Awe CheckSearch and speed
Core Web Vitals: the three numbers Google judges speed by
LCP, INP and CLS in plain words: the “good” thresholds, how visitor data differs from lab data, and what to fix first.
Checked against sources: 30 September 2026
- 2.5 sLCP: the “good” threshold
- 200 msINP: the “good” threshold
- 0.1CLS: the “good” threshold
Three metrics and their thresholds
Google evaluates the 75th percentile of real Chrome visitors over 28 days: three of every four people must fall inside the threshold.
- LCP, speed of the main block: good under 2.5 seconds.
- INP, responsiveness to clicks and typing: good under 200 milliseconds.
- CLS, layout jumps while loading: good under 0.1.
Lab runs versus real people
PageSpeed shows both. A lab run uses a throttled processor and often looks worse than reality. Real-user data exists only for sites with traffic. When it exists, Awe Check trusts it more.
What to fix first
- Images: AVIF or WebP, width and height set, lazy loading below the fold.
- Third-party scripts: ads and chat widgets most often ruin INP, so defer them.
- Fonts: your own, font-display: swap, no more than two families.
- Ad and image slot heights: reserve the space and CLS drops to near zero.
Frequently asked
What matters more for search, mobile or desktop metrics?
Google indexes the mobile version, so look at mobile first.
Why does PageSpeed give different numbers for the same site?
Run-to-run variance can be large. Take at least three runs and look at the middle.
