Is your business showing up in ChatGPT? A 10-minute self-check

“I don’t even know if ChatGPT knows we exist.”

It’s a fair worry. You can see your Google rankings. You can read your reviews. But what an AI tells a customer about you happens in a private conversation you never see.

The good news is that you can look for yourself, and it takes about ten minutes. This guide walks you through the whole check. It covers how to set up so your results are honest, which five questions to ask, how to record what comes back, and how to read the results. At the end, you’ll know whether AI names you, whether it gets your details right, and who it recommends instead.

If you’re new to the idea of AI visibility, start with What is GEO? A plain-English guide for business owners. Then come back here to test your own business.

1. Why should you check what AI says about your business?

Because customers are already asking. When someone asks an AI for a recommendation, the answer they read may be the only place your business gets considered. A quick check tells you whether you’re named, whether the details are right, and which competitors AI suggests instead.

This isn’t a future problem. In BrightLocal’s 2026 survey of US consumers, 45% had used ChatGPT or another AI tool to find local business recommendations, up from 6% the year before. AI is now the third most popular source of business recommendations, behind only Google and Facebook.

The answer itself is what matters, because people rarely click past it. Pew Research Center found that Google users clicked a link inside an AI summary on just 1% of visits. If the AI’s answer doesn’t include you, most people will never go looking.

There’s a second reason to check: AI can get things wrong. It might list an old address, a service you stopped offering, or confuse you with a business that has a similar name. You can’t fix what you haven’t seen.

2. What do you need before you start?

A browser, a private window, a simple table to record answers, and ten minutes. Test in three tools: ChatGPT, Google’s AI Mode, and Perplexity. Stay logged out, or use ChatGPT’s Temporary Chat, so your own history doesn’t nudge the answers toward you.

Time: about 2 minutes.

Why a private window matters. AI tools personalise. If you’ve spent months chatting with ChatGPT about your business, it may mention you simply because it remembers you. Your customers don’t have that history. Open a private or incognito window, or use ChatGPT’s Temporary Chat, which doesn’t use your saved memories. Now you’re seeing something closer to what a stranger sees.

Why three tools. Each AI tool draws on different sources and behaves differently. ChatGPT is the biggest standalone assistant. Google’s AI Mode and AI Overviews sit inside the search engine most of your customers already use. Perplexity shows its sources clearly, which makes it a useful window into where AI finds its information. If you have time, add Gemini or Microsoft Copilot.

Set up a simple table. A spreadsheet or a sheet of paper works. You’ll fill in one row per answer:

Question	Tool	Run	Named?	Position	What it said about you	Correct?	Competitors named	Sources cited
Q1	ChatGPT	1	Yes	2nd of 4	“Known for fast call-outs”	Yes	A, B, C	Review site, directory

It looks like a lot of columns, but most answers take less than 20 seconds to log.

3. Which five questions should you ask?

Ask the questions a real customer would type, in their own words. Cover five angles: a recommendation, an urgent problem, a comparison with a competitor, a direct question about your business, and a question about your reputation. Always include your area, or your niche if you don’t serve a local market.

Time: about 3 minutes.

Each question tests something different:

1. The recommendation question. “Who’s the best [service] in [area]?” This is the big one. It tests whether AI puts you on the shortlist at all.
2. The problem question. “My [problem] needs fixing today. Who can help in [area]?” Real customers often describe a problem, not a service. This tests whether AI connects your business to the situations you actually solve.
3. The comparison question. “[Your business] or [competitor]: which is better for [need]?” This shows how AI weighs you against a rival, and why.
4. The direct question. “What do you know about [your business name] in [city]?” This tests whether AI recognises you, and whether its facts are right.
5. The reputation question. “Is [your business] reliable? What do customers say?” This shows how AI summarises your reviews and reputation.

Here’s how those questions might look for three different businesses:

Question	A plumber in Cape Town	An accountant in Johannesburg	An HVAC company in Phoenix
Recommendation	Who’s the best plumber in the southern suburbs?	Who’s a good accountant for small businesses in Johannesburg?	Best AC repair company in Phoenix?
Problem	My geyser burst. Who can come out today in Rondebosch?	I’m behind on my SARS returns. Who can help me catch up?	My AC stopped working in 110 degree heat. Who can come today?
Comparison	[You] or [competitor] for a bathroom renovation?	[You] or [competitor] for a growing startup?	[You] or [competitor] for a new AC install?
Direct	What do you know about [you] in Cape Town?	What do you know about [you] in Johannesburg?	What do you know about [you] in Phoenix?
Reputation	Is [you] reliable? What do reviews say?	Is [you] trustworthy? What do clients say?	Is [you] any good? What do customers say?

