How to Get Started with AI Visibility: A 3-Question Check

Your boss wants an answer about AI visibility. Reddit gives you twenty things to do. This guide gives you three questions to start with.

Picture this: your boss is breathing down your neck about AI visibility.

Can you tell me whether our brand or competitors shows up in ChatGPT?

You say you will look into it. Then it hits you: you have no idea where to begin.

So you do what most of us would do. You open ChatGPT, search Reddit, skim a few guides, and before long you have a dozen tabs open.

One person says to track hundreds of prompts. Another says to map personas. Someone else says to compare models, fix structured data, check crawler access, publish llms.txt, and buy a dashboard.

Most of that advice is not wrong. It is simply arriving before you know what problem you are looking for. You end up knowing more terms, but you are no closer to answering your boss.

If that sounds familiar, this guide is for you. Take some pressure off. You are not trying to build a perfect measurement system today. You are trying to get your bearings.

---

How to Get Started with AI Visibility

People use "AI visibility" to mean a lot of things: mentions, recommendations, positions inside a list, or simply whether an assistant knows a company exists.

You do not need to untangle all of that terminology before you begin. For this first check, you should try to understand these three things:

1. Does the assistant understand your brand?
2. Does it consider your brand alongside known competitors?
3. Does it bring up your brand when you have not mentioned it first?

These may sound similar, but they are different outcomes.

An assistant may recognize your brand when prompted but never recommend it in an unbranded answer. It may understand your product but describe it for the wrong audience. It may list your brand as a credible alternative to one competitor while consistently recommending another.

That is why finding your name once is not enough. You want to see how the answer changes as you remove information from the question.

---

Run the Three-Question Check

Do not overthink the setup. Choose:

• One audience: the type of person or company you want to reach (usually your key ICP).
• One problem: a meaningful problem your product or service genuinely addresses.
• One competitor: an established option buyers may evaluate instead of you.

Choose a situation a real customer might bring to a sales call. "Small accounting firms that need to shorten monthly close" is more useful than "businesses that need better software." Use the same audience and problem in all three questions.

Before you ask: Run each question in a separate new conversation. Question 1 names your brand; asking all three in one thread can carry that context into Questions 2 and 3. If available, use ChatGPT Temporary Chat, Gemini Temporary Chat, another private-chat mode, or test without signing in. This reduces context carryover, although it cannot remove every source of variation.

Question 1: Does AI understand your brand?

What is [brand], who is it for, and how does it help [audience] solve [problem]?

You supplied your brand name, so this does not test whether the assistant would find you independently. It asks something more basic: when you introduce your brand, does the assistant understand it?

Look for:

• the correct company, product, category, and audience;
• a clear connection to the selected customer problem;
• current rather than outdated information; and
• an accurate description of strengths, limitations, and use cases.

If the assistant cannot identify your brand, confuses it with another company, or invents basic details, save the answer before changing anything. Recognition is the first issue to investigate.

If it recognizes your brand but assigns the wrong audience, category, or value, you may have a framing problem instead.

Question 2: Does AI consider your brand an alternative?

What are the best alternatives to [competitor] for [audience] trying to solve [problem]?

This question names your competitor but not your brand. It checks whether the assistant associates you with the same buyer problem and competitive set.

Look for:

• whether your brand appears;
• where it appears in the list;
• whether it is a credible alternative or an afterthought;
• whether the stated reason matches your real strengths; and
• which competitors receive stronger treatment.

Not seeing your brand can feel like an immediate verdict. It is not. The assistant may not connect your brand to the problem, may rely on sources that emphasize other vendors, or may not have enough evidence to recommend it confidently.

Question 3: Does AI bring up your brand on its own?

What tools or services should [audience] consider for solving [problem]?

This is probably closest to the question your boss actually cares about. If a potential buyer asks for help without supplying a shortlist, does your brand enter the conversation?

Look for:

• whether your brand appears;
• whether it is merely listed or actively recommended;
• whether the reason given is accurate; and
• which competitors appear more prominently or receive stronger treatment.

Avoid asking only, "How can I solve [problem]?" That may produce general tactics without naming products or providers. If you want to test brand discovery, the question should make recommendations a reasonable part of the answer.

Run the same three questions in ChatGPT and Gemini using the same wording. If they disagree, you have not done anything wrong. Keep the results separate: the disagreement is evidence that one assistant should not stand in for all of them.

