Methodology: This check follows wSEO methodology 1.0. How the assessment works

Field score first, local culprit second

How to audit CLS and layout shifts

Field data shows the real-world scale while Chrome Performance identifies the cluster, timing and elements to repair. Test load and long sessions separately on mobile and desktop.

Auditing CLS through field score, layout-shift cluster and culprit elements
One green Lighthouse run cannot exclude late shifts experienced by real visitors.

Quick independent check

  1. Open PageSpeed Insights or Field metrics in Chrome Performance for the exact URL.
  2. Select mobile and desktop. Record URL/origin scope, period and 75th-percentile CLS.
  3. In a clean profile use Performance → Record and reload with throttling and the target viewport.
  4. Select the highest-scoring cluster in Layout shifts, then each diamond; save affected nodes and potential culprit.
  5. Record a separate post-load journey: scroll, consent, popup, carousel, form and messages.
  6. Enable Layout Shift Regions in Rendering and repeat the shift for visual confirmation.

How to rate the metric

Current CLS thresholds
75th-percentile CLSRatingAction
≤ 0.10GoodTest late journeys and preserve baseline
> 0.10 and ≤ 0.25Needs improvementStart with worst cluster
> 0.25PoorPrioritize cause discovery and repair

The threshold applies to the field distribution, typically at the 75th percentile. A local recording is a diagnostic example, not a substitute for real-user data.

What the trace means

Shift score depends on affected viewport area and movement distance. Nearby events are grouped into session windows/clusters; page CLS uses the largest cluster rather than summing every movement in an unlimited session.

An affected node may have been moved by another block above it. Potential culprit is a lead that must be confirmed using timing, DOM and removal of the suspected cause.

Discrepancies and false conclusions

  • Field and lab differ because of devices, networks, cache, personalization and session duration.
  • When URL data is insufficient, the tool may show origin data; do not present it as URL-specific.
  • A shift within 500 ms of recent input can have hadRecentInput and be excluded from CLS.
  • Scroll alone does not exclude a shift; late insertion during scrolling can count.
  • Autoplay carousels, font swap, ad refresh and SPA navigation require post-load recording.
  • One local zero does not prove every template and state is stable.

Automated pricing applies to one agreed cause and templates on the service page. Use the step-by-step guide for repairs.