AI Visibility Report

What is the AI Visibility Report?

The AI Visibility Report is a free tool that runs thirty weighted checks on one page, split into Technical SEO, AEO and GEO, for a score out of 100. It also has an AI model read the page, reports whether it recognises the brand and which competitors it names, and lists the failed checks to fix first.

How to use the AI Visibility Report

1. Enter your domain, such as example.com. The report reads the page at that address, so a bare domain means your homepage.
2. Press Generate report, then Sign up free to unlock to see the result.
3. Read the score out of 100 and the three pillar bars: Technical SEO out of 40, AEO out of 30 and GEO out of 30.
4. Start with Quick AI Wins and Improvement Opportunities, then use the full check breakdown to see the measured value behind each pass or fail.

What is the difference between SEO, AEO and GEO?

In this report, technical SEO asks whether a crawler can read and index the page, AEO (answer engine optimization) asks whether an engine can lift a clean answer out of it, and GEO (generative engine optimization) asks whether a generative engine has enough entity and evidence signal to name you. Technical SEO has twelve checks worth 40 points, AEO nine worth 30, and GEO nine worth 30.

The weights sum to 100, so the score is the raw total of the checks you pass. Keeping the pillars apart is the point: a page can score full marks on technical SEO and close to nothing on the other two, and one blended number would hide that.

What does each pillar check?

Technical SEO checks for no noindex, a title of 30 to 70 characters, a meta description of 120 to 170, exactly one H1, a canonical, HTTPS, a viewport tag, a lang attribute, an Open Graph title and image, a Twitter card, alt text on 80% of images, and five or more internal links.

AEO looks for extraction surface: FAQ schema or three question-style headings, Article or HowTo schema, an opening paragraph that reads as a definition, three or more H2 and H3 headings, lists or tables, a TL;DR block, an author byline, a reviewed-by line, and three or more links to other sites.

GEO looks for what makes you a nameable entity: Organization or Brand schema, Product or Service schema, breadcrumbs, three or more schema types, 1,000 words or more, comparison wording, figures such as percentages or prices, twenty or more distinct capitalised names, and a published or updated date.

What does AI know about my brand?

The report's AI section shows it. An AI model reads the page's title, meta description, H1, schema types and opening copy, then says what it would tell a buyer about the brand, which competitors it groups you with, and which topics and categories you own. To see whether a model names you when a buyer asks about your category, use the AI Brand Visibility Checker.

The Strong, Mixed or Weak label under Brand Reputation Check follows the overall score: 75 and up is Strong, 50 to 74 Mixed, under 50 Weak.

How is the AI visibility score calculated?

Each of the thirty checks carries a weight from 2 to 5 points, and a check either earns its full weight or nothing. There is no partial credit: a title of 72 characters fails the 30 to 70 character check outright. The points you earn in each pillar make that pillar's score, and the three pillar scores add up to the overall score with no rescaling.

The weight also sets the priority shown on each failed check. Checks worth 4 or 5 points are High, 3 points Medium and 2 points Low. Quick AI Wins shows up to three failed High checks, Improvement Opportunities shows up to six failed checks of any priority sorted by weight, and Key AI Strengths lists up to five passed checks worth 4 points or more. Because the checks are measured in code, the score only changes when the page does.

What does an AI visibility report look like?

The report opens with the brand name, taken from the page's og:site_name tag or, failing that, from the domain, plus the date it was generated, an AI Readiness label and a count such as 21/30 checks passed. Below that sit the score out of 100 and three bars for Technical SEO, AEO and GEO, each showing points earned, checks passed and a percentage.

Next comes the "How ChatGPT sees" card, marked Knows the brand or Limited knowledge with a confidence of high, medium or low, and a paragraph on what the model would say. Brand Reputation Check shows your meta description as the summary, with up to three industry categories. Then come Key Topics Detected, up to four Competitors, Key AI Strengths, Improvement Opportunities and the full list of thirty checks, each with the measured value behind it, such as 142 chars or 3 H2, 0 H3.

How is this different from AI visibility tracking tools?

AI visibility tracking tools usually send a fixed set of buyer prompts to several AI engines on a schedule and record how often your brand is mentioned or cited over time. This report does something narrower: it audits one page for the signals that make it easy to read, quote and attribute, and it asks one model what it would say about the brand from that page.

The two answer different questions. Tracking tells you whether you are being named; this report tells you what on the page to change so you are more likely to be. To test buyer questions against one engine, the ChatGPT Visibility Checker writes buying-intent queries for your category that never name your brand, asks ChatGPT each one, and shows whether your brand comes up in the answers.

Why might the report miss something my page has?

The report reads the HTML the server sends and does not run JavaScript, so anything added in the browser, including schema injected by a tag manager, is not seen. Schema types are read only from JSON-LD, at the top level or inside an @graph; microdata and RDFa are not counted, and a JSON-LD block that fails to parse adds nothing.

Several checks look for specific patterns. The definition check reads only the first paragraph. The TL;DR check looks for phrases such as "TL;DR", "In short", "Key takeaways" or "Quick answer" near the top of the page. The date check needs an article:published_time or article:modified_time meta tag, or a <time> element with a datetime attribute; a date written only as plain text does not count. If a check fails on a page you know has the feature, it is usually the markup the check expects that is missing.