Why Your Website Is Slow
Five causes explain almost every slow site we are asked to look at. None of them are mysterious, all of them are measurable in the browser you already have — and we fixed all five on this website, with the numbers to prove it.
Get a number before you touch anything
“Slow” is not a diagnosis. Before changing a line of code, find out what the page actually weighs, because the fix for a 12 MB page and the fix for a 300 KB page are completely different jobs.
Open your site in Chrome, press F12, go to the Network tab, tick Disable cache and reload. At the bottom you get two numbers that matter: how many megabytes crossed the network, and how long until the page was finished.
For a second opinion use PageSpeed Insights, which grades your real visitors on mobile hardware rather than your laptop. Ignore the score. Read the four measurements underneath it — the ones that matter are TTFB (how long the server took to start answering) and LCP (how long until the biggest thing on the page appeared).
What is actually wrong
In order of how much they usually cost you. The first one is responsible for more slow sites than the other four combined.
1. One image nobody sized
A single photograph straight out of a camera, or an animated GIF used as a background, can be ten times the weight of everything else on the page. Our own homepage was a 11.66 MB animated GIF, downloaded before anything appeared on screen. It compressed to nothing — GIF is already compressed, badly.
2. JavaScript written into the page
Code placed inside the page — rather than in a file the browser can cache — is downloaded again on every visit, on every page, and cannot be cached at all. Ours carried 883 KB of it in the HTML. As a separate file it is fetched once; as inline text it is fetched forever.
3. Fonts and stylesheets that block the first paint
Every extra stylesheet and font family is a round trip before anything can be drawn. Chains of @import inside stylesheets are the worst version, because the browser cannot even start the second request until the first finishes.
4. Nothing is reused between pages
If your header, footer, fonts and CSS are not cacheable, the second page a visitor opens costs as much as the first. That is invisible to a page-speed test, which only ever looks at one page.
5. No compression and no caching layer
Text compresses enormously. Our homepage was 973 KB of HTML and arrived as 207 KB once the server applied Brotli. If your host sends text uncompressed, you are paying for four times the bandwidth you need — and so is every visitor.
And the one people blame first
Hosting. It matters, but it is rarely the cause. A fast server cannot rescue a 12 MB image, and a slow server is a few hundred milliseconds — noticeable, but not the difference between one second and twelve.
What we changed, and what it weighed
This is our homepage, measured before and after — not a demonstration page, the actual site you are reading now.
The image change took one afternoon: convert the animation to WebP at 1280 pixels wide, keep the poster frame, replace the file. No redesign, no new layout, nothing moved on the page. It removed 9.2 MB from every single visit.
The order to fix it in
- Images first. Resize, convert to WebP, and check every one over 200 KB. This is usually 80% of the problem and 100% of the visible improvement.
- Then move inline scripts into files. One file, cached, loaded once — instead of the same code inside every page.
- Then fonts. One family is enough. Load it from your own server or preconnect to the font host, and never chain
@importinside another stylesheet. - Then turn on compression — Brotli if the host supports it, gzip otherwise. This is a switch, not a project.
- Then caching, so repeat visitors download nothing twice.
- Hosting last, and only if the numbers still say the server is the bottleneck.
Work down that list until the total transfer is under a megabyte. Stop there. Further work has diminishing returns and real risk of breaking things that currently work.
What people ask next
Does speed actually affect my Google ranking?
Yes, but modestly, and only as one signal among many. The bigger effect is commercial: a slow page loses visitors before your content ever loads. Fix speed for the person, and the ranking benefit arrives as a side effect.
Will a caching plugin fix my site?
Only if caching is the problem. On a site carrying a 10 MB image, a caching plugin changes almost nothing, because the image is cached too — it is simply enormous. Measure first, then choose the tool.
My developer says the site is fast. Who is right?
Whoever has a number. Ask for the total transfer size of the homepage with the cache disabled, and the LCP on a mobile connection. A page can feel instant on a fibre connection in an office and take nine seconds on a phone.
What is a good size for a homepage?
Under 1 MB for everything — HTML, CSS, fonts, images. That is achievable with one photograph and a compressed hero. Above 3 MB you are losing people on mobile data, and you will not notice because they leave silently.
Can you fix a site without redesigning it?
Usually, yes. That is the point of measuring: images, scripts, fonts and compression are all invisible changes. The layout does not move, and the page gets lighter.
Want a number for your site?
Send us the address and we will tell you what it weighs, what is heaviest, and what we would fix first — no obligation, and no redesign required.
