How to Check if ChatGPT Recommends My Business in 30 Minutes

Want to know how to check if ChatGPT recommends my business? You do not need a paid tool to find out whether ChatGPT or Google's AI mentions your business. You need five buyer questions, a notebook or spreadsheet, and about 30 minutes. Do it once to get a baseline, then repeat it every week.

This test is manual, so it is slow. It shows you the gap clearly, which is the first step.

TL;DR

• Write five questions a buyer would type, not your brand name.
• Ask each one three times in ChatGPT and once in Google's AI answer.
• Log whether your brand appears, and which competitors do.
• Repeat every week for four weeks and look at the trend.

Step 1: Write five buyer questions (10 minutes)

Write the questions a buyer would type, not your brand name. Use these patterns:

• Best [tool type] for [job]. Example: "best scheduling tool for a dental clinic"
• Alternatives to [competitor]
• Is [competitor] worth it, or should I pick something else
• How do I [task your product solves]
• What is the cheapest [tool type] for a small team

Choose the five that match a sale you would like to win. Write them in a spreadsheet, one per row.

Step 2: Set up your log (5 minutes)

Make four columns:

1. Question
2. Tool and date (for example "ChatGPT, 27 Oct")
3. Your brand named? (yes, no, or partly)
4. Competitors named (list them, in the order they appear)

Add a column for the link or source if the tool shows one.

Step 3: Ask each question three times in ChatGPT (10 minutes)

Use the same browser and the same account for every run. Ask the question exactly as written. Open a new chat for each question, so earlier questions do not shape the answer.

Record whether your brand appears in each run. If it appears in two of three runs, record "2 of 3". Note the order of the names, because the first names often get more attention.

Step 4: Repeat in Google's AI answer (5 minutes)

Search the same five questions in Google. Note whether an AI overview appears, and whether it names your brand or links to your site. Google does not always show an AI answer, so record "no AI answer" when that happens.

Step 5: Read the sources (5 minutes)

For each question, look at the sources the AI used, if it shows them. Ask yourself:

• Which sites were named?
• Are any of them yours?
• Are any of them places where you could ask for a listing or a mention?

This is the most useful part of the test. It shows you the outside sites that the AI trusts for your category.

An example log

Here is an illustrative log for a made-up dental clinic, so you can see the format:

A row like the first one tells you where to start. The second row shows a question where the buyer already sees alternatives, but not you.

What the results tell you

• Named in most runs for most questions: Keep your pages current and check again in a month.
• Named for some questions only: Write a clear answer page for the questions you are missing.
• Named in none: Start with the answer on your site, then ask for one mention on an outside source the AI cited.
• Competitors named, you are not: Read their pages and the sources. Look at what makes their answer easy to quote.

A study by SparkToro and Gumshoe found that the same prompt gave the same list of brands in fewer than 1 in 100 runs. So do not judge a single run. Look for a pattern across three runs and across several weeks.

Common mistakes in the test

• Changing the question between runs. Small wording changes can change the answer. Keep the exact text in your log.
• Using a logged-in history with many old chats. Old chats can shape the answer. Open a new chat each time.
• Checking only once. One result is a snapshot. The trend is the signal.
• Counting a mention of a competitor's name as your mention. Check the spelling and the context.

What a good result looks like

A good result is not a perfect score. It is a pattern you can act on. Look for three signs over the four weeks:

• Your brand appears for at least one of the five questions in most runs, and the share rises over time.
• The sources the AI names include at least one site you could ask for a mention.
• The competitors that appear often are the ones you can compare yourself with, so you know which claims to match or beat.

If you see none of these after four weeks, the problem is usually the answer on your site, not the test. Go back to the question that matters most to your buyers and make sure one page answers it directly.

Keep the log even when the results look bad. A record of weeks with no mention is still useful, because it shows you exactly when a change did or did not help. Many people stop after the first poor result. The businesses that improve are usually the ones that keep checking.

Repeat it every week

Run the same five questions at the same time each week. Keep the log. After four weeks you will see whether your share of mentions is rising, flat or falling. That trend is the signal to watch, not any one answer.

If you want the same test run for you each week, with the answers logged in one place, that is the job of a tracking tool. The mention rate guide explains the metric. For the reasons your results may change, read why AI answers change from one run to the next.

Start a free trial of AgenticSEO to run this check every week without the spreadsheet.

Check your AI visibility free: see which prompts name you, and which name your competitors. Start free