Page speed and local visibility: where the line really is
Speed is not a feeling, it has published numbers. Largest Contentful Paint should be 2.5 seconds or less, Interaction to Next Paint 200 milliseconds or less and Cumulative Layout Shift 0.1 or less, each measured at the 75th percentile of page loads. Anything else in a speed quote is decoration.
The three numbers worth memorising
There are exactly three stable Core Web Vitals, and each has a published good threshold. Largest Contentful Paint measures loading and is good at 2.5 seconds or less. Interaction to Next Paint measures responsiveness and is good at 200 milliseconds or less. Cumulative Layout Shift measures visual stability and is good at 0.1 or less.
INP became the stable interactivity metric in 2024, replacing First Input Delay. If a report you were sent still talks about FID, it is measuring something that has been retired.
What the percentile changes about the conversation
The thresholds are assessed at the 75th percentile of page loads, segmented across mobile and desktop. That single word rearranges the whole discussion: your fastest visit does not count, and neither does your worst. Three quarters of visits have to clear the bar.
So the middle bands matter. LCP between 2.5 and 4 seconds needs improvement and beyond 4 seconds is poor. INP between 200 and 500 milliseconds needs improvement, beyond 500 is poor. CLS between 0.1 and 0.25 needs improvement, beyond 0.25 is poor.
That gives you 3 metrics with 3 bands each, so 9 possible verdicts and one honest question to ask any supplier: which of the nine are we in, measured on real visits rather than on a test run from a fast connection.
| Metric | Good | Needs improvement | Poor |
|---|---|---|---|
| Largest Contentful Paint | 2.5 s or less | 2.5 to 4 s | over 4 s |
| Interaction to Next Paint | 200 ms or less | 200 to 500 ms | over 500 ms |
| Cumulative Layout Shift | 0.1 or less | 0.1 to 0.25 | over 0.25 |
Where a local business usually loses
Layout shift is the cheapest to fix and the most often ignored: give every image explicit width and height so the text stops jumping when the picture arrives. That alone moves many small sites out of the poor band for CLS.
After that, look at what loads before the page is usable. Map embeds, chat widgets and font files are the usual three, and each of them is a decision rather than a necessity.
Measure before and after, on the same page and over a mobile connection. A change that moves LCP from 4.2 to 2.4 seconds is worth reporting. A change that moves it from 2.4 to 2.3 is worth nothing, because the threshold it had to clear was already cleared.
Questions and answers
Is FID still a Core Web Vital?
Why does my test tool disagree with the field data?
Which one should I fix first?
Does passing guarantee better rankings?
Sources
- web.dev, Web Vitals The three metrics, their thresholds and the 75th percentile assessment
- Google Search Central, AI features in Google Search That no additional technical requirements apply for AI features
