Free tool
AI visibility check: can ChatGPT find your website?
Twelve questions, one score, one prioritised list: this check measures whether your website meets the technical and content conditions under which language models find, read and cite it.
Twelve questions about your website
Answer honestly. "Don't know" counts as partial, because uncertainty is itself a finding.
Your result
50
out of 100
Needs work. The foundations stand, decisive signals are missing.
What helps most
Sorted by weight: each point links the article that covers it in depth.
Does your robots.txt allow the AI services' search crawlers (OAI-SearchBot, PerplexityBot, Claude-SearchBot)?
+7 points openAllow the search crawlers in robots.txt; blocking them stops ChatGPT, Perplexity and Claude from indexing your site for search. GPTBot and ClaudeBot, by contrast, collect training data; whether you allow them is a separate decision.
Related articleDo your pages carry structured data (Organization, Article, FAQ) as JSON-LD?
+6 points openAdd JSON-LD for Organization, Article and FAQ; schema is the grammar AI answers build on.
Related articleCan your main content be read in the HTML without JavaScript?
+6 points openRender content server-side or prerender it; most AI crawlers do not execute JavaScript.
Related articleDoes every important page answer its core question in the first two sentences?
+5 points openWrite answer-first: put the answer before the explanation, not behind three paragraphs of introduction.
Related articleDo your texts contain verifiable figures, dates and primary sources rather than opinion only?
+5 points openBack claims with source and date; AI systems preferentially cite verifiable statements.
Related articleAre author, role and company identifiable on every expert article?
+4 points openAdd an author box with name, role and LinkedIn profile; E-E-A-T signals decide citability.
Related articleDo your pages show a visible date, and are they updated when laws or versions change?
+4 points openMaintain datePublished/dateModified and rewrite deadline articles after the deadline passes.
Related articleIs there a machine-readable summary of your offering at /llms.txt?
+3 points openCreate an llms.txt file: it takes minutes and bundles your most important pages for language models. Regular fetching by AI crawlers is barely documented so far, so it carries little weight here.
Related article · Generate llms.txt nowDo you answer the most common customer questions as standalone Q&A blocks?
+3 points openBuild FAQ blocks with real customer questions; they are the format AI overviews adopt most easily.
Related articleDoes your homepage load within 2.5 seconds on mobile (LCP)?
+3 points openCheck Core Web Vitals; Google uses them as a ranking signal and crawlers abandon slow pages.
Related articleDoes every page have a clean canonical URL without duplicates (trailing slash, http/https, www)?
+3 points openKeep canonicals and redirects consistent; duplicates dilute every ranking signal.
Related articleDo your own pages link to each other by topic, not only via the menu?
+3 points openSet internal links from body copy; they give crawlers and AI systems the topical context.
Related article
How the check counts
The twelve points cover three layers: access (may crawlers read the site), structure (can they understand it) and substance (is it worth citing). Access weighs most, because every later layer fails without it: if you block the search crawlers, you do not need an llms.txt. The checkpoints draw on the linked GEO article, the weights are our own choice; the check does not replace measurement. It shows which conditions are missing.
Common questions
What is a good score?
Above 80 is strong: your website meets the conditions under which language models find and cite it. Between 50 and 80, decisive signals are missing: usually structured data, answer-first copy or verifiable sources. Below 50 the website is effectively invisible to AI crawlers, regardless of content quality.
Why does "Don't know" count as partial?
Because uncertainty is itself a finding: if you do not know whether robots.txt blocks GPTBot, you have never checked crawler access. The check measures awareness as much as technology.
Does the check measure my actual visibility?
No, it measures the prerequisites. Whether ChatGPT and Perplexity actually cite your pages is shown by measurement via Search Console and Bing, described in the linked article. The check answers the question before that: whether you are citable at all.
What helps most?
In this order: allow the AI services' search crawlers, make the main content readable without JavaScript, add structured data and answer the core question of each page in its first two sentences. Together these four points are worth 48 of the 100 points. An llms.txt takes only minutes but counts for little, because regular fetching by AI crawlers is barely documented so far.
From score to strategy
The check shows the gaps, not the order in which they pay off for your business. An initial call settles that.
Book an initial call