Reachable, Extractable, Trusted: The 3-Minute Self-Check for AI Visibility

Most AI-visibility advice runs 2,000 words deep before it tells you where to actually look. Here's the short version: three checks, pass or fail, that tell you exactly which layer is holding your business back from AI recommendations.

Run through them in order. Each one is a hard dependency for the next, there's no point fixing your content if the AI can't reach the page, and no point building trust signals if the AI can't extract a clear answer once it arrives.

Check 1: Reachable, can the AI even get to your site?

This is the one businesses skip because it feels too basic to be the problem. It's usually the problem.

• Open yourdomain.com/robots.txt. Does it explicitly Allow GPTBot and OAI-SearchBot? These are two different crawlers, GPTBot feeds OpenAI's training data, OAI-SearchBot powers live ChatGPT Search citations. Blocking either one closes a door.
• Does your key content render without JavaScript? Most AI crawlers read the raw HTML they get on the first request. If your service pages only populate after a client-side script runs, the crawler sees a blank page.
• Is anything behind a login wall, a cookie-consent gate, or a bot-blocking security rule that also catches legitimate AI crawlers?

Fail this check and nothing else matters. For the full technical breakdown, the exact robots.txt syntax, which crawlers belong to which engine, and how to verify access, see how to get your website cited in AI search answers.

Check 2: Extractable, can the AI pull a clean answer out of the page?

Reachable isn't the same as readable. The AI can load your page and still fail to understand it.

• Does the first 50 words of your service page state who you are, what you do, and where you operate? Not a mission statement, a plain sentence an AI could quote directly.
• Do you have schema markup (LocalBusiness, Organization, FAQPage) telling machines explicitly what your business is, rather than leaving it to inference?
• Do your FAQ headers match the actual questions people type into ChatGPT, not the questions your marketing team thinks sound good?

A page that passes Check 1 but fails Check 2 gets crawled and ignored, the AI has the content but nothing citable inside it.

Check 3: Trusted, would the AI actually believe the answer?

This is the layer most businesses under-invest in, because it's the slowest to build and the hardest to fake.

• Is your business name, address, and phone number identical across your website, Google Business Profile, and every directory you appear in? Inconsistency reads as three separate, less-credible entities instead of one trusted one.
• Do your reviews contain specifics, what you did, where, for whom, or do they just say "great service"? Specific reviews are what AI models actually cite.
• Has anyone besides you said your name recently? A mention in a listicle, a local publication, or an active subreddit carries more weight with AI engines than anything on your own site.

For the full playbook on building this layer, NAP consistency, review generation, schema, and third-party mentions, see our complete guide to getting ChatGPT to recommend your business.

Score yourself

Three checks, three possible outcomes:

You fail at...	What's actually broken	Fix first
Reachable	AI crawlers can't load your pages	robots.txt, JS rendering
Extractable	AI loads your pages but can't quote them	Direct answers, schema, FAQ
Trusted	AI can quote you but won't recommend you	NAP consistency, reviews, mentions

Most businesses that feel invisible to ChatGPT assume the problem is Trust, reviews, reputation, authority. Often it's Reachable or Extractable, and those are the two-hour fixes, not the six-month ones.

For the full playbook once you know which layer is broken, see how to get cited by ChatGPT, Perplexity, Gemini, and Google AI Overviews.

Not sure which one you're failing? Run a free scan at lookelo.com and see exactly where ChatGPT, Perplexity, Gemini, and Google AI Mode stand on your business.