All articles

Website speed: what Google actually measures, and how to check it yourself

Three numbers tell you whether your site is fast. Here they are, where to read them for free, and what most often drags them down on a small-business website.

Stéphane Agoume3 min read

"My site is fast" is a claim nobody should have to take on trust — ours included. Site speed is measurable, with free public tools, in a few minutes. Here is how to run that check yourself, on your current site and on any proposal you are handed.

The three numbers that matter

Google publishes three measurements, the Core Web Vitals. They describe what a visitor actually experiences, not how powerful the server is.

MetricWhat it measuresGood score
LCPTime until the largest visible element (usually the top image or headline) is painted≤ 2.5 s
CLSLayout shifts — the button that jumps just as you tap it≤ 0.1
INPDelay between a tap and the page visibly reacting≤ 200 ms

These are part of Google's ranking signals. They weigh less than the relevance of your content, but they separate otherwise comparable sites — and above all, they describe what someone searching for you on a phone, outdoors, on an average network, actually goes through.

Where to read the numbers for free

  1. PageSpeed Insights (pagespeed.web.dev): paste your address and read the mobile tab, not the desktop one. Mobile is where most local small-business traffic comes from.
  2. Search Console, "Core Web Vitals" section, if your site is already connected: it shows measurements collected from your real visitors over 28 days.

One caveat: PageSpeed Insights shows two blocks. The top one, when present, comes from real visitors. The bottom one is a lab test, re-run every time — it varies between attempts. To compare two versions of a site, run the test three times and keep the middle value.

What slows a small-business site down, in order

Across the sites we audit, the causes almost always show up in the same order.

  • Unresized images. A 4 MB photo straight off a phone, displayed in a 600-pixel-wide frame. By far the biggest offender, and the easiest to fix.
  • Third-party scripts. Live chat, cookie banner, review carousel, ad pixel, fonts loaded from an external service: each adds a network round trip before the page appears.
  • Above-the-fold animations. A headline that fades in is counted by Google as painted at the end of the fade. Half a second of animation on your main headline is half a second of LCP given away for free.
  • Accumulated plugins. On plugin-based sites, every module installed "just to try" keeps loading on every page, including the ones where it does nothing.

What to do with the result

If your mobile score is under 50, there is almost always a quick win in images and third-party scripts, with no redesign. Between 50 and 90 the work gets structural: load order, code splitting, font choices. Above 90, speed is no longer the next project — content and local search are.

Either way, write down the score and the date before you change anything. Without a starting measurement, no improvement can be demonstrated — not by you, and not by whoever you hire.