Almost every website proposal mentions speed, and almost everybody has at some point seen a red number in a PageSpeed test and felt uneasy. It’s worth separating what genuinely changes rankings from what is optimising for the sake of a score.
The short answer
Speed is a ranking factor, but a weak one. Google has said so openly: it’s mostly used as a tie-breaker when two pages are roughly equal on content. Content that answers the question more precisely beats a faster page almost every time.
But speed strongly affects sales, and that’s the real reason to deal with it. Somebody waiting four seconds on a phone while out and about simply goes back.
What’s measured: three numbers, not a hundred
The rest of the report is diagnostics. These three numbers are what Google looks at:
| Measurement | What it is | Good value |
|---|---|---|
| LCP | how long until the largest element appears — usually the header image or main heading | under 2.5 s |
| INP | how long the page lags when you tap or click it | under 200 ms |
| CLS | how much the content shifts while loading | under 0.1 |
The third is the most underrated. CLS is that effect when you go to click and an image or banner loads above and shifts everything — so you click the wrong thing. It doesn’t cost you seconds, it costs you irritation and trust.
Why a score of 100 isn’t the goal
The tool that gives you a score measures one page, in laboratory conditions, with a simulated slow phone. That isn’t your visitor.
The real measurements come from real people’s browsers and appear in Search Console, under Core Web Vitals. If it says most URLs there are “good”, a score of 68 in a test shouldn’t worry you. When the two disagree, Search Console is the one that counts.
And one more thing: the journey from a score of 90 to 98 usually costs more effort than it returns. The journey from 35 to 75 almost always pays off.
What actually slows a site down, by frequency
From what we most often see in audits, in this order:
1. Images nobody resized. A 4 MB photo from a phone, dropped straight into the text. This is the cause in most cases and the easiest to fix — resize to the width it’s actually displayed at and save it as WebP. It often makes the page ten times lighter.
2. Too many plugins. With WordPress it’s a classic: 34 plugins, eight of which are used. Each adds its own CSS and JavaScript to every page, including the ones where it has no business being.
3. Tracking and chat scripts. Analytics, a pixel, a chat bubble, a map, two fonts from the cloud. Each one calls its own server before the page finishes.
4. Images with no set height and width. It doesn’t slow loading, but it causes content shift — that’s CLS. The fix is one attribute per image.
5. Cheap hosting. A server that takes a second and a half just to respond. No optimisation helps here; the only solution is to move.
The order worth working in
- Resize the images. Almost always the biggest single gain, and it takes one afternoon.
- Set width and height on every image, and reserve space for banners.
- Remove plugins you don’t use and defer chat and maps until the page has rendered.
- Turn on caching and compression on the server — that’s a setting, not a rebuild.
- Measure again after a week in Search Console, not in a test.
- Only then think about changing the theme, the framework or the hosting.
When speed is urgent
Drop everything and fix it immediately if any of this is true:
- loading on mobile takes over four seconds
- Search Console shows more than half of URLs are poor on mobile
- you have a shop and the basket or checkout is slow — that’s where money is lost directly
Otherwise: take one afternoon for the images and go back to content.
If you don’t know where to start, a technical audit goes through the site and returns findings sorted by what they cost — see what it looks like.
In short
- Speed is a ranking factor, but a weak one; content carries far more.
- For sales, speed matters, and that’s the real reason to sort it out.
- Watch three numbers: LCP, INP, CLS — and from Search Console, not from a test.
- Images are the cause in most cases and the cheapest to fix.
- A score of 100 isn’t the goal. The goal is that the page isn’t slow where the buying decision is made.