Core Web Vitals: what they are, and what fixed them on our sites
Core Web Vitals are three measurements Google uses for how a page feels: how fast the main content appears (LCP), how fast the page responds to a tap (INP), and how much the layout jumps (CLS). A page passes when LCP is 2.5 seconds or less, INP 200 milliseconds or less, and CLS 0.1 or less. Below: what each one means, the seven fixes that moved our own pages, and how the PageSpeed score differs from them.
What are Core Web Vitals?
Three measurements, each with a limit a page has to meet on phones and on computers.
- 2.5 sLCP · when the largest image or text block on the first screen appears
- 200 msINP · how long the page takes to react after a tap, click or key press
- 0.1CLS · how much the content moves while the page loads
“Good” limits · web.dev, Google · 75th percentile of page loads
Google counts the 75th percentile of page loads, and treats phones and computers separately. So three out of four visits on a phone need to meet the limit, not just your own visit on fast office Wi-Fi.
INP replaced the older First Input Delay (FID) on 12 March 2024. Guides that still list FID were written before that date.
What slowed our pages, and what fixed it
These are the fixes we made on our own website and three sample sites on 30 September 2026. Each row names the cause, the measurement it hurt, and what changed.
| Page · what slowed it | Hurt | Fix | Result |
|---|---|---|---|
| umberform.com a 3D scene hung the page for over 10 seconds on machines without a graphics card | LCP, TBT | show a still image when there is no graphics card | without a graphics card: 99–100, blocking 0 ms |
| all four an empty Tag Manager container loading 290 KB of scripts | weight | removed it | 290 KB less on every page |
| SALTHOLM analytics loading at the same time as the main photo | LCP | load analytics after the page | 88 → 96–97 in our own test |
| VORMLAND a 3D cloud of projects kept the page busy for 16 seconds | TBT | a grid of the same projects without a graphics card | 65 → 96, with the next fix |
| SALTHOLM, VORMLAND the page faded in on load, so the main photo appeared while the page was still invisible | LCP | no fade on load | measured together with the fixes around it |
| SALTHOLM text animations and smooth scrolling started before the first screen | LCP | start them after the page has loaded | 71 → 99, with the fixes above |
| umberform.com styles loaded as separate files | LCP | styles placed inside the page | 87 → 100; LCP 3.9 s → 1.2 s |
CLS was 0 on all four pages when we measured on 23 October 2026, so none of our fixes were about layout jumps. The usual causes are images without set dimensions, banners that appear late and fonts that swap in.
Two patterns stand out. First, the delay was often in waiting, not in downloading: on our home page, the main headline waited 1,330 ms for the styles, and 110 ms after the fix. Second, the biggest fixes were about the order of loading, not about removing features: the 3D scene, the animations and the analytics all stayed.
Our home page headline waited 1,330 ms for the styles. After one fix: 110 ms.
The 3D scene, the animations and the analytics all stayed. Only the order changed.
Do Core Web Vitals affect Google rankings?
They count, but Google gives no weight. Its own guidance says: “We highly recommend site owners achieve good Core Web Vitals for success with Search.” It adds that they align with what its ranking systems seek to reward.
So speed is part of a good page, not a shortcut past a better answer. It matters for visitors either way: on online shops, slow pages convert worse, as the studies on our redesign page show.
Is the PageSpeed score the same as Core Web Vitals?
No. The PageSpeed score comes from a test run on Google’s machine, called lab data. Core Web Vitals come from real visitors in Chrome, called field data. PageSpeed Insights shows both on one page: field data at the top, the lab score below.
Total Blocking Time, how long scripts keep the page too busy to respond, counts for 30%. INP is not in the score at all, because a lab test has no visitor to tap.
A new or small site often has no field data at all. On 23 October 2026, PageSpeed Insights showed none for TARNHOLM, our furniture sample site, because not enough Chrome users had visited it. Its usual lab score on a phone is 97. Until that changes, the lab score is the only number you have.
Why does my PageSpeed score change every time?
Because the test machine, the network and Google’s servers change between runs. Google’s own Lighthouse documentation suggests thinking of performance “as a distribution of scores, rather than a single number.”
Two habits make the numbers more useful:
- Run the test three to five times and look at the usual result, not the best or the worst one.
- Add something to the address, such as
?test=1, before each run. In our tests, the same address tested again within minutes returned the same saved report.
Where should you start?
Start with LCP on a phone. PageSpeed Insights names the element it measured, so you know exactly what to look at. Then check what loads before that element: scripts, styles, fonts and other images.
If the LCP element is an image, our guide to image size for a website covers widths, formats and which image to load first. If a fix needs changes to your site’s code, that is the kind of work an upgrade covers; see when a redesign is worth it. On a builder such as Wix or Squarespace, some of these fixes are in the platform’s hands, not yours.
We measure your main pages on a phone and list the fixes in order of benefit for the work. You get a short document, not a sales call.
Send it with the button on the right→
Questions about Core Web Vitals
What is a good Core Web Vitals score?
LCP of 2.5 seconds or less, INP of 200 milliseconds or less and CLS of 0.1 or less, for three out of four visits on phones and on computers.
Do I need a PageSpeed score of 100?
No. The score is a lab test, and it moves between runs. Our TARNHOLM page usually scores 97 on a phone. Aim for the three Core Web Vitals in the green, and a usual lab score in the 90s.
Why does PageSpeed say “no data” for real users?
Your page has not had enough visits from Chrome users yet. Field data needs a steady flow of visitors, so new and small sites often show only the lab result.
Can you check my site?
Yes. Send us your address for a free review: we measure your main pages on a phone and list the fixes in order of benefit for the work.