Free tool · AI visibility
Free AEO readiness score
To score a page for answer engines, check four things in the order an engine meets them: whether it can read the page at all, whether there is a passage short enough to quote, whether the page says who is making the claim, and whether an engine can tell what the page is. This runs 21 such checks worth one hundred points and shows every one of them.
Free. The whole score and every check need no account; comparing against a second page needs a free sign-in.
AEO depth layer
See the Side-by-side comparison
The Side-by-side comparison is computed from your own page and costs nothing to run. We ask for a free account because the expensive tools on this shelf stay free, and because it lets us tell you when your result changes. No card, no trial clock.
- The same twenty-one checks run against a second page you name, usually the one being cited instead of yours, with the gap reported per pillar.
- Only the checks where the two pages actually differ, so the eight rows that decide the outcome are not buried under thirty identical ones.
- Ordered by what each check is worth, so the top of the list is the biggest single move available rather than the loudest warning.
Signing in saves the details you entered above so we can follow up about your results. See our privacy policy for what we keep and why. The AEO Readiness Score result you already have stays free either way. Privacy policy.
The four pillars, and every check inside them
Printed before you enter a URL, because a score you cannot interrogate is not worth having. The pillars are in the order an engine meets them, a warning earns half its weight, and a check that does not apply to your page leaves the arithmetic entirely rather than counting against you.
Readable
25 pointsWhether an engine can get your words at all. Everything else is hypothetical until this passes.
| Check | Points | Threshold | Why an engine cares |
|---|---|---|---|
| The page answers with a 200 | 4 | present or absent | An engine that gets an error or an endless redirect has nothing to read and does not come back soon. |
| Nothing tells crawlers to skip the page | 5 | present or absent | A noindex in the meta tag or the X-Robots-Tag header removes the page from the indexes most assistants are grounded on. |
| There are enough words to answer with | 6 | 300 wordsour judgement, not a published Google limit; Google has said there is no word-count threshold | There is no word count that ranks. There is a floor below which the page contains no passage anyone could quote, and 300 is where we put it. |
| The words are in the HTML, not assembled in the browser | 5 | 1% of the response is visible textour judgement, not a published Google limit | We do not run JavaScript and most crawlers that feed assistants do not either. A page that is mostly markup and a loading state reads as empty to all of them. |
| The page declares its language | 2 | present or absent | One attribute, and it decides which language's index the page belongs in. |
| The whole page fits in one read | 3 | present or absent | A page so large it has to be cut short is a page whose later sections are read by nothing, including us. |
Extractable
30 pointsWhether there is a passage in here that answers something, short enough to be quoted whole.
| Check | Points | Threshold | Why an engine cares |
|---|---|---|---|
| The opening paragraph is short enough to quote whole | 8 | 60 wordsour judgement, not a published Google limit, from the block lengths our own five-engine runs kept quoting | The passage that gets quoted is the one that needs no editing. A long opener has to be cut, and a passage that needs cutting is usually taken from somewhere else. |
| The opening answers what the h1 promised | 5 | 30% shared meaningful wordsour judgement, not a published Google limit | A heading that asks something and an opening that talks about the company leaves nothing self-contained to lift. |
| At least one subheading is shaped like a question | 6 | present or absent | People ask assistants questions. A heading that matches the question is the cheapest way to be the passage that gets quoted, and a page of statement headings matches nothing that was asked. |
| Each question heading is followed by something quotable | 6 | 60 words in the block beneathour judgement, not a published Google limit | Asking the question is half the work. Answering it in the next sixty words is the half most pages skip, and it is the half that gets cited. |
| The typical paragraph is short enough to lift | 5 | median block under 60 wordsour judgement, not a published Google limit | When every block is long, an engine has to choose the boundary itself, and a passage it cut is one it is less confident quoting. |
Attributable
22 pointsWhether the page says who is making the claim, when, and with what specificity.
| Check | Points | Threshold | Why an engine cares |
|---|---|---|---|
| The markup names an author | 5 | present or absent | A model asked who said this reads the markup, not the byline you styled into the header. |
| The markup names a publisher | 5 | present or absent | Without it a claim from this page is attributed to a bare domain, or to nobody, which is the difference between being cited and being used. |
| The entity is linked to itself elsewhere | 4 | present or absent | sameAs is how a model connects the name on your page to the entity it already knows about. Without it your brand is a string. |
| The page says when it was written | 4 | present or absent | On any question where currency matters, a page with no readable date loses to one that has a date, without being worse. |
| Some sentences carry a specific number | 4 | present or absent | A page of general statements gives a model nothing it could not have written itself, which is the most common reason a page is read and not cited. |
Findable
23 pointsWhether an engine can tell what this page is and get from it to the rest of your site.
| Check | Points | Threshold | Why an engine cares |
|---|---|---|---|
| A title that describes the page | 5 | at least 3 meaningful wordsour judgement, not a published Google limit | It is the label every citation, result and share carries, and a template default labels every page of yours the same. |
| A meta description | 4 | present or absent | Without one the engine picks its own extract, which means the sentence representing your page is one you did not write. |
| A canonical pointing at this page | 4 | present or absent | A missing or inherited canonical splits the page's standing across every variant of its address, or hands it to a different page entirely. |
| Structured data describing this page | 6 | present or absent | Organization and Breadcrumb describe the site. Something has to describe the page, or the markup says nothing about what you want cited. |
| Links onward into your own site | 4 | 3 internal linksour judgement, not a published Google limit | A page nothing links out from is a dead end for a crawler, and a page nothing links to is reached late and rarely. |
What this measures, and what it cannot
The line matters more here than on any other tool on this shelf, because the whole category is full of tools that blur it.
