Ranking is one game. Getting quoted inside a generated answer is a different one, and it starts with a question nobody checks: is the crawler even allowed in? This scans your page across five pillars of citation readiness and hands you an ordered fix list.
Five pillars, scored separately, because a page usually fails on exactly one of them.
Can a model lift a clean, self-contained answer? We look for question-form headings, whether the sentence immediately after each one is the answer rather than a warm-up, a standalone "X is…" definition, paragraph chunking, lists and tables, and FAQ/HowTo schema.
Answer engines filter by scope before they rank by quality. We check for machine-readable location (LocalBusiness schema with address, geo or areaServed) and — if you supply a target region — whether that region appears in the title, the H1 and the body.
Models quote pages that commit. We count distinct concrete figures, sentences that attribute a claim to a source, hedging density ("it depends" is unquotable), and whether the page carries a verifiable modified date. Plus single-source completeness: can this page answer on its own, or will the engine stitch your page together with a competitor's and credit them?
Author markup, visible bylines, credentials and tenure, and sameAs links that resolve your organisation to a known entity rather than a name the model has never encountered.
The pillar that makes the other four moot when it fails. We parse your robots.txt properly — user-agent groups, longest-match path rules, Allow overriding Disallow — and evaluate 12 AI crawlers individually against your specific URL path: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, meta-externalagent, CCBot and Bytespider.
We also flag nosnippet/noarchive/max-snippet:0, which specifically forbid the excerpt an AI answer is, and whether your content exists in the raw HTML at all.
This scan does not execute JavaScript — on purpose, because GPTBot, ClaudeBot, PerplexityBot and CCBot do not either, so the raw HTML is what they genuinely see. If your content is injected client-side, this scan tells you they are looking at an empty page.
It also cannot tell you whether you are currently cited in a specific AI answer. That needs live querying of each engine on tracked queries, which is monitoring rather than auditing — that is Watchdog.
Treat the scan as a to-do list in order: access first, then how the page is structured to be quoted. If access is the problem, confirm each crawler with the AI Bot Access Checker. If the page is readable but vague, the Schema Builder helps state who you are and what you do in machine-readable form. Our AEO/GEO consultant page explains how we work through this for clients. The Chrome extension can run a check on any page you are viewing.