Key Takeaways
- You can test whether ChatGPT recommends your business in 15 minutes with four prompts in a temporary chat. No monitoring tool required for the first pass.
- Every test lands in one of three buckets: cited with a source, mentioned by name without a link, or absent. Each bucket has a different fix, and applying the wrong one wastes months.
- A citation means ChatGPT retrieved live data and attached a source to your name. A bare mention is fragile training-data residue that can vanish between runs.
- Only 1.2% of local businesses get recommended by ChatGPT, per a 350,000-location study. Over 90% of the 60+ businesses I have audited had zero AI visibility.
- The foundation fix costs $0 in software and 4 to 6 hours. Paid help is a one-time audit or an ongoing monthly retainer.
The 15-Minute Self-Test
Most owners have never once asked ChatGPT the question their customers ask it every day. That is the entire test. You are going to ask it the way a stranger with a problem would, then score the answer.
Setup matters. Open a temporary chat, or log out entirely, so your own history and memory do not tilt the results. Make sure web search is on. Then run these four prompts:
- The head query: "best [your trade] in [your city]"
- The buyer query: "I need a [trade] in [city], who should I call and why"
- The problem query: "recommend a [trade] near [neighborhood] for [the specific job you want more of]"
- The identity query: "who is [your business name] and are they reputable"
Run each prompt three times in fresh chats. ChatGPT answers are not deterministic, and a single run can flatter you or bury you by chance. Three runs per prompt gives you a real signal in about 15 minutes.
The first three prompts test whether you surface unprompted, which is where the money is. The fourth tests something more basic: whether ChatGPT knows you exist at all. When I audited over 60 local businesses across dental, med spa, roofing, HVAC, and plumbing, over 90% failed even that one. Not low visibility. Zero.
What a Citation Actually Looks Like
People score this test wrong because they treat any appearance of their name as a win. It is not. There are three distinct outcomes and they mean very different things.
Cited means your business is named in the recommendation and there is a source attached: a link chip pointing to your website, your Foursquare listing, or a directory profile. ChatGPT pulled live data about you and is willing to show its work. This is the outcome that survives repeat runs.
Mentioned but not linked means your name appears in the text with nothing behind it. ChatGPT half-knows you. Your name lives in some scraped list or stale dataset, but the model cannot confidently connect it to a website, an address, and a phone number. Mentions come and go between runs, which is exactly why they feel so inconsistent when owners test casually.
Absent means competitors got named and you did not, across all three runs. This is the most common result by far, and it is fixable, but the fix is bigger.
One scoping note: this test covers ChatGPT specifically. Perplexity and Google AI Overviews pull from different data sources, so a clean result on one platform tells you nothing about the others. The breakdown of who pulls from where is in our comparison of ChatGPT vs Perplexity vs Google AI Overviews for local visibility.
Your Three Possible Results
| Result | What it means | Your move |
|---|---|---|
| Cited | Named with a source link | Defend and expand coverage |
| Mentioned, no link | Known name, broken entity | Fix the data layer |
| Absent | ChatGPT has never heard of you | Build the foundation |
Pick your bucket honestly, using the worst of your three runs, not the best. Then apply exactly one of the three fixes below.
Choose the Defend Play If You're Cited
First, congratulations. You are in the 1.2%. Now understand that this position is rented, not owned. The businesses that stay cited do three things.
One, they widen query coverage. You got cited for "best [trade] in [city]", but buyers ask dozens of variants: emergency phrasing, price phrasing, neighborhood phrasing. Test those variants and publish pages that answer the ones where you drop out.
Two, they deepen structured data. A cited business with thin schema is an invitation for a competitor with rich schema to replace it. The specific types that matter are covered in which schema types actually move AI citations for local businesses. Most of it is a few hours of work.
Three, they log results monthly. Fifteen minutes, same four prompts, same scoring. When a citation slips to a mention, you want to know that month, not two quarters later when the phone slows down.
Choose the Entity Fix If You're Mentioned but Not Linked
This is the best bad news you can get. ChatGPT already wants to recommend you. It just cannot verify you. The fix is data hygiene, and it is boring on purpose.
Start with your name, address, and phone. They must match exactly, character for character, across Foursquare, Bing Places, Apple Maps, and your own website footer. "Suite 200" on one listing and "#200" on another is enough inconsistency to keep the entity fuzzy. Foursquare is the one that matters most here, because ChatGPT leans on it for location data.
Then make your website state the obvious. A plain sentence on your homepage that says what you do, who you serve, and where you operate, plus LocalBusiness schema carrying the same details. Models cannot infer what you never wrote down.
Give it 30 days, then re-run the test. Most mentioned-not-linked businesses convert to cited on data hygiene alone, without spending a dollar. If you are weighing whether the follow-through is worth your time at all, the math is laid out in is AI search optimization worth it for a small local business.
Choose the Foundation Build If You're Absent
You are with the 90%+. The upside of starting from zero is that the first moves are cheap and known.
Week one: claim Foursquare, Bing Places, and Apple Maps. Fill every field, especially the long description and category selections on Foursquare. Fix NAP consistency everywhere. Add LocalBusiness schema to your site. That is 4 to 6 hours and $0 in software.
Weeks two through eight: publish content that answers the exact questions buyers ask ChatGPT. Not generic blog posts. Decision-stage pages: what the job costs, how to choose a provider, what goes wrong when it is done cheap. Those are the pages browsing-enabled answers cite.
