← ALL POSTS
July 20, 2026 5 MIN READ

How to Check If ChatGPT Actually Recommends Your Business

Most people test their AI visibility by typing their own name. That's the vanity check, not the real one. Your customers don't know your name, they describe their problem. Here's how to test the way they actually search, with prompts you can copy and a way to read the results.

I found you through ChatGPT.every third new client lately

Not Google. ChatGPT. The first few times I honestly didn’t believe it. Then it kept happening.

So I did what you’d probably do. I opened ChatGPT and typed my own name to see what it said about me. It knew who I was, said something nice, and I closed the tab feeling pretty good about myself.

The trap
Typing your own name is just Googling yourself with extra steps.

Every person who found me through ChatGPT had one thing in common: none of them typed my name. They couldn't have. They didn't know it yet.

That’s the whole game with AI visibility. It’s not about what ChatGPT says when someone already knows you. It’s whether ChatGPT brings you up to a stranger who’s describing the exact problem you solve. Two completely different questions, and almost nobody tests the second one.

Test like a customer, not like the owner

When you need something you’ve never bought before, you don’t type a brand. You type your problem.

“Best noise-canceling headphones for an open-plan office.” “Someone who can fix a leaking shower in Brooklyn this week.” “Accountant who actually gets e-commerce.”

You let the AI hand you a shortlist. Then you pick from the names it gave you.

Your buyers do the exact same thing about you. So typing your own name tells you almost nothing. Of course ChatGPT knows your name, it read your website. That’s the easy part. The real test is whether it puts you on the shortlist when your name never comes up.

This is short-tail versus long-tail, moved into AI.

Nobody small wins “web designer.” Too broad, too crowded, the big directories and agencies own it. But “Webflow designer for a SaaS founder who cares about SEO”? That’s specific. That’s winnable. And it happens to be exactly how someone with a real budget and a real problem actually asks.

The long tail is where you can show up. It’s also where the people who’ll actually pay you are looking. So that’s where you test.

The prompts to actually run

Open ChatGPT, Claude, Perplexity, and Google’s AI mode. Yes, all of them. They don’t agree with each other, and your clients don’t all use the same one. Run these in each, and swap the brackets for your real situation.

Start with the problem prompts. This is a stranger trying to find someone like you:

Who's the best [what you do] for [specific type of client]?
I run a [type of business] and I need [the outcome you deliver]. Who should I hire?
I'm looking for someone to [specific job] near [your city]. Any recommendations?
What are the best options for [your category] if I care about [the thing that makes you different]?

Then the comparison prompts, for someone already deciding:

What should I look for when hiring a [your role]?
Who specializes in [your specific thing] for [your specific audience]?

And only at the very end, the vanity one:

What do people say about [your name]? Are they any good?

Run that last one too. It’s not useless, it tells you whether the AI has your basic facts straight. Just don’t mistake it for the real test. If you only exist when someone types your name, you’re findable. You’re not recommendable. Those are different, and only one of them brings you new clients.

One tip: run each prompt two or three times, and reword it. AI doesn’t give the same answer twice. One try is a coin flip, not a read.

How to read what comes back

You’re really looking at four things.

Check 01
Do you show up at all?

Actually named, not lumped into “a few local studios.”

Check 02
Is it specific and correct?

“A Paris web designer” is weak. “A Webflow specialist who helps founders get found by AI” means the machine actually understands you.

Check 03
Are competitors there, without you?

That one stings. It's also the single most useful thing you'll learn all week, so sit with it instead of clicking away.

Check 04
Where is it pulling from?

Your site, a review site, a podcast, a directory. Those sources feed the machine, so they're the levers you can actually pull.

Perplexity and Google’s AI usually show their sources, and ChatGPT will too if you ask. So look at what they cite, then go feed those places.

Roughly, you’ll land in one of three spots, and the whole job is climbing this ladder:

Invisible
Findable
Recommended
You don't come up for the problem at all.
You appear, but only once your name is typed.
You get handed to strangers who never knew you existed.

If you come up empty

Good. Now you have a map instead of a feeling.

If ChatGPT can’t recommend you to someone describing your exact job, it’s not because AI is broken or out to get you. It’s because nothing out on the web tells it clearly what you do, who you do it for, and that real humans trust you. That’s fixable. It’s most of what I write about here, and honestly it’s most of what I do for clients.

But you can’t fix what you never measured. So before you touch your website or your content, run the test. Not your name. The thing your best future client would type at 11pm, the night they finally decide to deal with this.

Your homework
Go find out if you show up.

Keep reading

Rather hand this off? Here’s how I do AI visibility for clients.