Methodology
AI search readiness
These checks look for technical access issues and content clarity on one submitted URL. They do not predict ChatGPT, Gemini, Perplexity or Google AI Overview rankings, and they do not produce a GEO score.
How we count results
Every check is one of: pass, fail, warning, not applicable, or unknown. Each result is labelled measured, heuristic, or owner-confirmed. Unknown checks stay visible and never count as passes.
The report line is “N verified checks passed; N issues; N recommendations; N unverified” — derived from the same rule registry. Training-crawler rules and llms.txt are informational and excluded from those totals.
Scope is always single-page. We state “site-wide linking not assessed.” A one-page finding is not a whole-site assessment.
Checks in this release
Page retrieval
We record HTTP status, the redirect chain, final URL, content type, and whether the response looks like a login or bot challenge. A successful fetch proves access from this scanner, not from every AI provider.
Search-crawler policy
We parse robots.txt for this page path against search-discovery agents (Googlebot, OAI-SearchBot, PerplexityBot). An unavailable robots.txt is unverified, not an automatic block. Training bots are listed separately and never lower this result.
Indexing directives
We read HTML meta robots and the HTTP X-Robots-Tag header, including contradictions. A crawler that cannot fetch the page may never see a noindex directive.
Snippet restrictions
We report nosnippet, max-snippet and data-nosnippet evidence. Impact is explained in context; providers do not treat these tokens identically.
Canonical consistency
We compare the declared canonical with the final URL and check whether the target responds. A missing self-canonical is a recommendation, not an automatic AI blocker.
Readable main content
When rendering is available, we compare initial HTML with rendered content. Critical copy missing from the first response is a compatibility concern, not proof every engine cannot read JavaScript.
Google eligibility evidence
Without authorized Search Console data the result is “not verified.” Public HTML cannot prove indexing, AI Overviews inclusion, or the Search generative AI control.
Page purpose
A heuristic: we quote whether the page states its product, service or question. This is not a ranking factor.
Structured-data consistency
We validate JSON-LD syntax and compare material claims such as prices with visible copy. We do not invent ratings, reviews, authors or an “AI schema.”
Offer completeness
On pages that look commercial, we look for scope, price or a quote path, limitations and a next step. Buyer-clarity recommendations only.
What this is not
- Not a promise that AI products will cite or mention your site.
- Not a measurement of consumer ChatGPT, Claude, or Google AI Mode answers.
- Blocking GPTBot or omitting llms.txt is not a search-readiness failure.
- We do not assign a probability of appearing in AI answers.