What our AI visibility check covers — and what it doesn't
Our homepage shows an AI visibility score, and every report includes it as its own dimension. A sceptical buyer, agency or journalist will ask: "what exactly do you check?" This page answers that question in full. Every item listed is something the crawler actually measures today — we don't publish checks the product doesn't perform yet.
What we check (the A1–A10 checklist)
The checks below follow the framework that AI-search experts and the major platforms (OpenAI, Perplexity, Anthropic, Google, Apple, Microsoft) document publicly. Each item names its evidence basis so you can verify it independently.
A1 — AI search crawler access in robots.txt
Why it matters: If OAISearchBot, PerplexityBot, Bingbot, Googlebot or Applebot are blocked, your site can't appear in ChatGPT search, Perplexity, Copilot, Google AI Overviews or Siri/Spotlight.
What we check: Checks robots.txt for explicit Allow/Disallow directives on 14 agents: 9 AI search crawlers (OAI-SearchBot, ChatGPT-User, PerplexityBot, Claude-Web, Claude-SearchBot, CCBot, Bingbot, Googlebot, Applebot) and 5 training crawlers reported separately.
Evidence: OpenAI, Perplexity, Anthropic, Apple, Google, Microsoft public documentation.
A2 — Training-vs-search split
Why it matters: Blocking training crawlers (GPTBot, ClaudeBot, Google-Extended) is a legitimate choice and does not remove you from AI search. Blocking search crawlers does.
What we check: Reports training-crawler blocks as 'Informational — your choice', not as errors. Search-crawler blocks are flagged as 'High' severity. Only search-crawler blocks affect the AI visibility score.
Evidence: OpenAI (GPTBot vs OAI-SearchBot), Google-Extended explainer, Applebot-Extended documentation.
A3 — Firewall / CDN blocking
Why it matters: Cloudflare and similar services can block bots even when robots.txt allows them — a silent cause of AI invisibility.
What we check: Detects Cloudflare challenges, CAPTCHAs and access-denial pages (403, 503 challenge walls) during the crawl. Flags blocked pages so you know the crawl couldn't reach them.
Evidence: Cloudflare, Google: 'allowed… by any CDN or hosting infrastructure.'
We re-fetch a sample of pages with each AI crawler's published user-agent string, so a CDN/firewall that allows us but blocks GPTBot/PerplexityBot/etc. is flagged as a High finding.
A4 — Indexability signals
Why it matters: Google says AI Overviews only use pages that are indexed and can show a snippet.
What we check: Checks: noindex tags (pages excluded from search), canonical URL presence, sitemap coverage (parses sitemaps to discover pages).
Evidence: Google Search Central.
nosnippet and max-snippet directives are on the roadmap — not yet checked.
A5 — Content in text form
Why it matters: Google lists 'important content is available in textual form' as a best practice. AI crawlers read raw HTML — content loaded only by JavaScript or embedded in images may be invisible.
What we check: Detects JavaScript-rendered shells (React, Next.js, Vue, Remix) and, on Gold tier, diffs the raw HTML against a headless-browser render to flag critical elements served only via JS.
Evidence: Google Search Central.
Content inside images or PDFs (rather than HTML text) is on the roadmap.
A8 — Sitemap presence (Bing readiness)
Why it matters: Copilot answers come from Bing's index. A sitemap helps Bing discover your pages.
What we check: Checks that at least one sitemap is referenced in robots.txt. The full crawl also parses sitemaps for URL discovery.
Evidence: Microsoft Learn.
Bing Webmaster Tools and IndexNow recommendations are on the roadmap.
A9 — Internal linking
Why it matters: Google lists internal linking as a factor for AI features too. Orphan pages and broken links waste crawl budget and confuse visitors.
What we check: Reports orphan pages (no internal links point to them) and broken internal links (return 404, 410, 5xx or are unreachable).
Evidence: Google Search Central.
A10 — llms.txt
Why it matters: llms.txt is a proposed standard (llmstxt.org) for machine-readable site guides. Google says it isn't needed for AI Overviews, but several LLM platforms look for it.
What we check: Checks whether /llms.txt exists and contains meaningful content (≥20 characters). Shown as 'Informational' in findings — not scored heavily.
Evidence: llmstxt.org, Google Search Central.
Answer-friendly structured data
Why it matters: FAQ, HowTo and Article schema (JSON-LD) make your content quotable — LLMs can extract Q&A pairs and structured facts directly.
What we check: Reports the percentage of crawled pages carrying FAQPage, HowTo, Article or QAPage schema. Flags when a site has JSON-LD but none of it is answer-friendly.
Evidence: Schema.org, Google Search Central.
What we don't check yet (roadmap)
| nosnippet / max-snippet directives | We check noindex but not the snippet-control directives Google says matter for AI Overviews eligibility. |
|---|---|
| Content in images or PDFs | We detect JS-only content, but do not yet flag pages whose key information is inside hero images or downloadable PDFs rather than in HTML text. |
| NAP consistency across pages | We do not yet extract business name/address/phone from structured data and cross-check for consistency — a trust signal for AI answers. |
| LocalBusiness / Organization schema validation | We detect schema presence and parse types, but do not yet validate that Organisation schema fields (name, address, phone) match visible page text. |
| Bing Webmaster Tools / IndexNow guidance | We check sitemap presence, but do not yet generate a specific Bing-readiness recommendation. |
| Recommended robots.txt / JSON-LD generator | On the roadmap: a copy-paste robots.txt block and LocalBusiness JSON-LD snippet pre-filled from the crawled site. |
Check your site's AI visibility now — free snapshot, in seconds.
Audit my site free Buy a report — from €29
Also see: what the full audit covers · AI search visibility explained