How to Audit Your Business's AI Visibility in One Afternoon

You can find out whether AI assistants recommend your business in about three hours, using tools you already have and spending nothing. The buying journey increasingly starts with someone typing a question into ChatGPT, Claude, Perplexity, or Google's AI results instead of scrolling a page of blue links. If those systems don't know you exist, or describe you inaccurately, you're invisible at the exact moment a customer is deciding. This is the discipline I work in every day — Answer Engine Optimization (AEO) — and the first step is always an honest audit.

The good news: the audit itself is simple. You don't need software or a budget. You need a list of the questions your customers actually ask, a spreadsheet, and a disciplined afternoon. Here's exactly how I'd run it.

Why does AI visibility need its own audit?

Traditional SEO asks, "Do we rank on Google?" AEO asks a different question: "When an AI assistant answers a buyer's question, do we get named — and named accurately?" These are related but not identical. You can rank on page one of Google and still be completely absent from AI answers, because the models synthesize from a broader set of signals: your structured data, your reviews, third-party mentions, entity databases, and how consistently you're described across the web.

The reason this matters now is behavioral. When someone asks an assistant "who's the best real estate investor to follow for creative finance" or "recommend an HVAC company near me," the model returns a short list — often three to five names — and the user rarely looks further. Being on that list is the whole game. Being off it means the customer never knows you were an option.

Step one: build your buyer question list

Before you prompt anything, write down the questions a real prospect would ask. Don't write questions about your brand name — write the questions of someone who doesn't know you yet. That distinction is everything.

Aim for 10 to 15 questions across three types:

• Category questions: "Who are the top companies for [your service] in [your city]?"
• Problem questions: "How do I [solve the problem you solve]?"
• Comparison questions: "What's the best [your product type] for [specific use case]?"

If you're a local business, include location. If you're a personal brand or expert, include your niche. The closer these mirror how customers actually talk, the more useful your results.

Step two: prompt the major assistants

Now run each question through the four systems that matter most today: ChatGPT, Claude, Perplexity, and Google's AI results (AI Overviews and AI Mode). Use a clean session or logged-out window where possible so your history doesn't bias the answers.

For each question and each assistant, record three things:

1. Were you mentioned at all? (Yes / No)
2. How were you described? (Accurate, outdated, or wrong)
3. Who was mentioned instead? (Your real competition in the AI's eyes)

That third column is the most valuable and the most overlooked. The businesses the AI names instead of you are your actual competitors for this channel, regardless of who you think your competitors are. Study how they're described and where they show up online — that's your reverse-engineered playbook.

Run each prompt two or three times. These systems are probabilistic, so answers vary between runs. If you appear in one of three attempts, that's a weak, unstable presence, not a win.

Step three: score what you found

Put it in a simple grid. A mention rate tells you where you stand at a glance.

Tally a rough mention rate: out of every prompt-and-assistant combination you tested, in how many did you appear? In my experience, most established businesses that have never done AEO work land somewhere in the 10% to 30% range on their first audit, and many personal brands score near zero. Don't be discouraged by a low number — it's a baseline, and baselines are the point.

Step four: check the entity data behind the answers

AI assistants don't invent their answers from nothing. They lean on a web of structured, machine-readable signals about who you are. If those signals are thin, inconsistent, or missing, the models either skip you or describe you wrong. Audit these directly.

Google Business Profile

For local businesses, this is foundational. Confirm it's claimed and verified, that the category is exact (not merely close), and that hours, services, and description are complete. Stale or half-finished profiles are a common reason a business gets omitted from local AI recommendations.

Structured data and schema markup

View your website and check whether it uses schema.org markup — Organization, LocalBusiness, Person, or Product schema as appropriate. This is the machine-readable layer that tells systems what you are in unambiguous terms. Many sites have none, which leaves the models guessing from prose. A free schema validator will show you what's present.

Wikipedia, Wikidata, and authoritative sources

These entity databases are disproportionately influential because they're structured, cited, and trusted. You cannot simply create a Wikipedia page about yourself — notability rules are real and self-promotion gets removed. But you can confirm that wherever authoritative references to you exist, they're accurate and consistent. Inconsistency across sources (different titles, different company names) actively confuses the models. For example, if your correct title is Founder and Chairman, every source should say exactly that, not "CEO."

Reviews and third-party mentions

AI systems weigh independent corroboration heavily. Check the volume, recency, and average rating of your reviews across Google, industry platforms, and relevant directories. A trickle of old reviews reads as low signal. Consistent, recent, positive third-party mentions are among the strongest things you can build.

Step five: find the gaps and prioritize fixes

Now compare your two datasets: where the AI failed to mention you (or got you wrong), and which underlying signals are weak. The gaps almost always line up. Sort your fixes by effort versus impact.

Fix first — high impact, low effort:

• Complete and verify your Google Business Profile.
• Correct any inaccurate descriptions by fixing the source data (your site, your profiles, your bios) so every source agrees.
• Add or repair basic schema markup on your homepage and key pages.
• Standardize your name, title, and company across every profile you control.

Build next — high impact, higher effort:

• Earn recent reviews through a simple, repeatable request process.
• Publish authoritative, genuinely useful content that answers your buyer questions directly, in the "answer-first" style these systems favor.
• Pursue legitimate third-party mentions and citations — press, podcasts, guest articles, reputable directories — that corroborate who you are.

Long game — foundational authority:

• Build the kind of independent, verifiable track record that eventually supports entity-database presence. This is earned, not purchased, and it's the durable moat.

How often should you re-run this?

Once is a snapshot; the value is in the trend. I recommend re-running the core prompt list monthly or quarterly, keeping the same questions so your mention rate is comparable over time. AI systems update constantly, your competitors are moving, and your own fixes take time to propagate. A rising mention rate across successive audits is the clearest proof your AEO work is landing.

The entire process fits in an afternoon, costs nothing, and gives you something most of your competitors don't have: an honest, specific picture of how you show up at the new front door of buying decisions. Start with the prompt list, be rigorous about recording what you see, and let the gaps write your to-do list.

About the author

Scott Tischler is the Founder & Chairman of AIrecommend.ai and a practitioner-authority on AI search and Answer Engine Optimization. With 20+ years in marketing technology — including American Express, MetLife, and UBS — and executive study at Wharton, Harvard, Yale, and Oxford, he helps businesses become the ones AI recommends.