How to tell if AI is recommending your competitors instead of you

You don’t need a consultant to find out whether AI assistants are sending customers to your competitors. You can check it yourself in about five minutes. Here’s how.

The test

Open ChatGPT, Perplexity, or Google’s AI mode. Now ask it the questions a real customer would ask when they’re ready to buy — not questions about your company, questions about the thing you sell. For example:

• “Who makes the best [your product] in [your city]?”
• “Best [your category] for [your customer’s situation]?”
• “I’m looking for [what you offer] — any recommendations?”

Ask three or four variations, the way a normal person would type them. Then read what comes back and look for one thing: whose names appear.

Reading the result

If a competitor is named and you aren’t — that’s the most important marketing fact you’ll learn this quarter. An AI just made a recommendation to a ready-to-buy customer, and you weren’t in it. That’s not a fluke; it’s a pattern that repeats every time someone asks.

If nobody is named — the AI is giving generic advice because it doesn’t trust any business in your space enough to name one. That’s an opening. The first company to build the right signals becomes the default answer.

If you are named — good. Now check how you’re described. Is it accurate? Flattering? Specific? Or is it vague and a little off? The way the AI describes you is the way millions of potential customers will hear about you first.

Why it says what it says

The AI isn’t judging who’s best. It’s naming who it can most confidently read, trust, and repeat. That usually comes down to a few unglamorous things happening behind the scenes:

• Whether your website spells out what you are in a format machines understand (structured data).
• Whether your key facts — what you do, since when, what makes you different — are consistent everywhere and confirmed by outside sources.
• Whether there’s a clean, quotable description of you sitting somewhere the AI has read.

A competitor beating you here usually isn’t better at the actual work. They’re just more legible to the machine. That’s fixable — and it’s a very different project from “do more marketing.”

What to do with what you found

1. Write down the exact prompts and answers. Screenshot them. That’s your baseline — you’ll want to compare in a few months.
2. Note who got named and how they were described. That’s your real competitive set, as the AI sees it.
3. Look at your own site the way a machine would. Can it even tell what you are, cleanly, without you explaining it out loud? Most sites can’t.

If the test stung a little, that’s useful — it means there’s ground to take, and it’s takeable. Making a business legible to the machines that now make the introductions is a specific, technical job with a real result at the end of it.

Run the test and didn’t like the answer? Send us the screenshots. That’s a great place to start.