Why site speed matters, and how to measure it
· 2 min read
Speed is not a technical detail, it is a sales problem. How to read the measurement, and the three usual culprits.
A slow site is a shop with a door that sticks. Everything inside may be in order; nobody bothers going in. Speed is also a measurement Google uses directly in ranking.
How to measure
Enter your address at pagespeed.web.dev. It is free and needs no account. Read the Mobile result; the desktop result is almost always better and will mislead you.
The 0-100 score alone means little. Look at the three measurements underneath:
- LCP (Largest Contentful Paint). When the main content appears. Under 2.5 s is good, over 4 s is poor.
- CLS (Cumulative Layout Shift). How much the page jumps while loading. Under 0.1 is good, over 0.25 is poor.
- INP (Interaction to Next Paint). How fast it answers a tap. Under 200 ms is good.
The three usual culprits
1. Fonts loaded from someone else's server
The page can sit blank until the font arrives from a third-party server. That alone adds seconds. The fix is simple: serve fonts from your own server and preload them.
2. Files served uncompressed
Stylesheets and scripts are plain text and shrink to about a fifth when compressed. If the server has compression switched off, every visitor downloads the full size. It is a single server setting.
3. Unscaled images
A photo displayed at 400 pixels is very often loaded at 3000 pixels wide. Images should be prepared at the size they are used and saved in a modern format.
About the jumping
What usually drives CLS up is this: a font or an image arrives late, takes up space, and pushes everything below it down. The fix is to reserve that space in advance — give images a width and height, and match the fallback font's metrics to the real one.
When to worry
If mobile LCP is above 4 seconds or CLS is above 0.25, you are losing customers. When those two improve, you notice it not in the score but in the number of enquiries.