Technical SEO Checklist

Crawl access and robots rules

Confirm that production is not blocked in robots.txt and that no accidental noindex remains from staging. Confirm that CSS and JavaScript needed to render main content are not disallowed. A blocked resource can make Google see a poorer page than users see.

If you use a staging site, it should be locked (auth or noindex plus robots disallow). Mixing staging and production in one Search Console property is how test titles leak into results.

• robots.txt does not disallow important sections
• HTML has no accidental noindex or nofollow on money pages
• XML sitemap is listed in robots.txt only if you intend it
• Crawl budget is rarely the first problem on a small site — index bloat is

Indexation: what should be in Google, and what should not

Index the pages that deserve to rank: services, locations that are real, products, guides. Do not index thank-you pages, internal search results, faceted combinations that create infinite URLs, or cart and account URLs.

Use Search Console’s URL Inspection and page indexing report. If Google has chosen a different canonical than you intended, fix the duplicates rather than arguing in a comment thread.

HTTPS, host version and canonical URLs

Serve HTTPS everywhere. Redirect HTTP to HTTPS, and pick one host (www or bare) and stick to it. Mixed content (HTTP assets on HTTPS pages) still causes console noise and occasional browser warnings.

Every indexable page should have a self-referencing canonical unless you have a deliberate duplicate that should consolidate. Canonical tags are hints, not magic; they work best when the duplicates really are duplicates.

Sitemaps, status codes and redirects

XML sitemaps should list canonical, 200 URLs you want indexed — not redirects, 404s or noindex URLs. Split large sitemaps. Keep lastmod honest if you include it; fake freshness helps no one.

Redirect chains (A to B to C) waste crawl time and dilute signals. Prefer single hops. Use 301s for permanent moves. Soft 404s (a 200 page that says not found) confuse users and bots. After a redesign, a mapped redirect file is not optional — it is the SEO project.

Site architecture and internal links

Important pages should be reachable in a few clicks from the homepage, with descriptive anchors. Orphan pages (no internal links) often remain under-indexed. Pagination, filters and tags need a policy so they do not spawn thousands of thin URLs.

HTML should use a logical heading order and real nav elements. If the main links exist only in a JavaScript widget that never renders as links, you have a discovery problem.

Performance and Core Web Vitals

Slow, unstable pages are a ranking and conversion problem. Measure field data where you have it, and lab data for debugging. web.dev’s Vitals overview (https://web.devexplorevitals) is the canonical explainer of LCP, INP and CLS.

Technical SEO’s job here is to remove obvious blockers: uncompressed images, blocked rendering, layout shift from ads and embeds. Deeper tuning is covered in blog/how to improve website speed and blog/core web vitals explained.

Structured data without theatre

Use structured data that matches visible content: Organization, Article, Product, FAQ only when the FAQ is on the page. Do not mark up fake reviews or prices that users cannot see. Invalid markup is worse than none.

Rich results are a possible side effect, not the strategy. Google documents eligibility in Search Central; eligibility changes. Keep JSON-LD in sync with the page.

Search Console and log-level hygiene

Verify the property, submit sitemaps, watch coverage, experience, and manual actions. Set up email for critical issues. If you have server logs, sampling a week of Googlebot hits tells you whether important sections are crawled at all.

Hreflang, canonicals and sitemaps should agree. Contradictions are a common international bug — see blog/international seo guide.

International and duplicate-language traps

Parameter URLs, print versions, and tracking parameters should be canonicalised or ignored in Search Console. Language versions need distinct URLs and a targeting method, not a JavaScript translate widget that never changes the HTML.

If you are not ready for true multi-region SEO, do not generate twenty thin country pages. That pattern is called out in our services copy for a reason: it helps nobody.