A technical GEO checklist covers the engineering layer that decides whether generative engines can access, fetch, parse, and trust your site, before any content is considered. It has five areas: rendering (is content in the HTML response), crawler access (are AI bots allowed and served correctly), structured data (does schema validate and describe the right entities), delivery hygiene (canonicals, sitemaps, status codes, speed to first byte), and observability (can you see which AI crawlers visit and what they fetch). Most technical GEO failures are silent: the site works for humans and Googlebot and returns nothing to an AI crawler. The free scan runs the rendering, access, and schema checks on any URL in a minute.
Technical GEO is where content strategy goes to die quietly. A team can produce excellent, citable articles on a site whose technical layer hands AI crawlers an empty response, and the articles will never be cited, and no one will know why, because every test the team runs uses a browser. This checklist is the engineering pass that has to clear before content spend makes sense, and it is written for whoever owns the stack, not the copy.
Every item below is a place where crawler parity breaks, parity meaning that what an AI crawler receives matches what a browser renders, in content, in access, and in structure.
Area one: rendering (checks 1 to 4)
1. Fetch key templates with JavaScript disabled and confirm the main content, headings, and key facts are in the raw HTML; if they are injected client-side, a crawler that does not execute the required JavaScript receives an incomplete page or an empty application shell, the failure detailed in the empty-shell audit. 2. Confirm server-side rendering, static generation, or pre-rendering covers every template that matters, not just the homepage. 3. Confirm critical content is not deferred behind lazy-load, infinite scroll, or tab components that need interaction. 4. Confirm no content-negotiation logic serves a stripped variant to non-browser user agents. Parity test: diff the no-JS fetch against the rendered DOM; the diff should be styling, not substance. Crawler behavior varies, so test with the documented user agents of the engines you care about.
Area two: crawler access (checks 5 to 8)
5. Audit robots.txt for the AI user agents you intend to allow (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others) and confirm none are blocked by accident, a decision worth making deliberately per the crawler-blocking strategy guide. 6. Confirm your CDN, WAF, or bot-protection layer is not rate-limiting or challenging AI crawlers with interstitials they cannot pass. 7. Confirm important pages are not behind login, paywall, or cookie-consent gates that hide content from non-interactive fetches. 8. Confirm AI crawlers receive the same status codes and canonical URLs as browsers, not redirected to a stub. Parity test: the crawler gets the page, ungated, with a 200.
Area three: structured data (checks 9 to 12)
9. Validate every JSON-LD block; a single syntax error can void the whole graph. 10. Confirm entity types match reality: Organization or Person for who you are, Article or BlogPosting for content, Product with real identifiers for products, not generic WebPage everywhere. 11. Confirm schema is in the initial HTML response, not injected after load where non-rendering crawlers never see it. 12. Confirm @id anchoring links the graph coherently and points at a real entity home. Parity test: the schema a validator sees on the raw response is complete and correct. The eight properties that move citations are the minimum to verify.
Area four: delivery hygiene (checks 13 to 16)
13. Canonical tags are present, self-consistent, and point at live URLs, especially across duplicated or multi-brand content. 14. The XML sitemap lists only canonical, indexable, live URLs with honest lastmod dates. 15. Time to first byte is reasonable, since slow responses raise the odds of fetch failure or incomplete retrieval under constrained crawling. 16. No redirect chains, soft 404s, or mixed status codes on important paths. Parity test: the crawler reaches the intended canonical URL fast, on the first hop.
Area five: observability (checks 17 to 20)
17. Server logs capture user agents, so you can see which AI crawlers visit; log analysis for AI bots is the reference method. 18. You can distinguish training crawlers from live-retrieval fetchers, since they mean different things. 19. You track fetch errors per crawler, because a crawler that keeps hitting 403s stops coming. 20. You have a dated baseline to compare against after changes. Parity test: you can prove, from logs, what the crawler received. Without observability, every other area is a guess.
How do you run the technical pass fast?
The free scan covers the rendering, access, and structured-data checks on any URL with no signup, returning each failure with a fix category, and it tells you specifically which issues are developer work versus copy-paste. That separates the technical layer from the content layer before you spend on either. Once the technical rungs pass, a free account's 10 credits shift the work to content, where the Content Creator diagnoses and closes the gaps the technical layer was hiding. Technical GEO does not replace content relevance; it establishes whether the content can be accessed and interpreted at all. Technical first, content second, always in that order. The business-level version of this list, covering readiness across all four layers, is the GEO readiness checklist.
A stack that passes every test the team owns is still untested for AI if every one of those tests uses a browser. The technical GEO checklist earns its place for one reason above the others: it is the only test in most organizations that fetches the site the way the engines do. That single difference is why sites with clean audits, fast scores, and strong rankings can still be invisible to AI, and why the fix is rarely more content and usually one engineer's afternoon.
Sources
- Google Search Central, JavaScript SEO basics: rendering and what crawlers can process. developers.google.com
- OpenAI, GPTBot and OAI-SearchBot: the crawler user agents and access rules. platform.openai.com
- Schema.org validator: for the structured-data checks. validator.schema.org
- Website AI Score, free scan: rendering, access, and schema checks on any URL. websiteaiscore.com
- Website AI Score, server log analysis for AI crawlers: the observability layer. View article
