BrandReco
All posts

4 min read

Why doesn't ChatGPT recommend my brand?

The five reasons an AI assistant leaves you out of "best tool for X" answers — none of them need a marketing team to fix — and how to see which apply to your site.

You've tried it yourself: you ask ChatGPT, Gemini, Claude or Perplexity for the best tool in your category, and the answer names three or four brands — and none of them is you. Sometimes it names a competitor you know you beat. Sometimes it describes you, but gets the facts wrong.

The good news is that this is rarely mysterious. When an assistant leaves a brand out, it's almost always one of a few concrete, checkable reasons. Here are the five we see most, and what to do about each.

1. The assistant can't read your site

Assistants that look things up on the web — ChatGPT search, Claude's search, Gemini's AI answers, Perplexity — each fetch pages with their own crawler, and many sites block them in robots.txt without realising it. A rule written years ago for "all bots", or a security plugin's default list, is enough.

The crawlers that matter for answers are things like OAI-SearchBot and ChatGPT-User (ChatGPT), Claude-SearchBot and Claude-User (Claude), Google-Extended (Gemini answers) and PerplexityBot. If they're blocked, the assistant can't cite you, can't open your site when a person asks for you by name, and falls back to whatever other people have written about you.

One nuance worth knowing: GPTBot is OpenAI's training crawler, not the one ChatGPT uses to answer. Blocking training crawlers is a legitimate choice — it just shouldn't be confused with being visible in answers.

What to do: check your robots.txt and make sure the answer crawlers above aren't disallowed. The free check at brandreco.com/free-report does exactly this for your domain in about a minute.

2. There's no plain, parseable statement of what you do

Assistants describe a brand from what its pages say. If your homepage leads with a clever tagline and never states, in one plain sentence, what the product is and who it's for, the assistant has to guess — and guesses become the wrong category, the wrong customer, or confusion with a similarly named company.

What to do: put one sentence on the homepage, high up, that a stranger could repeat: what you do, for whom. Publish an llms.txt — a short plain-text file at /llms.txt describing your site for AI assistants — and add Organization structured data to your homepage so the facts are machine-readable too. These are small, one-off changes.

3. You're missing the pages the question is answered from

When someone asks "how much does X cost", "X vs Y", or "best tool for [use case]", the assistant looks for a pricing page, a comparison page, or a use-case page. If a competitor has one and you don't, they win that question by default — not because they're better, but because there was something to read.

What to do: look at which of these pages your competitors have that you don't — pricing, comparison, about, use cases, docs or help, testimonials, case studies — and add the missing ones, even in a short, simple form. A one-screen pricing page with plain numbers beats no pricing page every time.

4. The sources the assistant trusts don't mention you

Answers aren't built from your site alone. Ask an assistant what it relied on and you'll often see review sites, directories and comparison articles — the places where other people describe tools in your category. If you're absent from the sources it cites, you're absent from the answer.

What to do: find out which sources the assistants actually cite for your category's questions, and get onto them honestly — claim the profile, ask real customers for real reviews, answer the question in the community where it was asked. No spam, no fake reviews; they don't work and they get found out.

5. Nobody is checking, so nothing changes

Most founders don't know how the assistants describe them today, have no baseline against competitors, and no clear next step — so the problem stays invisible while competitors quietly teach the AI about their space.

What to do: ask the assistants the real questions your buyers ask, record the answers, and treat each gap as one small task. That's the whole idea behind BrandReco: it runs those questions across ChatGPT, Gemini, Claude and Perplexity, compares your site's coverage with your competitors', and turns what it finds into a few specific actions a day — with the wording or code already drafted — so you can fix things in about fifteen minutes a day, without a marketing background.

Start with the one-minute check

If you only do one thing after reading this, run the free AI-visibility check on your own domain. It tells you whether the assistants are allowed to read your site, whether you have an llms.txt, and whether a sitemap is declared — the first three items on this list — with no account needed.

Want to see how AI agents describe your brand?

The free check takes about a minute — no account, no card.

Why doesn't ChatGPT recommend my brand? — BrandReco