How do I check if my business shows up in ChatGPT and other AI search tools?

You sit down, pour a coffee, and check whether ChatGPT knows your business exists. You type your name into the chat box. ChatGPT writes back a confident paragraph describing what you do, where you are based, and even mentions a service or two. You exhale. You are clearly visible.

That test told you almost nothing useful.

A 2025 Semrush study of over 1,000 consumers found 37% now start their search with an AI tool rather than Google. For professional services and B2B buyers the proportion is higher again. Those buyers are not typing your name. They do not know your name yet. They are typing what they need.

The right test is whether ChatGPT, Perplexity, Gemini, Claude, Grok and DeepSeek would name you when a real prospect asks for help. This post walks you through that test and what to do next.

Key Takeaways

• Around 37% of consumers now start their search with an AI tool, not Google, and the shift is sharper for professional services and B2B.
• Google gives ten links and lets the buyer choose. AI gives one or two named recommendations. You either appear or you do not.
• Typing your own business name into ChatGPT tests recognition, not recommendation. Real customers do not know your name yet.
• The Cross-Engine Visibility Test Method uses five need-based queries across six AI tools, scored as Recommended, Mentioned or Missing.
• AI referral traffic converts at around 14.2% versus 2.8% for Google because buyers arrive already decided. Visibility in AI is now a revenue lever, not a vanity metric.

Why is the brand-name test the wrong test?

Most owners run the same first test. They type their business name into ChatGPT. A paragraph comes back, and they conclude they are visible.

The problem is you have just asked the AI a different question to the one your customer asks. When you type your own name, the AI treats it as a request for information about your business and writes a description. That confirms recognition. It says nothing about whether the AI would surface you to a stranger asking for a recommendation. Those are two different jobs the model performs, and the second one drives revenue.

The GetRecommended.io scan deliberately excludes brand-name queries from the visibility score for this reason. Brand queries return informational content, not recommendations, so they do not measure the thing you care about: whether the AI puts you forward when someone is choosing.

Real customers do not know your name yet. They know what they are trying to fix. They type "best accountant for hospitality clients in Brisbane" or "employment lawyer for unfair dismissal small business Melbourne". The AI decides who to name. If you are not in the answer, you are invisible to that buyer, even if the AI knows you exist when prompted directly.

How is AI search actually different from Google, and what does it mean for you?

Google and AI search engines do different jobs and reward different things. Treat them like the same channel and you will keep losing buyers without knowing why.

Google works like a phone directory. You tell it what you want, it hands you ten links, you click around and choose. The buyer does the decision-making work. Being seventh on the page still means you are visible.

AI search works like asking a well-informed friend for a recommendation. You describe your situation. The friend names one or two specific businesses with a sentence of context. There is no list. If you are not in the answer, you are not in the conversation. There is no page two.

ChatGPT now handles roughly 12% of Google's total search volume. AI tools also send around 190 times less traffic to websites than Google does, because people are getting their answer inside the AI tool and not clicking through. For businesses chasing pageviews, that sounds catastrophic.

Here is the part most owners miss. The buyers who do contact you after an AI recommendation are a different quality of prospect. Bain and Company research found around 98% of consumers verify an AI recommendation before making contact, and roughly 45% verify by Googling the business name directly. Semrush data puts the conversion rate of AI-referred traffic at around 14.2% compared to 2.8% for traffic arriving from Google search. These are not browsers. They are buyers.

So what does this mean for you? Visibility in AI is the new top-of-funnel for buyers who are already decided. Missing from those answers is missing the most pre-qualified prospects in your market.

The Cross-Engine Visibility Test Method

This is the test that replaces the brand-name check. The Cross-Engine Visibility Test Method uses five need-based queries across six AI engines, scored against a three-tier outcome. Run it once and you will know where you stand. Run it monthly and you will see your fixes working.

The method has three components.

Component 1: Five need-based queries. Write five queries in the language a motivated buyer would type. The format is "what the buyer needs + their location or qualifier". Specificity matters. Broad queries return generic answers.

