SEO Audit

Ranked, evidence-backed findings beat a 100-item checklist. Fix what blocks indexing first, then what wins clicks, then what improves long-term authority.

Step 1 — Establish the baseline

Collect (paste, do not paraphrase):

curl -sI https://example.com | head -20            # status, redirects, headers
curl -s https://example.com/robots.txt
curl -s https://example.com/sitemap.xml | head -50
curl -sL https://example.com | rg -o "<title>.*</title>|<meta name=\"description\"[^>]*>|<link rel=\"canonical\"[^>]*>|<h1[^>]*>.*</h1>"
npx unlighthouse --site https://example.com         # optional, CWV per page

From source: find how metadata is generated (Next.js generateMetadata, Astro layouts, Rails helpers), how the sitemap and robots are produced, and which routes are dynamic.

Ask the user for: target keywords or topics, the top pages by business value, Search Console access or export if available, and the geographic/language markets.

Step 2 — Technical crawl and index checks

• One canonical host and protocol; HTTP→HTTPS and www/non-www redirect with a single 301, no chains.
• robots.txt does not block CSS/JS or important paths; sitemap referenced.
• Sitemap contains only 200, canonical, indexable URLs; lastmod is real; split above 50k URLs.
• No accidental noindex (meta or X-Robots-Tag) on money pages; staging is noindexed.
• Canonical tags are absolute, self-referencing, and consistent with pagination and parameters.
• Soft 404s return real 404; deleted pages 410 or redirect to the closest equivalent.
• Rendering: critical content is in the HTML response (SSR/SSG), not injected only client-side. Test with curl vs browser.