How to check whether ChatGPT recommends your business

The short answer: ask it the questions your customers would ask, write the questions down first, and ask the same ones every time. One casual "who's the best plumber near me?" tells you almost nothing, because AI answers change from one try to the next. A fixed set of questions, asked the same way each month, tells you whether you're showing up and whether that's changing.

This is the method behind our AI search visibility work. You can run a simple version of it yourself.

Why does this matter for a small business?

More people now ask an assistant instead of typing into a search box. ChatGPT can search the web and link to the sources it used. Google shows AI-written summaries above some of its normal results. In both cases the customer gets one answer instead of ten links, and the businesses named in that answer get the call.

If you're not in the answer, you're not in the running. And unlike a search results page, there's no page two to scroll to.

How do you build a question set?

Write fifteen to twenty questions a real customer would ask before hiring someone like you. Write them before you start testing and don't edit them afterwards. The value comes from asking the same questions every time, so changing them resets your baseline.

Cover three kinds of question:

• Buying questions. "Who's the best electrician in College Station?" or "Recommend a bookkeeper for a small restaurant in Austin."
• Comparison questions. "Is it worth hiring a local SEO firm or doing it myself?"
• How-to questions in your field. "How often should a water heater be flushed?"

Buying questions matter most, because they're the ones where the assistant names a business. How-to questions show whether the assistant trusts you as a source.

Which assistants should you check, and how?

Start with the ones your customers actually use. For most small businesses that means ChatGPT, Google's AI answers, and Perplexity. If you're not sure, ask a few customers how they found you last time.

How you ask matters as much as what you ask. A few habits keep the results honest:

• Start a fresh chat for every question. Earlier questions in the same conversation shape the answer.
• Don't mention your business in the question. "Who's the best roofer in Austin?" tests visibility. "Is Smith Roofing good?" doesn't.
• Ask from where your customers are. Location changes local answers, so test from your service area when you can.
• Write down the date with every result. These tools update often, and a dated record is the only way to see a trend.

Run the full set twice when you start. Answers vary between runs, and two passes show you which results are stable and which are noise.

What should you record for each answer?

Four things, and never a simple yes or no:

1. Were you cited? Did the answer use your website as a source?
2. Were you recommended? Did it tell the customer to contact you?
3. How were you described? The words it used about you, copied exactly.
4. Who got recommended instead? The competitor it named, if it didn't name you.

Keep cited and recommended separate. A business can be quoted on every how-to question in its field and still lose every buying question to a competitor. Being the source isn't the same as getting the customer, and a report that counts both as a win hides the problem.

Why do assistants recommend one business over another?

Nobody outside these companies knows exactly, and anyone who claims to is guessing. But a few things are clearly within your control, and they line up with what Google already says about local results: relevance, distance and prominence, helped by complete and accurate business information.

In practice that means:

• Your details match everywhere. Name, address, phone and hours should be identical on your website, your Google Business Profile and every directory. Conflicting details make it harder for any system to be confident about who you are.
• Other sites mention you. Local news, industry directories and partner sites give assistants something to build a picture from.
• Your website states the facts plainly. What you do, where you work, your license number if you have one, and your prices if you can publish them. A page that answers a question in its first sentence is easier to quote.
• Structured data describes your business. It lets machines read your details exactly instead of guessing.

There's an overlap with good SEO here, but much of it happens off your website. That's why we run AI search visibility as its own service rather than folding it into a normal SEO retainer.

What do you do with the results?

Look at the "who got recommended instead" column first. If the same competitor keeps winning your buying questions, look at what they've published that you haven't. Often it's something simple: a clear service page, a published price range, or a license number.

Then check the "how were you described" column for anything wrong or out of date. An assistant repeating an old address or a service you no longer offer usually means a listing somewhere still says so. Fix it at the source and check again next month.

Can anyone guarantee you'll be recommended?

No. Nobody controls what an AI says, and there's no paid placement inside these answers. What you can control is whether the facts about you are consistent, easy to find and easy to read. That's what the assistants work from.

So measure it honestly, fix what's in your control, and keep asking the same questions. If you'd rather we ran it for you, our AI search visibility service does exactly this, with the cited and recommended numbers reported separately every month. Or start with a free audit of the site itself.