AI visibility check is a method of testing whether AI engines can find, extract, and cite content from a client's site. It involves running the site through a set of technical and content audits to see what blocks discovery and what makes citation more likely.
You need this when a client asks whether their site shows up in ChatGPT, Perplexity, or Google's AI Overviews. You also need it when you are deciding whether to add AI-search visibility as a service line.
This guide walks you through the full process. By the end, you will have a complete audit checklist and scoring rubric you can run on any client site.
What AI engines look for when deciding whether to cite a site
AI engines need to trust what they extract. They look for named sources, structured data, and content that can be verified.
If a page makes a claim without naming where it came from, the engine cannot verify it. If the page has no schema markup, the engine cannot tell what the content means. If the site has no robots.txt or sitemap, the engine may not crawl it at all.
Your job is to check all three layers: technical access, structured data, and content quality.
Prerequisites: what you need before you start
You need access to the client's site, Google Search Console, and any CMS login. You also need a way to check schema markup, run a crawl, and test accessibility.
If the client has no Search Console access set up, add it first. You cannot audit technical SEO without crawl data.
Set aside 30-45 minutes for the first audit. After that, you can run the same checklist in 15 minutes.
Step 1: Check technical access
Open the site's robots.txt file by adding /robots.txt to the domain. Check whether any important pages are disallowed. If you see Disallow: / with no user-agent specified, the entire site is blocked.
Next, check the sitemap. Look for a link in robots.txt or add /sitemap.xml to the domain. Open it and confirm that key pages are listed. If the sitemap is missing or outdated, AI engines may miss new content.
Finally, check the site's meta robots tags. Open the source code of a few key pages and search for <meta name="robots". If you see noindex or nofollow, those pages will not be crawled.
Common mistake: Agencies often check robots.txt but forget to check meta robots tags on individual pages. A single noindex tag can block an entire section.
Step 2: Check structured data
Structured data is defined as machine-readable markup that tells AI engines what your content means. Without it, engines have to guess.
Use Google's Rich Results Test or Schema.org validator. Paste in the URL of a key page. Look for Organization, Article, FAQPage, Product, or LocalBusiness schema.
If the page has no schema, add it. If the schema is present but incomplete, fill in the missing fields. AI engines prefer schema that includes author, datePublished, and publisher fields.
Also check for an llms.txt file. This is a plain-text file that tells AI engines what the site is about and where to find key content. It lives at the root domain, like robots.txt.
Common mistake: Agencies add schema but never validate it. Broken schema is worse than no schema because it sends conflicting signals.
Step 3: Check content quality and citations
AI engines cite content they can verify. That means every factual claim needs a named source.
Open the client's top 5 pages by traffic. Read through each one. For every sentence that makes a claim, ask: where did this come from?
We suggest flagging any sentence that uses vague attribution phrases without naming a specific, checkable source. For example, if a sentence says "According to research" but does not name which research, flag it for revision. If it references what "most businesses" do but cites no source for that prevalence claim, flag it as well.
Rewrite flagged sentences to either name the source or remove the claim. AI engines will not cite content that borrows the authority of evidence while naming nothing a reader could check.
Common mistake: Agencies assume that vague phrases like "industry data" or "experts say" are good enough. They are not. Name the source or delete the claim.
Step 4: Check readability and structure
AI engines extract content in chunks. If a paragraph is too long or a sentence is too complex, the engine may skip it.
Check the client's top pages for paragraphs longer than 60 words. Break them into smaller chunks. Check for sentences longer than 20 words. Simplify them.
Also check for headers. Every page should have at least 4-6 H2 headers and 2-3 H3 subheaders. Headers help engines understand the structure and extract the right sections.
Common mistake: Agencies focus on keyword density and ignore structure. AI engines care more about clear hierarchy than keyword repetition.
Step 5: Check accessibility
Accessibility refers to a set of technical standards that make content usable by people with disabilities. AI engines use the same standards to parse content.
Run the site through an accessibility checker. Look for missing alt text, low-contrast text, broken ARIA labels, and keyboard navigation issues. These are the same issues that block AI engines from understanding images, tables, and interactive elements.
AISO Studio uses axe-core for real code-level WCAG findings and includes screen-reader simulation for accessibility testing. If you are running audits manually, use the WAVE browser extension or axe DevTools.
Common mistake: Agencies treat accessibility as a compliance checkbox. It is also a content-quality signal.
AI Visibility Audit Checklist and Scoring Rubric
Use this checklist on every client site. Score each section out of 100, then average the scores for a final AI visibility score.
Technical Access (0-100 points)
- Robots.txt allows crawling of key pages (25 points)
- Sitemap.xml exists and is up to date (25 points)
- No noindex or nofollow tags on key pages (25 points)
- Site loads in under 3 seconds on mobile (25 points)
Scoring: 1 point per item completed, multiplied by 25.