Examples:

• "small business accountant Brisbane who specialises in hospitality"
• "employment lawyer Melbourne unfair dismissal small business"
• "web designer for tradies Sydney"
• "bookkeeper for medical practices Canberra"
• "financial planner for women nearing retirement Adelaide"

If you sell to one type of customer in one location, vary the angle on that same buyer. If you sell across categories, vary the wording.

Component 2: Six AI engines. The same query lands differently across engines because each weights different signals. Test all six.

• ChatGPT. The most-used AI tool globally. Use the default chat mode.
• Perplexity. Checks live websites in real time, so reflects your current presence faster than ChatGPT. A strong proxy for whether your site is readable to AI.
• Gemini. Google's AI tool. Connects directly to your Google Business Profile, Maps and reviews.
• Claude. Growing share among knowledge workers and professional services buyers. Tends toward considered answers.
• Grok. X's AI tool. Lower share for most service businesses, but worth a check.
• DeepSeek. Free, growing in adoption across Asia-Pacific. Useful as a tiebreaker.

Component 3: Three-tier scoring. Every answer falls into one of three categories. Score each query against each engine.

• Recommended. The engine names you with positive context, like "Smith & Co is a strong option for hospitality bookkeeping in Brisbane." This drives pre-qualified buyers to you.
• Mentioned. Your name appears without a clear endorsement. The engine knows you exist but does not back you. Better than missing, but it does not move buyers.
• Missing. You do not appear at all.

Five queries times six engines gives you 30 data points. The pattern across those cells tells you which signal is weakest and what to fix first.

How do you actually run the test in 30 minutes?

Block 30 minutes. Open one browser. Have a notes app open beside it.

1. Minutes 0 to 5: Write your five queries. Use the format above, in customer language, not your own. If you are not sure what your customers type, lift the wording from recent enquiry emails or a sales call transcript.

2. Minutes 5 to 10: Open the six engines in tabs. ChatGPT, Perplexity, Gemini, Claude, Grok, DeepSeek. Signed-in or signed-out is fine as long as you stay consistent. Personalisation matters far less in AI tools than in Google.

3. Minutes 10 to 25: Paste each query into each engine. Read the answer carefully. Score it: Recommended, Mentioned, or Missing. Note any competitors named instead of you. Those are the businesses winning the buyer in that moment.

4. Minutes 25 to 30: Tally the patterns. Count how many of your 30 cells are Recommended, Mentioned, or Missing. Note which engines are weakest and which queries return the worst results. That tally is your starting position.

A worked example. A Brisbane accounting firm focused on hospitality runs the test. ChatGPT shows Mentioned in two of five. Perplexity is Missing across all five. Gemini is Recommended in one and Missing in four. Claude, Grok and DeepSeek are Missing across the board. Perplexity Missing means their live website does not signal hospitality clearly enough. Gemini partial means the Google Business Profile is incomplete. ChatGPT Mentioned but not Recommended means the engine knows they exist but cannot find enough third-party signal. The other three Missing means the broader entity signal is thin. That diagnosis tells them what to fix first.

How do you read the three common result patterns?

Most owners' results land in one of three patterns. Each points at a different fix.

Pattern 1: Missing on one or two engines, visible on the others. A tool-specific gap. Missing from Perplexity almost always means your live website presence is thin: few mentions on other sites, outdated content, or weak positioning copy on the homepage. Missing from Gemini almost always points at a Google Business Profile that is incomplete or not verified. Work backwards to the signal that engine weights most.

Pattern 2: Missing across all six engines. The engines cannot identify who you are or cannot find outside confirmation. The fixes are foundational. Rewrite your homepage headline so it states clearly what you do and for whom. Add schema markup. Create an llms.txt guide file. Get listed in credible directories for your sector. Slowest to fix, biggest payoff.

Pattern 3: Mentioned but not Recommended on most engines. The engines know you exist but do not have the confidence to endorse you. Common causes: thin or stale Google Reviews, few mentions on other websites, or not enough written content in your specialty. Read how AI search engines actually use your Google Reviews for the deeper look on the review side.

