Free tool · AI Content Citability Checker
Paste any page or draft and get an instant score for how easily ChatGPT, Perplexity, Gemini and Google AI Overviews can quote and cite it. You'll see exactly what to fix, from direct answers and question-led headings to facts, sources and readability.
The checker
Paste the text of a page, a draft from your CMS or the page's full HTML. Everything runs in your browser and nothing is stored.
Your result
0out of 100
Fixes first, then improvements, then passes.
Not scored, but they help search engines and AI systems trust the page.
Send us the page and your score, and we'll reply personally with the three changes we'd make to get it cited. Free, with no obligation.
The guide
AI answer engines don't cite whole pages. They retrieve passages: a definition, a step, a statistic, a short answer under a heading. The pages that get cited are the ones where those passages are easy to find, easy to lift out and worth trusting.
In practice, citable content has five qualities:
For most pages, the single biggest improvement is moving the answer to the top. Here's the same opening written two ways.
In today's fast-paced digital landscape, businesses are always looking for new ways to stay visible. Search is changing and there are many factors to consider. In this article, we'll explore everything you need to know about answer engine optimisation and why it matters.
Answer engine optimisation (AEO) is the practice of structuring content so AI assistants and search features can quote it as a direct answer. It focuses on clear definitions, question-led headings and short, factual passages.
The first version spends three sentences saying nothing quotable. The second could be lifted straight into an AI answer, with your page cited as the source.
Google says there are no special requirements for appearing in AI Overviews: a page needs to be indexed and eligible to show in Google Search with a snippet. From there, AI Overviews draw on pages that answer the query well, so the qualities this checker measures (direct answers, clear structure, specific and well-sourced information) all apply.
ChatGPT, Perplexity and other AI assistants use their own search indexes and crawlers, but reward the same things. If AI crawlers can't reach your site at all, none of this matters, so check that first with our AI Search Readiness Scorecard.
It can. Neither Google nor the AI assistants rule content out because it was written with AI. What they reward is usefulness, accuracy and originality.
The problem with most AI-written content is that it's generic: it repeats what's already published, so there's no reason to cite it over the original. First-hand experience, original data and a clear point of view are what make any content worth citing, however it was drafted.
The checker runs thirteen checks in three groups. Each one is scored as a pass, a partial pass or a fail, and weighted by how much it affects whether AI can quote your content.
Whether the first paragraph gives a concise answer, roughly 15 to 60 words, and uses the key terms of your target question if you entered one.
How to fix: Rewrite your opening as a one or two sentence answer to the page's main question. Save background and context for after the answer.
Whether the opening paragraphs contain a definitional sentence, such as “X is a…” or “X refers to…”.
How to fix: Add a plain definition of your main topic early on. Definitions are among the passages AI quotes most often.
How many of your H2 to H4 headings are phrased as questions people actually ask.
How to fix: Rephrase some section headings as questions, for example “How much does it cost?” instead of “Pricing”, and answer each one in the first sentence below it.
Whether the content is broken into sections often enough (about one heading every 300 words) and, if you pasted HTML, has a single H1.
How to fix: Add descriptive H2 and H3 headings to break long sections up. If you pasted HTML, make sure the page has exactly one H1.
Average and longest paragraph length. Long paragraphs bury answers and are harder to quote accurately.
How to fix: Split long paragraphs so each makes one point in roughly 40 to 70 words, and make sure each one makes sense when read on its own.
Whether steps, options and comparisons are formatted as lists or tables rather than buried in prose.
How to fix: Turn steps, features, pros and cons or comparisons into bulleted lists, numbered steps or a table.
Whether the content includes a set of questions with short answers, or FAQ schema.
How to fix: Add an FAQ section with three to six real questions your audience asks, each answered in two to four sentences.
How many numbers, percentages, prices, dates and measurable details appear per 500 words.
How to fix: Replace vague claims with specifics: numbers, timeframes, prices, named tools and concrete examples.
Links to external sources if you pasted HTML, or references to studies, data and named sources in the text.
How to fix: Back up key claims by linking to original research, official documentation or data, and name the source in the sentence.
The Flesch Reading Ease score, a standard measure based on sentence length and word length.
How to fix: Use shorter words and sentences where you can. Aim for a Reading Ease of 50 or more for most business content.
The average number of words per sentence.
How to fix: Keep most sentences under 20 words. Long sentences often split naturally at “and”, “but” or “which”.
Common filler phrases that add words without adding meaning, such as “in today's fast-paced world” or “it's important to note”.
How to fix: Delete filler phrases or replace them with a specific point. If a sentence still works without the phrase, cut it.
Total length. Very short pages rarely cover a topic well enough to be the best source.
How to fix: Expand thin pages to cover the questions a reader would ask next. For most topics, 600 words or more gives enough room.
Each check is worth a set number of points, 100 in total. A pass earns all of them, a partial pass (“Improve”) earns half and a fail earns none. Direct answers, question-led headings, paragraph length, lists, facts and sources carry the most weight, because they have the biggest effect on whether a passage can be lifted and trusted.
AI systems would struggle to find a clean, quotable answer here. Start with the biggest wins: lead with a direct answer, add question-led headings and break up long paragraphs.
The building blocks are there, but answers are buried or unsupported. Work through the fixes marked “Fix” first, then the ones marked “Improve”.
Well structured and easy to quote. The remaining gains usually come from more specific facts, named sources and an FAQ.
This content gives AI systems exactly what they look for. Make sure the page is indexed and crawlable, then focus on building the authority that gets it chosen.
Content is AI-citable when AI assistants and search features can easily find a passage in it that answers a question, lift that passage out accurately, and trust it enough to name your page as the source.
No. It doesn't try to tell whether content was written by a person or by AI. It checks how easy your content is for AI systems to quote and cite.
Either works. Pasting the page's HTML, from View Page Source, gives the most accurate result, because the checker can see your real headings, lists, tables, links and schema. Plain text and Markdown work well for drafts.
No. The checker measures how citable the content on the page is. Whether it's actually cited also depends on whether AI crawlers can access your site, how authoritative your brand is, and what competing pages say. Our AI search optimisation work covers all of these.
Long enough to answer the question fully, and no longer. AI cites passages, not word counts, so a focused 800-word page with clear answers often does better than a rambling 3,000-word guide.
They often do, because short question-and-answer pairs match how people prompt AI tools. Keep the answers specific and useful. FAQs padded out only for search engines add little.
Schema helps search engines and AI systems understand what a page is about and who published it. It won't make weak content citable, but it removes ambiguity, which is why the checker reports it as a bonus signal when you paste HTML.
No. The analysis runs entirely in your browser. Nothing you paste is sent or saved. If you request a review, we receive your score, the list of fixes and the details you enter, but never your content.
Keep going
Is your website AI-friendly? Fifteen questions show how likely ChatGPT, Perplexity, Gemini and Google AI Overviews are to recommend your brand, with a prioritised fix list.
What are your buyers asking AI? Generate the prompts to test in ChatGPT, Perplexity and Gemini, with a free sheet to track whether AI mentions you.
Help search engines and AI understand your site. Generate valid JSON-LD for your organisation, location, articles, breadcrumbs and site name, checked as you type.