A trade website rebuilt where it was failing: Google mobile speed from 65 to 96

The starting point

An established mobile tuning business with a website built elsewhere. It looked fine on a laptop. On a phone, where its customers are, it was slow enough that Google scored it 65 out of 100. The site had a page for each area it covers, which is the right idea, but the pages were a paragraph of links with a title that named the town and nothing else. Nobody searches for a town on its own; they search for the job and the town.

The business did not want a new website, and it did not need one. It needed the parts that were failing rebuilt in place.

What we did

• Found the cause of the speed. The background photos had been saved in a format meant for graphics, not photographs, which made each one many times heavier than it needed to be. Same pictures, converted properly, and the homepage stopped waiting on them.
• Rebuilt the location pages into real pages: what the business does in that area, which towns it drives to, what the equipment is, and a title that says the service and the place. One page each for Wrexham, Chester, the Wirral, Shropshire and Manchester.
• On 3 September 2026, rewrote the opening paragraphs so that the first sentence states the facts a customer and a search engine both need: the service is fully mobile, the equipment is dealer-level, and the named towns are the ones covered.
• Resubmitted the changed pages to Google and left the pages that were already earning clicks alone. That is a rule, not a preference.

What happened, with dates

The mobile speed score went from 65 to 96, measured with PageSpeed Insights on the live site after deployment in August 2026. Anyone can repeat that measurement, and we would rather they did. The location pages and the 3 September openings are too recent to have a Search Console reading: we read it monthly, and new pages on an established domain usually take weeks to be re-crawled and months to settle. When the figures exist, they go on this page with the same caveats.

What it cost the client

Access, facts and a little checking. We needed the site's code and hosting, the list of towns actually driven to, the plain facts about the equipment and the service, and the photos already on the site. The owner's time went on answering those questions and reading the new pages for anything untrue. We did not ask for new photography, a logo or a brand workshop, because none of it would have moved the score.

What we would do differently

Convert the photos on day one, before writing a word. That single change took the score from 65 to 96 and needed no copy, no approval and no design. We did it in the same week as the location pages, which meant the speed result and the content changes went live together, and that makes it harder to say which change did what in Search Console. Speed first, measure, then content.

What a builder can take from this

Photos of extensions and loft conversions are the heaviest thing on a builder's website, and they are usually uploaded straight from the phone at full size. That alone can be the difference between a score in the sixties and one in the nineties. Location pages are the second trap: a page per town with nothing on it but the town's name is what Google's spam rules call a doorway, and a page that says what you build there, with the jobs you have done nearby, is not. Both are fixable on the site you already have. If you want to know which of the two your site has, start with the free audit and quote.