Website speed: how much it matters for visitors and for Google

"Is my site fast enough?" is a question that hides two different questions, with answers of very different weight: how much it matters to visitors, and how much it matters to Google. It's worth separating them, because mixing them up almost always leads to overrating the second and underrating the first.

For visitors: the part that weighs more

Whoever opens your site decides whether to stay before reading a single word, just from how fast something appears to look at. That's not an impression: it's measurable with a real test, traffic split evenly between two versions of the same page.

Rakuten 24, a Japanese convenience store chain, did exactly that: a month-long test, half of visitors on the original page and half on a faster version, everything else identical. The result, published by Google: +53% revenue per visitor and +33% conversion rate, purely from making the page faster.

Vodafone Italy measured the same mechanism with a similar test, on roughly a hundred thousand clicks a day per version: by cutting the loading time of the page's main content by 31%, sales rose 8%.

Both are huge companies, and the exact percentages don't transfer one-to-one to a small business site: the point isn't "expect the same number", it's that the direction is the same everywhere it's been measured with a real comparison rather than a guess. A faster site keeps more people, and more people who stay turn into more enquiries.

For Google: a real signal, but not the one you think

Here the temptation runs the other way: assuming a fast site "wins" on Google. It doesn't, and Google says so itself, in its own official documentation: it always seeks to show the most relevant content, even when the page experience is sub-par. There's no single signal dedicated to speed: ranking systems look at a range of signals tied to the overall page experience.

In practice: a slow but genuinely useful page can still rank well. Between two competing pages with similar content, the faster one has an edge. It's a real factor, but a minor one next to writing something that answers the search better: speed doesn't make up for weak content, it gives a nudge upward to content that's already strong.

It's worth knowing how widespread the problem still is: according to the Web Almanac by HTTP Archive, based on Chrome data collected in July 2025, only 48% of mobile sites and 56% of desktop sites pass every threshold Google considers "good". Most sites in the world, today, still don't get there: this isn't a race lost before it starts.

The thresholds themselves are public: a site is "good" if it loads within 2.5 seconds, responds to taps within 200 milliseconds, and doesn't jump around while it loads. Above 4 seconds, Google rates it as "poor".

What usually makes a site slow

Three causes cover most cases, and none of them requires rebuilding the site.

Oversized photos. A photo shot on a recent smartphone often weighs several megabytes, and if it's uploaded as-is, without resizing it for the web, it alone can double a page's loading time.

Poor hosting. If the space where the site lives takes too long to answer the first request, every page starts out behind before the browser even begins downloading anything: what a good hosting plan should include is covered in full in another article.

Too many third-party scripts. Plugins, tracking tags, chat widgets, embedded maps: each one adds something the browser has to download and run, and they pile up over time as nobody removes the ones that stopped being useful.

How to find out if your own site is slow

With PageSpeed Insights, a free tool from Google: enter the site's address and look at the mobile score, not desktop, since most people searching do so from a phone. The result also shows which of the three thresholds is failing, which is already half the work of knowing what to fix.

If you run a place to stay, the same check is the third of seven in a list built specifically to work out whether a website that isn't bringing bookings needs fixing or rebuilding: worth running all seven, not just the one on speed.

Where to start

A red score on PageSpeed Insights is already costing you visitors, regardless of what Google does or doesn't do about it: that's the part of the problem that weighs more, and in most cases it's fixable without touching the site's structure. If you want to know which of the two situations you're in, get in touch and I'll tell you after checking it myself, not just from an automated score.