Core Web Vitals
Google PageSpeed performance with AI recommendations
Open the page in the appMeasured metrics
- LCP (Largest Contentful Paint): time until the largest visible element is rendered. Target: < 2.5s.
- FID / INP (Interaction to Next Paint): responsiveness to first user interactions. Target: < 200ms.
- CLS (Cumulative Layout Shift): visual stability (no unexpected content movement during loading). Target: < 0.1.
- TTFB (Time to First Byte): delay before first byte sent by server. Target: < 800ms.
- Performance Score: global Google Lighthouse score out of 100.
Mobile and desktop
PivotSEO measures Core Web Vitals separately for mobile and desktop, because Google uses mobile score for ranking (mobile-first indexing). A toggle lets you switch between views. Results are cached to avoid reload on each visit.
CrUX field data: what your real visitors experience
At the top of the page, the “Field data — real Chrome users (28 days, p75)” card shows Chrome UX Report (CrUX) data: the real p75 measured from your visitors over the last 28 days — the value under which 75% of experiences fall. The five metrics (LCP, INP, CLS, FCP, TTFB) are shown separately for mobile and desktop, with the percentage of “good” experiences under each value and a CWV passed / failed (mobile) badge. This field measurement — not the lab test — is what Google actually uses for ranking.
Lab vs field: two complementary measurements
- Lab test (PageSpeed / Lighthouse): a simulation under fixed conditions (standardized device and connection). Perfect for diagnosing and immediately verifying the effect of a fix.
- Field data (CrUX): the aggregate of your visitors' real Chrome sessions over 28 days. This is the verdict that counts for Google — a site can pass the lab and fail the field (or the reverse) depending on your audience's actual devices and connections.
- Reaction delay: after an optimization, the lab improves immediately, but field data takes up to 28 days to reflect the change (rolling window).
Claude recommendations
After analysis, a « Generate AI recommendations » button launches Claude to produce a prioritized list of quick wins (< 1h dev), mid-term (1-5 days) and long-term (refactor). Recommendations are contextualized based on your specific metrics.
