How to check if your business shows up in ChatGPT (and what to do if it doesn't)

For years the question was whether you ranked on Google. There’s another one now, almost nobody in your sector is asking it yet, and it decides deals: when a buyer asks ChatGPT to recommend a company like yours, are you on the list?

More and more purchases start there. Someone asks a model for a supplier and gets two or three names back. If yours isn’t among them, you don’t exist for that buyer, and nothing tells you. No traffic drop. No alert. You’re just not in the conversation.

The good news is that checking takes ten minutes and costs nothing. Here’s how to do it properly, what the answers actually mean, and what to do with what you find.

How to check, step by step

You don’t need a paid tool for the first look. You need ten minutes and the discipline to do it right, because most people do it wrong: they type their company name, see that ChatGPT knows who they are, and relax. That measures nothing.

Start a clean session. No history, incognito, or logged out. Your usual account already knows who you are and will tell you what you want to hear. You need to see what a stranger sees.

Ask about the service, not your brand. This is the whole thing. Don’t ask “what do you know about [your company]”. Ask what a buyer who’s never heard of you would ask: “which companies would you recommend for [your service] in [your city]”, “best [what you do] provider for a mid-sized company”, “alternatives to [your best-known competitor]”. If you have to name yourself to appear, you didn’t appear. You just introduced yourself.

Run 10 to 15 different questions. Two won’t do. You’ll fool yourself in either direction. Cover the real variants: by city, by customer type, by the problem someone actually has (“my X broke, who do I call”), by comparison. These are the questions people ask before they buy.

Ask each one several times. Models aren’t deterministic. The same question can give you different answers two attempts in a row. Showing up once doesn’t mean you always show up, and missing once doesn’t mean you never do. What counts is frequency, so run each question three or five times and note how often you appear. That percentage is your real number, not a yes or a no.

Record three things per answer: whether your brand comes up, where (first, buried in a list, last), and how they describe you (accurately, vaguely, or plain wrong). A spreadsheet is enough.

Then do it again in Gemini, Perplexity and Claude. Almost nobody bothers, and it’s the step that matters most. Each model recommends a different set of brands. You can be everywhere in ChatGPT and invisible in Perplexity. There’s life beyond ChatGPT, and checking one model is like checking one search engine and calling it a day.

Mentioned isn’t the same as cited

Here’s a distinction almost no article explains properly, and it changes what you do next.

A mention is when the model names you in the text, with no link. That comes from training: if the internet has talked about you consistently, the model has you in its head and says your name. It’s brand awareness. No traffic, but you’re in the room.

A citation is when the model links your site as the source of what it’s saying. That happens when it searches in real time and finds your content clear and credible enough to lean on. Awareness plus qualified traffic from people clicking through.

The difference matters because the fixes are different. If nobody mentions you, your problem is authority: the sources the model trusts don’t talk about you. If they mention you but don’t cite you, your problem is your content: your site isn’t built for a model to use as a source. Working out which one you have is half the diagnosis.

Reading what you found

With the answers in front of you, you’re in one of these. They’re not exclusive: you can be in one with ChatGPT and another with Perplexity.

You show up, described well. Good. You have presence. The job now is defending it and stretching it into more question types, because your competitors want that space.

You show up, described wrong. Worse than it looks. The model is misrepresenting you to buyers. It might have you doing something you don’t do, or quoting prices from three years ago. Fix this before anything else: a wrong mention attracts customers who were never a fit.

You don’t show up, but your competitors do. The most common one, and the most fixable. It doesn’t mean your company is worse. It means your competitors are easier for a model to read: more presence in the sources it checks, better structured content, more external mentions. That’s a visibility problem, not a quality one, and visibility is work you can do.

Nobody shows up, not you or them. That’s an opening. The space in your category is empty. Whoever fills it first gets the recommendations for the next few months.

