Does ChatGPT Recommend Your Business? Ask It and Find Out

TL;DR. Paste the interview prompt into ChatGPT, Claude, or Gemini and answer its six questions. It writes you a five-prompt test script personalized to your business. Run the script in fresh chats on at least two engines, fact-check the answers with prompt 2, and extract who's getting cited with prompt 3. Three outcomes: you're recommended (protect it), you're absent (seven fixes, mostly DIY), or the AI is wrong about you (correct the sources). If you'd rather not do the runs yourself, the $300 AI Visibility Audit does all of it across engines and hands you a priced fix list, credited in full toward any work after.

What the AI says about you is your new first impression

Your customers changed where they ask the first question. It used to go into Google; now it goes to ChatGPT, Gemini, or Perplexity, and the answer that comes back is your first impression, before your website, before your reviews, before you know a buyer exists. This isn't theory for us: one of our own recent leads found us by asking Gemini for help with their problem. Not Google. The conversation started with whatever Gemini said about us, and we weren't in the room for it.

Here's the strange gap. Search this exact question and every result tells you how to get recommended. Marketing tips, all the way down. Nobody answers the literal question: what does the AI say about you, today, when a real customer asks. That's what the three prompts below find out. No dashboard, no email gate, no subscription: a fresh chat and ten minutes. If you want the bigger picture on why this channel now matters, is SEO dead after AI covers it; short version, Google still matters, it just stopped being the only front door.

Prompt 1: the AI visibility interview (it writes your test script)

Paste this into ChatGPT, Claude, or Gemini. It asks you six short questions, one at a time, then hands you a five-prompt test script built from your actual business: your customer's money question, your city, your competitors, your name. Generic tests miss because customers don't ask generic questions.

You are a search and AI visibility specialist. Interview me about my business, then write the exact test script I need to find out what AI engines say about me.

Rules for the interview:
- Ask ONE question at a time and wait for my answer. No multi-part questions.
- Ask at most 6 questions. Skip any you can already answer from what I have said.
- Push back once when an answer is vague. "Everyone within driving distance" is not a service area.

Cover, in this order: my business name exactly as customers say it; my city or service area; my core service in plain words; my top 3 competitors by name; the one money question a customer asks right before buying from me, in the customer's own words; and where customers currently find me.

Then output MY TEST SCRIPT with exactly these five prompts, each personalized from my answers, each under a numbered heading:
1. My customer's money question, word for word
2. "best [my service] in [my city]"
3. "[my business] vs [my closest competitor]: which one should I pick and why?"
4. The near-me variant: "[my service] near me" with my neighborhood or city named
5. "should I use [my business name]? give me an honest assessment"

End the script with two instructions: run each prompt in a FRESH chat in at least two of ChatGPT, Gemini, and Perplexity, and save every raw answer, because the next prompt in this guide needs them. Keep the whole script under 250 words. No analysis, no advice, just the script.

Copy this prompt

The script it writes is the whole point. Testing "best plumber in Austin" tells you something. Testing the exact question your customers ask right before they spend money tells you whether the machine is sending them to you or to the shop across town.

How to run your test script

Three rules, and all three change your results.

Fresh chats only. If you test in the account where you've been chatting about your business for months, the model knows you and gets polite. Your customers get the cold answer, so that's the one you need to see. Open a fresh chat with memory off, or use a logged-out window.

At least two engines. They don't agree. Each reads different sources, so being recommended by ChatGPT says nothing about Gemini. Perplexity shows its citations right on the answer, which makes prompt 3 much easier. Ten extra minutes buys you the real map.

Save the raw answers. Copy the full text, sources included, into a doc with today's date on it. Prompt 2 needs the raw text, and next month's run needs a baseline to compare against.

Prompt 2: assume it's wrong until proven right

AI answers sound equally confident when they're right and when they're inventing your opening hours. So if the engine mentions you at all, don't celebrate yet. Paste its answer into this prompt, in a fresh chat, and let a second AI tear it apart. Wrong hours, a stale price, a service you dropped two years ago, or a mix-up with a similarly named business in another state: all of it loses real customers, quietly.

