For years, search visibility had a familiar shape. A buyer typed a query into Google, scanned a page of links and decided where to click.
That journey has fractured. As I explored in The New Shelf Position, the digital shelf is no longer simply a page of search results. It is the shortlist systems buyers use to research, compare, and decide.
Today, the same buyer might ask ChatGPT for a shortlist, use Google AI Mode to compare vendors, read a conventional search result, check LinkedIn, and then ask a colleague for reassurance. Discovery no longer happens in one place, and neither does trust.
This creates a new problem for B2B companies: your website can rank for relevant keywords, and your brand can still be absent from the answer.
The question is no longer only, “Can people find us?” It is also:
- Can search engines and AI systems understand what category we belong to?
- Can they connect our claims to credible evidence?
- Can they retrieve the right information when a buyer asks a specific question?
- Do we give that buyer enough confidence to choose us?
That is what an AI visibility audit should test. As I explained in What a Marketing Audit Actually Reveals, the purpose of an audit is not to produce a longer list of activities. It is to identify the constraint preventing those activities from producing a result.
What is a B2B AI visibility audit?
A B2B AI visibility audit is a structured evaluation of whether a company is easy to understand, trust, retrieve and recommend across traditional search and AI-powered discovery.
It examines more than rankings. It looks at the clarity of the company’s positioning, the evidence behind its claims, the accessibility of its content, its authority beyond its own website and the consistency of the information machines use to interpret the brand.
The audit should answer four questions:
- Understand: Is it immediately clear what the company does, for whom and why it is different?
- Trust: Are its claims supported by proof that exists both on and beyond its website?
- Retrieve: Can search and AI systems access the right passage for a specific buyer question?
- Choose: Does the complete experience reduce risk and make the next step obvious?
I use these four stages because visibility without clarity creates noise, authority without retrievability remains hidden, and discovery without trust rarely creates demand.
First, ignore the mythology around GEO
There is already too much advice treating generative engine optimization as a collection of secret formatting tricks.
The reality is less theatrical and more useful.
Google says the same foundational SEO practices used for conventional search apply to its AI features. A page must be crawlable, indexed and eligible to appear with a snippet. Google also says there is no special AI schema or machine-readable file required for inclusion in AI Overviews or AI Mode.
OpenAI similarly gives publishers explicit controls for search discovery through OAI-SearchBot, separate from GPTBot controls related to model training.
In other words, the fundamentals still matter. But the standard of clarity has risen.
AI systems often assemble answers by retrieving and synthesizing information from multiple sources. Google describes a “query fan-out” process in which its AI features may issue several related searches across subtopics. A company therefore needs more than one page targeting one keyword. It needs a coherent body of evidence that can answer the different questions surrounding a buying decision.
There is no guaranteed formula for being cited or recommended. There is, however, a disciplined way to make a brand easier to interpret and substantiate.
The four-part B2B AI visibility audit
1. Understand: can the market and the machine explain you?
Start with the simplest test: ask five people to read your homepage for ten seconds and explain what your company does.
Then ask several AI search tools the same questions:
- What does [company] do?
- Who is [company] designed for?
- What category does [company] compete in?
- How is [company] different from its main alternatives?
You are not testing whether the response flatters you. You are testing whether the answer is accurate and consistent.
If the results vary wildly, the problem is often not the model. It is the underlying signal.
Many B2B homepages use broad language such as “transforming the future of work” or “unlocking innovation through AI.” Those statements may sound ambitious, but they do very little classification work. They do not identify the customer, the problem, the product category or the outcome.
Audit the following:
- Does the homepage state the category in plain language?
- Is the ideal customer explicit?
- Is the primary problem named in the customer’s vocabulary?
- Is the differentiated value concrete rather than aspirational?
- Are products, services and use cases separated clearly?
- Do the About page, social profiles, press bios and directory listings describe the company consistently?
The goal is not to flatten the brand into robotic copy. It is to create a stable core that can be understood everywhere, then build personality and emotion around it. As I argued in Why B2B Brands Are Afraid of Emotion—and Pay for It, technically accurate messaging is not necessarily memorable messaging.
A useful test: complete this sentence without jargon: “[Company] helps [specific customer] achieve [specific outcome] by [distinct approach].” If leadership cannot agree on the answer, the visibility problem begins with positioning.
2. Trust: are your claims supported by independent evidence?
A company’s website tells the market what it wants to be known for. The wider web helps determine whether that claim is believable.
This is where SEO, digital PR and brand strategy stop being separate activities.
Suppose a technical SaaS company says it is the leading solution for a regulated industry. An AI system—or a cautious buyer—will look for corroboration. Are there named customers? Detailed case studies? Recognized experts? Relevant media mentions? Conference appearances? Documentation? Reviews? Original research? Clear authorship?
Audit your evidence across five areas:
- First-party proof: case studies, customer outcomes, methodology, product documentation and original data.
- Third-party validation: reputable media coverage, expert citations, reviews, awards and analyst or industry recognition.
- Expert identity: named authors, leadership biographies, relevant experience and a visible connection between people and claims.
- Freshness: current product information, recent examples and visible update dates where they genuinely matter.
- Consistency: the same core facts, category and proof points across the website and credible external sources.
Avoid the temptation to manufacture volume. Fifty low-quality syndications do not create the same trust as one relevant, independently earned source. Authority is not a numbers game; it is contextual.
For B2B brands, the most valuable evidence often comes from the work itself: a measurable customer result, a useful dataset, a distinctive framework or a specialist point of view that others choose to reference.
3. Retrieve: can the right answer be found at the right moment?
A brand may be clear and credible yet remain difficult to retrieve.
This happens when useful knowledge is buried in PDFs, locked inside videos, hidden in JavaScript interfaces or spread across vague pages that each try to say everything.
