Can AI See Your Website? How to Check (and What to Fix First)

Here's a test most business owners have never run: can ChatGPT, Claude, and Perplexity actually read your website? Not "is it online" — can an AI crawler fetch it, parse it, and extract facts confident enough to repeat to a customer? We've graded hundreds of local business sites, and a shocking share fail this test without knowing it. Invisible isn't a ranking problem; it's a plumbing problem.

The five ways websites go dark to AI

1. Blocked crawlers. Security plugins, CDN "bot fight" modes, and copy-pasted robots.txt files routinely block GPTBot, ClaudeBot, PerplexityBot and friends. The site works fine in a browser; to AI, it returns a locked door. This is the most common failure and the fastest fix.

2. JavaScript-only content. Many AI fetchers read raw HTML and don't execute scripts. If your text only appears after JavaScript renders — common with cheap page builders and some site platforms — an AI crawler sees an empty shell. Your beautiful site is, to the machine, a blank page.

3. No structured data. Without LocalBusiness, Service, or FAQPage schema, the AI must infer your hours, service area, and offerings from prose. Inference is where models get things wrong — or decline to answer, and recommend the competitor whose facts are machine-readable.

4. Buried answers. AI systems extract passages that directly answer questions. If your pricing page opens with three paragraphs of brand story before mentioning a number, the extractable answer may effectively not exist.

5. Slow or flaky hosting. AI crawlers have short timeouts and shallow patience. A site that takes eight seconds to respond gets sampled less, cited less, trusted less.

The ten-second check (and the ten-minute one)

We built a free tool that runs the whole inspection at once: the AI Visibility Checker. Give it your URL and it scores you 0–100 across eight checks — AI-crawler access in robots.txt, JSON-LD schema, FAQPage markup, llms.txt, answer-first structure, question headings, meta description, and content depth — with a plain-English fix list. It's the same audit we run at the start of client projects, minus the invoice.

Want the manual version? Three spot checks: (1) open yourdomain.com/robots.txt and look for AI bot names under "Disallow"; (2) in your browser, view the page source and search for a sentence from your homepage — if it's not in the raw HTML, scripts are hiding it; (3) ask ChatGPT "what does [your business] in [your city] do?" and see whether it answers from your site or shrugs.

What to fix first

Work the failures in the order above — access, rendering, schema, structure, speed — because each layer depends on the one before it. Perfect schema means nothing if crawlers are blocked; brilliant answer-first copy means nothing if it only exists in JavaScript. Most sites clear the whole list in an afternoon: a robots.txt edit (generator here), a schema block (generator here), and a rewrite of two page openings.

Why bother this month, not someday

AI recommendations compound. Models re-crawl, re-verify, and slowly build confidence in businesses whose facts check out from every angle — which means the visibility you establish now is the recommendation engine's memory later. Every month your site stays dark to AI, a competitor's gets more legible. Run the check, fix what it flags, and if the flags run deeper than an afternoon — a rebuild-grade rendering problem, say — that's literally what we do.