How the AI Search Visibility Checker Works: The 15 Checks Behind the Score
The AI Search Visibility Checker reads one page, runs 15 checks on it and gives it a score from 0 to 100. The score tells you whether AI can reach the page, pull a clean answer out of it and tell who published it.
It does not tell you whether ChatGPT or Perplexity will cite you, because nobody can see that. It checks the part you control, which is the first part of AI search visibility: whether the page can be retrieved and quoted at all.
The AI Search Visibility Checker runs 15 weighted checks on one page and gives a score from 0 to 100. Here is every check, what it is worth and what a low score means.
What the AI Search Visibility Checker Looks At
You paste in one URL. The checker fetches that page the way an AI retrieval tool would, reading the raw HTML, the robots.txt file and the llms.txt file. It does not look anything up in a ranking database.
That matters because every result is something you can change. A low score is never a mystery. It is a short list of things on your own page.
The 15 Checks in the AI Search Visibility Checker
Each check is pass or fail. A pass earns the full points for that check, and a fail earns none. All 15 add up to 112 points.
| # | Layer | Check | Points | What it asks |
|---|---|---|---|---|
| 1 | Retrievability | AI crawlers are allowed | 12 | Does robots.txt let the major AI crawlers in, or does it turn them away? |
| 2 | Retrievability | Content is in the HTML | 12 | Do at least 250 words arrive in the raw page, before any script runs? |
| 3 | Retrievability | Canonical URL declared | 6 | Does the page say which URL is the real one? |
| 4 | Extractability | Exactly one H1 | 8 | Is there one clear heading that names the page? |
| 5 | Extractability | Usable title | 6 | Is the title between 15 and 70 characters? |
| 6 | Extractability | Meta description | 5 | Is there a description between 50 and 200 characters? |
| 7 | Extractability | Answers before it explains | 10 | Is the first paragraph after the H1 a full answer of at least 25 words? |
| 8 | Extractability | Sectioned with H2 headings | 5 | Are there at least two H2 headings? |
| 9 | Entity clarity | Structured data present | 12 | Is there any JSON-LD on the page? |
| 10 | Entity clarity | Publisher identified | 10 | Does the structured data name an Organization or a Person? |
| 11 | Entity clarity | Linked to outside references | 8 | Does the publisher point to profiles elsewhere with sameAs links? |
| 12 | Entity clarity | Page type declared | 6 | Does the page say what it is, such as an Article, FAQPage or DefinedTerm? |
| 13 | Machine-readable signals | llms.txt published | 5 | Is there an llms.txt file at the site root? |
| 14 | Machine-readable signals | Open Graph metadata | 3 | Does the page repeat its title and summary in Open Graph tags? |
| 15 | Machine-readable signals | Question and answer markup | 4 | Are real questions and answers marked up as FAQPage or QAPage? |
Three checks are worth 12 points each: AI crawlers allowed, content in the HTML and structured data present. Those three decide most of the score.
How the AI Search Visibility Checker Turns 15 Checks Into One Score
The score is the points a page earned, divided by 112, times 100, rounded to a whole number. A page that passes everything scores 100.
Here is a real one. Our own glossary page failed three checks: answers before it explains (10 points), sectioned with H2 headings (5) and question and answer markup (4). That is 19 points lost, so it earned 93 of 112. Divide and round, and the score is 83.
| Score | Verdict | What it means |
|---|---|---|
| 85 to 100 | Strong | The page states what it is, who published it and what it answers, so AI can use it without guessing. |
| 65 to 84 | Workable | The basics are in place, but AI still has to infer things the page should have said outright. |
| 40 to 64 | Weak | Enough is missing that AI is rebuilding the meaning from the text alone, which is where mistakes start. |
| 0 to 39 | Poor | AI has very little to work with beyond raw text. |
The Four Layers of the AI Search Visibility Checker
The 15 checks sit in four layers, and the layers run in order. Each one only matters if the one before it works.
- Retrievability, 30 points. Can AI get to the page at all? This layer holds the crawler check, the content-in-HTML check and the canonical check. If robots.txt turns AI crawlers away, nothing else on the list matters.