---

How to Read the Answers Together

If the results look messy, that is normal. Use the patterns below to name a likely starting point—not to pass judgment on your brand.

Question 1 is missing, confused, or factually wrong; your brand is absent from Questions 2 and 3. That may indicate a recognition or brand-identity problem.

Question 1 recognizes your brand but assigns the wrong audience, category, value, or limitations. That may indicate a framing or positioning problem.

Question 1 is accurate, but relevant peers appear while your brand is absent from Questions 2 and 3. That may indicate a competitive association, authority, or discovery problem.

Your brand appears in Question 2 but not Question 3. That may indicate the brand is known as an alternative, but unprompted discovery is weak for this problem.

The same competitor dominates Questions 2 and 3. That may indicate possible competitor displacement.

Your brand appears in all three but competitors receive stronger endorsements. That may indicate a recommendation-strength problem rather than simple absence.

Your brand appears accurately and receives relevant treatment in all three. That is a promising baseline for this specific test.

Do not turn this into a "two out of three" score. A brand that appears only when named has a different problem from a brand that appears unprompted but is described inaccurately. The same score would hide that difference.

Also remember that an answer shows what happened, not automatically why it happened. If your brand is absent, the cause might be unclear first-party positioning, outdated third-party descriptions, stronger competitor coverage, the wording of the question, or ordinary response variation.

Do not ask the assistant, "Why didn't you mention my brand?" and treat its explanation as evidence. It can produce a plausible reason without having access to the internal cause of its earlier response.

---

Record What You Saw

You do not need a dashboard. For this first pass, capture the same fields for each of the three questions:

• provider and test date;
• whether the brand appeared (Question 1 is prompted by design);
• whether the description was accurate, partly accurate, or inaccurate;
• whether the brand was listed, recommended, or discouraged;
• which main competitors were mentioned; and
• important wording or caveats.

Keep the full responses with those notes. "Absent" cannot show whether the assistant recommended a competitor, misunderstood your category, or answered without naming any vendor.

---

What This Quick Check Cannot Tell You

Once you have the answers, you may want to turn them into a percentage or a final verdict for leadership. Pause there.

Three questions cannot establish:

• your overall AI visibility;
• how consistently your brand appears across audiences, providers, and question variations;
• why one brand or source was selected over another; or
• which website or third-party change will fix the problem.

This does not make the exercise useless. It keeps the conclusion proportional: you have three observations that point toward what to investigate next, not a comprehensive measurement.

---

What to Investigate Next

Let the pattern guide your next step.

• If the assistant does not understand your brand: Check whether your website and important third-party profiles clearly agree on your company, product, audience, category, and current capabilities.
• If your brand is understood but described incorrectly: Save the exact language and investigate where it may come from. The guide to brand framing in AI search provides a deeper framework.
• If your brand is missing from the competitor or unbranded answers: Examine which competitors appear and which sources support them. Start with competitor displacement and the difference between organic and prompted visibility.

---

Start Small

If all you do today is run these three questions and save the answers, that is enough for a first pass.

You will have moved from "I have no idea where to begin" to specific observations: ChatGPT understands us but frames us incorrectly. Gemini sees us as an alternative but does not bring us up unprompted. One competitor appears everywhere we expected to appear.

That is not the end of an AI visibility analysis. It is a much more confident beginning—and a far more useful answer than a generic score.

---

Run the Check, Then Come Back

Run the three questions, save the complete answers, and then come back here. Once you can see whether your problem looks like recognition, framing, competitive displacement, or unprompted discovery, choosing the next step becomes much easier.

If you want to investigate the problem yourself, use Viziquo's six-step AI visibility audit process to expand the check across representative audiences, questions, providers, framing, and competitors.

If you would rather bring in outside help, use How to Evaluate a Provider for an AI Visibility Audit to compare methodologies, deliverables, provider coverage, and the claims vendors make about their results.

You do not need to choose between those paths today. Run the three questions first. Come back with the evidence, and choose the path that fits what you found.

---

Key Takeaways

• Start with one audience, one customer problem, and one competitor. Run the three questions in separate temporary, private, or signed-out conversations where available.
• Read the pattern across the answers. Do not reduce it to a score.
• Treat the result as a starting point that tells you what to investigate next—not as a complete measurement of your AI visibility.