See whether AI platforms can find, understand, and recommend your business with a manual audit covering entities, content, and more.
I spend a lot of my time these days with independent hoteliers and small business owners discovering something uncomfortable: ranking well on Google doesn’t necessarily translate to visibility when a customer asks ChatGPT instead.
The numbers back them up. SOCi’s 2026 Local Visibility Index found that ChatGPT recommends just 1.2% of them, compared with 35.9% visibility in Google’s local 3-pack — roughly a 30x gap. In retail, SOCi found only about 45% overlap between the brands most visible in traditional local search and those AI platforms recommend, meaning a majority of retail brands winning the map pack were absent from AI answers entirely.
Enterprise brands can address that gap with a platform subscription, a vendor, and a line item in next year’s budget. For most of the small businesses and independent hoteliers I work with, none of that exists. There’s no budget for an AI visibility platform. There’s barely a budget for the website.
That doesn’t mean nothing can be done.
The audit has to be done by hand and done well enough to actually move the needle.
Here’s the version I’ve been teaching small business owners and hoteliers to run on their own, with nothing but a browser and a free AI assistant.
Why the old audit doesn’t catch this
A standard technical SEO audit checks crawlability, page speed, backlinks, and on-page optimization. All of that still matters. None of it tells you whether an AI system can find, trust, or cite your business because AI visibility depends on a different, wider set of signals:
- Entity consistency across every place you appear.
- Whether your facts are specific enough to extract cleanly.
- Whether anything backs up the claims on your page.
Entity authority is the foundation that most small business sites are missing entirely. AI trust is built on consistency. If your website contradicts your Google Business Profile, that’s entity drift, and it quietly tanks your confidence score with every platform that notices.
Most small businesses have never audited for this at all because the tooling to do it usually assumes an enterprise budget and an entity graph nobody’s built yet.
That’s the execution gap. Most organizations haven’t created an entity map tied to business priorities or audited whether their key entities are underdefined or inconsistently reinforced.
Some still optimize content briefs for keyword clusters rather than the structured, detailed entity coverage that large language models actually depend on.
That’s true at the enterprise level. It’s also true — doubled — for a 40-room independent hotel or a neighborhood bakery.
What to check in your AI visibility audit
Paid AI visibility tools generally run structured prompts across multiple AI platforms, track citations and mentions over time, benchmark against competitors, and flag inconsistencies across your entity footprint automatically. That ongoing, automated, cross-platform tracking is genuinely hard to replicate by hand, and it’s the real value a paid platform provides.
But the diagnostic part, the part that tells you where you stand today and what to fix first, doesn’t require automation. It requires being methodical about it. Here’s the manual version, broken into what I’d consider the essential checks.
1. Can AI systems actually reach your site?
Before anything else, check your robots.txt file at yourdomain.com/robots.txt for any line disallowing GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. Then check whether your homepage requires clicking through a language selector, a location picker, or a cookie wall before real content loads.
These block AI crawlers the same way they’d stall a first-time human visitor who can’t click through, except nobody notices because Google’s crawler usually gets a pass — these interstitials don’t extend to less established bots.
2. How complete is your entity inventory?
List every distinct thing about your business that an AI system might need to describe: the business itself, every product or room type, staff, policies, and location context.
Then check which of those actually have a documented, specific set of facts somewhere, rather than just a passing mention in marketing copy. Location and proximity facts — what’s nearby, how far, how to get there — are the single most commonly skipped category, even on otherwise well-documented sites.
3. Is your content specific or generic?
AI systems extract facts more reliably from unambiguous statements than from marketing language that leans on vague pronouns and empty adjectives.
Paste a page into an AI assistant and ask it directly: “Flag any sentence where a pronoun or vague phrase doesn’t clearly point to what it’s describing.” A page full of flagged instances is a page an AI system struggles to cite cleanly.
4. Can every claim be traced to a source?
An unverifiable superlative — “the best in town,” “world-class service” — is worse for AI trust than no claim at all because it can’t be checked against anything. List the claims on your key pages that use words like best, most, only, or award-winning, and ask where each one actually comes from.
5. Are you saying anything an AI can’t get anywhere else?
This is the information gain question, and it’s the one most small business sites fail hardest. Search a few of your own sentences in quotes and see how many competitors could’ve written the exact same thing.
A specific number, a named process, a first-party statistic your business alone has access to — that’s what separates content an AI system has a reason to cite from content it can skip because 10 other sites say the same thing.
6. Do you have structured data?
Right-click your homepage, view page source, and search for “application/ld+json.” If nothing appears, there’s no structured data on that page at all. If something does appear, check whether it’s a complete, deeply nested entity or a flat, thin one.
As one recent framework for enterprise entity strategy puts it, the real questions are whether your JSON-LD reflects your actual business ontology or sits flat, whether you’re using sameAs to connect to Wikidata and the Knowledge Graph, and whether your schema is kept in sync with your CMS and Google Business Profile or left to drift on its own. Small businesses can ask themselves the same three questions, scaled down.
7. Do your facts match everywhere?
List every place your business appears online — your website, Google Business Profile, Bing Places, Apple Maps, Facebook, any relevant directory — and check your hours, phone number, and key policies across all of them side by side. This is the single highest-leverage free fix available, and it’s also the one most consistently ignored.
8. Are you connected to a recognized knowledge source?
Search your business name in Google and check for a Knowledge Panel. Search it on Wikidata. Check whether your structured data includes a sameAs field pointing to either.
Entity resolution, confirming which real business a name actually refers to, carries real weight with AI systems, and most small businesses have never checked whether they’ve achieved it at all.
9. What do AI systems actually say about you, right now?
Ask ChatGPT, Claude, Perplexity, and Google directly, in a fresh, private browser session, a specific question about your business: hours, a policy, what’s nearby.
Compare the answer to what’s actually true. This is the step that turns everything above from theory into a measurable baseline, and it’s the one I’d tell any small business owner to do first if they only have 10 minutes.
Score it, then go deeper
Running through those nine checks and scoring yourself 0 to 2 on each gives a rough but genuinely useful baseline, out of 18.
For anyone who wants to go further without spending anything, the same E-E-A-T framework long used to judge content quality translates well here:
Are your basic trust signals (a secure site, real contact information, written-out policies) actually in place?
Does your content show real, first-hand experience rather than generic description?
Are the people behind your facts actually named and credentialed?
Are you recognized by sources beyond your own website?
When a manual audit isn’t enough
A manual audit is a snapshot, not a monitoring system. It tells you where you stood on the day you ran it.
AI systems change what they know about a business over time, sometimes for reasons outside your control, and tracking those shifts at scale is genuinely difficult to do by hand, especially across multiple markets or locations.
That’s a fair trade-off for a business with no budget, not a permanent state. The manual version gives you a real starting point, and for most small businesses right now, that starting point doesn’t exist.
A rough score today beats a perfect score you never measured because the tool cost more than a month of marketing spend. Start with the free version. Fix what it turns up. If the business grows to need continuous, automated tracking across dozens of locations and prompts, that’s a good problem to have — and a better one than not knowing where you stand.
