What are Core Web Vitals, and which thresholds must you hit?
Core Web Vitals is Google’s way of reducing “is this page a good experience?” to three measurements. All three are collected in real visitors’ browsers, not in a lab.
The three measurements
| Metric | What it measures | Good | Poor |
|---|---|---|---|
| LCP | Largest content painted on screen | ≤ 2.5 s | > 4.0 s |
| INP | Response time to an interaction | ≤ 200 ms | > 500 ms |
| CLS | Content shifting while it loads | ≤ 0.1 | > 0.25 |
Google evaluates these at the 75th percentile: at least 75% of your visitors must be in the “good” band. So your average can look fine while you still fail.
LCP — when the page counts as “open”
Usually the hero image or the largest heading block. The cause is rarely the server; more often it is late-discovered images, render-blocking CSS/JS and heavy themes.
INP — “I clicked and nothing happened”
INP replaced FID in 2024 and is stricter: rather than the first interaction, it measures the worst interaction across the visit. JavaScript that occupies the main thread for long stretches is punished here.
CLS — the page that jumps
Most often caused by images with no declared dimensions, fonts that load late, and ad or cookie boxes injected between content. It is usually a cheap fix: writing width and height on images is often enough.
How to measure
- PageSpeed Insights — gives both real-user data (where available) and a lab test. Free.
- Search Console → Core Web Vitals — groups your whole site by real visitor data. This is the one that matters.
- Chrome DevTools → Lighthouse — fast feedback while building, but remember it is measuring your own machine’s speed.
Why speed is a commercial matter
- 53% of mobile visits are abandoned if a page takes more than three seconds (Google).
- Going from a 1-second to a 3-second load raises bounce probability by 32%; going to 10 seconds raises it 123% (Google/SOASTA, 2017).
- A 0.1-second speed-up lifted retail conversions 8.4% and average order value 9.2% (Deloitte & Google, “Milliseconds Make Millions”).
This is why we treat speed work as conversion work rather than technical debt. For a concrete example, read our WordPress case study: we took a cart page from 2.9 seconds to 0.95.
Where to start
- Find the page group that is failing in Search Console.
- Measure a single URL from that group in PageSpeed Insights and see which metric is failing.
- Follow the metric: LCP means images and render blockers, INP means JavaScript load, CLS means dimensions and fonts.
- Make one change, measure again the same way. If there is no gain, revert it.
If you are not sure where to start on your own site, leave us the address and we will scan it and come back with a priority list.