Resources ·

How to check what AI says about your business

You can check what AI says about your business yourself, free, in about twenty minutes: open ChatGPT, Google Gemini, and Perplexity, ask each one the questions a customer would ask — without naming your business first — and read the answers, including the sources they cite. This guide gives you the exact questions, what to look for in the answers, and how to tell a real problem from normal noise. No tools to buy, nothing to sign up for, and no catch: this is the honest first step whether you ever hire anyone or not.

Which assistants to ask

Use the assistants your customers actually use. The free tiers are enough:

  • ChatGPT — make sure web search is on (it usually is for questions like these; you’ll see it searching)
  • Google Gemini — including a plain Google search, where AI Overviews now answer many local questions
  • Perplexity — the most citation-forward of the group; every answer shows its sources
  • Claude and Meta AI if you have another ten minutes

Two habits make the results more trustworthy. Start a fresh conversation for each round, so earlier questions don’t color later answers. And remember the assistant may factor in your location — that’s fine, it’s what your customers’ assistants do too.

The four kinds of questions

Customers ask assistants four different shapes of question, and each one tests something different about your business. Replace the bracketed parts with your own trade and town.

Discovery — the big one. Ask the way a stranger would, without naming yourself:

“Who does [mobile detailing] in [Spokane]?” “Best [dentist] near [Boise] for [same-day crowns]?”

This tests whether you exist in the answer at all.

Comparison — conditions attached. Real customers qualify:

“Who does [tint] in [city] and is open Saturdays?” “Which [med spa] near me takes walk-ins?”

Pick the conditions you actually win on. If you’re open Saturdays and the answer names only competitors who aren’t, you’ve found something specific and fixable.

Factual — now name yourself. This stops testing visibility and starts testing accuracy:

“What are [Your Business Name]’s hours?” “What services does [Your Business Name] offer, and what’s the phone number?”

Reputation — the trust check:

“Is [Your Business Name] reputable? What do reviews say?”

How to read the answers

For each answer, three passes:

1. Are you there? On discovery and comparison questions, are you named at all? Who is named instead? Don’t over-read a single miss — but note it.

2. Is each fact right? On the factual questions, check every claim — hours, phone, address, services, prices — against reality. Sort each one into correct, wrong, or missing. Wrong and missing are different problems: wrong means a source somewhere holds a bad fact; missing means no source states it at all.

3. Where did it come from? This is the pass most people skip, and it’s the one that turns findings into fixes. Expand the citations on each answer and note which sources the assistant read — your site? Your Google profile? A directory you’ve never heard of? A competitor’s page? When an answer is wrong about you, the citation trail usually shows you exactly which source fed it the bad fact. When a competitor gets named instead of you, the trail shows which source earned it for them.

One more test: turn the web off

If the assistant lets you disable web search, ask about your business with it off. You’ll most likely get vagueness, an honest “I don’t have information about that,” or a confident guess. This isn’t a trick — it’s the clearest demonstration of the underlying fact: AI has no memory of your business. Everything useful in the earlier answers came from sources retrieved in the moment. That’s why the sources are the whole game.

What people usually find

The findings run to type:

  • Absence. Discovery questions answered entirely with competitors. Usually a sources problem — you’re missing from what the assistant reads — not a quality-of-your-work problem.
  • Stale facts. Old hours, a previous address, a dead phone line — usually traceable through the citations to one out-of-date listing.
  • Phantom details. Services you don’t offer, staff who don’t exist. These get called hallucinations, but when you trace them they usually start from one bad source — we once traced a nonexistent practitioner to a single misspelled name in a patient review (the write-up is on our results page).
  • A split identity. The assistant seems unsure which listing is you, or describes two versions of your business. Look for duplicate profiles.
  • A competitor with better sources. Their citation trail is deep — profiles, reviews, press, a thorough website — and yours is thin. Unwelcome, but it’s a to-do list, not a verdict.

Before you conclude anything, ask twice

Assistants vary their answers. The same question, asked twice, can name different businesses. So don’t build conclusions on one run: ask each discovery question two or three times, ideally on different days. A business that’s absent from every run has a real gap. A business that appears in two runs out of three doesn’t have a crisis — that’s ordinary variance. This is also why you should be skeptical of any screenshot — ours included — presented as proof on its own: single answers are anecdotes. Patterns are data.

When this is enough — and when it isn’t

For many businesses, this DIY check is genuinely enough to act on. If you found a wrong phone number on a directory, fix the directory. If you found a duplicate Google listing, resolve it. If assistants cite a directory you’re not on, get on it. You don’t need to pay anyone to do those things, and doing them will help.

Where DIY runs out is the same place it does everywhere: repetition and follow-through. Checking once tells you where you stand today; knowing whether you’re improving means running the same questions the same way, month after month, keeping dated transcripts, checking every claim against a verified record, and being statistically honest about whether a change is real or noise. That’s a measurement discipline more than a task — and it’s the part we sell, so weigh our take on it accordingly.

If you’d like a second pair of eyes on what you found, the free check is exactly that: we show you what three assistants answer about your business, which sources they cited, and the one gap we’d fix first — yours to keep, whatever you decide.

← All resources

See what AI says about your business

On a short call, we show you what three assistants answer about your business today, which sources they cite, and one obvious gap — free, no obligation.

Free AI Visibility Check