- Home
- Technical and performance
- Website Speed Test
Website Speed Test
Measure your pages’ speed and find out what actually slows them down, with the data to back it.
Data from the Google PageSpeed Insights and CrUX APIs. The Core Web Vitals come from real Chrome users, aggregated over twenty-eight rolling days. Free and unlimited.
What this tool is for
“My site is slow” is an observation, not a diagnosis. This tool separates two things people systematically confuse: what a test machine measures in a lab, and what your visitors actually live through.
The score out of a hundred, the one everybody quotes, is a simulation. A standardised machine loads your page once, on a theoretical connection. That is useful for diagnosis and for testing a fix, but it is not what Google takes into account to rank your pages.
What it does take into account is the Core Web Vitals measured on real Chrome users, aggregated over twenty-eight days. That data cannot be simulated. It is why a site showing 95 in the lab can be rated “poor” by Google: the simulation was not seeing the problem visitors live through on an entry-level phone over mobile data.
The most expensive confusion: spending weeks pushing a lab score from 82 to 94 without ever looking at the field data — which will not have moved, because the real bottleneck was elsewhere.
How to read the results
Three metrics count, and each describes a different moment of the experience.
LCP measures the time before the largest visible element paints, generally the main image or heading. It answers “when does the page appear?”. Below 2.5 seconds is good; past 4 seconds is poor. The cause is nearly always an image too heavy, a server slow to answer, or a font blocking the display.
INP measures the delay between a click and the page’s visible reaction. It answers “does the page respond when I use it?”. Below 200 milliseconds is good; past 500 is poor. The cause is always JavaScript monopolising the processor — consent banners, tracking scripts, third-party widgets.
CLS measures the content shifts during loading, those jumps that make you click the wrong thing. Below 0.1 is good; past 0.25 is poor. The cause is nearly always an image with no declared dimensions, or a box inserted above the content after the fact.
Each metric appears twice: field and lab. When the two diverge, believe the field. You can follow those three values over time with the Core Web Vitals checker.
Common mistakes
Measuring the home page and drawing general conclusions. The home page is often the most carefully built page on the site. Your visitors mostly arrive on inner pages, which have never been optimised. Measure the pages that actually receive traffic.
Testing only on desktop. It is comfortable, but Google has indexed the web from its mobile version since 2019. A score of 92 on desktop and 41 on mobile is a common case — and it is the 41 that counts.
Tackling the hardest thing first. Rewriting your JavaScript is long, risky, and often returns less than converting your images to WebP. Start with the images: the biggest gain for the least effort.
Stacking tracking tools. Every banner, every advertising pixel, every chat widget adds JavaScript that degrades the INP. The gain from one more tracking script rarely exceeds what it costs in lost visitors.
Confusing score with ranking. No optimisation will lift a page whose content does not answer the question asked. Speed separates two equivalent pages, it does not replace relevance.
Going further
- Core Web Vitals CheckerReal field data from users
- Page Size CheckerTotal weight of a page and breakdown by resource type
- Security Headers CheckerAudit the five HTTP headers that protect your visitors
Frequently asked questions
Why does my score differ on every run?
The lab score is simulated on a shared machine whose load varies from one second to the next. A five-to-ten-point gap between two successive runs is perfectly normal and means nothing. Never compare two isolated scores: look at the trend across several measurements, or better, rely on the field data covering twenty-eight days of real visits.
Should I aim for 100 out of 100?
No, and chasing it is counter-productive. The score is a weighted average of lab metrics; Google does not use that number as a ranking factor. What it uses is the Core Web Vitals measured on your real visitors. A site at 78 with all three Web Vitals green is better placed than one at 95 with a poor INP.
What is the difference between lab data and field data?
Lab data is a simulation: a standardised machine loads your page once, on a theoretical connection. Field data comes from real Chrome users, aggregated over twenty-eight days. Only the latter counts for rankings. The lab serves to diagnose and to test fixes; the field tells you where you actually stand.
I have no field data, is that serious?
It simply means your page does not receive enough visits for Google to build a reliable sample. That is neither a penalty nor a technical fault. In that case rely on the lab data and the recommendations, bearing in mind that they often overstate slow-connection problems.
Does speed really influence ranking?
Yes, but less than people say. Core Web Vitals are a confirmed ranking factor of moderate weight: between two pages of comparable relevance, the faster one goes ahead. On the other hand, no technical optimisation will lift a page whose content does not answer the question asked. Speed separates, it does not qualify.
Where do I start when everything is red?
With images, in almost every case. They make up most of a page’s weight and are the simplest to fix: conversion to WebP or AVIF, dimensions declared in the HTML, lazy loading for everything below the fold. That single set of moves often resolves both the LCP and the CLS.