The Three Questions the AI Visibility Scorecard Actually Answers

Most B2B companies that run AI visibility audits come out the other side with a lot of data and almost no strategic clarity.

They have rankings for how their blog content performs in AI citations. They have lists of queries where their name appears vs. doesn't. They have traffic estimates from AI-driven referrals. They have schema markup scores and llms.txt completeness reports.

None of that tells the company what it actually needs to know.

AI visibility isn't a single score to optimize. It's the answer to three distinct questions, and if you don't separate those questions cleanly, you end up trying to solve all three at once with interventions that don't fit any of them.

Here are the three questions, in the order they matter.

QUESTION 01

Do I exist?

The first and most basic question: when an AI is asked about your company directly, can it answer?

Type your company name into ChatGPT. Type it into Perplexity, Gemini, Claude. Does the AI return a coherent, specific description of what you do? Or does it hedge, substitute a generic placeholder, or fail to recognize the name entirely?

An existence problem usually shows up as:

• AI descriptions that hedge ("I don't have specific information about...")

• Generic placeholders ("X is a software company that helps businesses...")

• Confusion with similarly-named companies

• Outdated information presented as current

• Lower-ranked AI platforms simply failing to recognize the name

If you're failing Question 1, nothing else matters. Working on accuracy or competitive positioning before AI can even find you reliably is premature optimization.

Companies failing Question 1 typically need entity signal work at the foundational level — getting basic, consistent, indexed content about the company into the sources AI systems treat as authoritative.

QUESTION 02

Am I represented accurately?

Once AI knows you exist, the second question is: what specifically does it say?

Ask the AI to describe your company in detail. Ask who your customers are. Ask what your product does. Ask about your pricing, your founders, your history, your positioning.

Then read the answer critically. Not for whether you exist in it — you passed Question 1 — but for whether the specifics are right.

An accuracy problem usually shows up as:

• Invented features you don't have

• Named customers you never signed

• Pricing that's years out of date or flatly wrong

• Positioning that doesn't match how you describe yourself

• Claims about acquisitions, pivots, or events that didn't happen

• Quotes attributed to your founders that don't exist

This is the question most companies don't audit carefully, because they're relieved just to find they exist in AI at all. But inaccuracy is arguably more dangerous than absence, because the buyer doesn't know the AI is wrong.

A buyer who asks AI about you and gets a confidently wrong answer will show up to a call with expectations you can't meet, questions you can't answer, and beliefs about your product that your sales rep has to spend the first ten minutes correcting.

Companies failing Question 2 need targeted content work — specifically, authoritative third-party content that corrects the wrong signals AI is currently reading.

You can't correct AI's perception by telling it it's wrong. You correct it by providing stronger, more authoritative signals than the ones it's currently using.

QUESTION 03

Am I competitive?

The third question is different from the first two. Questions 1 and 2 are about you in isolation. Question 3 is about you in comparison.

Ask the AI category-level questions. "What are the best [your category] tools in 2026?" "Who should I talk to for [your use case]?" "What are the top vendors in [your space]?"

Now pay attention to two things: which companies the AI names, and where you fall in that list.

A competitive problem usually shows up as:

• Your name appearing in describe-me queries but not in recommend-me queries

• Appearing only in long-tail or narrowly-specified queries, not broad category queries

• Being ranked below competitors you consistently beat in head-to-head deals

• Being named as an "alternative" rather than a primary option

• Missing from top-5 lists while competitors are consistently included

Competitive failures are often the most diagnostic failures, because they reveal what AI systems collectively believe about your positioning in the category — not what any individual AI says about you, but what the consensus is.

Companies failing Question 3 need a different kind of work than Question 1 or Question 2. They need to change AI's understanding of where they belong in the category — which usually requires building authority signals specifically within the category: analyst coverage, review site presence, category-specific thought leadership, comparative content that positions them against the category leaders.

Why These Three and Not Others

A lot of AI visibility scorecards measure things that aren't actually predictive of business outcomes. Common ones include:

Content indexation rates — how much of your website AI has crawled. Tells you volume, not impact.

Schema markup completeness — whether your structured data is correctly formatted. Technical but not strategic.

llms.txt compliance — whether you have the file and whether it's well-formed. Important but not diagnostic.

Traffic attribution from AI — how much traffic AI referred to your site. A lagging metric, not a leading one.

Each of these measures something real. But none answers what the CMO actually needs to know, which is: is my company winning, losing, or invisible in the conversations AI is now mediating?

AI visibility isn't a single score to optimize. It's the answer to three distinct questions.

The three questions map to distinct interventions, and they tell you what to do next in an order that makes strategic sense.

Question 1 tells you whether the basic infrastructure exists. If not, fix that first.

Question 2 tells you whether the signal is clean. If not, correct the signal before investing in competitive positioning.

Question 3 tells you where you rank against the field. If not where you want to be, build category authority.

Skip any of those steps and you waste resources optimizing things that don't matter yet.

What a Real Audit Against These Three Questions Reveals

When I run a Scorecard audit for a company, the pattern is usually predictable.

Most mid-market B2B companies pass Question 1 — they exist in AI at a basic level. AI can describe what they do, even if the description is generic.

Most fail parts of Question 2. There are specific things AI gets wrong about them — wrong customers named, wrong features described, wrong positioning suggested — and they've never audited the specifics closely enough to know.

Most fail Question 3 hard. In category-level queries, they're either absent entirely or ranked below competitors they consistently beat in head-to-head sales situations.

The pattern tells you a sequencing story:

The "they have a basic presence but the specifics are wrong" companies need content corrections and authority-building around specific claims.

The "they exist but they're not competitive" companies need a strategic reposition in AI's understanding of their category.

The "they fail all three" companies need foundational work before anything else matters.

No scorecard that fails to distinguish these three problems can tell a company what to do next. Which is why most AI visibility audits produce a lot of data and almost no direction.

The Point of the Framework

The Scorecard isn't a tool to generate metrics.

It's a framework to sort companies into the three buckets that matter, so the next intervention is strategically sound rather than randomly reactive.

Question 1: Do I exist?

Question 2: Am I represented accurately?

Question 3: Am I competitive?

Answer those three in that order. Let the answers determine where you invest. Skip the intermediate metrics that don't map to strategic action.

Most companies measure everything and decide nothing. The Scorecard decides nothing and measures only what matters.

That distinction is the whole point.