Key takeaways

  • Run checks in order: access, crawl, index, render, speed, structure, schema. Earlier problems mask later ones.
  • AI crawler access is now a technical SEO check, not an afterthought.
  • Most traffic losses after a redesign trace to redirects, canonicals or rendering — all on this list.

1. Access and crawler control

  1. robots.txt exists, is valid and doesn't block CSS, JS or key sections.
  2. AI crawlers (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) are allowed or blocked deliberately — not by accident.
  3. XML sitemap exists, is referenced in robots.txt, contains only canonical, indexable 200 URLs.
  4. No staging or development URLs are indexable.
  5. HTTPS everywhere, with HTTP → HTTPS redirects and no mixed content.
  6. One canonical hostname (www or non-www) with the other redirecting.

2. Crawl efficiency

  1. Crawl stats in Search Console show no spikes in errors or server response times.
  2. Parameter and faceted URLs are controlled (canonical, noindex or blocked) so they don't consume crawl budget.
  3. Pagination uses crawlable links, not only JavaScript.
  4. Internal links to 404s, redirects or noindexed pages are fixed.
  5. Orphan pages (no internal links) are either linked or removed.
  6. Click depth to key pages is three or fewer from the homepage.

3. Indexation

  1. Indexed page count matches intent; “Crawled – currently not indexed” is small and explained.
  2. Every indexable page has a self-referencing canonical, or a correct canonical to its master.
  3. noindex is not applied to pages you want to rank (check templates, not just pages).
  4. Duplicate titles and descriptions are resolved.
  5. Thin pages (tag archives, empty categories, test pages) are noindexed or improved.
  6. hreflang is correct and reciprocal if you serve multiple languages or regions.

4. Rendering

  1. Main content is present in the initial HTML, or server-side rendered, for key templates.
  2. The rendered page (Search Console URL Inspection) matches what users see.
  3. Lazy-loaded content that matters for ranking is still discoverable.
  4. No critical content is behind tabs or accordions that require JavaScript to render.

5. Speed and Core Web Vitals

  1. LCP under 2.5s on mobile for key templates — check the LCP element and fix that asset specifically.
  2. INP under 200ms — identify long tasks from third-party scripts and defer them.
  3. CLS under 0.1 — set dimensions on images, embeds and ads; avoid late-loading fonts shifting text.
  4. Images are compressed, correctly sized, and served in WebP/AVIF.
  5. Fonts are preloaded and limited to the weights used.
  6. Unused CSS and JS are removed; render-blocking resources are deferred.
  7. Caching and a CDN are in place.

6. Structure and on-page

  1. One H1 per page, followed by a logical H2/H3 hierarchy.
  2. URLs are short, readable and stable.
  3. Titles are unique, under ~60 characters, and lead with the topic.
  4. Internal links use descriptive anchor text and point to the pages you want to rank.
  5. Breadcrumbs are present and marked up.
  6. Images have descriptive alt text where meaningful, empty alt where decorative.

7. Structured data

  1. Organization schema is site-wide with logo, URL and sameAs links.
  2. Page-type schema (Article, Service, Product, LocalBusiness, FAQ, HowTo) matches visible content exactly.
  3. All schema validates with no errors in the Rich Results Test.
  4. Schema is updated when content changes.
  5. Search Console enhancement reports show no errors.
Need it done?Our Technical SEO service runs this list, prioritises the findings and implements the fixes.