What does AI say about your business?
There is no reason to pay anyone for results you can see yourself. This page gives you the exact questions to ask ChatGPT, Google, or Claude, a simple way to run them so the answers actually mean something, and a plain way to read what comes back. If you find gaps and want to learn how to fix them, that is what a discovery call is for.
One answer tells you almost nothing.
Ask an AI assistant who the best in your trade is, and you get an answer. Ask again tomorrow and you may get a different one. The assistants rotate which sources they lean on from run to run: Google data one time, Yelp the next, then review sites, directories, and business websites. A single answer is weather, not climate.
What matters is consistency. Out of many answers, how often are you the recommendation: 0% of the time, 10%, 50%, 70%? That rate is your starting point. The gap between your rate and the businesses at the top of your market is the gap worth closing.
Pick one assistant, fresh chat
Use the assistant your customers use, and start a brand new chat for every run, so your history and earlier questions do not tilt the answers.
Run all 12, three days in a row
Same questions, same assistant, once a day for three days. It takes about 15 minutes a day.
Tally after every run
For each answer, mark whether you were named, whether you were recommended, and who was recommended instead of you.
The 12 questions your customers are asking.
Fill in the brackets first: your trade, your town, your main service, and your business name. Then copy each question into the chat, one at a time.
Say your business is Mike’s Plumbing in Boise, and your main service is water heater installation:
Question 2 becomes“I need a plumber in Boise. Who do you recommend?”
Question 5 becomes“How much does water heater installation cost in Boise?”
Question 11 becomes“What do you know about Mike’s Plumbing in Boise?”
Who is the best [your trade] in [your town]?
I need a [your trade] in [your town]. Who do you recommend?
What are the top-rated [your trade] companies in [your town]?
Who offers [your main service] in [your town]?
How much does [your main service] cost in [your town]?
Which [your trade] in [your town] is best for [a common job in your trade]?
Who is the most trusted [your trade] in [your town]?
Which [your trade] in [your town] has the best reviews?
I’m comparing [your trade] options in [your town]. Who should I avoid, and who stands out?
Is there a [your trade] in [your town] that specializes in [your specialty]?
What do you know about [your business name] in [your town]?
Is [your business name] a good choice for [your main service] in [your town]?
Tally it up.
After each run, fill in that day’s columns. Named means your business appears anywhere in the answer. Recommended means the assistant actually puts you forward as a choice, not just a mention. In the last column, write who got the recommendation when it was not you. Those names are your real competition in AI answers.
| Question | Day 1 | Day 2 | Day 3 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Named? | Recommended? | Recommended instead | Named? | Recommended? | Recommended instead | Named? | Recommended? | Recommended instead | |
| 1. Best in town | |||||||||
| 2. Who do you recommend | |||||||||
| 3. Top-rated companies | |||||||||
| 4. Who offers your service | |||||||||
| 5. What your service costs | |||||||||
| 6. Best for a common job | |||||||||
| 7. Most trusted | |||||||||
| 8. Best reviews | |||||||||
| 9. Who to avoid, who stands out | |||||||||
| 10. A specialist like you | |||||||||
| 11. What AI knows about you | |||||||||
| 12. Are you a good choice | |||||||||
Your recommendation rate
Count every Recommended mark across all three days, divide by 36 (12 questions, 3 days), and multiply by 100. That percentage is your recommendation rate. Do the same count for Named if you want the full picture.
As a rough guide to what the rate means:
- 0%For these questions, AI does not know you exist. It cannot recommend a business it cannot describe.
- Up to 25%You surface occasionally, when the assistant happens to lean on a source that knows you.
- 25 to 60%You are in the mix. You get recommended regularly, but someone else is the safer answer more often.
- 60% +You are close to the default answer in your market. Protect it, because rates move as sources change.
Take the worksheet with you.
The worksheet puts the 12 questions, the three-day protocol, and the tally grid on two printable pages, so you can work at the kitchen table instead of at a screen.
Download it now
Free PDF, no sign-up, no strings. Print it, fill in your brackets once, and use the same sheet all three days.
Download the worksheet (PDF)Or have it emailed
We will send the same PDF to your inbox so it is there when you sit down to run the questions.
What the gaps usually mean.
It describes you differently every time, or gets basic facts wrong.
That is a clarity problem. The assistant does not hold one clear picture of who you are, what you do, and where you work, so it improvises.
It knows your name, but never recommends you and never points to your website.
The evidence for you is scattered. Your website, your Google profile, your reviews, and your listings do not agree with each other, so the assistant leans on a business that is easier to check.
The same competitors keep taking the recommendation.
Right now, they are the understandable option in your market. The fix is not a trick or a hack. It is making your website, Google profile, reviews, and directories tell one consistent story, and keeping it current.
One honest note: nobody can promise you will be recommended every time. Anyone who promises that is selling you something.
Found gaps you want to close?
Schedule a discovery call. On the call, we run the measured version of this same panel for your business and walk through what fixing the gaps would take, in plain language, before you spend anything.
Schedule a discovery callA note on method: these are the same questions we use when we measure a client’s starting point, and the same ones we run again after a build to check progress. Answers vary from run to run and from person to person. That variation is the entire reason for the three-day protocol: you are measuring a rate, not collecting a verdict.