We Scored 58/100 on Google's Speed Test. Here's the Cheap Fix That Got Us to 100.
Our product pages scored 58/100 on Google's speed test — the same test that quietly decides how many visitors stick around. Here's the low-cost, no-framework fix that took us to 100, and what your own score might be costing you.
The number that should worry you
Our own product pages scored 58 out of 100 on Google's speed test — the same test Google uses, in part, to decide how fast your site loads for a customer standing in a car park on 4G. If you've never run your own site through that test, you don't actually know your number either, and it is very unlikely to be better than ours was.
The usual answer to a bad score is "rewrite the whole site" — hand it to an agency, pick a framework, spend weeks and a five-figure invoice on a rebuild. We didn't do that. We changed eleven files. The pages are now 100 / 100 / 100 / 100.
Why this matters more than it sounds
A slow site isn't a technical embarrassment, it's a leak in your funnel. Every extra second before your page is usable is a customer who closes the tab before they've seen your price, your case study, or your phone number. They don't complain. They just don't convert. You never see the ones you lost.
The fix, for us, cost nothing beyond the time to do it — no new hosting, no agency invoice, no framework licence. That's the point: most of what wrecks a small business site isn't expensive to fix. It's just never been looked at.
What we actually changed (no framework required)
Three things did almost all of the work:
- We stopped autoplaying a video in the hero. A 240 kb looping clip was the first thing the browser had to load, and on a throttled connection it pushed our "page feels ready" moment to 3.4 seconds. We replaced it with a lightweight graphic. That moment is now 0.6 seconds.
- We stopped borrowing fonts from someone else's server. Every extra outside connection your site makes before it can show text is time your visitor spends staring at a blank screen. Hosting our own fonts and trimming the file to the characters we actually use cut that file from roughly 140 kb to 38 kb.
- We fixed our headings. Our page had three "main heading" tags and no sub-headings underneath — technically meaningless to a search engine and confusing to anyone using a screen reader. We gave it one clear topic and a proper structure underneath. That's the entire SEO fix.
None of it needed a rebuild. It needed someone to read the audit and fix what it actually flagged, in order, and then stop.
What it looked like once real visitors saw it
Lab scores are lab scores. So we checked 30 days of real visitor data on the live page. The p75 load time — the number that describes what 3 out of 4 real visitors experienced — came out at 1.4 seconds, comfortably inside Google's own "good" threshold. Bounce rate dropped and time on page went up. We didn't run a formal split test — we're a small team, same as most people reading this — but the direction of the signal didn't need one.
What to do with this if you run a small business
You don't need to read a full Lighthouse audit to act on this. You need one number: open PageSpeed Insights, paste in your homepage, and look at the score. If it's below 80, you are very likely losing enquiries you'll never know about — not because your product is wrong, but because the page never finished loading in the time a stranger was willing to give it.
Reply to this post with your score and we'll tell you, honestly, whether your fix looks like our eleven files or something bigger — no pitch, just the answer.