Performance is a crawl problem too
Googlebot’s crawl rate tracks how quickly your origin answers. Slow TTFB and heavy HTML do not only hurt humans — they throttle how often deep URLs get revisited.
That is why AuditHQ treats document timing and LCP HTML signals as first-class findings (AHQ-PRF-*), scored and listed next to classic SEO checks — not buried under a single PageSpeed number.
Start with TTFB, then the head
Time to first byte — if the HTML arrives late, every paint metric moves right. Prefer CDN/edge cache, cut multi-hop redirects, and reduce origin work before you obsess over AVIF.
Parser-blocking <script src> in <head> — scripts without defer/async hold first paint. Keep JSON-LD as type=application/ld+json; move or defer everything that is not required before paint.
LCP image priority — fetchpriority=high on the hero, never loading=lazy on the LCP candidate, and preload only when the URL is known in the first HTML.
Lab measurements vs field Core Web Vitals
Field CrUX still decides the Core Web Vitals pass/fail. Lab measurements and AHQ-PRF findings tell you what to change on this deploy.
AuditHQ shows Performance, Accessibility, Best practices, and Privacy observations in Lab measurements, with a Fix-first spotlight for the highest-severity open checks — so you do not confuse a privacy observation with a Best-practices fail.
Ship one slice with an AI agent
Copy the Lab or SEO Fix-first AI fix prompt when the spotlight names TTFB or a blocking script. The paste includes Observed evidence and the AuditHQ fix hint — your agent should not invent a full redesign.
Re-run the audit after deploy. When the rule flips to pass, move to the next Fix-first item or the full Lab measurements section paste.