Why does ChatGPT recommend my competitor and not me?

Usually because the wider web describes them as belonging to your category and does not describe you that way. Models draw on training data and live retrieval, both of which reflect third-party sources rather than your own marketing. It is rarely a judgement about product quality.

How do I check if ChatGPT mentions my brand?

Ask the questions your buyers would ask, not questions about your brand by name, and ask them repeatedly across several assistants. Answers vary between runs, so a single check is not a measurement. Counting mentions across many samples is what produces a reliable answer.

Can I ask ChatGPT to remember my business?

No, not in any way that affects other people. Memory features are specific to your own account and conversations. They do not change what the model tells anyone else.

How long does it take to start being mentioned?

If you surface through live retrieval, changes can appear within weeks of new pages and citations being picked up. If your category is answered mainly from training data, it is tied to model training cycles and takes considerably longer.

How to get your brand mentioned in ChatGPT

ChatGPT names a brand either from what it learned in training or from a live web search made during the answer. To be mentioned more, you need to be consistently described across the sources those two paths draw on — which means third-party citations and a clearly-defined entity, not just a better website.

Updated 19 July 2026

First, find out whether you are mentioned at all

Most people asking this question have not actually checked properly. They have asked ChatGPT about their brand once, seen something reassuring, and stopped.

That is not a measurement, for three reasons.

Asking about yourself is the wrong question. Ask "tell me about Acme Cigars" and the assistant will oblige — it is answering a prompt about a named entity. That tells you nothing. The question that matters is the one a buyer asks: "where can I buy Cuban cigars online in the UK". Being named there is visibility. Being described when asked directly is not.

One answer is not a result. These systems are non-deterministic. The same question produces different brands on different runs. A single check is a single sample, and a single sample of a random process tells you very little.

One assistant is not the market. ChatGPT, Gemini, Perplexity and Google's AI Overviews disagree with each other constantly. Being named in one says little about the others.

Doing this properly means asking many buyer-phrased questions, repeatedly, across providers, and counting. That is what an AI visibility measurement is.

Why you are not being named

Assuming you have checked and the answer is bad, the causes fall into a short list.

You are not described anywhere the model learned from

This is the most common and the least comfortable. If the web's discussion of your category — roundups, comparisons, forums, review sites, trade press — does not mention you, then a model trained on that web does not know you belong in the answer. Your website is not the fix, because the model never read your website.

Your name is ambiguous

Brands with generic names get their mentions attributed elsewhere. We hit this ourselves: our own name contains a common phrase, and our first measurements credited us with mentions that were nothing to do with us — the words appeared, the brand did not. We had to build a filter separating an attributable mention from a coincidental phrase match. If your brand shares a name with a common noun, a place or another company, assume you are being both over- and under-counted.

Assistants cannot read your site

If your content requires JavaScript to render, or your server refuses AI user-agents, then the retrieval half of the system cannot use you. This is checkable in minutes and blocks everything downstream. A free scan will tell you.

You are absent from the sources that get retrieved

When an assistant searches mid-answer, it pulls from a fairly narrow set of sources it treats as reliable for that category. If your competitors are on those and you are not, they get named and you do not — regardless of who has the better product or the better site.

What to do, in order

Measure properly first. Buyer-phrased questions, many samples, multiple assistants. Everything else is guesswork without it, and you cannot tell whether a change helped if you never had a baseline.
Fix readability. Server-rendered content, no crawler blocks, clean HTML, structured data. Cheap, fast, and it gates everything.
Make your entity unambiguous. One consistent name and description everywhere. If your name is generic, pair it with a qualifier consistently ("Acme Cigars, the London tobacconist") so the association has something to attach to.
Earn citations on retrieved sources. Identify what actually gets cited in your category — the assistants will tell you if you ask them for sources — and work to be present and accurately described there. This is slow, unglamorous and the highest-leverage thing on the list.
Publish genuine answers. Buying guides, comparisons, honest FAQs. Content that answers the question rather than describing your company.
Re-measure. Same questions, same method, later. Anything else is a story.

What will not work

Asking ChatGPT to remember you. Chat memory is per-user and does not affect what anyone else is told.
Keyword-stuffing your site. Models are not matching keywords.
Buying a mention. There is no placement to buy.
One-off PR. A single article rarely shifts an aggregate impression built from thousands of sources.