How Fast Does Your Website Need to Be?

Somebody will sell you a speed score. It is a clean number, it goes up when they work on it, and it makes a tidy line on an invoice.

It is also not the thing that costs you jobs.

What actually matters

Someone standing in a wet kitchen taps your result on a phone, on mobile data, possibly with two bars of signal. The only question that matters is how long until they see something that tells them they are in the right place.

Under two and a half seconds and you are fine. Past that, people start going back and tapping the next result, and they do not come back later to see if it improved.

Google's own research on why speed matters makes the same point in more detail: abandonment climbs steeply with load time, and it climbs fastest on mobile connections.

Why your score is not the target

Run your site through PageSpeed Insights and you get a number out of a hundred. Useful, and routinely misread.

A score is a lab test on a simulated device. It is a diagnostic that tells you where to look. It is not what your customer experiences, and chasing the last ten points usually means optimising things no human would ever notice.

What is worth reading on that report is the Core Web Vitals section, which Google explains in its Web Vitals guide. Those measure three things a person genuinely feels: how long until the main content appears, how quickly the page responds when tapped, and whether things jump around while loading.

That third one is underrated. A page that shifts just as you reach for the phone number is infuriating in a way no score captures.

What actually makes a trade site slow

In the order I find them:

1. Photos straight off a phone. A 4MB image of a finished bathroom, displayed at 600 pixels wide. One of these can outweigh the entire rest of the page.
2. A page builder doing work you never asked for. Loading the scripts for sliders, popups and animations on every page, whether that page uses them or not.
3. Plugins. Each one small, all of them together not small.
4. Fonts. Four weights of three families, each a separate download, each blocking text from appearing.
5. Chat widgets and tracking. Third-party scripts you cannot optimise because they are not yours.

Notice what is not on that list: how the site looks. Design almost never causes this.

The two-minute check

Before you pay anybody to fix anything:

• Open your site on your phone, on mobile data, with wifi off
• Count how long until you can read something useful
• Try to tap the phone number without waiting for anything to settle

If that feels fine, your speed problem is probably somebody else's sales pitch. If it feels slow, now you know, and you knew without a report.

When it is worth doing something

Fixing images is almost always worth it, costs little, and often halves the load on its own. Stripping plugins you do not use is free. Both are worth doing on any site.

Rebuilding purely for speed is rarely worth it. Rebuilding because the site is slow and buries your phone number and has one page listing eleven services is a different matter, because then speed is a symptom rather than the problem. That is the point at which a new website stops being an expense and starts being a fix.

The honest summary: fast enough is a real standard and most trade sites miss it, usually for boring reasons that take an afternoon. Perfect is a vanity metric, and nobody has ever chosen a plumber because his site scored 98.