Capability gap · diagnostic

Why doesn't ChatGPT recommend my business?

There's rarely one cause — it's usually a combination of three things: whether assistants can fetch and read your pages at all, whether your content is shaped so a generated answer can quote it, and whether anyone else on the web talks about you in a way an assistant's retrieval pass would surface. None of these are fixed by a trick; they're closer to prerequisites.

1. Crawl access — can the bots even reach you?

Assistant-search products (ChatGPT's web-search tool, Perplexity, Gemini's grounding) fetch live pages using their own crawlers or a shared retrieval index. If your robots.txt blocks GPTBot, PerplexityBot, ClaudeBot, or Google-Extended — sometimes a default in page-builder or CDN templates, not always a deliberate choice — those systems structurally can't read your pages, independent of anything else about your content. Check yours with our AI crawler robots.txt generator and checker.

2. Content shape — is there an answer to quote?

Assistant answers are generated by pulling short, quotable spans out of fetched pages. Pages that bury the direct answer under marketing copy, require JavaScript to render the actual text, or never state the plain fact a buyer is asking about (price, feature, comparison, direct claim) give a generation model nothing clean to lift. This is a content-authoring problem, not a technical one — the fix is usually adding a short, direct "answer capsule" near the top of the relevant page, in plain text.

3. Third-party signal — does anyone else vouch for you?

In our own probe runs (see benchmarks), review-roundup and third-party sites are frequently cited alongside or instead of a vendor's own domain — a domain owner's own marketing page competing against an independent article discussing the same category. Assistants weight this kind of third-party mention because it looks like independent corroboration, not self-promotion. A brand-new domain with no outside mentions has less for a retrieval pass to find in the first place, regardless of how good the site itself is.

4. Structured data and machine-readable signals

Schema.org markup (Organization, FAQ, Product) and an llms.txt file don't guarantee a citation, but they remove ambiguity about who you are and what you offer — useful when a retrieval system is deciding between several plausible sources. Check your current setup with the Organization schema generator or the FAQ schema generator.

What we won't claim

We don't have a formula that predicts whether a specific assistant will cite a specific domain for a specific query — nobody outside the model providers does, and the answer changes with phrasing, model version, and the live retrieval pass. What we can do is show you, with dated evidence, whether you're currently showing up for real buyer-phrased intents, and where the structural gaps above (crawl blocks, missing schema, thin content) actually sit on your own pages.

Frequently asked

Is this the same as SEO?

Related but not identical — see AEO vs. SEO: what actually changes for the mechanical differences.

If I fix all four causes above, will ChatGPT recommend me?

We can't promise that, and you should be skeptical of anyone who does. These are prerequisites that make citation more plausible, not a guarantee — assistant answers are generated fresh per query and vary with phrasing and model version.

How do I know if I'm currently being cited at all?

You can't reliably find out by asking the assistant yourself — see why ChatGPT can't self-report its own citations. Repeated external sampling is the only way to check.

Related

Once you understand why, the next question is what to actually do about it: How to get your business recommended by ChatGPT. For the crawler-access piece specifically: Do AI chatbots crawl your website? Which bots, and how to check.