What a slow website actually costs you

Picture the person you want. She is standing in her kitchen at ten past eight on a Tuesday evening, holding a phone in one hand and a mug in the other, and she has finally admitted the units have to go. She types kitchen fitters near me. She taps the third result down, because the first two were adverts and she has learnt to skip those.

Then she waits. The screen stays white. A grey rectangle appears where a photograph is going to be. Something shifts down the page just as her thumb comes to rest, and she taps a menu she did not want. She goes back. She taps the fourth result instead.

That is the whole event. It took about four seconds, she will never think about it again, and it cost somebody a nine-thousand-pound kitchen.

She did not decide your website was slow. She decided the other one was there.

Where does the three-second rule come from?

The figure people repeat traces back to Google's own research into mobile page speed. The finding that stuck was this: as a mobile page goes from one second to three seconds to load, the probability that someone leaves rises by around 32%. Push it to five seconds and the increase is far steeper.

It is worth being precise about what that says, because it is routinely mangled. It is not a claim that 32% of your visitors leave. It is a claim about the increase in probability of a bounce across a population of pages. It is an aggregate drawn from a very large sample, and your particular site with your particular audience will not match it exactly.

What survives the caveats is the direction and the steepness. Every second you add costs you a meaningful slice of the people who were, a moment ago, genuinely interested. Not browsers. Not tyre-kickers. People who typed the words and chose your link.

Which speed measurements actually matter?

Google publishes three numbers about how a page feels to use, and they are the ones worth knowing because they are the ones being measured on you whether you look or not.

Largest Contentful Paint is how long the main thing takes to appear — usually your hero photograph or your headline. Under 2.5 seconds is good. Most small business sites we audit in Surrey land between four and eight, and the reason is nearly always the same: one enormous unresized photograph taken on somebody's phone and uploaded exactly as it came off the camera.

Interaction to Next Paint is how long the page takes to respond when someone taps. Under 200 milliseconds is good. This one punishes sites carrying a lot of script — a page builder, four tracking tags, a chat widget, a cookie banner from a plugin, a review carousel. Each one seemed free when it was added.

Cumulative Layout Shift is the measure of things moving after they appear. Under 0.1 is good. It is the one that produced the mis-tap in the kitchen: an image without declared dimensions arrives late, everything below it jumps, and the thumb lands on the wrong thing.

What is a faster website worth in real money?

Percentages are easy to nod along to and impossible to act on, so do the arithmetic in the units of your own business.

Say four hundred people a month reach your site from search. Say your average job is £4,000 and one in twenty enquiries becomes a job — conservative for a kitchen fitter, pessimistic for a roofer. If your site currently converts 2% of visitors into enquiries, that is eight enquiries and roughly £16,000 of work a month.

Now suppose speed and clarity together lift that to 3%. Not a transformation; a competent rebuild routinely does better. That is twelve enquiries, and £24,000. The difference is eight thousand pounds a month from the same four hundred visitors, the same van, the same two lads, and the same amount of advertising.

At 2% today8 enquiries · £16,000

At 3%12 enquiries · £24,000

At 4%16 enquiries · £32,000

This is why we measure a site before we say anything about how it looks. An opinion about your logo is worth nothing next to a number about your load time, and the number is free — Google's tool is public, it is at pagespeed.web.dev, and you can run your own site through it in the time it takes the kettle to boil.

What are the three most common causes?

• The hero photograph is enormous. Four megabytes is common. Resized and served properly it should be about 180KB — roughly one twentieth the size, and visually identical on a phone. That single change is usually most of the load time.
• There is no phone number in the header. More than half your visitors are on a mobile, holding a telephone, wanting to ring you. Making them hunt is a self-inflicted wound.
• Nothing has been tested on a bad connection. Your site is fast on your desk on fibre. Test it on a phone, on mobile data, in a car park. That is the condition your customers are actually in.

Do you need an agency to fix this?

You do not need an agency to fix most of this. Resizing images, adding dimensions to them, removing two tracking scripts you forgot you installed and moving to decent hosting will take a competent developer a day and will move your numbers more than any redesign. If you have someone who looks after your site, send them this and ask them to do it.

What we do, if it is ever useful, is rebuild the whole thing in five working days for £499 plus VAT — five pages, written for you, on your own domain, and yours to keep. Every price we charge is published, so you can see it before you speak to anyone.

But run the test first. The number is free, it is Google's rather than ours, and whatever you decide to do about it, it is worth knowing what the woman in the kitchen is waiting for.

Questions we get asked