What to do if you’re not there

The check is the photo. Changing it is work, and it has a name: GEO (Generative Engine Optimization), getting a model to treat you as a reference. Three fronts.

Authority and outside presence. The sources a model checks need to talk about you: press, industry directories, reviews, reference sites. Models cross-check before they recommend anyone. If nobody credible mentions you, the model won’t risk putting you in an answer.

Content a model can actually use. Clear, direct answers to the questions your customers ask, with verifiable numbers, properly structured. A model can’t interpret vague. If your site says you offer “integrated high-value solutions”, it has no idea what you sell. If it says what you do, who for, and how, with numbers, it can cite you. And watch the JavaScript: if your content only exists after the page renders, some crawlers never see it.

The technical base. Crawlers need permission to read you, schema markup should identify your company as an entity, and your information has to be consistent everywhere. Small things that decide whether you make it into the answer.

None of this is a trick or a switch. There’s no form to fill in to get listed in ChatGPT. It’s slow work, similar in logic to SEO but with different levers, and it compounds.

When doing it by hand stops working: tools

The manual check is perfect for the first photo and for an occasional look. It has a ceiling, though: it doesn’t scale. Tracking 30 or 50 questions a week, comparing month to month, or watching several competitors at once stops being realistic by hand.

That’s what these are for. They fire your questions at the models on a schedule, collect the answers, and give you numbers: how often you’re mentioned, in what tone, and what share of recommendations you’re taking versus your competitors. The ones people actually use in 2026:

Otterly.ai and Peec AI are the accessible end, and the ones most marketing teams and agencies land on. They track your brand across ChatGPT, Perplexity, Gemini and others, with clear dashboards and competitor comparison.

LLM Pulse covers mentions, citations and sentiment across ChatGPT, Perplexity, Gemini and Google AI, including AI Mode and AI Overviews. It also analyses which sources get cited and gives you prioritised fixes, which is the part that turns a dashboard into a plan.

Profound is aimed at large companies: a lot of data depth and prompt volume, priced accordingly.

Semrush and Sistrix have both added AI monitoring to their suites, which makes sense if you already live in one of them.

One warning about all of them. Measuring isn’t appearing. A tool will confirm week after week what you already knew, which is that you’re not there. It’s the thermometer, not the medicine. The data tells you what to fix first. It doesn’t do the fixing.

Frequently asked questions

How do I know if my business shows up in ChatGPT?

Open ChatGPT in a session with no history and ask it 10 to 15 questions a customer would ask before hiring you, about the service and the city, not about your name. Run each question a few times, since the answers vary. Note whether your brand appears, where, and how it’s described. Then repeat in Gemini, Perplexity and Claude, because each model recommends a different set of brands.

Why doesn’t my business show up in ChatGPT if I rank first on Google?

They’re different systems, so your Google position doesn’t carry over automatically. That said, ranking well on Google makes it considerably more likely you’ll show up in a model: the sources it checks and the authority signals it reads overlap heavily with what SEO already works on. It’s a strong head start, not a guarantee.

What’s the difference between being mentioned and being cited?

A mention is your name in the text with no link, which is brand awareness. A citation links your site as the source, which also brings traffic. They’re fixed differently: mentions depend on your authority and outside presence, citations depend on how your content is structured.

How often should I check?

Every one to three months is reasonable if you’re actively working on visibility. Models update unevenly and answers move, so a periodic photo tells you whether you’re improving, slipping, or watching a competitor take your space.

Do I need a paid tool for this?

Not to start. The free manual method is enough for the first photo and for an occasional check. Tools earn their place when you want to track a lot of questions continuously, measure change over time, or keep an eye on several competitors at once.

We can help you get there

If you’ve run the check and your business isn’t showing up, or it’s showing up worse than your competition, that’s fixable. We’ve been doing GEO since 2024 as a specialist agency, built into every SEO project. Tell us about your case and we’ll give you an honest read on where you stand in AI answers today and what can be done about it.