A GEO readiness checklist is a set of checks that tell you whether a website is ready to be cited by generative engines like ChatGPT, Gemini, Claude, and Perplexity. The checks fall into four groups: can engines read the site, can they understand its structure, can they verify who it is, and do they currently cite it. Below is the working checklist, 20 checks, in the order to run them, with what a pass and a fail look like. You can run the whole list by hand in an afternoon, or let the free Website AI Score scan run the core checks on any URL in about a minute with no signup. Readiness is a sequence, not a score: each group depends on the one before it.
GEO readiness is the honest first question before any generative-engine work: is this site even in a state where optimization can pay off. Most are not, and the reason is almost never content. It is a broken layer underneath, unreadable delivery, chunk-hostile structure, an unverifiable entity, that makes every downstream effort return nothing. A readiness checklist exists to find that layer before you spend on the layers above it.
Think of it as a ladder with four rungs, each a prerequisite for the next, and run it bottom-up, recording the first failed dependency before moving up, because a fail on rung one makes rungs two through four irrelevant until it is fixed.
Rung one: can engines read the site? (checks 1 to 5)
1. Fetch a key page without JavaScript and confirm the main content is present in the raw HTML; if it is not, you have the empty-shell problem and nothing else on this list matters yet. 2. Confirm robots rules are not blocking the AI crawlers you want (GPTBot, ClaudeBot, PerplexityBot, Google-Extended) by accident. 3. Confirm important pages return 200 and are not gated behind logins, interstitials, or cookie walls that swallow the content. 4. Confirm the sitemap lists your canonical URLs and nothing dead. 5. Confirm no critical content lives only in images, PDFs, or embedded widgets an engine cannot parse. Pass: the crawler receives what the browser shows. Fail: any gap between the two.
Rung two: can engines understand the structure? (checks 6 to 10)
6. Each important page states its direct answer within the first hundred tokens, per the inverse pyramid rule. 7. Heading hierarchy is clean, one H1, logical H2 sections, each section self-contained enough to be quoted alone. 8. Tables and lists are real HTML, not images or div soup, so they survive extraction. 9. No section depends on a previous one to make sense, because engines lift passages, not pages. 10. Key facts (prices, specs, dates, credentials) appear as plain text, not only in interactive elements. Pass: a random paragraph, read in isolation, still answers something. Fail: the meaning only exists at page level. The content-level version of these checks, run against a specific target phrase, is the pre-publish AEO checklist.
Rung three: can engines verify who you are? (checks 11 to 15)
11. Structured data is present, valid, and parses without errors in a validator. 12. The organization or person behind the site is declared with the schema properties that move citations, not just a generic WebPage type. 13. There is a true entity home, one page that unambiguously defines who you are in machine-readable form. 14. Your name, description, and identity agree across your site and the external profiles engines trust, anchored with sameAs. 15. Authors of expert content are named, credentialed, and consistent. Pass: an engine can confirm you exist and are who you say from independent sources. Fail: you are a claim with no corroboration.
Rung four: do engines currently cite you? (checks 16 to 20)
16. Query the major engines for your core topics and record whether you appear. 17. Record how often across repeated queries, since a single result is noise. 18. Check how the engines describe you, and whether the description is accurate and current. 19. Check whether competitors appear where you do not. 20. Record a dated baseline so next month's run measures change. Pass: cited, accurately, on the topics you own. Fail: absent, or present and described wrong. This rung is the only one that needs the engines themselves, which is why reading citation patterns over time is the readiness measure that matters most and a layer many technical checklists do not measure. Treat it as an observation, not a deterministic pass or fail: it records what the engines currently do, and that can change without the page changing.
How do you run this in a minute instead of an afternoon?
The free scan runs the core of rungs one through four on any URL, no account: it fetches the page as a non-rendering crawler, checks structure and schema, and scores current standing across six LLMs, returning the failing checks with a category for each fix. That gives you the readiness verdict in about a minute. The work of closing the gaps, restructuring content, writing the missing direct answers, filling the entity layer, is what the 10 free credits on a new account and the Content Creator are for: diagnose a page's specific gaps for one credit, generate or rewrite the fix for four, then re-scan to move up the ladder. A readiness checklist identifies the prerequisites for citation; it cannot guarantee citation for every query.
A checklist run once, ticked, and forgotten is the way most readiness work ends, and the ladder does not hold that way. A framework update can drop you off rung one, a template change off rung two, a redesign off rung three, without anyone noticing until rung four goes dark. The sites that stay citable are the ones that re-run the ladder on a schedule, because readiness is a state you maintain, not a certificate you earn.
Sources
- Google, AI features and your website: Google's stated requirements for appearing in AI Overviews and AI Mode. developers.google.com
- Schema.org validator: for rung-three structured-data checks. validator.schema.org
- OpenAI, GPTBot documentation: crawler access rules for rung one. platform.openai.com
- Website AI Score, free scan: runs the core readiness checks on any URL in about a minute. websiteaiscore.com
- Website AI Score, five citation patterns: the rung-four measurement over time. View article