Google explicitly recommends that important content be available in text, accessible through internal links and supported by structured data that matches what users can see. These are not glamorous tactics, but they determine whether information can be discovered and interpreted.
Map content to the questions a real buying committee asks:
- What does the product do?
- Who is it for—and who is it not for?
- How does it work?
- How is it different from an alternative?
- Can it integrate with the buyer’s existing stack?
- Is it suitable for a regulated or complex use case?
- What does implementation involve?
- What evidence supports the claimed result?
- What does it cost, or how is pricing determined?
- What are the risks and limitations?
Then check whether each answer has a clear, indexable home.
Strong retrieval does not require writing an article for every slight variation of a keyword. It requires a deliberate content architecture: authoritative core pages supported by focused explanations, comparisons, use cases, case studies and FAQs where they genuinely help.
Each page should have one dominant purpose. Use descriptive headings, answer the central question early, define specialist terms, link related evidence and make important claims easy to quote without removing their context.
Structured data can help search engines interpret information, but it must reflect the visible page. Adding schema to weak or contradictory content does not repair it. Likewise, an llms.txt file may be experimented with by some publishers, but Google states that no special AI file is required for its AI search features. Treat machine-readable extras as supporting infrastructure, not as a substitute for a coherent site.
4. Choose: does visibility turn into a commercial decision?
Marketing teams can become so focused on being cited by AI that they forget the buyer.
A mention is not a strategy. Traffic is not a result. Even a perfect AI-generated description of your company has limited value if the next interaction creates doubt.
Audit the decision experience:
- Does the page match the promise that brought the visitor there?
- Is the value proposition specific to the visitor’s context?
- Are proof points close to the claims they support?
- Does the company acknowledge reasonable limitations and trade-offs?
- Is there a clear next step for buyers at different levels of intent?
- Can a visitor understand what will happen after making contact?
For complex B2B sales, “choose” does not always mean purchase. It may mean adding the company to a shortlist, sharing a page internally, requesting a technical conversation or returning when the timing is right.
Design for that progression.
A simple AI visibility scorecard
Score each statement from 0 to 2:
- 0: absent or inaccurate
- 1: partially present or inconsistent
- 2: clear, supported and current
Understand
- Our category, audience and core outcome are explicit.
- Our positioning is consistent across important online profiles.
- Search engines and AI tools describe us accurately.
Trust
- Our most important claims are supported by specific evidence.
- Relevant third parties validate our expertise or results.
- Our authors and experts have clear, credible identities.
Retrieve
- Important content is crawlable, indexable and available in text.
- Buyer questions have clear answers on dedicated or logically structured pages.
- Internal links connect claims, explanations and supporting evidence.
Choose
- Our proof is visible at the moment a buyer needs reassurance.
- Our next steps are clear and appropriate to the buyer’s intent.
- The website makes it easy to compare, share and continue evaluating us.
The maximum score is 24.
- 0–8: Your brand is difficult to interpret and easy to omit.
- 9–16: The foundation exists, but gaps are weakening discovery or trust.
- 17–20: Your visibility system is strong, with specific opportunities to improve.
- 21–24: You have a coherent foundation; focus on monitoring, new evidence and high-value topic expansion.
The number is directional, not scientific. Its value is in forcing honest discussion across brand, content, SEO, PR and conversion—functions that buyers and machines experience as one system.
How to test AI visibility without fooling yourself
One prompt in one tool proves very little. AI responses can vary by model, location, freshness, wording and whether live web search is used.
Build a repeatable prompt set instead:
- Define 15–25 questions across awareness, comparison, risk and purchase intent.
- Test them on the AI and search experiences your buyers are likely to use.
- Record whether your brand appears, how it is described, which sources are cited and whether important competitors appear.
- Separate branded prompts from non-branded category prompts.
- Repeat the same test monthly or quarterly.
- Pair citation tracking with business outcomes such as qualified enquiries, influenced pipeline and assisted conversions.
Do not grade only for mentions. Grade for accuracy, relevance, source quality and whether the answer represents a meaningful buying situation.
What to fix first
If the audit uncovers dozens of weaknesses, do not respond by publishing dozens of generic articles.
Fix the system in this order:
- Correct technical access problems. Important pages must be crawlable, indexable and internally linked.
- Clarify the category and positioning. Machines cannot resolve what the business itself has not defined.
- Strengthen high-intent pages. Improve product, service, solution, comparison and case-study content before expanding the editorial calendar.
- Close the evidence gap. Add concrete proof and pursue credible third-party validation.
- Build topical depth around buyer decisions. Publish material that resolves real questions, not content created simply to hit a volume target.
- Measure the complete journey. Monitor search, AI referrals, citations, branded demand and qualified commercial outcomes together.
This is why I increasingly see SEO, GEO, AEO, content and PR as one visibility system. Each discipline solves a different part of the same problem. I explain those roles in more detail in SEO, GEO, and AEO: The Three Disciplines That Now Control Whether Your Brand Gets Found and SEO Is Not Enough Anymore: What GEO and AEO Actually Change.
The real objective is not to rank. It is to become the obvious choice.
AI search has changed the interface, but not the commercial reality underneath it.
Buyers still need to understand what you do. They still need evidence. They still need the right information at the right moment. And they still need a reason to trust the decision.
The companies that win the next phase of discovery will not be the ones that produce the most content or chase every emerging GEO tactic. They will be the ones that make their expertise clear, their evidence accessible and their value easy to retrieve.
That is the standard I would use to audit visibility now: understand, trust, retrieve, choose.
I work with founders and B2B leadership teams to connect positioning, SEO, AI visibility, content and PR into one growth system. If your company is doing credible work but remains difficult to find, explain or choose, start a conversation.