- Extractability, 34 points. Once AI is on the page, can it lift a clean answer out? This layer looks at the H1, the title, the description, the first paragraph and the H2 headings.
- Entity clarity, 36 points. Can AI tell who published the page and what kind of page it is? This is the biggest layer, because an answer engine deciding whether to trust a claim needs to know whose claim it is.
- Machine-readable signals, 12 points. These are backup statements of the same facts. They help when something else goes wrong, so they carry the fewest points.
Which Part of AI Search Visibility the Checker Covers
AI search visibility has three parts: the page is retrieved, the page is cited and the page is accurately represented. The checker works on the first part directly. If crawlers are blocked or the content is not in the HTML, the page cannot be retrieved, and nothing else you do will fix that.
For the other two parts, the checker tests whether the page is built so they can happen. A clear first answer and a named publisher make a page easier to cite. Structured data that matches the visible text makes it harder to describe wrongly.
What the checker cannot see is the outcome. For that you ask the AI tools directly, and the guide to how to check AI search visibility walks through it.
What the AI Search Visibility Checker Does Not Tell You
- Whether ChatGPT, Perplexity or Google AI Overviews will cite the page. That depends on the question, the tool and the competing pages that day.
- Where the page ranks. For rankings and clicks, use the numbers in What Is Search Visibility? and see what a good search visibility score looks like.
- Whether the answers AI gives about you are true. Only a person reading those answers can judge that.
- How the rest of the site is doing. The checker scores one page at a time, so check your most important pages and templates.
We Ran the AI Search Visibility Checker on Our Own Glossary Page
The glossary definition of AI search visibility is the strongest page on this site, and it scored 83. Workable, with gaps. I expected to find content gaps.
The gaps were not in the content.
The first paragraph after the H1 was a one-line note listing other names for the term, so the checker never found an answer there. The page had fewer than two H2 headings. And the glossary template had no way to show or mark up an FAQ at all.
All three were template problems, not writing problems. So the fix went into the template: the other-names line stopped being a paragraph, the template learned to print an FAQ and mark it up, and the page itself was rewritten into clear sections. I ran the checker on the live page afterward, and it scored 100, a Strong rating, up from 83.
The lesson is the useful part. One template change moves every page built on that template, which is why a low score on one page is worth following back to its cause.
What to Fix First When the Checker Score Is Low
Fix the earliest layer that failed. A page that blocks AI crawlers has a retrieval problem, and better headings will not help it. A page that lets crawlers in but buries its answer has an extraction problem. Work down the layers in order, then run the checker again and watch the number move.
Check a page now
Paste in any URL and get the score, every pass and fail, and the fix for each failure. It is free.
Open the AI Search Visibility CheckerWhere the Checker Fits Among the AI Symantix Tools
The checker is the free, instant, one-page tool. The AI Readiness Score looks wider, across five areas of a site. AI bot data shows which AI crawlers actually came and what they fetched. The Search Visibility Index covers the search half. And the AI Symantix Audit puts a person on your whole site for a written review.
Frequently Asked Questions
How does the AI Search Visibility Checker work?
You enter a URL and the checker fetches the page the way an AI retrieval tool would. It runs 15 pass or fail checks across four layers, adds up the points the page earned out of 112 and shows a score from 0 to 100 with the fix for every failure.
What is a good score on the AI Search Visibility Checker?
A score of 85 or more is Strong, 65 to 84 is Workable, 40 to 64 is Weak and below 40 is Poor. Strong means the page states what it is, who published it and what it answers, so AI can use it without guessing.
Which checks count the most in the AI Search Visibility Checker?
Three checks are worth 12 points each: whether AI crawlers are allowed in robots.txt, whether the content is in the HTML and whether structured data is present. Answers before it explains is next at 10 points, tied with naming the publisher.
Does a high score mean AI will cite my page?
No. A high score means the page is built so AI can reach it, quote it and tell who published it. Whether an AI tool cites it depends on the question, the tool and the competing pages at that moment, and no checker can see that.
Why did my page score low when it ranks well?
Rankings reward links, relevance and engagement. The checker rewards being explicit: a clear first answer, a named publisher and structured data that matches the page. Many high-ranking pages have no structured data at all, which is the gap the checker shows.