You can find out whether ChatGPT recommends your business in about ten minutes, with nothing more than a browser and a notepad. The method is simple: ask the AI engines the questions your customers actually ask, then count how many of the answers name you. Not "does it know my name" but "does it bring me up when a stranger asks for someone like me". That is the only version of "does ChatGPT recommend my business" that matters for new customers.
Most owners who run this test for the first time get zero mentions. That is normal, it is not a verdict on your business, and it is fixable. The point of the test is to get a baseline you can repeat next month, and to learn which websites the engines trust when they answer questions about your trade and your town.
Here is the test, what to write down, and how to read the result.
Why ask customer questions instead of your own name?
An AI engine is a tool such as ChatGPT, Perplexity or Google's AI Mode that answers a question in a written paragraph instead of a list of links. If you type your business name into one of them, it will usually tell you something, because it can search the web and find your site. That proves the engine can find you. It does not prove the engine recommends you.
A recommendation happens when someone who has never heard of you asks "who does this in my town" and your name comes back. Those are the answers that bring in new work, so those are the questions to test. The brand question still belongs in the set, but as a control. It tells you whether the engine has anything accurate to say about you at all.
The ten-minute test, step by step
Set a timer and work through this in order.
- Write five questions a customer would ask. Use the words a real person would use, not your marketing language. Cover the five types listed below.
- Open a fresh chat in each of three engines. ChatGPT, Perplexity, and Google (AI Mode, or the AI Overview that appears at the top of a normal search). Turn web search on wherever there is a switch. A fresh chat matters, because a conversation that already knows your business will keep bringing it up.
- Ask each question in each engine. Five questions, three engines, fifteen answers. Paste the question exactly as you wrote it.
- Write down two things per answer. Which businesses were named, in order, and which websites were cited as sources. Perplexity shows numbered sources. Google's AI answers link out to the pages they used. ChatGPT with search on shows its sources too. The cited sources are the important part, and most owners skip them.
- Save the list of questions. Repeat the same five next month, unchanged, so the results are comparable.
The five question types:
- Service plus city: "Who does residential roof repair in Prosper, Texas?"
- A problem: "My AC is blowing warm air in Celina. Who can come out today?"
- A comparison: "Should a small business in Frisco hire a local web designer or use a site builder?"
- A "best" question: "Best family dentist in Frisco TX"
- Your brand: "Is [your business name] any good?"
Keep them short and natural. If a question reads like an ad, a customer would never type it, and the answer tells you nothing.
A simple scorecard
Count the answers that named your business and write the result as "named in N of 15". A few lines in a notebook is enough:
- Question 1, service plus city: ChatGPT no, Perplexity no, Google no
- Question 2, problem: ChatGPT no, Perplexity no, Google no
- Question 5, brand: ChatGPT yes, Perplexity yes, Google yes
That example scores 3 of 15. Next to each line, note who was named instead and which sites were cited. Two pages of notes is all the test produces, and it is worth keeping, because the comparison with next month's run is where the real information is.
Resist the urge to turn this into a ranking. These engines give a different answer every time you ask, and the order of names in a paragraph is not a position on a list. "Named in N of 15" is the honest number.
What does the result mean?
0 of 15. This is the most common result for a small business, and it is not a sign that anything is broken. The engines mostly draw on directory listings, review sites, local news and well-structured business websites. If you are thin in those places, you won't be named, whatever the quality of your work. It is fixable, and the cited sources tell you where to start.
Named only on the brand question. The engines know you exist and can describe you, but they do not put you forward when a stranger asks. This usually means your site says who you are but not clearly what you do, where, and for whom, in plain text an engine can quote. Our post on how to show up in AI search answers covers the site changes that help.
Named on one or two customer questions. You are on the engines' radar for part of your work. Look at which question types you appeared for and which you didn't. Often a business is named for its main service but not for the problem questions, because nobody has written a page that answers the problem.
The cited sources. This is the list to act on. If Google's answer about plumbers in your town cited Yelp, a Nextdoor thread and a competitor's FAQ page, then you want a complete Yelp profile, a presence in that neighborhood conversation, and an FAQ page of your own that answers the same questions better. The engines are telling you, in writing, where they look.
Common mistakes that make the test lie to you
- Asking only once. One run is a snapshot of one moment. Answers change day to day, which is why you repeat the same five questions monthly and look for the trend.
- Asking in a chat that already knows you. If you spent the morning asking ChatGPT to rewrite your service page, it will happily name you in the next question. Start a fresh chat every time, and consider logging out or using a private window.
- Treating one answer as a rank. Being listed third in a paragraph on Tuesday means nothing on Wednesday. Count mentions across the set, not positions.
- Using your own name in the prompt and calling it a win. "Is Smith Plumbing good?" returning a paragraph about Smith Plumbing is the engine finding your website, not recommending you. Only the customer questions count toward a real result.
- Skipping the sources. Writing down the business names without the cited websites throws away the half of the test that tells you what to do next.
The engines tell you where they look, so the sites they cite are your to-do list.
Want it run for you?
If you would rather not spend the ten minutes, or you want a bigger sample, we run this check free for businesses in Frisco, Celina and Prosper. We write 20 to 45 questions for your trade and your city, run them across several AI engines, and send you a one-page report: how many answers named you, who was named instead, which sources the engines relied on, and the three fixes that would matter most. The details are on our AEO page.
How Deca Digital helps
We are a small studio in Celina, Texas. We build websites that state the facts an AI engine needs in plain text, write the question-and-answer pages the engines like to cite, clean up directory listings so they match, and add the structured data that tells engines who you are. See our services for what that covers. If your site also has a chat assistant, we make sure its answers match what the pages say.
If you'd like a second pair of eyes on your test results, or want us to run the full check, book a free discovery call on our contact page.