An AI visibility audit checks four pillars worth 25 points each: Technical Health, Content Quality, Authority and Trust, and AI Presence. Most businesses fail on the first two before the rest can compound.
You ran the test. You typed "best [your service] near me" into ChatGPT and waited for your business to appear. It did not. Three competitors did, two of whom you have never heard of. So you went looking for an answer and ended up reading about audits, scans, and scores, and now you want to know what one of those things actually checks before you spend any time on it.
Fair question. Most articles about AI visibility audits stay vague. They list trends, tell you AI search is changing, and suggest you "optimise for it" without naming what gets measured.
This post is the opposite. It walks through exactly what a proper AI visibility audit looks at, in the order it looks, with examples of what passes and what fails at each pillar. By the end, you will be able to predict your own score within a few points, and you will know which pillar to fix first.
Key Takeaways
- An AI visibility audit checks four pillars, each worth 25 points: Technical Health, Content Quality, Authority and Trust, and AI Presence. The score is the sum, out of 100.
- Technical Health checks whether AI engines can crawl, parse, and confirm your business through schema markup, page speed, and an llms.txt guide file.
- Content Quality checks whether your headings, FAQ structure, and per-service pages give AI engines something specific to extract and cite.
- Authority and Trust checks Google Reviews, Google Business Profile completeness, and third-party mentions. This is the pillar most owners underweight.
- AI Presence is the output check: which AI engines actually name your business for real customer queries, and which competitors get named instead.
Why does an SEO audit miss what AI engines see?
A standard website audit checks whether Google can find your pages and rank them. It looks at keywords, backlinks, indexability, and page speed. Useful work. It is also designed for a different question than the one AI engines are answering.
Google ranks ten links and lets the user pick. AI picks one answer and names a few businesses inside it. The signals overlap, but they diverge in the places that decide whether you get named or skipped. An SEO audit will tell you that your page about acupuncture loads quickly and ranks for "acupuncture Brisbane". It will not tell you whether ChatGPT mentions you when someone asks for "the best acupuncturist for migraines in Brisbane".
That second question is the one your prospects are actually typing. Conductor's analysis of 21.9 million queries found AI Overviews now appear in 25% of Google searches, up from 13% twelve months ago. Ahrefs found AI features reduced click-through rates for top-ranking content by 58%. The traffic that used to come from a Google ranking is being absorbed into single AI answers, and those answers either name your business or they do not.
The audit you need is not a faster version of the old one. It checks different signals, in a different order, against a different output.
What goes wrong when you skip the audit and just start fixing things?
You fix the wrong layer. Most businesses owners I talk to have already done some work on their AI visibility before they get to a scan. They have rewritten a homepage, added an FAQ, asked for more reviews. The reason none of it has shifted their AI presence is almost always the same: they fixed the layer they could see, not the layer that was actually blocking them.
A common pattern: an owner spends two months publishing four new blog posts, certain that "more content" is the answer. The scan then shows the real blocker is missing schema markup and a Google Business Profile that has not been claimed. The new content is not invisible to AI because it is bad. It is invisible because nothing on the site confirms what business is publishing it. Princeton's GEO research found structured data appears in 61% of AI-cited pages. Without it, AI engines fall back to guesswork, and they make conservative guesses about businesses they cannot confirm.
Another common pattern: the owner has good schema, a healthy Google Business Profile, and detailed service pages. They are still missing from AI answers. The audit shows their business name appears on almost no third-party sites: no industry directories, no professional association pages, no media mentions. The trust pillar is empty. AI engines need outside confirmation, not just a good website.
The audit exists so you do not waste a quarter fixing the wrong thing. It tells you which of the four pillars is dragging your score, and which one to start on this week.
The Four-Pillar Visibility Scorecard
The Four-Pillar Visibility Scorecard is the structure the GetRecommended.io scan uses to measure AI visibility. Each pillar is worth 25 points. The score is the sum, out of 100. The pillars are checked in the order they affect each other, because fixing a later pillar before an earlier one is mostly wasted effort.
