages live in PageSpeed Insights.

If your own team cannot do this reading, technical auditing is part of our SEO service; if the fixes are in the site’s code, our web and software service takes over. You will find where Core Web Vitals sit in a technical audit in the technical SEO checklist, and a short definition in the glossary.

One step you can take today: open the mobile report in Search Console and enter the example URL of the reddest row in PageSpeed Insights. Write down the three 75th percentile values in the field section and which measure fails its threshold. To sort the numbers, you can use our Core Web Vitals checker.

Sources

This article is general information, not legal, financial or technical advice. Platforms, their rules and the law change over time, so some details may have changed since 15 November 2026 or contain small errors. Before you make a decision for your business, check the current official sources or write to us.

Tell us about
your project.

Fill in the form or write to us directly, and we will agree on a day to talk.

Your first meeting is with Deniz Sezer, founder of Detmore.

First meeting free · Reply within one working day · Written proposal in 2 to 3 days · NDA ready

What do you need? pick as many as you like

Name, email and phone are required.

Contact

We reply within one working day.