AI Visibility Audit: How to Find Out If ChatGPT and Perplexity Know Your Business Exists

Quick answer: Run ten specific queries across ChatGPT, Perplexity, Claude, and Google AI Mode. Score the results in four buckets: named, paraphrased, mentioned in context, or invisible. The pattern that emerges tells you whether you are a default recommendation in your category, an occasional mention, a fringe presence, or completely absent. The full audit takes 30-45 minutes. This post walks through the exact prompts and how to interpret what you find.

---

Why You Cannot Just Search for Your Business Name

The first instinct when someone hears "are you visible in ChatGPT?" is to open ChatGPT and type their business name. That tells you almost nothing useful.

AI tools will name your business if you ask about your business specifically. They will say "yes, [Your Business Name] is a company that does X." This is not visibility. This is the model regurgitating data when prompted with the data.

Real visibility is whether the AI recommends you when a customer asks for a recommendation in your category. That is a fundamentally different test.

The audit below runs that test systematically. Ten prompts. Four AI tools. Forty data points. Ninety minutes maximum. You will know exactly where you stand.

---

The 10 Prompts to Run

Replace `[your category]` with your service or product type, `[your city]` with your primary geography, and `[specific use case]` with a narrow segment of customers you serve.

Foundational Recommendation Prompts

Prompt 1: "Who is the best [your category] in [your city]?"

This is the most basic recommendation query. If the AI does not name you here, you have core visibility issues. If it does, that is the foundation for everything.

Prompt 2: "I need a [your category]. Can you recommend someone in [your city]?"

The conversational version of prompt 1. Some AI tools respond differently to formal versus conversational phrasing. Run both.

Prompt 3: "What [your category] companies are good for [specific use case]?"

This tests whether you are tied to your specific niche or just generally known. A medical practice might be cited for "primary care" but not for "primary care for diabetic patients." The latter is the higher-value query.

Comparison Prompts

Prompt 4: "Compare top [your category] options in [your city]."

This tests whether the AI groups you with the other reputable options. Being mentioned in a comparison list is meaningful even if you are not named first.

Prompt 5: "What are the differences between [Competitor A] and [Competitor B] in [your city]?"

If you should be in this comparison and you are not named, that tells you the AI does not see you as a peer to those competitors. That is data to act on.

Contextual Prompts

Prompt 6: "I am a [your ideal customer description]. What [your category] should I work with in [your city]?"

This tests whether the AI matches your business to the right customer profile. Your service description should make it clear who you serve. If the AI cannot tell, your entity description is too vague.

Prompt 7: "What [your category] do experts recommend for [specific scenario]?"

This is the credibility test. The AI is being asked to cite expert opinion. If your business shows up here, the AI views you as a credible authority. If not, you are not in the expert mention layer yet.

Brand and Authority Prompts

Prompt 8: "What do you know about [Your Business Name]?"

This is the entity test. The AI's response tells you exactly how it represents your business. A clear, accurate description means your entity is sharp. A vague or wrong description means your entity needs work.

Prompt 9: "Who founded [Your Business Name] and what do they do?"

This tests the depth of the entity. AI tools that know the founder, year founded, and core services have a strong entity. Tools that only know the business name have a weak entity.

Prompt 10: "What is [Your Business Name] known for?"

This tests how the AI characterizes your specialty. A precise answer ("known for AI implementation for small and mid-market businesses in Miami") is what you want. A vague answer ("an AI company") means the AI does not understand your differentiation.

---

How to Score the Results

Run each prompt in each AI tool. Score the response in one of four buckets.

Named

The AI explicitly names your business in the response. This is the highest tier. Track which prompts produce named results and which do not.

Paraphrased

The AI describes a business that fits your description without naming you specifically ("there is a Miami-based AI consulting firm that specializes in small and mid-market implementation"). This means the entity is clear but the model is not citing the name with confidence. Often a near-miss that becomes a named result with more directory and review work.

Mentioned in Context

You appear in a list with several other businesses, often a long list. You are visible but not the recommended choice. Track which prompts produce this and look for the position (first, third, last in the list).

Invisible

You are not mentioned at all. The most common bucket for businesses that have not done GEO work. If most of your results are invisible, you have foundational work to do.

---

What Each Pattern Means

Pattern A: Named on most prompts

You are a default recommendation in your category. Your entity is sharp, your directory presence is strong, and your review language reinforces your specialty. This is the goal state. Maintenance work is enough at this point.

Roughly 5-10% of businesses we audit land here without our help. Usually the ones that have either invested in SEO and content for years and benefit from the spillover, or the ones that have an obvious entity advantage (founder is a public figure, brand has been around for decades).

Pattern B: Named on some prompts, missed on others

You are visible for the prompts that match your strongest signals. This is the most common pattern for businesses that have done some GEO work but not the full program. Common gaps: strong on city-level queries but weak on use case queries, or strong on broad category queries but weak on specific service queries.

