You don't need software to find out whether ChatGPT recommends your company. You need five minutes, a notepad and a little discipline. The discipline matters most, because one ChatGPT answer can mislead you in either direction.
This guide gives you the exact test, explains why you should repeat it, and shows how to read what you find. It is the same method behind our Who AI Recommends pages.
Why one answer isn't enough
In our November 2026 study we asked ChatGPT three customer-style questions for each of 390 trade-and-city combinations in Texas, each question twice (2,238 answers in total).
- In 48% of markets, the same top business was named in all six answers.
- In about 78% of markets, the top business was named in at least five of the six.
- But on average only 1.4 businesses per market were named in at least five of six answers, and in about 22% of markets no business was named that reliably.
Put plainly: ChatGPT often has a favorite, and then it rotates the rest of the list. If you ask once and don't appear, you may just have drawn a bad roll. If you ask once and do appear, you may have drawn a good one. Asking several times shows you the pattern.
This is a snapshot of ChatGPT's search through OpenAI's API on one day, in Texas. The app on your phone may behave a little differently. Treat the result as a useful signal, not a verdict.
The test
Step 1: Write the questions your customers actually ask
Use plain language with your city. We use these three:
- "Who is the best [your trade] in [your city]?"
- "Can you recommend a good [your trade] in [your city]? Give me a short list."
- "What are the top-rated options for a [your trade] in [your city]?"
Add one or two of your own, such as "who should I call for a leaking roof in [city]" or "emergency plumber near [neighborhood]".
Step 2: Ask each one at least twice, in a fresh chat
Open a new chat each time, so earlier answers don't influence later ones. Ask each question twice (three times is better). If you're logged in, ChatGPT may use your past chats, so try once logged out or in a private window too.
OpenAI says ChatGPT can share your location with trusted third-party providers to supply local information (OpenAI help), so results can differ by where you are. Run the test from your business area if you can.
Step 3: Record what you see
Make a simple table for each question:
| Run | Businesses named (in order) | Were you named? | Sites it cited |
|---|---|---|---|
| 1 | |||
| 2 | |||
| 3 |
Write down the business names and the website links shown under the answer. The links matter as much as the names, because they show where ChatGPT looked.
Step 4: Count the pattern
- Named in most runs: you have a reasonably stable presence for that question. Keep an eye on it.
- Named in some runs: you're on the edge. This is the group where improving your sources can make the most difference.
- Never named: look at who is named and which sources are cited, then see below.
How to read the sources
The sites ChatGPT cites are the most useful part. In our study the most-cited were Angi (78.5% of markets), ContractorsRanked (71.8%), Birdeye review pages (52.3%), BestProsInTown (41.5%) and Thumbtack (37.4%). Yelp, Facebook and Google domains barely appeared. Your results will vary by trade and city.
Ask yourself:
- Are you on the sites it cites? Search for your business on each one. Do you have a claimed profile with correct details? See the directories ChatGPT reads.
- Do those sites show recent reviews for you? Review volume and freshness tend to go with being named. See how many reviews and how recent.
- Are the details right? Wrong phone numbers, old addresses and closed hours confuse everything. ChatGPT builds answers from what the sites say.
- Is your own website readable? Check that your site isn't blocking ChatGPT's search crawler. See how to let OAI-SearchBot in.
- Who is named instead of you, and why? Look at their listings and reviews with the same questions. Why ChatGPT recommends your competitor walks through the usual reasons.
Repeat it monthly
Answers change. Run the same questions on the same day each month and keep the table. After you fix something (a listing, a batch of reviews, your website), you'll be able to see whether the pattern moves, without guessing. No fix is guaranteed to change the result, and a few weeks may pass before sources update.
Test the other AI tools too
Do the same with Google (look at the AI summary above results and in AI Mode) and Perplexity. Only about 5% of citations overlap between ChatGPT, Perplexity and Gemini, according to Yext's research, so you can be visible in one and missing in another.
What this test can't tell you
- It can't tell you how many customers use AI to find you. Most still ask friends and search Google. A 2026 roofing survey found 74% of homeowners follow a recommendation from a neighbor or family (Roofing Contractor).
- It can't prove why you're missing. It shows you symptoms, such as missing listings. The cause is often several things together.
- It can't be gamed. Don't write fake reviews or buy "AI ranking" services that promise results. Nobody can guarantee what an AI tool recommends.
FAQ
What if I'm named but for the wrong services or city? That's a data problem. Check your profiles and website for old descriptions and service areas.
What if ChatGPT says I'm closed or gives the wrong number? Fix the source: your Google profile, Bing Places, Apple Maps and the directories. Make every listing match.
Is there a faster way? Yes. Our free scan runs the questions several times, lists who's named and the exact websites read, and checks your website too.
Skip the spreadsheet. Run a free scan at localpickai.com. You'll see whether ChatGPT names you, who it names instead, which websites it read, and what to fix first. Free, no signup, about a minute.
