Last Updated on July 19, 2026
Your website is probably slower than you think, because you have never watched a stranger wait for it.
In July 2026 we loaded 24 real UK business websites in a real Chrome browser on a fast broadband connection. Roofers, electricians, plumbers, pest controllers, skip hire, drainage, accountants. The kind of firms where one phone call is worth hundreds or thousands of pounds.
The results, on a connection far better than the average customer’s phone:
- Median load time: 2.2 seconds
- 10 of 23 measurable sites took over 2.5 seconds
- 7 took over 4 seconds
- The slowest, a Leeds roofing firm, took 8.6 seconds
- Only 3 sites loaded in under 1 second
If a site takes 4 seconds on fast broadband, it takes a lot longer on a phone on 4G outside a job. That is where your customers are standing.
The mistake
Business owners treat speed as a technical nicety. Something the web person mentions, like tyre pressure.
It is a revenue setting. A customer with a burst pipe or a leaking roof opens three sites from search. The one that appears first gets read first. The one still loading gets closed. You will never see that lost enquiry in any report. It just never arrives.
What is actually slowing UK business sites down
Beyond timing the sites, we measured what was in them. Three patterns caused nearly all the damage.
1. Slow hosting. Before your page can even start appearing, the server has to respond. Google recommends under 200 milliseconds. Eleven of the 24 sites missed that, and 8 took over half a second. The worst case, a Leeds roofer, made visitors wait 4.2 seconds before receiving a single byte. Nothing on that site can be fixed until the hosting is. That is the true cost of the £3-a-month hosting deal.
2. Uncompressed images. The heaviest site we tested, a Sheffield pest control firm, weighs 8.3 megabytes and carries 83 images on its homepage. Another, a Bristol plumber, has 206 images on one page. For comparison, the lightest site in the test, another Sheffield pest control firm, delivers its whole homepage in 187 kilobytes. Same trade, same city, one site 44 times heavier than the other. The heavy one is not 44 times more persuasive.
3. Plugin and script bloat. The median site made 60 separate requests to load one page. The worst made 151. Every chat widget, slider, font pack and tracking pixel you bolt on is another queue your customer stands in.
What did not matter
Platform. The fastest site in the test loaded in 0.8 seconds and runs WordPress. So does the slowest, at 8.6 seconds. WordPress, Squarespace and custom code all produced fast and slow sites.
Fast sites stay fast because their owners keep junk out of them, whatever the platform.
What to fix first (by stage)
Start. Do one thing: put your postcode into your own site on your phone, on mobile data, and count the seconds. If you get past four, you have a problem worth solving before any marketing spend. Compress your images (free tools do this in minutes) and ask your host what your server response time is. If they cannot answer, that is your answer.
Build. Add caching, lazy-load images below the fold, and delete every plugin you cannot justify in one sentence. Then re-test. The goal is boring: under 2.5 seconds on a phone, reliably.
Scale. Stop guessing. Tie speed to money: track enquiries per hundred visitors before and after the fixes. If a fix does not move enquiries or rankings, it was decoration.
The takeaway
Ten of the 24 real UK business sites we tested are quietly turning customers away at the door, and their owners are paying for marketing to send more people to that door.
Speed is the cheapest marketing improvement most UK businesses will ever make, because you have already paid for the visitors you are losing.
Not sure where your site stands? Run it through our Whito Pro. It takes minutes and it is blunt.
Method: 24 UK business websites across seven sectors, loaded fresh in a real Chrome browser on fast broadband in July 2026. Load time, server response, page weight, request count and image count captured from the browser’s own performance data. One site did not report a complete load time and is excluded from load-time counts. Cross-domain resources can under-report size, so page weights are conservative. On mobile connections, real-world times will be worse than reported here. Sites anonymised; full dataset held by Whito.
