Free AEO Checker
Score any page against 18 answer engine checks and see whether it is built to be quoted by ChatGPT, Perplexity, Google AI Overviews and Copilot. No email required.
Check any page for free
Enter a URL. The full score, every category and every failed check appear on this page, with no email required.
What is AEO?
AEO (Answer Engine Optimisation) is the practice of structuring content so AI systems can confidently cite and summarise it.
As more users turn to AI for answers, being cited becomes as important as ranking in traditional search.
Entity Optimisation
Help AI systems recognise and understand your business entity.
Structured Data
Implement schema markup that AI can parse and trust.
Direct Answers
Structure content to answer common questions clearly.
Authority Building
Strengthen trust signals that AI systems rely on.
What the AEO checker measures
The iDigitGroup AEO checker is a free answer engine optimisation tool. It fetches one URL, reads the HTML your server returns, and scores it against 18 checks in four categories: entity recognition, structured data, content quality for AI, and authority signals. The result is a score out of 100 and a list of the checks that failed.
Answer engine optimisation, or AEO, is the work of making a page easy for an AI system to read, trust and quote. Search engines rank pages. Answer engines such as ChatGPT, Perplexity, Google AI Overviews and Copilot assemble an answer and cite a handful of sources. A page can rank respectably and still never be quoted, because the passage a model needs is buried in prose, the business behind it is not declared as an entity, and nothing on the page says who wrote it or when.
This page documents every check the tool runs, what each one looks for in your HTML, how to fix it, and exactly how the four category scores combine into the number you see. Nothing here is withheld. You can score your own page by hand from this page alone, and the tool will agree with you.
The four categories and all 18 checks
Each category is scored out of 100 from its own checks, then the four are combined by weight. Points are shown as pass points, then the partial credit a warning earns where partial credit exists.
Entity recognition
25 per cent of the final score
Whether an AI system can work out which real organisation this page belongs to. Answer engines resolve pages to entities before they decide who to cite, so a page with no declared entity is a page with nothing to attribute an answer to.
Organisation schema
25 points, no partial credit
Any JSON-LD block on the page typed Organization, Corporation or LocalBusiness.
Publish one Organization block sitewide carrying name, url, logo and sameAs links to the profiles you actually control. One canonical block beats several competing ones.
LocalBusiness schema
25 points, or 10 if absent
A LocalBusiness type, or any schema type whose name contains Business.
If you serve a place, add LocalBusiness with the real address, telephone, opening hours and geo coordinates. Never invent a location you do not trade from.
Brand consistency
25 points, or 10 for a partial match
The title tag against the first H1. It passes when either string contains the first 20 characters of the other.
Make the title and the H1 name the same thing. Fighting for two different keywords in those two fields costs you the check and confuses the entity signal for no gain.
WebSite schema
25 points, or 5 if absent
A JSON-LD block typed WebSite.
Add a WebSite block with name, url and a publisher reference pointing at your Organization node, so the two are linked rather than floating separately.
Structured data
30 per cent of the final score, the heaviest category
Whether the facts on the page are readable by machine as well as by eye. This category carries the most weight because schema is the cheapest and most reliable way to hand an answer engine something quotable with attribution attached.
FAQ schema
25 points, or 10 if question headings exist without the schema
A FAQPage or Question type. Partial credit applies when the page carries three or more question style headings but no matching schema.
Mark up the questions the page genuinely answers. Never mark up a question the visible page does not answer, and never mark up an FAQ that is not visible to a reader.
HowTo schema
20 points, or 5 where lists are present without it
A HowTo type. Partial credit applies when the page carries three or more lists.
Where the page teaches a real sequence, add HowTo with a step for each visible step. Where it does not teach a sequence, leave this one failing rather than faking it.
Service schema
20 points, no partial credit
A Service or Product type.
One Service block per service page, with name, description, areaServed and a provider reference back to your Organization node.
BreadcrumbList schema
15 points, or 5 if absent
A BreadcrumbList type.
Emit a BreadcrumbList that matches the breadcrumb a reader can see. It tells an answer engine where the page sits, which matters when it has to choose between two pages of yours.
Article schema
20 points, or 5 if absent
An Article, BlogPosting or NewsArticle type.
On editorial pages add headline, author, datePublished, dateModified and publisher. Update dateModified only when the content actually changed.
Content quality for AI
25 per cent of the final score
Whether there is a passage on the page worth lifting. Models quote spans, not whole pages, so a page wins here by answering a specific question in a specific place rather than by being long.
Direct answers
25 points at five or more, 15 at two to four, 5 below that
H2 and H3 headings that read as questions, either because they contain a question mark or because they open with an interrogative such as what, how, why, when, where, which or who.
Rewrite two or three headings as the question a customer would type, then answer each one in the first sentence or two beneath it. Everything else in the section can follow the answer.
Content depth
25 points at 1,500 words or more, 15 from 500, 5 below that
Visible word count after scripts, styles, noscript blocks, SVG and iframes are stripped out.
Add sections that answer questions the page currently leaves open. Padding a thin page to hit a number does nothing, because the check rewards depth only as a proxy for coverage.
Structured content
25 points at five or more, 15 at two to four, 5 below that
The combined count of lists and tables in the HTML.
Turn a prose comparison into a table and a described sequence into an ordered list. Both are far easier for a model to extract cleanly than the same content as a paragraph.
