AI brand visibility is whether ChatGPT, Perplexity, and Google AI Overviews mention, recommend, or correctly describe your brand when a buyer asks a question you should win. You can measure it in 30 minutes with the 15 prompts below and a spreadsheet. No tool, no subscription, no login beyond the 3 free chat windows.
Most business owners have asked ChatGPT about themselves once, got a vague or wrong answer, and stopped there. That single check tells you almost nothing. AI brand visibility is a pattern across many questions, and you only see the pattern when you run the same set of prompts, the same way, across all 3 engines.
This is the set I use. Copy it, swap in your details, and you will know in half an hour where you stand and what to fix first.
In this guide
What AI brand visibility actually means
AI brand visibility has 3 layers, and most people only test the shallowest one.
Layer 1 is recognition: does the engine know you exist and describe you accurately when someone types your name? Layer 2 is recommendation: does the engine bring you up on its own when someone asks a category question, with no brand named? Layer 3 is preference: when someone compares you to a competitor, which way does the engine lean?
"Does ChatGPT know my business" is a Layer 1 question. It matters, but it is not where the money is. Buyers who already know your name are going to find you anyway. The revenue is in Layer 2, the category prompts, where a buyer with a budget and no brand in mind asks for a recommendation and the engine hands out 3 to 5 names. If yours is not one of them, you were never in the running.
Each engine gets to those names differently. I broke down the mechanics in how ChatGPT, Perplexity, and AI Overviews choose their sources. This article is the measurement side: what to ask, and how to read the answers.
How to run the test in 5 steps
- Open 3 fresh chats. ChatGPT with search turned on, Perplexity, and a Google search that triggers an AI Overview. Use a fresh chat or an incognito window for each so your own history does not color the answers.
- Fill in your variables once. You need 6: your brand name, your category in plain buyer language (not your tagline), your main customer type, your city or "United States" if you sell nationally, your top competitor, and a realistic budget number your buyer would type.
- Run all 15 prompts in each engine. That is 45 answers. Paste each one into a spreadsheet with 4 columns: mentioned (yes or no), position if mentioned (1st, 2nd, 3rd, or later), the URL it cited for you if any, and whether what it said was accurate.
- Score each answer 0, 1, or 2. The rubric is in the scoring section below. Total it per engine. The maximum is 30 per engine, 90 overall.
- Repeat monthly, same prompts, same order. One run is a snapshot. The second run is the first real data point, because it shows direction.
The 15 prompts
Square brackets are your variables. Type the prompts as a real buyer would, without your brand name in the category prompts. That restraint is the whole point.
Category prompts: does the engine bring you up unprompted?
- What are the best [category] for [customer type] in [city or United States]?
- Which [category] brands are worth the money in 2026?
- I need [product or service] for [specific use case]. My budget is [budget]. What should I get?
- Who are the most trusted [category] companies in [city]?
- What [category] do people on Reddit recommend?
Prompt 5 is not a joke. Reddit consensus drives a large share of what ChatGPT and Perplexity surface for product and service questions, so asking the engine to channel Reddit tells you whether you exist in the conversations those engines trust.
Brand prompts: does the engine know you and get you right?
- What is [brand] and who is it for?
- Is [brand] legitimate and trustworthy?
- What do customers say about [brand]?
- How much does [brand] cost?
- Who runs [brand] and where are they based?
Score accuracy hard here. An engine that confidently gives your old pricing, a discontinued product, or the wrong founder is worse than one that says it does not know, because the buyer believes it.
Comparison prompts: which way does the engine lean?
- [Brand] vs [competitor]: which is better for [customer type]?
- What are the best alternatives to [competitor]?
- Is [brand] worth it compared to cheaper options?
Prompt 12 is the one that hurts. If the engine lists 5 alternatives to your biggest competitor and you are not one of them, you have found the exact gap a competitor's buyers are falling through.
Verification prompts: can the engine prove what it said?
- What is [brand]'s address, phone number, and opening hours? (For national brands: what is [brand]'s website and how do I contact them?)
- What sources did you use for that answer about [brand]? Give me the links.
Prompt 15 shows you which pages the engines are actually reading about you. If the cited links are a Reddit thread, a directory listing, and a 2023 news mention, and not your own site, you know why the answers are stale.
