How to Check If AI Recommends Your Business (Free 60-Second Test)

You can find out right now whether AI search engines recommend your business. It takes about 60 seconds, costs nothing, and the answer is rarely what people expect. Most owners assume they show up. They don't.

Key takeaways

• The fastest way to check: open ChatGPT, Claude, Perplexity, and Microsoft Copilot, and ask the exact question a buyer would type. If your name does not appear in the first answer, you are not in the recommendation pool.
• "Invisible" is the most common result. Even owners with strong Google rankings and a 10-year track record often find zero mentions across all four engines.
• One ranking is not enough. Each engine pulls from different sources, so you have to check all four. Showing up in Perplexity but not ChatGPT means you are missing the bigger audience.
• Branded searches lie. Searching your own name will always return your site. The real test is the discovery-intent question, the one a stranger would ask before they know you exist.
• The free AI Visibility Check automates this test across 8 discovery-intent questions per engine and sorts the result into one of four outcomes: Invisible, Mixed, Winning, or Empty-niche.

Why this check matters more than your Google ranking

Your Google ranking tells you how you do when someone types keywords into a search box. The AI recommendation check tells you something different. It tells you whether AI engines name you when a buyer asks for a recommendation in plain English.

These are two different games now. Buyers do not type "best business coach Denver" anymore. They ask ChatGPT, "I run a 5-person agency and I'm stuck at $40K months. Who should I talk to?" Whoever ChatGPT names is who gets the discovery call.

I built BakingSubs to test this shift in a niche I had no reputation in. It now has 162,500 Microsoft Copilot citations to date, with 112,500 of those landing in just the last three months. No ads, no backlinks, no social. The reason it works is that the site was built to be cited by AI engines, not just ranked by Google. Most expert-led sites were not.

So before you spend another month writing content, posting on LinkedIn, or running ads, do the check. The result decides what you should do next.

The 60-second manual test (do this right now)

Open four browser tabs. One for ChatGPT, one for Claude, one for Perplexity, one for Microsoft Copilot. In each one, ask the same question. The question matters more than the engine.

The right question is what a buyer would type, not what you would type. Buyers do not search for your category. They describe their problem and ask who can help.

Here is the template:

"I'm a [specific buyer description] and I need help with [specific problem]. Who are some good [coaches / consultants / specialists] who work with people like me?"

Plug in your actual buyer. If you coach mid-career women leaving corporate to start agencies, the question is: "I'm a mid-career woman leaving a corporate VP role to start a marketing agency. Who are some good coaches who work with founders like me?"

Now read the answer. There are four possible results, and each one means something different.

Result 1: You are invisible

The engine names three to five people. None of them are you. This is the most common outcome. If you got this, the rest of the post is for you.

Result 2: You are named, but mixed in

The engine names you alongside three or four competitors, with no clear differentiation. You are in the pool, but you are not the recommended option. This is better than invisible. It still loses most of the discovery calls.

Result 3: You are the lead recommendation

The engine names you first, with specific detail about who you work with and what makes you different. The other names are listed as alternatives. This is what winning looks like. Very few sites hit this.

Result 4: The engine has no good answer

The engine gives a vague response, names general categories instead of specific people, or admits it cannot recommend anyone. This is an empty-niche result, and it is the best opportunity in the post. You can own the answer by being the first credible expert the engines pick up.

Why you have to check all four engines, not just one

Each engine pulls from a different mix of sources. ChatGPT leans heavily on long-form authoritative content. Claude weights author signals and tends to be more conservative about naming specific people. Perplexity surfaces sources prominently and updates faster than the others. Microsoft Copilot, in my experience with BakingSubs, has the most aggressive citation behavior. It will recommend sources the other three skip.

This means you can show up beautifully in Perplexity and be completely invisible in ChatGPT. I see this constantly. A consultant runs the check, finds her name in Perplexity, feels relieved, and stops. Then six months later she wonders why her pipeline is still dry. The reason is that her buyers use ChatGPT, not Perplexity, and she never checked.

You also cannot assume the engines see your site at all. Claude in particular is selective about which sources it weighs. A clean homepage with no author bio and no topical depth will be ignored, even if it ranks well on Google.

