Case study
A trade site that took eleven seconds to load, fixed in one week
A trade business a few miles from Wrexham had a site built on a page builder. On a phone on a normal signal it took eleven seconds before a customer could do anything with it. Its area pages were 72% identical to each other. Half of its pages had never been indexed by Google. We spent one week on it, changed no design and wrote no new marketing copy, and the same site became usable in 2.4 seconds. The caveat belongs at the top: speed and duplication are things we measured directly, so they are facts. Rankings and enquiries are not on this page, because a week is nowhere near long enough to claim them.
By Daniel Stoychev, Webso Digital · 4 September 2026
11.0s → 2.4s
until the page could be used on a phone
65 → 96
PageSpeed Insights mobile score
72% → 29%
text shared between area pages
44
pages resubmitted for indexing
What the audit found
Three problems, none of them the design
The owner thought the site looked fine, and it did. What it could not do was arrive quickly, say anything different from one town page to the next, or get its pages into Google's index.
- Weight. The page loaded a full page-builder runtime, several font families and a set of photographs at their original camera size before anything appeared. On a phone that is eleven seconds of a blank or half-drawn screen.
- Sameness. The pages for each town it covered had been made by copying one page and swapping the place name. Measured with the place names masked, so that only the structure is compared, they shared 72% of their text.
- Absence.Roughly half the pages were not in Google's index at all. Some were not linked from anywhere a crawler could reach; the near-identical ones were the kind Google routinely drops.
What we changed, in order
One week, no redesign
- Days one and two: the images. Every photograph resized to what the layout actually uses, served in a modern format, with its width and height in the markup so the page stops jumping while it loads.
- Day three: the code the page waits on. Fonts cut to what the design uses and loaded so text appears immediately, and the scripts that did not need to run before the page was drawn moved out of the way.
- Days four and five: the area pages. Each one rewritten around what is actually different about that town, which meant the roads people arrive by, the kind of work that comes from there, and honest travel time. Duplication fell to 29%.
- Day six: the plumbing. Internal links so every page can be reached, a sitemap that lists only live pages, and 44 pages submitted for indexing.
- Day seven: measurement. Search Console and analytics checked, and a baseline written down so the next month can be compared with something.
What we did not do
The design stayed exactly where it was
There was no rebuild, no new brand, no new photography and no fresh sales copy. That was deliberate. A site that already brings in enquiries is not a place to experiment: the rule we work by is that anything currently earning gets evidence before it gets changed. Everything in the list above is a technical fault with a measurable before and after, which is a different kind of decision from “we would style it this way”.
It also keeps the result honest. When the only variables are weight, duplication and indexing, the numbers cannot be explained away by a new look or a new offer.
Where each figure comes from
- Load times and the 65 to 96 score: Google PageSpeed Insights, mobile, before and after our deployment. Free, and anyone can repeat it.
- The 72% and 29%: our own comparison of the area pages with the place names masked, so structure is compared rather than geography.
- The 44 pages: Search Console, the same account the owner can see.
- No ranking or enquiry figures appear here. A week is too short, and quoting them would be the kind of claim this page exists to avoid. The client is not named, which is our rule for every case study.
What this means in Wrexham
Most local sites we open have at least two of these three faults
Page-builder sites are the norm for trades around Wrexham, and they fail in the same three ways: too heavy for a phone on a Welsh signal, town pages made by find-and-replace, and pages Google has never indexed. None of that needs a redesign to fix, and the checks are free. If you want to know which of the three your site has before you spend anything, ask and we will send the measurements with the method next to each one, the same way they appear above.
The other case on the results page is the opposite situation: a business with no site at all, built from nothing and followed for sixty days.
Want the same three checks run on your site?
Send the address and the towns you work in. You get the load time, the duplication figure and the indexed count, with the method next to each, and a price if you want the work done.