How to Find Out What People Ask AI About Your Expertise (If You're a Coach or Consultant)

You find out what people ask AI about your expertise by farming the questions from the places people already ask you: your sales calls, conversations with your own customers, Reddit, YouTube, social media, and what people ask your competitors. You cannot read ChatGPT's logs, and you don't need to. The engine is not making anything up. It builds its answer by asking a bunch of smaller questions a person with context about your business would ask, and those are questions you have been hearing for years.

Here's how I explain it to a coach or consultant, and the order I'd do it in.

Key takeaways

AI is not making stuff up. Half the time it literally goes to Google, asks a bunch of related questions, gets context about your business, and folds the answers into one reply. The questions it asks are the questions your buyers ask.
The best source is your own experience: sales calls, customers in person, Reddit, YouTube, social media, and the questions people put to your competitors. The better you get at farming those, the more questions you have.
More questions is the point, because each one is a page where you can show real first-hand knowledge, and that is what gets you named.
Once you have the list, ask the engines each question and see who gets named. That is the free check, and it tells you which ones to write first.

What the engine is actually doing with your question

Ask ChatGPT or Claude "what is the best ice cream shop in New York City" and watch what it does before it answers. It looks up Brooklyn, Manhattan, Hoboken, the Bronx, the West Village, the East Village, gelato shops, reviews, and then folds all of that into one answer with a few names in it.

That is the whole secret to this question. The reality is that AI is not making up stuff. It pulls from different sources. Half of the time it literally just goes to Google and asks a bunch of different questions related to the question a person might ask. It gets context about your business, figures out what a person who has context about your business might ask, and then returns an answer.

So when someone asks it "should I hire a leadership coach", it is quietly asking what a leadership coach costs, what the first session looks like, how long it takes, how you know if it worked, and who does it well for someone in this situation. If you can answer all of those back questions the AI needs to ask in order to create that synthesized answer, why wouldn't AI pick the person it got all of its answers from?

Where to farm the questions from

The correct way to find them is to pull from your existing experience, the places where people already ask you things. In the order I'd go:

Your sales calls. What does the first ten minutes of a call sound like? Write down the questions word for word, the way the person said them, not the way you would phrase them. "How do I know if I need a coach or just a course" is a question. "Coaching vs. courses" is a keyword.
Your own customers, in person. What people ask you when you are speaking with them, at an event, after a session, in the follow-up email. The questions they ask after buying are the ones the engine uses to judge whether you know what you're doing.
Reddit. Search your subject and read how strangers phrase the problem before they know your category exists. This is where the wording comes from that nobody would ever type into a keyword tool.
YouTube. The comments under any video in your subject are a list of unanswered questions, and the video titles that get views tell you which ones people are actively trying to solve.
Social media. Same thing, shorter. What gets asked in replies and DMs.
What people ask your competitors. Their reviews, their FAQ pages, the comments on their posts. A question a buyer asked them is a question a buyer will ask AI, and right now the competitor's page is the answer.

The better you get at farming those questions from all those different sources, the more questions you'll have, and the more likely you'll be to show up in ChatGPT for that question, because you're showing your real first-hand knowledge when it comes to answering it. That is the mechanism. Not more keywords. More questions you can answer better than anyone, because you have actually been asked them.

What it looked like when I ran it on myself

I sell an AI visibility product, and the first time I ran my own free check on my own business it came back zero out of eight. What that made me realize was the questions that were shown there. I looked at them and thought, "I'm not sure how I would create content that would show up for that." I did not have pages that answered what a buyer actually asks before hiring someone like me. I had pages about what I do.

That is the gap this exercise finds. What AI says about me doesn't speak to my skill or my ability. It just speaks to what content AI is able to find on me. The same is true for you, and the question list is the map of what it cannot find.

Then check which ones name someone else

Once you have twenty or thirty questions, ask ChatGPT and Perplexity each one, in the buyer's words, and write down who gets named. Some will name a competitor. Some will name nobody in particular. A few might already name you. The ones that name a competitor are the pages to write first, because a buyer is asking, an answer is being given, and it is not you.

The free Visibility Check does the first eight of these for you: it asks ChatGPT and Perplexity eight buyer questions for your niche and shows how many name you versus your competitors. It is the fastest way to see whether the exercise above is urgent or merely useful. The AI Visibility Tracker then keeps the full list, checks every question every week, and tells you the day an engine starts naming you.

And then you write the page. One question, one page, with the answer first and something on it that came from actually being asked. If you want the structure I use, it's here.

Frequently asked questions

Can I see the actual questions people type into ChatGPT?

No. There is no Search Console for ChatGPT. But you don't need the logs, because the engine builds its answers from the same questions your buyers already ask you. Farm those and you have the list.

How many questions do I need?

More than you think, and more than you'll have on day one. Every question you can genuinely answer from experience is a page you can be named for. Start with the ones from your last five sales calls and add from Reddit, YouTube and your competitors' pages as you go.

What if my coaching is too personal to reduce to questions?

Then your sales calls are full of them and you have stopped hearing them as questions. "Will this work for someone who's tried therapy" is a question. "How long before I notice anything" is a question. Write them down exactly as people say them.

Do I need a tool for this?

Not to find the questions. Your inbox and your call notes are the tool. Where software helps is asking every question across every engine every week and telling you when an answer changes, which is the part nobody keeps up by hand.

If you've farmed your list and can't tell which ones to write first, shoot me a message on LinkedIn with three of them. I'll tell you which one I'd start with.