Structured Data (0-100 points)
- Organization schema present and valid (20 points)
- Article schema on blog posts with author and datePublished (20 points)
- FAQPage schema on FAQ sections (20 points)
- LocalBusiness schema on location pages (20 points)
- llms.txt file present at root domain (20 points)
Scoring: 1 point per item completed, multiplied by 20.
Content Quality (0-100 points)
- Every factual claim names a source (30 points)
- No vague phrases like "studies show" or "research indicates" (20 points)
- No prevalence claims without named sources (20 points)
- Content updated in the last 90 days (15 points)
- Byline and author bio present (15 points)
Scoring: Award points based on percentage of pages that pass each check.
Readability (0-100 points)
- Average paragraph length under 60 words (25 points)
- Average sentence length under 20 words (25 points)
- 4+ H2 headers per page (25 points)
- 2+ H3 subheaders per page (25 points)
Scoring: 1 point per item completed, multiplied by 25.
Accessibility (0-100 points)
- All images have descriptive alt text (20 points)
- Text contrast meets WCAG AA standards (20 points)
- All interactive elements are keyboard-accessible (20 points)
- No broken ARIA labels (20 points)
- Screen-reader test passes on key pages (20 points)
Scoring: 1 point per item completed, multiplied by 20.
Final AI Visibility Score
Add the five section scores and divide by 5. The result is the client's AI visibility score out of 100.
Example:
- Technical Access: 75
- Structured Data: 60
- Content Quality: 80
- Readability: 90
- Accessibility: 70
Final Score: (75 + 60 + 80 + 90 + 70) / 5 = 75 out of 100
Include this score in the client report. Flag any section below 60 as high-priority.
Troubleshooting: what to do when the audit finds problems
If the technical access score is low, start with robots.txt and sitemap. These are the fastest fixes and unblock everything else.
If the structured data score is low, add schema to the top 10 pages by traffic. Use a plugin like Yoast or Rank Math if the client is on WordPress. If the client is on a custom CMS, write the schema manually and add it to the page template.
If the content quality score is low, rewrite the flagged sentences. Either name the source or remove the claim. If your client's content strategy allows for hedged language, test whether it reduces citation rates in your own audits before applying it broadly.
If the readability score is low, break up long paragraphs and add headers. This is the easiest fix and has the biggest impact on AI visibility.
If the accessibility score is low, start with alt text and contrast. These are the most common issues and the easiest to fix.
Frequently Asked Questions
Question: What is an AI visibility audit?
An AI visibility audit is a process of checking whether a site meets the technical and content standards that AI engines use to decide what to cite. It covers technical access, structured data, content quality, readability, and accessibility.
Question: How is AI visibility optimization different from SEO?
AI visibility optimization means setting up content so AI engines can extract, verify, and cite it. SEO focuses on ranking in search results. The two overlap in technical areas like schema and crawlability, but AI visibility puts more weight on named sources and content verification.
Question: What does a visibility zero score mean?
Visibility zero means the site is either blocked from crawling, has no structured data, or has content that cannot be verified. It is the starting point for most audits. Fix technical access first, then add schema, then rewrite unsupported claims.
Question: How do I check AI local search visibility?
AI local search visibility refers to how often a business appears in AI-generated answers to location-based queries. Check for LocalBusiness schema, Google Business Profile integration, and location-specific content with named sources. Also check that the site's NAP (name, address, phone) matches across all platforms.
Question: What is an AI local visibility report?
An AI local visibility report is a document that shows how a local business performs in AI search. It includes schema validation, citation analysis, and a breakdown of which queries trigger the business name. AISO Studio generates white-label AI local visibility reports with agency branding.
Question: How do I track AI search brand visibility over time?
AI search brand visibility means how often a brand is named in AI-generated answers. Track it by running the same set of queries every month and logging whether the brand appears. Also check citation counts in tools that monitor AI engine output. AISO Studio includes automated WCAG scoring with risk flags and tracks changes over time.
Key Takeaways
- AI visibility check is a method of testing whether AI engines can find, extract, and cite a site's content.
- Start with technical access: check robots.txt, sitemap, and meta robots tags.
- Add structured data to every key page using Organization, Article, FAQPage, or LocalBusiness schema.
- Rewrite any sentence that makes a claim without naming a source.
- Break long paragraphs into chunks under 60 words and add headers.
- Run an accessibility audit using axe-core or WAVE to catch issues that also block AI engines.
- Use the scoring rubric to generate a final AI visibility score out of 100.
- Flag any section below 60 as high-priority and fix it before moving to the next client.
See it on a client's site
If you want to see what this looks like on a real client site, AISO Studio offers a free 7-dimension content audit at aiso.studio/audit. You get three free audits with no account required.
The platform also includes a 14-day full-access trial with no credit card required. You can run the full audit checklist, generate white-label PDF reports, and publish directly to WordPress. Every factual claim in generated content is verified, and sentences that claim evidence without naming a source are reported.
Start with the free audit and see the output. If it fits your workflow, try the 14-day pass and run it on your full client roster.