The GEO checklist
Start with the website work supported by provider documentation: accessible pages, useful answers and accurate business information. Then measure actual visits and appearances. A technical readiness check is a starting point, not a promise of AI citations.
Start with crawlable, indexable pages
Check important URLs load successfully, that internal links reach them, and that robots rules or noindex directives do not block content you want indexed. Google says pages must satisfy its search requirements before they can appear in its generative AI search features. Eligibility does not guarantee inclusion.
Separate search access from training access
Read each provider's crawler documentation before changing robots.txt. Anthropic distinguishes ClaudeBot for model development from Claude-SearchBot for search and Claude-User for user-directed retrieval. Allowing or blocking a training crawler is not a universal switch for AI search. Make the access decision for each purpose explicitly.
Publish information worth using
Answer the questions your customers actually ask, with direct explanations, original evidence and clear headings. Keep business identity, product scope, contact details and prices consistent. Improve the page for readers rather than splitting it into disconnected fragments to chase a hypothetical citation.
Use structured data when it matches the visible page
Accurate structured data can describe a business, product or article. Keep it aligned with the facts visitors can read. Google says there is no special schema required for generative AI search. Adding FAQ markup or extra schema does not guarantee citations.
Treat llms.txt as optional documentation
An llms.txt file can be maintained for services that use it, but Google states it does not use these files for search visibility or rankings. A missing file is not evidence that your website is invisible to AI. Prioritise access, useful content and accuracy first.
Earn relevant references
Share original research, practical examples or tools with audiences who can use them. Keep real company profiles accurate. A directory listing is useful when it helps people discover or verify your product; a collection of low-quality links is not a shortcut to guaranteed rankings.
Measure the result rather than the checklist
Compare complete periods in Search Console and your website analytics. If testing an AI answer, record the provider, question, date and response. A single answer is one observation, not a stable market-share measure. Track visits and useful enquiries alongside any technical scores.
Frequently asked questions
Do I need llms.txt to appear in Google AI search?
No. Google states that its search systems do not use llms.txt or special AI text files for visibility or rankings. Other services may use them for their own purposes.
Do I have to allow every AI training bot?
No. Training, search and user-directed retrieval are different purposes. Read each provider's crawler guidance and choose the access rules you want.
Can a readiness scan prove I get AI citations?
No. A scan can inspect public website signals. Actual citation claims require recorded responses from the relevant provider.
Will this checklist increase traffic?
It can help remove technical friction and improve useful content. No traffic increase is guaranteed. Measure visits and enquiries over complete, comparable periods.
Turn website findings into an action plan
Create an account to start a Full Audit of your website, with supporting evidence and a prioritised action plan.
Start Full Audit