Performance

Web speed: what Google measures and what to fix

Three acronyms, three thresholds and five fixes. Most of what else gets written about this is surplus for a small business.

appnweb Published 4 August 2026 Reviewed 18 October 2026 6 min read

The short answer

Google measures three things: how long the largest thing on screen takes to appear (LCP, good under 2.5 seconds), how long the page takes to respond when you tap something (INP, good under 200 milliseconds), and how much the content moves while it loads (CLS, good under 0.1). On a small-business site, ninety per cent of the problem is unoptimised images, fonts and third-party scripts. Fixing those three usually does it.

The three acronyms, in plain words

LCP is how long the biggest visible element takes to paint: usually the header photograph or the headline. If it drags, the person sees a blank screen and decides the site is broken.

INP measures how long the page takes to react when you press something. It replaced the old FID and it punishes script-heavy sites hardest: everything looks fine, but pressing the menu button does nothing for half a second.

CLS measures how much the content jumps while loading. It is why you go to tap a link, a banner slides in above, and you tap something else instead. It is fixed by reserving space for whatever is going to appear, and it is nearly always the cheapest of the three.

The five fixes that move the most

On a small-business site the breakdown is nearly always the same, and it is worth attacking in this order because the effort is very unevenly distributed: the first two usually matter more than the other three combined.

Images are first and by far the largest. A three-megabyte header photograph uploaded straight off a phone is the entire problem on many sites. Converted to a modern format and resized to the dimensions it is actually displayed at, that same photo weighs a hundred kilobytes.

Third-party scripts are second: every advertising pixel, every floating chat, every embedded map brings its own code and its own latency. On a small site, two of them already show.

Images in a modern format, at the size they are actually shown
Fewer third-party scripts, and the survivors loaded late
Self-hosted fonts rather than fetched from another domain
Reserved space for images, ads and the cookie notice
Hosting on a content network rather than a single server

What is measured versus what is felt

There are two kinds of measurement and they get confused constantly. PageSpeed Insights gives you a lab score, simulated on a theoretical device. Field data comes from people who have genuinely visited your site, and that is what Google uses.

You can have a lab score of 95 and poor field data, usually because your visitors are on a mid-range phone with patchy signal and you tested on fibre. The score that counts is the field one.

And a useful reminder: speed is a tiebreaker, not a dominant factor. A fast page with thin content does not beat a slow page with good content. What slowness does is make people leave before reading the good content.

Where the figures come from

The three Core Web Vitals are LCP, INP and CLS, and Google sets the "good" threshold at 2.5 seconds for LCP, 200 milliseconds for INP and 0.1 for CLS.

Source Google — web.dev, Web Vitals

Related services

Compliance

Accessibility in e-commerce

Sectors

Keep reading

WordPress or a static site
When WordPress is right, when it is not, and the year-three question that decides for you.
Local SEO for a neighbourhood business
The four things that decide whether you appear on the map, and the per-neighbourhood page trap that sinks whoever uses it.
Cookies, GDPR, and how to keep measuring
What the law requires, why rejecting must be as easy as accepting, and how to measure visits with no banner.

Common questions

If your question isn’t here, write to us and we’ll answer it. Contact

How fast should my site be?

Under two and a half seconds for the first thing people see, measured on a phone with ordinary 4G rather than on your desktop with fibre. Under a second is excellent and entirely achievable on a static site.

Does the cookie notice affect it?

Considerably, in two ways: the script takes time, and the banner appearing suddenly shifts the content and spikes CLS. Reserving its space and loading it properly costs little and shows.

Is more expensive hosting worth it?

It depends where you are starting. Moving from saturated shared hosting to something decent makes an enormous difference. Moving from decent to very expensive, almost none: beyond that point the bottleneck is your images and your scripts, not the server.

What if my site is WordPress?

It can be optimised a great deal: caching, images, removing plugins nobody uses. What does not go away is that every page is built when it is requested. Good numbers are reachable with work; the work just has to be repeated periodically.

We will measure yours

Send us the address and we will tell you your three numbers, what is causing them, and which fix changes most for least work. No commitment.

Or directly: hola@appnweb.site; Monday to Friday, 09:00 – 18:00 (CET)