Two tips. First, use the words your customers use, not your industry’s words. South Africans say “geyser,” Americans say “water heater,” and the British say “boiler.” Second, keep your questions the same every time you run this check. That way you can compare results month to month.

4. How do you run the check?

Ask each question twice in each tool, starting a fresh chat each time. That gives you 30 answers. For every answer, record whether you were named, where you appeared, what was said, which competitors came up, and which sources were cited.

Time: about 4 minutes.

Why twice? Because AI answers change from one run to the next, far more than most people expect. A study by SparkToro and Gumshoe had hundreds of volunteers run the same prompts through ChatGPT, Claude and Google’s AI nearly 3,000 times. The chance of getting the same list of recommendations twice was under 1 in 100. The chance of getting the same list in the same order was closer to 1 in 1,000.

That sounds chaotic, but there’s a useful pattern underneath it. Even when the exact lists changed, the tools kept drawing from a fairly consistent group of brands. So a single answer tells you very little, while a handful of answers starts to show you whether you’re in that group.

A few habits keep your results clean:

• Start a new chat for every run. Otherwise the AI builds on its earlier answer.
• Don’t correct the AI mid-check. If it gets something wrong, note it and move on. Arguing with it changes the conversation and your results.
• Click the sources. In Perplexity and Google’s AI, the cited sources are right there. In ChatGPT, look for the source links under or beside the answer. Note which websites keep appearing.
• Take screenshots of anything surprising. Especially wrong details. You’ll want them when you fix things.

5. How do you read your results?

Look for patterns across the 30 answers, not single results. Your business will usually land in one of four places: named and accurate, named but described wrongly, left out while competitors are named, or not recognised at all. Each one points to a different fix.

Time: about 1 minute to spot the pattern.

Start with one simple number: in how many of the 30 answers were you named? Then read down the other columns. Most businesses fall into one of these four patterns.

Named, and it’s right. Well done. Now protect it. Look at the sources cited when you were named. Those are the pages AI relies on to vouch for you, so keep them accurate and current. Keep reviews coming in steadily, because recent reviews matter to both customers and AI.

Named, but it’s wrong. This is more common than you’d think, and it’s the most urgent to fix. AI tools do make mistakes. In a large study of news questions coordinated by the European Broadcasting Union and the BBC, 20% of AI answers contained major accuracy problems, including invented or outdated details. That study was about news, not businesses, but the lesson carries over. AI repeats what it finds, even when it’s wrong. An Ahrefs experiment made the point sharply. A researcher invented a company and planted false claims about it online. Nearly every AI tool repeated those claims as fact, even though the company’s own FAQ denied them. The fix is to find where the wrong detail lives, such as an old directory listing or an outdated article, and correct it at the source. Fixing only your own website often isn’t enough.

Competitors named, but not you. Your sources column is now your roadmap. Which review sites, directories, articles and forums does AI cite when it recommends your competitors? Those are the places AI trusts in your market, and you need to be there too. Being talked about across the web matters a lot. In an Ahrefs study of 75,000 brands, the number of times a brand was mentioned across the web was the factor most closely linked to appearing in Google’s AI Overviews. It was a far stronger link than backlinks.

AI doesn’t know you. If the direct question comes back empty, or AI confuses you with someone else, start with the basics. Make your business name, address, phone number and services identical everywhere. Complete your Google Business Profile and your key directory listings. The Google Business Profile mistakes that quietly cost you calls is a good place to start.

The number of answers you appeared in is a rough version of what we call presence, the first of five signals we measure. You can read about all five in The 5 signals that decide whether AI recommends you.

6. What else can you check in five more minutes?

Two quick technical checks tell you whether AI can reach your website and whether it’s already sending you visitors. Make sure your robots.txt file doesn’t block OpenAI’s search crawler, and look in Google Analytics for visits that came from chatgpt.com.

Check that AI can read your site. Every website has a file at yourwebsite.com/robots.txt that tells automated visitors what they’re allowed to read. OpenAI’s publisher guidance is clear: if you want your content to appear in ChatGPT’s search answers, you shouldn’t block its crawler, called OAI-SearchBot. Open the file and look for OAI-SearchBot, or for a rule that blocks all bots. If you see “Disallow: /” under either, ask whoever manages your website to take a look. Some security plugins and hosting setups block AI crawlers by default without telling you.

Check whether ChatGPT already sends you visitors. OpenAI also says ChatGPT adds a tag, utm_source=chatgpt.com, to links it sends people to. In Google Analytics 4, open the traffic acquisition report and look for chatgpt.com as a source. You can also look for perplexity.ai. Even a small number tells you that AI is already recommending you to some people.

7. What can’t a DIY check tell you?

It’s a snapshot, not a measurement. Thirty answers from one location on one day can’t tell you your true share of AI recommendations, how that changes over time, or how every engine treats you. For that you need many more questions, run many more times, across every tool.