Diagnose the pattern before you start fixing. Throwing effort at the wrong layer is the most common reason owners spend money on AI visibility and see nothing change.

Practical steps for the next 14 days

Run the test, score the results, and move on the weakest layer.

1. Day 1: Run the full Cross-Engine Visibility Test Method. Block 30 minutes. Five queries, six engines, three-tier scoring. Save the spreadsheet. This is your baseline.

2. Day 2: Note the competitors named instead of you. These are your real AI competitors, often a different list to your Google competitors. Look at their websites. What signals are they sending that you are not?

3. Day 3 to 5: Diagnose the pattern. Match your result to one of the three patterns above. Mostly Missing means work on identity and positioning. Mostly Mentioned means work on trust signals. Split by engine means work on that engine's preferred signal first.

4. Day 6 to 10: Run a structured audit. For the full picture across the four pillars (Technical Health, Content Quality, Authority and Trust, AI Presence), run a free GetRecommended.io scan. The scan gives you a numbered visibility score and names the competitors AI engines recommend instead of you.

5. Day 11 to 14: Fix the highest-impact item first. Do not try to fix everything at once. For most service businesses the fastest-moving fix is the homepage headline plus a complete Google Business Profile. For others it is a sweep of recent Google Reviews. Do the one thing fully before moving to the next.

6. Day 30: Re-run the same test. Same five queries, same six engines. Score again. Compare to your baseline. Movement on even two or three of the 30 cells is real progress. If nothing has moved, the fix you chose was not the weakest layer. Re-diagnose and try again.

You will not get to all-Recommended in 30 days. You will see the pattern shift and know your fixes are working.

Frequently asked questions

Is AI search replacing Google for finding local businesses?

Not entirely, but it is changing how buyers use Google. Bain and Company research found around 80% of search users rely on AI summaries at least 40% of the time. The bigger shift is in where decisions are made. AI tools are increasingly where buyers decide which specific business to contact. Google is becoming a verification step rather than a discovery tool.

Why does typing my own business name into ChatGPT not work as a test?

When you type your own business name, the AI treats your query as a request for information about your business and tells you what it knows. That is not how a potential customer uses it. Customers ask need-based questions like "who should I use for X in Y" and the engine decides which businesses to name. Testing your own name checks recognition, not recommendation. The GetRecommended.io scan excludes brand-name queries from the visibility score for the same reason.

How long does it take for AI engines to reflect changes to my website?

It depends on the engine. Perplexity checks live websites and can reflect changes within days. ChatGPT relies more on its training data and indexed snippets, and typically takes two to six weeks for changes to show. Background code changes, like schema markup and an llms.txt guide file, tend to surface faster than rewritten body copy because they are processed differently.

Should I test in incognito mode or signed out?

It makes little difference for AI tools. Unlike Google, ChatGPT and Perplexity do not personalise results based on your browsing history in the way Google does. What matters is the wording of the query. Use customer language, not your own business name, and keep the wording the same each time you re-test so you can compare like with like.

How often should I re-run the visibility test?

Once a month while you are actively making changes to your site, reviews and listings. Once your fixes are in place and the results have stabilised, quarterly is enough. Test the same queries each time so you have a clean comparison. Random testing makes it impossible to know whether a change moved the dial or whether the engine simply gave a different answer that day.

The Bottom Line

If you have been testing your visibility by typing your own business name into ChatGPT, stop. That test was always going to feel reassuring. It was never going to tell you whether buyers are being pointed your way.

The Cross-Engine Visibility Test Method takes 30 minutes and gives you the answer your brand-name test could not. Five customer-language queries across six engines, scored Recommended, Mentioned or Missing. The pattern across the 30 cells shows which signal is weakest. Fix that layer, re-test in a month, and watch the cells shift.

This week, block 30 minutes and run the test yourself. When you want a structured second opinion that names the competitors AI engines pick instead of you, run a free GetRecommended.io scan. You can also check the scan FAQ, and read why ChatGPT does not recommend your business for the deeper diagnosis on each weak layer.