Skip to content
AI Search (GEO)

AI Citability Score

Scores your page against the traits shared by pages that get cited in AI answers: clear question headings, short self-contained paragraphs, lists, tables, numeric data and outbound sources.

  • Sign-in requiredMembership is free
  • Open on the Pro plan100 runs a day
  • Saved to your historyReachable from your account

Aracı çalıştırın

1 zorunlu alan · sonuç birkaç saniyede hazır

Pro pakette · günde 100 analiz

Üyelik ücretsizdir ve yarım dakika sürer. Formu şimdiden doldurabilirsiniz; giriş yaptığınızda bu sayfadan ayrılmazsınız, yazdıklarınız yerinde kalır.

What does this tool look at?

Can ChatGPT, Perplexity and Google AI Overviews read you and cite you? Readiness checks for the new generation of search.

  • Use of question-shaped headings
  • Share of paragraphs short enough to quote
  • List, table and numeric data density
  • Author, date and source signals (E-E-A-T)

What this tool measures

AI answers do not read a page end to end and summarise it. They lift passages that stand on their own. A well-written page that flows as one long block therefore gets quoted less than a weaker page built from separable parts.

The tool scores your page on nine measures: question-shaped headings, the share of paragraphs between 20 and 90 words, the number of lists and tables, density of numeric statements, an author signal, a date signal, the presence of JSON-LD, outbound citations and total word count.

The thresholds come from structural traits shared by pages that AI answers cite often. Two question headings pass and one warns; at least 40 percent of paragraphs should sit in the 20 to 90 word band; five numeric statements pass and two warn; 600 words pass, 300 is the warning line.

How to use it

  1. 1

    Submit a single page

    The measurement works per page, not per site. Start with your most visited or most important guide; homepages naturally score low against these criteria.

  2. 2

    Read the checks, not the score

    The total is a quick signal. The information lives in which measure failed: each of the nine points at a different fix, and you do not need all of them at once.

  3. 3

    Turn headings into questions

    Write “How much does a website cost?” instead of “Pricing”. The check looks for a question mark or an opening interrogative in the heading.

  4. 4

    Put a direct answer under each heading

    The short paragraph right after a heading is the text most likely to be quoted. Answer first, save the context and nuance for the paragraphs after it.

  5. 5

    Anchor claims in numbers

    The numeric counter recognises figures paired with units: percentages, currency, counts, people, days, hours, years. “Fast delivery” does not count, “six working days on average” does.

When to use it

  • When asking ChatGPT a question in your field returns competitors as sources and never you.
  • Before publishing a long guide, to review its structure while editing is still cheap.
  • When a page ranks well in classic search yet never appears in the AI Overviews box.
  • While setting a writing standard for a content team that needs measurable targets.

Reading the result

  • Low quotable paragraph ratio

    If fewer than 40 percent of paragraphs sit between 20 and 90 words, the text is either long blocks or one-line fragments. Split blocks at one idea each and merge the fragments.

  • No lists or tables

    Two lists or tables pass the check. Turning a procedure into a numbered list and a comparison into a table lifts readability and quotability at the same time.

  • Author or date signal missing

    The tool looks for an author meta tag, an author field in JSON-LD, article:published_time, or a time element carrying datetime. Printing the date in visible text alone is not enough; it must be marked up.

  • No outbound citations

    Two external links pass. You do not have to link to a competitor: official institutions, standards documents and research sources make claims verifiable and satisfy the check.

  • Thin content

    Below 600 words warns and below 300 fails. This is not an instruction to pad. A page that genuinely covers its topic reaches that range on its own; if it does not, the scope is too narrow.

FAQ

Questions about the AI Citability Score

Didn't find your answer?

Send us your question and we'll reply the same day.

Ask us

No. Neither Google nor OpenAI publishes such a score. The criteria were collected from structural patterns that recur in frequently cited pages. The aim is to acquire those traits, not to raise a number.

No. Structure is necessary, not sufficient: the page also has to be relevant, reachable and not blocked to AI crawlers. Use the AI Crawler Access Checker for that second condition.

They are, when they answer a narrow question completely. Definition pages and single-question help articles get cited often despite being short. A low score matters when the page's purpose is broad.

It is not. The audit reads the HTML source. If your content assembles in the browser the score drops, and that reflects reality: most AI crawlers also read without executing JavaScript.

The tool found the problem. Now for the fix

We run technical SEO, content and AI visibility work end to end. Let us review your site and come back with a clear list of what to do.

Monday – Saturday 09:00 – 18:00