| Question | Answered here? | Where it is answered |
|---|---|---|
| Could an engine read this page at all? | Yes, measured | This tool, the Readable pillar |
| Is there a passage short enough to quote? | Yes, measured | This tool, the Extractable pillar |
| Does the page say who is making the claim? | Yes, from the markup | This tool, the Attributable pillar |
| Is my page better prepared than a rival's? | Yes, with a free account | The side-by-side comparison |
| Does ChatGPT actually cite my page? | No | Ask the engines: the AI visibility checker |
| Which questions do my buyers ask? | No | Answer Radar |
| Am I allowed to be crawled by AI bots? | No | The AI crawlability checker |
This is a structural read of your own page. It tells you whether an engine could answer with it, which is the part you control. For what the engines actually say, the AI visibility checker asks them, the AI crawlability checker confirms they are allowed to read you, and the on-page SEO checker covers the search-index half of the same page.
How to score a page for answer engines
Step 1
Enter the page you want an assistant to cite
Not your homepage. Pick the page that answers a specific question a buyer would ask, because that is the page an engine has a reason to quote.
Step 2
Read Readable first, then Extractable
They are in the order an engine meets them. A page that fails Readable has no Extractable score worth reading, because nothing got that far.
Step 3
Fix the opening paragraph before anything else
It is the single heaviest check here at eight points, and it is usually the cheapest to change. An answer in the first sixty words is the passage that gets lifted.
Step 4
Then compare against the page being cited instead
Sign in and enter a second address. The same twenty-one checks run against both, and the result is the specific list of things that page does and yours does not.
When you outgrow this tool
A readiness score tells you whether one page could be cited. Answer Radar tells you which questions your buyers actually ask, whether you are named in the answer, and which competitor is named instead.
Scoring one page is free and complete on its own. Linkeddit is what you use when you need to know which questions your buyers ask, and which competitor gets named in the answer.
FAQ
AEO Readiness Score questions
What the four pillars measure, how this differs from an on-page audit, and the question a tool like this cannot answer.
What is an AEO readiness score?
It is a measure of whether an answer engine could use your page, as opposed to whether a search index would rank it. Four pillars, in the order an engine meets them: Readable, worth 25 points, whether it can get your words at all; Extractable, worth 30, whether there is a passage in there short enough to quote; Attributable, worth 22, whether the page says who is making the claim; and Findable, worth 23, whether an engine can tell what the page is. The full catalogue is printed below.
How is this different from your on-page SEO checker?
The on-page checker asks whether a page is well built for a search index: title, meta, canonical, links, images. This asks whether an assistant could answer a question with it. They overlap, and they disagree more often than you would expect. A page can pass thirty-five on-page checks and still contain no self-contained passage anyone could quote, which is the failure this tool exists to find.
Why is the opening paragraph worth eight points?
Because it is the passage most often lifted, and because a long one has to be cut. An engine quoting a page prefers the passage that needs no editing, and a passage it had to cut is one it is less confident attributing. Sixty words is our threshold, taken from the block lengths our own five-engine runs kept quoting, not from anything anyone has published.
Can you tell me whether ChatGPT actually cites my page?
No, and no tool that reads your HTML can. This is a structural read of your own page: it measures whether an engine could extract an answer, not whether one did. Those are different questions and conflating them is the most common dishonesty in this category. For what the engines actually say when someone asks about your category, you need to ask them, which is what the AI visibility checker and Answer Radar do.
My page is a React app and scored badly on Readable. Is that wrong?
No, it is the finding. We read the HTML your server returns and do not execute JavaScript, and most of the crawlers that feed assistants behave the same way. If less than one percent of your response is visible text, the page reads as a loading state to all of them. That is a real problem with a real fix, server rendering or pre-rendering, not an artefact of our tool.
What is in the layer that needs an account?
A side-by-side comparison. You give it a second address, usually the page being cited instead of yours, and the same twenty-one checks run against both. It reports the gap per pillar and the specific checks where the two pages differ, ordered by what each one is worth. The free score is complete on its own: signing in adds a second page, it does not unlock part of the first.
Why is the whole score free when other tools gate the breakdown?
Because gating a breakdown is charging someone to see how a number they were already shown was calculated, and a score you cannot audit is not worth having. Every weight and threshold on this page is published before you enter a URL, and the arithmetic is shown beside the result. The thing behind the sign-in is an additional page, which is a genuine addition rather than a piece removed.
More free tools
Every tool on the shelf is free, with no card and no trial. Most need no account at all. See all free tools.
Build the JSON-LD properties that make a page attributable rather than eligible for a rich result, with the reason each one matters.
Find out what is stopping a page being indexed, which is the half of the question anything can actually answer.
See the answer an engine can build from your page, with every sentence labelled by the element it came from.
Whether each answer engine can reach and quote one specific page, checked per engine against the bots it actually uses.