Pillar 1: Technical Health (25 points). This pillar checks whether AI engines can crawl, parse, and confirm your business at the code level. The scan looks at four things. Schema markup (Organization, LocalBusiness, Service, FAQPage) is the foundation, and missing schema is a measurable visibility deduction. An llms.txt guide file tells AI crawlers which pages on your site actually matter. Page speed and crawlability decide whether AI engines can read your pages at all. Canonical URLs and sitemap hygiene stop duplicate content confusing the engines. This pillar is mostly a developer fix, and it is the fastest one to move.
Pillar 2: Content Quality (25 points). This pillar checks whether your content gives AI engines something specific to extract and cite. The scan looks at three things. Question-led H2 and H3 headings match the format AI engines produce when answering queries, so they extract from this structure reliably. A FAQ section with clearly labelled questions and answer-first paragraphs is the single highest-extractability content pattern. Per-service pages, not bundled service descriptions, let AI engines surface a specific service for a specific query. A page that lists "acupuncture, physiotherapy, massage" as one block dilutes the signal for all three.
Pillar 3: Authority and Trust (25 points). This pillar checks whether outside sources confirm your business exists, operates, and delivers. The scan looks at three things. Google Reviews are checked for presence, recency, completeness, and review-text patterns, because Google Reviews are a key trust signal for AI search engines. Google Business Profile completeness correlates with 2.8 times higher AI search appearance, according to SE Ranking's 2.3 million-page study. Third-party mentions in industry directories, professional association sites, and credible local sources confirm what your homepage claims. This is the slowest pillar to build and the one most owners underweight.
Pillar 4: AI Presence (25 points). This is the output pillar. The scan runs real customer queries (not your business name) across ChatGPT, Perplexity, Gemini, and other engines, and records whether your business appears, where it appears, and which competitors AI recommends instead of you. Brand-name queries are excluded because they return informational content rather than a recommendation. Non-brand customer queries are the ones that drive new business, so they are the ones the score uses. This pillar is diagnostic. It tells you whether the work you have done on the first three pillars has reached the engines yet.
The pillars stack. Pillar 1 confirms who you are. Pillar 2 gives engines material to extract. Pillar 3 backs you with outside evidence. Pillar 4 measures what reached the answer. Working on Pillar 4 directly is impossible; you can only move it by improving the first three.
Practical steps to run your audit in the next 14 days
A two-week sequence that walks each pillar in order. You can finish the first pass in an afternoon, but the trust pillar is the one that compounds over months.
-
Day 1. Run the free AI visibility scan. This gives you the baseline score across all four pillars in about two minutes, names which competitors AI engines recommend instead of you, and tells you which pillar is dragging the score most. Start here so you are not guessing about which layer to fix.
-
Day 2 to 3. Audit Pillar 1: Technical Health. Open your site in a browser, view page source, search for "application/ld+json". If it is missing, schema markup is the single fastest fix. Add Organization schema (or LocalBusiness if you have a physical location), Service schema for each service, and FAQPage schema for any FAQ section. Add an llms.txt file at the root of your domain naming your three to five most important pages. A developer can do this in two to four hours.
-
Day 4 to 6. Audit Pillar 2: Content Quality. Read your homepage and your top three service pages. Rewrite topic-label headings as question headings ("Employment Law Services" becomes "What does an employment lawyer do for small businesses?"). Move the answer to the first sentence of each section. If your services are bundled on one page, split them into individual pages. AI engines extract per-service answers, and bundled pages get skipped.
-
Day 7 to 10. Audit Pillar 3: Authority and Trust. Check your Google Business Profile completeness in Google's dashboard. Fill any blank fields. Search for your business name across the top five industry directories and professional associations in your sector. Note where you are missing. Check your Google Review profile for recency and pattern clarity, not just count. If your last review is older than 90 days, build a request into your service-completion flow this week.
