SEO

Diagnose a Core Web Vitals issue hurting SEO

A prompt that separates genuine ranking-relevant CWV problems from noise in PageSpeed reports, and prioritizes fixes by realistic dev effort.

Works with Claude / GPT1,320 uses 4.5

The prompt

seo-core-web-vitals-diagnosis
[yourdomain.com/blog/] pages are failing the Largest Contentful Paint threshold in
Search Console's Core Web Vitals report: mobile LCP averages 4.2s against a 2.5s
target, affecting an estimated 65% of blog URLs. CLS and INP are both passing. The
site runs WordPress with a page builder theme and lazy-loaded images.

PAGESPEED INSIGHTS REPORT (paste the mobile report for a representative URL): [paste
here]
HOSTING SETUP: [shared hosting, no CDN currently]
THEME/BUILDER: [Elementor on a shared-hosting WordPress install]

Diagnose this like a developer who has to justify the fix cost, not just report the
failing metric:
1. Identify the actual LCP element on the sampled page (usually the hero image or
   first heading) from the PageSpeed report, since generic "optimize images" advice
   is useless if the real bottleneck is server response time (TTFB) instead.
2. If TTFB is a meaningful chunk of the 4.2s (say, over 800ms), state plainly that
   image optimization alone won't fix this and shared hosting or a lack of caching is
   the real constraint, even though that's a harder, costlier fix to recommend than
   "compress images."
3. If the LCP element is an image, check whether it's being lazy-loaded when it
   shouldn't be, since lazy-loading the above-the-fold hero image is a common
   page-builder misconfiguration that directly delays LCP.
4. Rank the fixes by effort vs. impact: quick wins (exclude the hero image from
   lazy-load, add a CDN, enable a caching plugin) versus structural fixes
   (theme/builder replacement, hosting migration) and be honest that the structural
   ones may deliver more improvement per dollar even though they're a bigger ask.
5. Note that CLS and INP passing doesn't mean they're safe long-term. If any planned
   fix (like adding a CDN or lazy-load changes) could regress them, flag it before it
   ships, not after the next crawl.

PageSpeed reports are easy to generate and easy to misread: a 4-second LCP number alone doesn’t tell you if the fix is a $12/month CDN or a full hosting migration. This prompt forces identification of the actual bottleneck element and an honest effort estimate, so the fix that gets scoped is the one that actually works.