The fix is targeted. Identify the dimension where you are missing (use case, customer type, service specialty) and add structured content, review language, and entity descriptions that reinforce that dimension.

Pattern C: Paraphrased without naming

The AI knows a business like yours exists. It cannot confidently name you. This usually means your entity is mostly clear but you are not the strongest entity in your category, so the AI hesitates to put your name on the recommendation.

The fix is volume. More directory consensus, more reviews with specific language, more structured content. You are close. Three to four months of focused work usually moves you from paraphrased to named.

Pattern D: Invisible across most prompts

The AI does not know you exist or does not have enough confidence in the entity to surface you. This is the starting state for most small businesses that have not run a GEO program.

The fix is foundational. Build out the entity (consistent NAP, clear about page, founder bios, service descriptions). Get into directories. Improve review language. Add structured Q&A content. The full GEO playbook covered in our GEO guide is what gets you out of invisible and into named.

---

How to Run the Audit Efficiently

Forty data points sounds like a lot. The audit itself takes 30-45 minutes if you do it efficiently.

Open four browser tabs: ChatGPT (chat.openai.com), Perplexity (perplexity.ai), Claude (claude.ai), and Google AI Mode (search.google.com with AI Mode enabled).

Create a simple spreadsheet with prompts down the left column and the four AI tools across the top. Twenty-six rows, ten data cells per row, plus your scoring column. Forty data points fit on a single screen.

Run prompts in order. Switch between tabs. Note the result in the spreadsheet. Each prompt takes two to three minutes per tool. Total time: 30-45 minutes.

If the AI produces a long response, you do not need to read all of it. Search for your business name and your category descriptions. The score is fast once you know what you are looking for.

---

What to Do With the Results

After scoring, look at the patterns.

If You Have Mostly Named Results

Maintain. Quarterly re-audits to make sure your position holds. Keep producing structured content. Keep training your team to ask for specific reviews. The position you have is defensible but not permanent.

If You Have Mixed Results

Identify the gaps. Is it a specific category of prompt where you are missing? A specific AI tool where you are weaker? The patterns tell you exactly where to invest the next round of work. Often a 60-day focused project closes the gap.

If You Have Mostly Paraphrased Results

You are close. The work is reinforcement: more directories, more specific reviews, more structured content. A 90-day focused program typically moves you from paraphrased to named.

If You Have Mostly Invisible Results

You need the foundational GEO program. Schema, directory cleanup, review language overhaul, structured Q&A content production, Bing indexing. The work is significant but well-defined. The 90-day implementation we run at Mi Assist AI is built for this case.

---

How Often to Re-Run the Audit

Quarterly is enough for most businesses. AI models update their entity data on rolling cycles. Your visibility shifts as your work compounds and as competitors do their own work. Quarterly re-audits catch shifts in time to respond.

Run a re-audit immediately after any major change: a name change, a new location, a service expansion, a major review push, a content production sprint. You want the data to verify the change improved your position.

---

What This Tells You About AI Search Strategy

The audit produces a baseline. The baseline is what every subsequent decision is measured against. Without a baseline you cannot tell whether GEO investment is working.

The other thing the audit tells you: GEO is measurable. The marketing channel that has been hardest to track historically (AI tool recommendations) becomes a thing you can run a test against and see results in your own data. That is rare for any marketing investment, especially in 2026.

If you have not run an AI visibility audit on your business, run it this week. The data will change how you think about your marketing investment.

---

FAQ

Q: How do I check if my business shows up in ChatGPT? A: Run a recommendation query like "what is the best [your category] in [your city]?" in ChatGPT. If your business is not named in the response, you are not currently a default recommendation. The full ten-prompt audit in this post gives you a complete picture.

Q: Do I need to run the audit in all four AI tools? A: Yes. Each tool draws from different data sources. ChatGPT pulls heavily from Bing. Perplexity does live web search. Claude has different training data. Google AI Mode integrates Google Search. Visibility in one does not guarantee visibility in the others.

Q: How long does the AI visibility audit take? A: 30-45 minutes if you run it efficiently with a spreadsheet. Less if you focus on the highest-value prompts only.

Q: Can I automate the audit? A: Yes, with tools like Profound, Otterly, or custom scripts that hit the AI APIs. For most small businesses, manual is faster than setting up automation. Automation is worth it when you are running quarterly audits across many service lines or geographies.

Q: What if my business is invisible in all four tools? A: That is the starting state for most businesses that have not run a GEO program. The fix is foundational: schema, directories, review language, structured content, and Bing indexing. Plan for a 90-day implementation.

---

Want us to run the full audit for your business and give you a remediation plan? Book a free AI visibility audit and we will run the ten-prompt test across all four tools and identify the three highest-impact fixes in the first thirty minutes.