How to check whether ChatGPT recommends your company: a 10-minute test for managers

Your customers are already asking AI assistants who to trust. Whether your company is in the answer is something you can check yourself, today, without a technician. This is the test we would run first with any company.

You need a spreadsheet or a blank page, ten minutes, and access to at least two assistants (for example ChatGPT and Gemini). Open a new conversation in each one so earlier chats do not colour the answer.

Step 1: write the eight questions your buyer would ask

Think of a prospective customer who does not know you yet. Write eight questions in the words they would use, not in your industry's vocabulary. Cover four types, two of each:

• Discovery: "Who are the best [your service] companies in [your city]?"
• Comparison: "What is the difference between [your company] and [main competitor]?"
• Trust: "Is [your company] reliable? What do people say about it?"
• Decision: "What should I check before hiring a [your service] provider in [your city]?"

The discovery questions matter most, because they are the ones where a buyer who has never heard of you can meet you for the first time.

Step 2: ask, and record

For each question and each assistant, write down four things:

1. Was your company named? Yes or no.
2. Which companies were named instead?
3. What did it say about you, word for word? Copy it.
4. What sources did it show, if it showed any?

Do not judge yet. The value of the test is in the record.

Step 3: repeat once

Ask each discovery question a second time in a fresh conversation. Assistants do not always answer the same way twice, so a company that appears in one answer out of two is in a different position from one that appears in both. Mark which results were stable.

Step 4: read the results

Three patterns cover most outcomes.

You are not named on discovery questions. The assistant did not treat your company as one of the answers. The test cannot tell you why. It tells you where to look: how clearly your site says what you do and where, and whether independent sources describe you the same way.

You are named, but described wrongly. Wrong services, wrong city, an outdated fact. This is often easier to fix than absence, because it points to specific pages or profiles that say something different from your own site.

You are named on trust questions but not on discovery ones. The assistant knows you when asked directly, but does not offer you to strangers. That is common for companies with a good reputation and little published material explaining what they do.

What this test does not tell you

It is a snapshot, not a measurement. Eight questions and two assistants cannot show how often you appear across all the ways your buyers phrase things, and it cannot tell you what to change. We say this plainly because a quick test can create false comfort or false alarm. Use it to decide whether a deeper look is worth it.

If you would like us to run the same questions at larger scale and against your competitors, that is what our audit does. You can ask for it below; we reply with a one-page summary.