Then re-test at day 30 and day 60. If you have moved from absent to mentioned, keep going yourself. If nothing has moved, that is the point where paying for help makes sense. Before you compare quotes, read what a monthly plan should include. If you want to see what audit money actually buys before committing, read what an AI search content audit costs in 2026.
Choose none of the paid options if your Google Business Profile is a mess, your reviews are under 20, or your website is broken on mobile. Fix the basics first. AI visibility built on a weak foundation does not convert the calls it generates.
Want the test run for you, with the fix mapped out?
We run this exact audit against your business, across ChatGPT, Perplexity, and Google AI Overviews, and hand you the bucket you are in plus the specific fix list. The call is free and you keep the findings either way.
Book a free AI visibility audit
Prefer to look before you talk? Fine. Run the four prompts yourself first, then bring your results to the call and we will tell you honestly whether you need help or just a checklist.
Frequently Asked Questions
What prompts should I use to test whether ChatGPT recommends my business?
Use a temporary chat with memory off so your own history does not skew the answer. Run four prompts: 'best [your trade] in [your city]', 'I need a [trade] in [city], who should I call and why', 'recommend a [trade] near [neighborhood] for [specific problem]', and 'who is [your business name] and are they reputable'. The first three test whether you surface unprompted. The fourth tests whether ChatGPT knows you exist at all. Run each prompt three times, because answers vary between runs. Record whether you were cited with a source, mentioned by name only, or absent. That single result decides which of the three fixes you need.
What does a real ChatGPT citation look like?
In a browsing-enabled answer, a citation is your business named in the recommendation with a source chip or link attached, usually pointing to your website, your Foursquare listing, or a directory page. A mention is your name in the text with nothing attached. The difference matters. A citation means ChatGPT retrieved live data about you and can defend the recommendation to the user. A bare mention means you exist somewhere in old training data or a scraped list, which is fragile and can disappear between runs. Always test with web search enabled, because that mode drives the large majority of local recommendations in 2026.
Why does ChatGPT mention my business but not link to my website?
Broken entity association, almost every time. ChatGPT knows your name from a directory or aggregator but cannot confidently connect that name to your website, address, and phone number. Common causes: name, address, and phone mismatches across Foursquare, Bing Places, Apple Maps, and your own site; a business name that changed at some point; a website with no LocalBusiness schema; or site copy that never plainly states who you are, what you do, and where you operate. Fix the data layer and the link usually follows within a few weeks. This is the cheapest fix of the three, because the platform already wants to recommend you.
How much does it cost to go from invisible to cited?
The foundation is nearly free. Claiming Foursquare, Bing Places, and Apple Maps, fixing name-address-phone consistency, and adding LocalBusiness schema costs $0 in software and 4 to 6 hours of your time. Paid help comes as a one-time audit or an ongoing retainer for citation network maintenance, schema work, and decision-stage content. Most owners should DIY the foundation first, re-test after 30 days, and only pay for ongoing work if the self-test still shows a gap they cannot close themselves.
How often should I re-run the self-test?
Monthly at minimum. The test takes about 15 minutes once you have your four prompts saved. Also re-test after any material change: a website migration, a rebrand, a new location, or a burst of new reviews. ChatGPT answers are not stable, so a single clean run in January tells you nothing about April. Keep a simple log with the date, the prompt, and the result bucket: cited, mentioned, or absent. Three months of that log shows you trend direction, which is the number that actually matters. A business moving from absent to mentioned is winning even before the first citation lands.
Where does ChatGPT actually get its local business data?
Primarily from Foursquare's location data, Bing's index, and live web content pulled through its search tool. It does not use Google's local data layer at all. No Google Business Profile, no Google reviews, no Google Maps rankings. That is why businesses that dominate Google Maps are routinely invisible in ChatGPT, and vice versa. If your Foursquare listing is unclaimed or wrong, you are handing ChatGPT bad data at the source. The platforms differ enough that we broke down the full source map in a separate comparison of ChatGPT, Perplexity, and Google AI Overviews.
How long until my fixes show up in ChatGPT answers?
Two to eight weeks for most data-layer fixes. Claimed and corrected listings propagate on the aggregators' schedules, not yours, and ChatGPT's retrieval picks up changes on its own cadence. Schema and website changes that get crawled can influence browsing-enabled answers faster, sometimes within days. Content plays, like publishing pages that directly answer the questions buyers ask ChatGPT, take 4 to 12 weeks to start earning citations. Do not judge any fix in under 30 days, and re-run the same prompts you used for your baseline so you are comparing like with like.
Is a ChatGPT recommendation actually worth real money?
Yes, and the scarcity is the point. Only 1.2% of local businesses get recommended by ChatGPT, based on a study of 350,000 business locations. I have audited over 60 local businesses across dental, med spa, roofing, HVAC, and plumbing, and more than 90% had zero AI visibility. The buyers who do arrive from a ChatGPT recommendation come pre-sold, because the platform vouched for you instead of listing you as one of ten ads. For a business with a $3,000 to $15,000 customer lifetime value, even 2 to 5 extra inquiries a month from AI answers changes the year.
About the author
Matthew Johnson is the founder of Pleiades Consultancy. He previously scaled his own marketing agency to multiple six figures before serving as CMO of an Amazon agency, where the client base tripled from 15 to 45 active clients during his tenure. He worked with some of the largest names in e-commerce, including Ridge Wallet, HexClad, BK Beauty, The Woobles, Walkize, Lonely Planet, and Obvi. He now works with service businesses to maximize their client acquisition and visibility through AI search with ChatGPT, Claude, Gemini, Perplexity, and Bing Copilot.