The same prompts for a DTC brand, a coach, and an HVAC company
The template is generic on purpose. Here is how 4 of the prompts read once the variables are filled, for the 3 kinds of business I work with most.
| Prompt | DTC skincare brand | Business coach | HVAC company, Phoenix |
|---|---|---|---|
| 1. Best [category] | What are the best fragrance-free moisturizers for sensitive skin in the US? | Who are the best business coaches for solo consultants in the US? | What are the best AC repair companies in Phoenix? |
| 3. Use case + budget | I need a moisturizer for eczema-prone skin under $40. What should I get? | I need help building a funnel for a $2,000 coaching program. Budget is $3,000. Who should I hire? | My AC is blowing warm air and I need it fixed today in Phoenix. Who should I call and what will it cost? |
| 7. Trust | Is [brand] a legitimate skincare company? | Is [coach name] a legitimate business coach? | Is [company] a licensed HVAC contractor in Arizona? |
| 12. Alternatives | What are the best alternatives to CeraVe for sensitive skin? | What are the best alternatives to [big-name coaching program]? | What are the best alternatives to [largest local competitor] in Phoenix? |
Notice the HVAC version of prompt 3. It contains urgency, a symptom, a city, and a price question. That is how a real homeowner types at 4 PM in July, and it is the query your service page has to answer if you want to be the name the engine hands back.
How to score what comes back
Every answer gets 0, 1, or 2.
0: your brand is not mentioned, or the engine says it does not know you, or what it says is wrong. 1: you are mentioned, but after competitors, or with a thin or partly outdated description, or with no link to your site. 2: you are mentioned in the top 3, described accurately, and the engine cites a page you control.
Total per engine, out of 30. Then look at the 3 totals side by side rather than the overall number, because the engines diverge and the gap is the diagnosis. A Perplexity total far above a ChatGPT total for the same 15 prompts almost always means a Bing indexing problem, because Perplexity runs its own retrieval and ChatGPT search leans on partner indexes. A strong AI Overviews total with weak ChatGPT and Perplexity totals means your Google rankings are carrying you and nothing else is.
There is no industry benchmark for this, and anyone selling you one is guessing. Use your own first run as the baseline and measure against it.
What the results tell you to fix
The pattern in your zeros points at the fix. Here is the mapping I use.
Zeros on prompts 1 to 5, decent scores on 6 to 10. The engines know you but never recommend you. That is a third-party problem, not a website problem. You need real mentions on the pages engines already trust for your category: review sites, Reddit threads, industry roundups, local news. Your own blog posts will not fix this.
Wrong facts on prompts 6 to 10. The engines have stale or conflicting information about you. Fix your About page, your Organization and Person schema, your Google Business Profile, and every directory listing so they all say the same thing. Then add a visible "Last updated" date to your key pages.
Losing prompts 11 to 13 to the same competitor every time. You do not have a comparison page and they do, or theirs answers the question better. Write the honest "[brand] vs [competitor]" page yourself, with a real table, before someone else writes it for you.
Wrong or missing answers on prompt 14. For a local business this is the most expensive zero on the sheet. Your Google Business Profile, your site footer, and your Bing Places listing disagree with each other, or one of them is missing. Fix NAP consistency first.
Prompt 15 cites everything except your site. Your pages are not answering questions in a form the engines can lift. Rewrite your top 5 pages to state the answer in the first 100 words. I walk through a real before and after in How to Get Cited by ChatGPT.
If you want the shorter version of this test, the 5-question AI search visibility audit takes 10 minutes and catches the biggest problems. The 15 prompts are what you graduate to once you are fixing things and need to measure movement.
What this test does not tell you
3 honest limits. First, answers vary between runs. Ask the same prompt twice and you can get a different list, so score the pattern across 15 prompts, not any single answer. Second, this does not give you volume. It tells you whether you appear, not how many people asked. Third, it is a manual snapshot. If you need daily tracking across hundreds of prompts, that is what the paid tools are for, and I compared 9 of them in Best AI Search Visibility Tools Compared. Run the 15 prompts first. Most businesses find out they do not need a tool yet.
The verdict
AI brand visibility is measurable today, for free, in 30 minutes, with 15 prompts and a spreadsheet. The number matters less than the pattern: where your zeros cluster tells you whether the problem is third-party mentions, stale facts, a missing comparison page, or a page that never states its answer.
Run it once for the baseline. Fix the biggest cluster. Run it again in 30 days. That loop is the whole strategy, and it beats every dashboard until you have earned the right to need one.
Want me to run the 15 prompts on your brand and read the results with you?
I do this on a free 30-minute strategy call for US businesses. You bring the brand name and your top competitor, I bring the spreadsheet, and you leave knowing which of the 3 engines is the real problem. No pitch deck.