Run the same question through all four. Write down what each one said. If your name appears in zero or one, you have a visibility problem worth fixing.

Three real-feeling check scenarios

The manual test feels abstract until you watch it play out. Here are three composite scenarios. Each one ran the 60-second check. Each one got a different result.

The coach: Yasmin in Austin

Yasmin is a leadership coach in Austin who works with engineering managers stepping into director roles. She has been coaching for 9 years, has a clean website, and ranks on page 1 of Google for "engineering leadership coach Austin."

She opened ChatGPT and typed: "I'm a senior engineering manager at a Series B startup, about to be promoted to director, and I need a coach who actually understands the technical-to-leadership transition. Who do you recommend?"

ChatGPT named three coaches. None were her. Two of the three had less experience and weaker sites. One of them was based in another state.

Yasmin's failure case was subtle. Her site had strong copy but no Person schema (the hidden tag that tells AI engines this page is about a real human expert, not a brand page). She also had no individual case studies, only generic testimonials. ChatGPT had nothing specific to extract about who she actually worked with.

This is the Invisible result for a coach who looks competitive on the surface. The Google ranking did not save her.

The consultant: Daniel in Toronto

Daniel runs a 2-person operations consulting firm in Toronto. His buyers are private equity portfolio company CEOs who inherited messy operations and need to clean up before the next round. He charges five figures per engagement and has zero interest in coaching.

He ran the check in Claude: "I just took over as CEO of a PE-backed industrial services company and the operations are a disaster. Who are good consultants for post-acquisition operational cleanup?"

Claude named two consulting firms and was vague about specific people. Daniel was not in the answer. But interestingly, when he ran the same question in Perplexity, his firm came up second. The difference was that Perplexity had indexed two long-form posts on his site about post-acquisition operations playbooks. Claude had not weighted them yet.

Daniel got the Mixed result. He was visible in one engine, invisible in three. His next move was to figure out why competitors were showing up where he wasn't.

The ecommerce founder: Priscilla in Portland

Priscilla runs a small ecommerce brand selling fermentation supplies (crocks, weights, starter cultures) to home cooks. She is the face of the brand. Her face is on the About page. She writes every blog post.

She ran the check in Microsoft Copilot: "I want to start fermenting my own kimchi and sauerkraut at home. Where should I buy good fermentation crocks and what starter culture do you recommend?"

Copilot named her brand twice in the response. Once for the crocks, once for the starter cultures. It cited her blog post on choosing the right crock size. Priscilla got the Winning result.

The reason was not luck. She had spent two years writing depth content on fermentation, structured around the questions her customers asked. Her site had the topical density Copilot needed to confidently recommend her. This is what the Citation Cluster Method does when it works.

Three buyers ran the same kind of check. Three completely different results. Each one tells the owner something different about what to do next.

The mistake that makes the check useless

The most common way owners run this check wrong is they search their own name. They type "tell me about [Their Name] coaching" and the engine, predictably, returns their site. They feel reassured. They close the tab.

This proves nothing. The engine returned a result because you gave it a unique entity to look up. Buyers do not know your name yet. That is the whole point of discovery search. The buyer types a problem description, and the engine decides whose name to surface.

The check only works with discovery-intent questions, the ones a stranger would ask before they know you exist. If you cannot resist checking your name, do it at the end as a sanity check. Never start there.

The other common mistake is asking only one question. Buyers ask 8 to 15 different questions during a typical research journey. They might search for credentials, then methodology, then pricing range, then specific situation matches. You can be the answer to one of those questions and invisible on the other seven.

This is why the AI Visibility Check runs 8 discovery-intent questions per engine. One question gives you a snapshot. Eight gives you a real picture of where you stand across the buyer's whole research path.

What each result actually means for what to do next

Now you have run the check. You have a result. Here is what each one tells you.

If you got Invisible across all four engines: This is the most common result, and it is also the most fixable. Invisible means the engines do not have enough specific, structured information about you to confidently make a recommendation. The fix is depth content on the specific problems your buyers describe, plus the structural signals (clear author identity, schema, internal linking between related posts) that let engines extract a clear answer. The 5 client-getting channels that still work in 2026 shows where to start.