The same SparkToro research that found how much AI answers vary also suggested what real measurement takes. To know an AI’s actual set of recommendations, you need to ask the same question many times, usually 60 to 100 times, and then average the results. Your ten-minute check does two runs.

Engines also differ more than you’d expect, even within one company. One Ahrefs analysis found that Google’s AI Mode and AI Overviews cited the same web pages only 13.7% of the time for the same question. And local answers depend on where the tool thinks you are. Your results in Claremont may differ from a customer’s results in Sandton.

None of that makes the self-check pointless. It’s the fastest way to spot big problems, like wrong details or complete invisibility, and to see which competitors you’re up against. Think of it as taking your own temperature. It tells you whether something’s wrong. It doesn’t give you a full diagnosis.

When you want the full picture, that’s what our free AI visibility audit is for. We test a large set of real buying questions for your business across every major AI engine, repeat them, and score the results across five signals.

What to do this week

• Run the check. Block out ten minutes, open a private window, and work through the five questions in three tools, twice each.
• Count your presence. Write down how many of the 30 answers named you. That’s your baseline.
• List every wrong detail AI gave about your business, with a screenshot of each.
• List the five sources AI cited most often, for you or for your competitors.
• Fix one thing. Correct the most damaging wrong detail at its source, or claim a listing on the site AI cites most for your competitors.
• Check your robots.txt for rules that block OAI-SearchBot.
• Put a reminder in your calendar to repeat the same check next month.

Frequently asked questions

Do I need a paid ChatGPT account to do this check?

No. The free version of ChatGPT is fine, and you can use it without logging in. Google’s AI Mode and Perplexity are also free to use. What matters more is testing in a private window, so your own history doesn’t influence the answers.

Why does ChatGPT give me a different answer every time?

AI tools generate each answer fresh, with some built-in randomness, so the list of names and their order change between runs. Research by SparkToro and Gumshoe found the same list came back less than 1 time in 100. That’s why you should look for patterns across many answers.

Can I just tell ChatGPT to recommend my business?

No. Telling ChatGPT about your business in a chat only affects your own conversations. It doesn’t change what it tells other people. AI builds its view of you from sources across the web, so the lasting fix is making those sources accurate, consistent and positive.

What should I do if ChatGPT says something wrong about my business?

Find where the wrong information comes from. Check the sources it cites, your directory listings, and older articles about you. Correct the detail at its source, and make sure your website states the right facts clearly. You can also use the thumbs-down button to flag the answer. Then check again in a few weeks.

How often should I repeat the self-check?

Once a month is plenty for most businesses. Use the same questions every time so you can compare results fairly. Check again sooner after a big change, like a move, a new service or a rebrand, because that’s when outdated details cause the most trouble.

Does this work for businesses outside the US?

Yes. The method works anywhere. Just phrase your questions the way your local customers talk, and include local place names. The sources AI trusts differ a lot between countries, which I cover in AI search in your market: SA, US, UK and Australia.

Are there tools that do this automatically?

Yes. AI visibility trackers run large numbers of questions across AI engines and log the results over time. Examples include Ahrefs Brand Radar and BrightLocal’s Local AI Visibility Tracker. They measure far more than a manual check can, but they cost money and still need someone to interpret the results.

Rather skip the spreadsheet?

The self-check shows you the surface. Our free audit shows you the whole picture: how often each AI engine names you, what it says, which sources it trusts, and who it recommends instead.

Book a free discovery call and choose the audit that fits: AI Visibility & GEO, SEO, or Local Business SEO. No pitch until you’ve seen the gap.

Book a discovery call

Related reading

• What is GEO? A plain-English guide for business owners
• The 5 signals that decide whether AI recommends you
• Google Business Profile mistakes that quietly cost you calls

About the author

Bantu Nonyukela is the founder of SearchLimelight. He has spent ten years in SEO, after starting out as a front-end developer, and is a Toptal-verified SEO expert. From Cape Town, he helps small and medium businesses in South Africa, the US, the UK and Australia get found in search and named by AI. Connect on LinkedIn

Sources

• BrightLocal, Local Consumer Review Survey 2026: brightlocal.com
• Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results,” July 2025: pewresearch.org
• SparkToro and Gumshoe, research on AI recommendation consistency, 2026: sparktoro.com
• European Broadcasting Union and BBC, News Integrity in AI Assistants, October 2025: ebu.ch
• Ahrefs, AI misinformation experiment, reported in the Ahrefs research roundup, May 2026: businesswire.com
• Ahrefs, “An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)”: ahrefs.com
• Ahrefs, analysis of AI Mode and AI Overviews citations: ahrefs.com
• OpenAI, Publishers and Developers FAQ: help.openai.com