The 10 best ChatGPT prompts to check a local business's AI visibility

Why it's worth typing this in before you pitch it

We analysed 10,000 US businesses across ChatGPT and Perplexity for our AI Visibility Report 2026, and the numbers make an uncomfortable case for checking before you assume. ChatGPT can identify 93% of businesses when asked directly, but of those it recognises, only 55% have their details described accurately. Worse, 41% of businesses aren't recommended at all for the kind of unbranded, "who should I use" question a real customer would actually ask.

That gap between existing and being recommended is exactly what these prompts are built to expose. You don't need a subscription or a scraper to start finding it, you need ChatGPT and about ten minutes. Below are ten prompts that surface the same gaps agencies are increasingly getting paid to fix, each one worth running on a client or prospect before you assume you already know the answer.

A quick note before you start: treat every answer as a single snapshot. AI answers shift between chats, models and even the same prompt asked twice, so one good result doesn't mean the job's done, and one bad one doesn't mean it's unfixable. That variability is itself worth showing a client.

Quick reference: what each prompt tests

The 10 prompts, and why each one matters

1. The straight awareness check

Prompt: "Tell me about [Business Name] in [City]. What do they do and what are they known for?"

This is the baseline. If ChatGPT can't answer this one with any confidence, or answers with details that sound generic rather than specific to the business, everything else on this list is secondary. Our data found ChatGPT recognises 93% of businesses on some level, so a blank or vague answer here is a genuine red flag, not just an edge case.

Or run it automatically: rather than typing this prompt into ChatGPT, Perplexity and Gemini separately for every prospect, run a free AI visibility check and get the awareness score across all of them in one go.

2. The unbranded discovery test

Prompt: "What's the best [service, e.g. plumber/dentist/hair salon] in [City]?"

This is the prompt that actually matters commercially. A business existing in ChatGPT's knowledge is one thing, being the answer to a question nobody asked it by name is another entirely. Our research found 41% of businesses aren't recommended for exactly this kind of unbranded query, even when ChatGPT clearly knows who they are.

Or run it automatically: the free AI visibility check tests unbranded recommendation queries as standard, no need to guess which phrasing might trigger a mention.

3. The head-to-head comparison

Prompt: "Compare [Business Name] and [Competitor Name] for [service] in [City]. Which would you recommend and why?"

This surfaces something a single-business prompt can't: the actual reasoning ChatGPT uses when it has to choose. You'll often see it cite review volume, service breadth, or opening hours, the same fundamentals our data shows genuinely move the needle on AI recommendation.

Or run it automatically: Insites' Competitor Audit runs this comparison at scale across up to three local rivals, with the gap colour-coded rather than buried in a paragraph of prose.

4. The accuracy audit

Prompt: "What are [Business Name]'s opening hours, address and phone number?"

Awareness and accuracy are two different scores, and our data shows the gap between them is wide: of the businesses ChatGPT recognises, nearly half have some detail wrong. Wrong hours or a dead phone number in an AI answer doesn't just look sloppy, it actively costs the business a customer who trusted the answer.

Or run it automatically: accuracy is one of four dimensions scored automatically in every free AI visibility check, cross-referenced against the business's actual listings.

5. The reputation check

Prompt: "What do people think of [Business Name]? What are they known for, good or bad?"

This is where review data quietly does most of the work. Our research found AI responds positively about 80% of the time when asked directly about an SMB, and that reputation score is built almost entirely on the review volume and sentiment sitting on Google and other directories, not on the business's own marketing copy.

Or run it automatically: the free check scores reputation as positive, neutral or negative per platform, so you don't need to interpret ChatGPT's tone by eye.

6. The real customer framing

Prompt: "I need a [service] near [neighbourhood/City] and I'm not sure who to use. Any suggestions?"

Phrasing matters more than most agencies expect. A prompt that reads like a genuine, slightly unsure customer often surfaces a different (and more honest) answer than a direct "what's the best X" query. Run both and compare; the difference itself is worth showing a client.

Or run it automatically: the free check standardises this so every prospect gets tested the same way, removing the guesswork around phrasing.

7. The niche and service-depth prompt

Prompt: "Which [service] in [City] specialises in [specific niche, e.g. emergency call-outs, cosmetic dentistry, wedding hair]?"

Generic category prompts test whether ChatGPT knows a business exists. This one tests whether it understands what the business actually offers beyond the category label, which depends heavily on whether that detail is clearly written on the website and Google Business Profile in the first place.

Or run it automatically: website AI readiness, including content clarity and structured data, is scored as part of the free check's awareness dimension.

8. The "why wasn't I mentioned" follow-up

Prompt: (after running prompt 2 or 6 without the business appearing) "What about [Business Name]? Why didn't you mention them?"

This follow-up sometimes gets ChatGPT to explain its reasoning directly: thin reviews, an incomplete profile, or simply not enough information available. It's not always reliable, ChatGPT can rationalise after the fact rather than reveal genuine cause, but it's a useful prompt for a sales conversation because it puts the gap in the client's own words.

Or run it automatically: the full audit ranks the actual issues behind a low score by priority, rather than relying on ChatGPT's after-the-fact explanation.

9. The consistency check

Prompt: run prompt 2 or 6 two or three times, in separate chats, a few minutes apart.

AI answers aren't static. The same unbranded query can return a different shortlist between attempts, which is a genuinely useful thing to show a sceptical client: this isn't a one-off ranking to game, it's an ongoing visibility problem that needs monitoring, not a single fix.

Or run it automatically: Insites' AI Visibility Monitoring re-runs these checks over time and tracks the trend, rather than treating one snapshot as the final word.

10. The source-tracing prompt

Prompt: "Where are you getting that information from? What are your sources?"

When ChatGPT does cite something, and increasingly it will, this shows you which listings, directories or the business's own website it's actually pulling from. If it's citing an outdated directory instead of the business's own site, that's a specific, fixable problem rather than a vague "improve your AI visibility" instruction.

Or run it automatically: the free check flags exactly which sources are being cited across each platform, so you're not stitching this picture together prompt by prompt.

Where this leaves you

Ten prompts and a bit of patience will get you a rough picture of how a business shows up in ChatGPT. What they won't easily give you is the same picture across Perplexity and Gemini too, a score you can hand to a client without explaining the methodology, or a repeatable way to check it again next month.

If you'd rather skip the copy-pasting, run a free AI visibility check and get the same answers, scored and client-ready, in under 60 seconds.

AI answers change between models, chats and over time, so treat any single result here as a snapshot rather than a permanent verdict. Re-testing periodically is part of the job, not a sign something went wrong.