Why Is My Business Not Showing Up In ChatGPT?

You ask ChatGPT to recommend someone who does the kind of work you do. It gives you a neat little list, and your business isn't on it. The same competitors can appear every time, which is especially puzzling when your business is well established, ranks well in Google and has plenty of information about it online.

The obvious reaction is to start looking for what you've done wrong. Is the website too vague? Can ChatGPT not find it? Have those competitors discovered some mysterious AI-search trick while you were getting on with the rather less mysterious job of running the business?

The frustrating part is that one missing result can't answer any of those questions. It shows you what happened for that particular search, but it doesn't explain why.

So before you rewrite half the website, publish another 20 articles or add whichever AI tactic is being recommended this week, look at what you actually asked. ChatGPT can recognise your business when you mention it by name and still leave it out when someone asks for a provider, a recommendation or help with a particular problem. Change the wording or context of the question and the answer can move again, which is why a few searches that show the same pattern tell you far more than one disappointing answer.

When someone says, “My business doesn't show up in ChatGPT,” my first question is usually: "Well, what did you ask it?"

It sounds like a small detail, but it completely changes what the answer means. There's a big difference between asking:

and asking:

In the first question, you've already handed ChatGPT the business name. It only has to find or pull together information about that specific business.

The second question gives it a choice. ChatGPT has to decide which businesses, people or sources seem to match the help described in the prompt. So it's a discovery query.

Which is how you can get an accurate description when you ask about your company by name, then disappear when you ask for businesses that do the kind of work you do. It's hardly reassuring if you were hoping to be recommended, but it shows that being recognised and being chosen are 2 different tests.

A branded query can show whether ChatGPT identifies the right business and what information it connects with it. A discovery or recommendation query gets much closer to what most business owners actually want to know: do I appear when someone is looking for what I sell?

Once you start asking that second kind of question, the answer becomes much more changeable than a simple visibility score suggests.

As soon as you move into discovery searches, “Does my business show up in ChatGPT?” becomes a much less tidy question.

Customers don't all describe the same need in the same way, and your business won't be an equally strong match for every version of it. Take an AI consultant. One person asks:

If you ask one question, don't appear, and immediately decide your website needs fixing, you've skipped most of the investigation.

We've already seen that a branded query and a discovery query are doing different jobs, and that 2 discovery questions can produce different recommendations. And this is how one search can turn into a website rewrite before you have any idea what the result means:

For an initial check, I'm trying to answer a fairly modest question:

When someone asks ChatGPT for the kind of help this business provides, what tends to happen?

You can learn a surprising amount from just a few checks:

• Ask about the business by name and see how ChatGPT describes it.

• Try several discovery questions that reflect the different ways a real customer might describe the need.

• Notice which businesses keep appearing instead of focusing only on whether yours appears once.

• When ChatGPT searches the web, look at the sources behind an answer that seems wrong, incomplete or out of date.

• Repeat the searches that matter most and see whether the same pattern holds.

For current public information, check whether ChatGPT is actually using the web. It can decide to search automatically when a question would benefit from web information, and you can also choose Web search yourself when you want to make the lookup explicit.[1] If it answers without searching, I wouldn't assume that response reflects the latest public information about the business.

This is still only the first layer. A proper DIY testing method soon becomes a piece of work in its own right, so I deal with that separately. For now, the aim is to move beyond “I didn't appear” and describe what is happening with a little more accuracy.

For example:

Once you can describe the pattern properly, the investigation stops spreading in every direction. You can ask a smaller set of questions that can actually lead somewhere.

This is where the investigation earns its keep, because producing a long list of things you could change is very easy.

There is no shortage of advice telling you to add content, rewrite pages, install schema, publish FAQs, check bots, claim profiles and generally keep yourself occupied for the foreseeable future. The harder question is which, if any, of those things the evidence actually supports.

What you find could point to:

• old or conflicting facts about the business that need correcting;

• a service or category the business wants to be found for but doesn't describe directly enough, either on the website or in other public sources;

• information that is missing, difficult to find or unavailable to search when it should be easy to reach;

• directories, profiles or other public sources telling an older or different story from the one the business tells now;

• a repeated gap around one kind of search, even though the business appears for other closely related questions;

• no convincing problem at all, in which case a major rewrite prompted by one disappointing answer would be a rather expensive guess.

Each of those findings calls for a different response. An outdated directory listing doesn't call for a website rewrite, and a service page that never quite says who the service is for won't be fixed by publishing another 20 blog posts. If the business already appears for the searches that genuinely fit it, there may be very little reason to change anything.

Sometimes you find a plausible gap without enough evidence to act on it yet. If competitors describe a service more directly, that gives you something specific to check across the website and other public sources. It still doesn't prove that changing one page or phrase will alter ChatGPT's recommendations.

This is why I treat “Why doesn't my business show up in ChatGPT?” as an investigation question. The missing result is the starting clue. It doesn't arrive with a diagnosis and a tidy list of fixes attached, inconvenient though that would be.

If you're happy to do the checking yourself, test the business across a well-chosen set of discovery questions and compare what you find. And if you'd rather have someone investigate the searches, the sources, the competitors and the website together, that is what my AI Visibility & Accuracy Audit is designed to do.

Either way, you want to come back with an answer to the question you actually asked:

"Why am I not appearing, and is there anything here that gives me a good reason to change something?"