Here is what an AI engine said when I asked about my business: [paste the full answer]

Here are the facts: [your business name, city, the services you actually offer, real hours, and your price range if it's public]

Your job is a fact-check, not a pep talk. Assume the answer is wrong until my facts prove otherwise, and say ACCURATE only on specific evidence, never on my enthusiasm or the model's confident tone.

Go claim by claim:
- List every checkable claim in the answer: hours, prices, services, service area, address, phone, reviews, years in business.
- Mark each claim VERIFIED (matches my facts or a source you can point to), WRONG (contradicts them), or UNVERIFIABLE (no way to confirm).
- Check whether the answer is mixing me up with a similarly named business. The tells: services I do not offer, a wrong city, a different industry.

Then output exactly:
1. One verdict line: ACCURATE, PARTLY WRONG, or WRONG BUSINESS
2. Every wrong or unverifiable claim, most damaging first
3. The likeliest source of the worst error: my own website, a stale directory listing, or a name collision with another business

Copy this prompt

Run it in a fresh chat too. A model that helped produce an answer will defend the answer.

Prompt 3: who's actually getting cited, and from where

This one turns a bruised ego into a to-do list. AI engines don't recommend from vibes. They assemble answers from sources they trust: directories, review platforms, local roundups, trade sites, your competitors' pages. Paste the raw answer from your "best [service] in [city]" test into this prompt and it shows you which businesses won and which sources put them there.

Here is the raw answer an AI engine gave for "best [my service] in [my city]": [paste the full answer, including any sources or links it showed]

Extract two lists. No commentary between them.

LIST 1: BUSINESSES MENTIONED
Every business named in the answer, in order of appearance, one per line. Then add one final line: MINE PRESENT: YES or NO.

LIST 2: SOURCES CITED
Every website, directory, publication, or review platform the answer cites, links, or clearly draws from, one per line. If no sources are shown, write NO SOURCES SHOWN, then list the 5 sites this kind of answer most commonly draws from for local service queries.

Close with one line telling me which single source from LIST 2 I should get present on first, and why that one.

Copy this prompt

List 2 is the entire game. Every site on it is a place your business either appears or doesn't, and getting present there is mostly grunt work, not magic. How to get cited by AI search walks through the mechanics.

Reading the results: the three outcomes

Every run ends in one of three places. Be honest about which one you're in.

You're recommended. The good problem. Protect it: keep hours, prices, and services current everywhere the engines read (your site, your Google Business Profile, every source from list 2), and don't delete or redesign the pages that earned the mention. Re-run the script monthly, because these answers move when their sources move.

You're absent. The most common result, and the most fixable. The engines aren't ignoring you; the sources they read just don't mention you, whatever your Google rank says. How to get your business recommended by ChatGPT is the fix list: seven fixes, and most of them you can do yourself with an hour a week. The GEO checklist puts them in working order.

The AI is wrong about you. The worst case, because a wrong answer works against you while you sleep. You can't edit ChatGPT, but you can correct the sources it reads: fix your own site first, then your Business Profile, then whatever stale directory listing the reality-check prompt pointed at. If it's confusing you with a similarly named business, make your footprint unambiguous: same name, same city, same service, spelled the same way, everywhere.

When to do it yourself, and when to pay someone

Do it yourself first. You just ran, for free, the diagnosis plenty of agencies would happily put on a retainer, and most of the seven fixes need your knowledge of your own business more than they need a specialist. If you've got an hour a week, the get-recommended guide plus this month's test script is a complete loop. Don't pay anyone just to find out what the AI says. You now know how to ask.

Pay for help in two honest cases. If you want the full picture without doing the runs, our $300 AI Visibility Audit runs every check for you across the engines and hands you the fix list with a price on each item, credited in full toward any work after, so the audit costs nothing if we end up doing the work. And if you're in a competitive local market where the answers keep shifting, Ongoing Rank from $500/month keeps you present and monitored. Take that only where it genuinely fits: a business with real local competition and no time, not a business that hasn't done the basics yet. Comparing outside help? Best GEO and AEO agencies is our honest read of the field, us included.