If you got Mixed (named alongside competitors): You are in the pool but not the lead. The fix here is differentiation specificity. The engines named you because they have some signal about your work, but not enough to put you first. You need more depth on the specific sub-niche where you actually win. Catching up to a competitor who is being recommended is a faster lift than starting from invisible.

If you got Winning: Do not get comfortable. AI engines update their recommendations as new sources publish. Winning today is not winning in six months. Your job is to keep publishing the depth content that earned you the spot, and to expand into adjacent questions where you are still invisible.

If you got Empty-niche: This is the biggest opportunity in the entire post. The engine has no good answer. That means the first credible expert who publishes serious depth on this problem will own the recommendation slot for the next year. Move fast.

Why this is not a one-time check

AI search is not static. The engines re-index, re-weigh, and re-recommend constantly. A competitor who was invisible last month can publish two strong posts and overtake you in six weeks. A new entrant can claim an empty-niche slot you were planning to take.

This is why I built BakingSubs the way I did. The 162,500 Microsoft Copilot citations did not arrive evenly. 112,500 of them landed in just the last three months, after I had been publishing consistently for over a year. The results compound, but only if the work is continuous.

For your business, this means running the check on a regular cadence. Monthly is plenty. Quarterly is the minimum. Track which questions you appear for, which competitors show up where you don't, and how those answers change over time. That is the dataset that tells you what content to publish next.

If running this manually every month feels like a lot of clicking, that is exactly the problem the AI Visibility Check was built to solve. It does the 8 questions across the 4 engines, sorts the result, and tells you what category you are in. You can do the same thing manually. The tool just removes the friction.

Frequently asked questions

How do I know if my business is showing up in ChatGPT?

The only reliable way is to ask ChatGPT a discovery-intent question that a buyer in your niche would actually type, and read the answer. Do not search your name. Search the problem your buyer is trying to solve, and see whether ChatGPT names you among the recommended experts. If your name does not appear in the first answer to a question a stranger would ask, you are not in ChatGPT's recommendation pool. Three specific ways to test this covers the full method.

What does it mean if AI engines do not recommend my business at all?

It usually means one of two things. Either the engines do not have enough specific, structured information about who you work with and what you do, or your content is too generic to be cited as a recommendation source. The fix is not more content in general. It is depth content on the specific problems your buyers describe, plus the structural signals that let engines extract a clear answer. The signs that you are invisible to AI search breaks down the most common causes.

Why does my competitor show up in AI search and not me?

Usually because they have more topical depth on the exact problems buyers describe, or stronger structural signals (clear author identity, schema, internal links between related posts), or both. Sometimes it is also that they have been publishing consistently for longer and the engines have weighted their content more heavily. The competitor advantage is rarely about size. AI often recommends the smaller competitor when the smaller site has more specific, citable content.

Do I need to check all four AI engines or is one enough?

You need to check all four. Each engine pulls from different sources and weights them differently. Showing up in Perplexity but not in ChatGPT means most of your buyers will never find you, because ChatGPT has the larger audience. The minimum useful check is the same discovery-intent question across ChatGPT, Claude, Perplexity, and Microsoft Copilot, and writing down the result for each.

How often should I run this check?

Monthly is the right cadence for most expert-led businesses. AI engines update their recommendations as new content gets indexed, so results from three months ago do not reflect where you stand today. Quarterly is the absolute minimum. If you are actively publishing content and trying to move from Invisible to Mixed or Winning, monthly checks let you see whether the work is paying off.

Run the check, then decide

The hardest part of fixing your AI visibility is knowing where you actually stand. Most owners skip this step and jump straight into producing content, hoping it will help. It usually doesn't, because they were solving the wrong problem.

Run the 60-second manual test now. Pick the discovery-intent question your real buyer would ask. Try it in all four engines. Write down what you saw. Then decide what kind of result you got, and what that means for your next move.

If you want the automated version that runs 8 questions per engine and sorts the result for you, the free AI Visibility Check is the fastest path. Either way, do the check before you do anything else.