-
Day 11 to 13. Audit Pillar 4: AI Presence. Pick five customer queries (not your business name) that a real prospect would type. Run them across ChatGPT, Perplexity, and Gemini. Record whether your business is named, which competitors are named, and what specific phrases the engines use. This is the output measurement and gives you a before-and-after benchmark.
-
Day 14. Re-run the scan. Compare to the Day 1 baseline. Pillar 1 should have moved noticeably if you implemented the schema and llms.txt fixes. Pillar 2 should have moved if you split bundled service pages. Pillars 3 and 4 will lag, because trust and presence compound over months. The 90-day re-check is where those pillars usually shift.
Most owners are surprised by which pillar is weakest. The visible problem (Pillar 4: not appearing in AI answers) is almost never the actual blocker. The blocker is usually in Pillar 1 or Pillar 3.
Frequently asked questions
What does an AI visibility audit actually check?
An AI visibility audit checks four pillars, each worth 25 points: Technical Health (schema markup, llms.txt, crawlability), Content Quality (question-led headings, FAQ structure, per-service pages), Authority and Trust (Google Reviews, Google Business Profile, third-party mentions), and AI Presence (which AI engines name your business for live customer queries). The total score is out of 100, and the audit names which competitors AI engines recommend instead of you so you can see the gap directly.
How is an AI visibility audit different from an SEO audit?
An SEO audit checks ranking signals: keyword density, backlinks, indexability, page speed. An AI visibility audit checks recommendation signals: whether AI engines can identify, verify, and confidently name your business in a single synthesised answer. A business can rank on page one of Google and still be invisible to AI engines that cannot confirm who it is or what it does. The signals overlap on content quality and structured data, but diverge on third-party platform presence, content extractability, and trust format.
What score do I need to be visible in AI recommendations?
There is no universal pass mark. What matters is how your score compares to the alternatives AI engines surface for your customer queries. A score of 60 may be enough if competitors score 40. The same score may not be enough if competitors score 75. The scan names the competitors AI engines recommend instead, which is the comparison that actually matters.
Why are brand-name searches excluded from the visibility score?
Brand-name queries (typing your business name into ChatGPT) return informational content about your business rather than a recommendation. They tell you AI knows you exist but not whether AI would recommend you to a stranger asking for help. The visibility score uses non-brand customer queries because those are the queries that drive new business. Including brand queries would inflate the score and hide the gap that matters.
Can I do an AI visibility audit myself?
Yes for most of it. Pillar 2 (Content Quality) is fully self-auditable: read your own pages and check headings and FAQ structure. Pillar 3 (Authority and Trust) is partly self-auditable: check your Google Business Profile and review count. Pillar 1 (Technical Health) needs a free schema validator or developer access. Pillar 4 (AI Presence) needs you to run the same query across ChatGPT, Perplexity, and Gemini and see whether your business appears. The free scan does all four pillars in about two minutes and produces the same scorecard a manual audit would, in less time.
The Bottom Line
An AI visibility audit is not a vague health check. It is four specific pillars, each worth 25 points, checked in the order they depend on each other. Technical Health confirms who you are. Content Quality gives AI engines something to extract. Authority and Trust backs you with outside evidence. AI Presence measures what reached the answer.
You do not need to fix all four pillars to start showing up. You need to know which pillar is weakest, fix that one first, and let the rest compound. The owners I see make the fastest progress are the ones who stop guessing about which layer is broken and start measuring it.
If you want to know your own scorecard before you spend another afternoon fixing the wrong thing, run the free AI visibility scan and look at which pillar is dragging your score. For more on the seven specific signals AI engines use, read how to get recommended by AI search engines. Common questions about scan results are answered on the scan FAQ. For the deeper picture on why AI search behaves differently to Google, Search Engine Land's 2026 AI local visibility report is the most rigorous public dataset available.