What Google actually measures when it measures your website

What does 43 out of 100 actually mean? Google’s PageSpeed Insights will happily hand your website a number like that, wrapped in a page of jargon, and leave you none the wiser about what to do on Monday. Here’s the plain-English version.

Behind the headline number, Google measures three things it calls Core Web Vitals — all definitions and thresholds below are Google’s own:

• Largest Contentful Paint — how long before the main thing on the page is actually visible. Not the spinner: the photo, the headline, the thing the page is for. Under 2.5 seconds is good. Think of it as how long the shop shutter takes to open.
• Interaction to Next Paint — when someone taps a button or opens a menu, how long before the page visibly responds. Under 200 milliseconds is good. Slower than that and the page feels like it’s ignoring you.
• Cumulative Layout Shift — how much the page jumps about while it loads. Under 0.1 is good. This is the one behind the universal experience of tapping a button just as something shoves it down the page.

The test is deliberately pessimistic

The number doesn’t come from your phone on your office wi-fi. It comes from a simulated mid-range phone on a deliberately throttled connection — because Google would rather measure the worst realistic case than the best one. Your customers include the person on the train with one bar of signal, and the test is built for them. It’s also why scores wobble: the same unchanged site can move several points between runs, because simulations vary.

That wobble is why our audit gives speed 30% of the overall score — measured and real, but not allowed to dominate two steadier judgements. The full weighting is in how we score your website.

What usually drags the number down

• Images uploaded straight off a phone — a 4MB photo doing the job of a 100KB one. The most common issue we see, and usually the cheapest to fix.
• Scripts that each seemed reasonable alone — the chat widget, the tracker, the slider, the fonts. Every one was “just one more”. Together they’re the queue your customer waits behind.
• Hosting chosen on price a long time ago, which adds a delay before the page even starts arriving.

What I’d do with this

Firstly, run PageSpeed Insights yourself — it’s free, and it’s the same Google measurement our audit uses, taken as-is. Secondly, fix images before anything else; it’s the best effort-to-result ratio on the list. Thirdly, be sceptical of anyone selling a guaranteed 100 — the last ten points cost more than the first forty, and your customers can’t tell the difference.

Or let our free audit run the measurement for you and put it alongside the two things PageSpeed can’t see: what your copy says, and how much of your admin is still manual.