Skip to content
Avolis

Why is my website slow

Where the missing seconds go: TTFB, page weight, scripts, caching, and when a hosting upgrade helps.

15 August 2026 · 4 min read

“Slow” is one word for at least five different problems, and the fixes are completely different. A page can be slow because the server takes too long to start answering, because the page itself is heavy, because some third-party script is stalling, because nothing is being cached, or because the server is on the other side of the world. Identify the right one and you can usually fix it in an afternoon. Guess, and you can spend a quarter on a bigger server and end up feeling exactly the same.

Where the time actually goes

Between the visitor typing in your address and the page being usable, the browser has to resolve the domain, open a connection to the server, wait for the server to start sending (this is the number that matters most), download everything the page needs, and render it. Each step has different causes and different fixes. Most of what people call “slow hosting” turns out to be one of the later steps.

The single most useful number is time to first byte, usually written as TTFB. It covers everything up to the point where the server starts sending: the network trip, the queue at the host, and the time your website or CMS actually spends building the page.

Here is the test that splits the problem in half. Check the TTFB when the site is quiet, and again when it is busy.

  • Slow only under load points at capacity: a shared host squeezing CPU, a small VPS running out, a database that only crawls with real traffic.
  • Slow even at 2am points at the site itself: too many database queries per page, a page rebuilt from scratch on every visit, static assets served without cache headers.

The page itself

Page weight is where most of the surprise lives. Sites are built by adding things, and nothing is ever removed. A photo exported at a few megabytes where a hundred and fifty kilobytes would do. A font loaded for two words. A video playing on the hero background just because it looked good in the demo. Each one is fine in isolation, and together they add up to a page that takes ages on a phone.

Third-party scripts are the quiet killers. The chat widget, the analytics, the ad tag, the booking embed. Each one is a download the visitor has to wait for, and if any one of those vendors has a bad day, your page waits with them. You do not control their performance. The honest fix is the boring one: remove or defer them, and re-test after every removal so you can see what each one actually cost.

Caching

If your page is assembled from a database and rebuilt on every visit, every visitor pays the full cost of the build. Cache the result and the server mostly sends a finished page, and the difference on modest hardware is significant. Static content (images, stylesheets, scripts) should be served from the browser on repeat visits without being re-downloaded. If it is, you are making everyone pay in full every time.

The distance question

A server in another continent adds a round trip before the server has done anything, and that is a floor, not a bug. If most of your visitors are in Australia and the host sits across the world, some of that time is physics. It is worth knowing where your hosting is. It is also one of the few things a new host can actually fix.

When upgrading hosting helps

Upgrading helps when the TTFB test pointed at capacity, and it is one of the cleaner fixes there. It does not help when the page is heavy or the site is uncached. A four-megabyte page on a bare metal server is still a four-megabyte page. The usual order of work is: measure, cut the page, cache the rest, and only then look at whether the hardware has headroom.

A checklist that works

  • Measure the TTFB when the site is quiet and again during normal traffic.
  • Look at total page weight and the ten largest files.
  • List every third-party script, disable them one at a time, and re-test.
  • Confirm static assets are cached in the browser and pages are not rebuilt needlessly.
  • Note where the hosting sits relative to your main visitors.

Hosting is one lever. It is often the wrong first lever, which is why the measurement comes first.

Discuss your infrastructure

Tell us what you're running today and what you need help with.