Use this when
pagespeed data, CrUX data, real user metrics, field data, CWV report.
Process
This team has no live PageSpeed Insights or CrUX API connection — PageSpeed/CrUX are public Google APIs keyed by an API key rather than a per-user OAuth account, and Brainwrite has no connector for them. Work from data the user exports and pastes in: a PageSpeed Insights JSON/HTML report, a CrUX API response, or a Search Console Core Web Vitals report screenshot.
From pasted PSI data, extract and interpret: lab scores (point-in-time Lighthouse) versus field data (28-day real Chrome user metrics) separately — never conflate the two; LCP, INP, and CLS at the 75th percentile with a Good/Needs Improvement/Poor traffic-light rating against 2.5s/200ms/0.1 thresholds respectively. Never reference FID; it was replaced by INP in March 2024 and removed from CrUX/PSI field-data tools in September 2024.
From pasted CrUX History data (if the user has a 25-week trend export), interpret trend direction per metric (improving, stable, degrading) and flag any metric trending toward Poor even if it's currently passing.
If the LCP element itself is identified in the data, connect the finding to a concrete fix (usually the hero image or a render-blocking resource) rather than reporting the number alone.
Report the metrics with traffic-light ratings, the lab-vs-field distinction stated explicitly, and specific fixes tied to whichever metric is weakest. Note plainly that this is working from a point-in-time export, not live monitoring.