Heading structure
25 points for a clean hierarchy, 15 for a partial one, 5 otherwise
Exactly one H1 together with three or more H2s passes. Two or more H2s without that clean H1 earns partial credit.
One H1 per page, H2 for each real section, H3 for subsections, and no skipped levels. Never use a heading tag for visual size alone.
Authority signals
20 per cent of the final score
Whether a model has any reason to trust the passage it just found. These are the E-E-A-T mechanics: a named human, an honest date, sources, and complete metadata.
Expert authorship
25 points, no partial credit
A rel=author link, an element with the author class, an itemprop of author, or an author property inside your schema.
Name the person who wrote it, give their credentials, and connect them with Person schema. This is the single most common zero on the whole checker.
Content freshness
20 points, or 5 if absent
A time element with a datetime attribute, an itemprop of datePublished, or datePublished inside your schema.
Publish honest dates and expose them in markup, not only as text. A date you refresh without touching the content is worse than no date at all.
External citations
25 points at three or more, 10 where other external links exist
Outbound links to .edu or .gov domains, Wikipedia or Google Scholar.
Cite a primary source wherever the page makes a claim that needs one. This check is deliberately narrow, so see the limitations below before you treat a low mark here as a verdict.
Internal linking
15 points at five or more, 8 at two to four
Relative links on the page, meaning links that begin with a slash or a dot.
Link to the other pages that cover the same subject. A cluster tells an answer engine which site owns a topic, and it is the cheapest of all these fixes.
Meta completeness
15 points for all three, 8 for a canonical or Open Graph alone
A canonical link, an og:title, and a twitter:card meta tag.
Put all three on every indexable page. Missing metadata rarely sinks a page on its own, but it is the sort of gap that appears alongside every other one.
How the AEO score is calculated
Each category is scored out of 100 from the points listed above, then the four category scores are combined as a weighted average and rounded to the nearest whole number. Nothing else is added, and no check is worth more than the points printed against it.
| Category | Weight | What it decides |
|---|---|---|
| Entity recognition | 25 per cent | Whether an answer engine can tell whose page this is |
| Structured data | 30 per cent | Whether the facts are machine readable and attributable |
| Content quality for AI | 25 per cent | Whether there is a passage worth quoting |
| Authority signals | 20 per cent | Whether the passage is worth trusting |
What the bands mean
AI-Ready
70 to 100
The page declares its entity, carries the schema that matches what it is, and answers questions in a form a model can lift. Remaining work is refinement.
Needs improvement
45 to 69
The page is readable but incomplete. In most cases one category is carrying the loss, and the structured data category is the fastest to move because it is weighted at 30 per cent.
Not AI-Ready
Below 45
Two or more categories are failing. Start with entity recognition, because the rest of the work has nothing to attach itself to until an answer engine knows who you are.
A URL the checker cannot fetch within 15 seconds returns a score of zero with a single message. That is a fetch failure, not a judgement on the page.
A worked example
Here is the arithmetic in full, on an invented service page, so you can follow the same steps on your own. The page carries Organization and WebSite schema but no LocalBusiness, and its title does not match its H1.
Entity recognition: 70
25 for Organisation schema, 10 for the missing LocalBusiness, 10 for the title and H1 mismatch, 25 for WebSite schema.
Structured data: 45
25 for FAQ schema, 0 for no HowTo, 0 for no Service schema, 15 for BreadcrumbList, 5 for no Article schema.
Content quality for AI: 100
1,700 words, six question style headings, four lists and one table, one H1 with five H2s. All four checks pass.
Authority signals: 63
25 for a named author, 5 for no marked up date, 10 for external links that are not .edu or .gov, 15 for nine internal links, 8 for a canonical and Open Graph without a Twitter card.
Weighted: (70 x 0.25) + (45 x 0.30) + (100 x 0.25) + (63 x 0.20) = 17.5 + 13.5 + 25 + 12.6 = 68.6, which rounds to 69 and lands one point inside Needs improvement. The cheapest way up is not more words, since content already scores 100. It is Service and HowTo schema, worth 40 points inside the heaviest category, which lifts structured data to 85 and the overall score to 81.
AEO checks against classic SEO checks
Most of these fields are familiar. What changes is what the field is being asked to do once an answer engine, rather than a results page, is the destination.
| What is checked | The classic SEO view | The AEO view used here |
|---|---|---|
| Title tag | Unique, keyword led, comfortably under 60 characters | Names the same entity as the H1, so the page can be attached to a known organisation |
| Word count | Enough depth to cover the search intent | Enough self contained detail that a passage can be quoted without the rest of the page |
| Headings | A keyword hierarchy for crawlers | Questions phrased the way people ask them, answered in the sentence directly beneath |
| Schema | Eligibility for rich results | Machine readable facts a model can lift with attribution intact |
| Outbound links | Risk to be managed and sometimes nofollowed | Citations that let a model verify a claim before repeating it |
| Internal links | Crawl paths and link equity | A topical cluster showing which site owns the subject |
| Author and date | A freshness signal | The evidence that decides whether an answer engine trusts the passage at all |
How to improve your AEO score
In the order that moves the number fastest on most sites. The first three are usually an afternoon of work.
- 1
Declare the entity once, sitewide
Add Organization and WebSite schema in the site template, and LocalBusiness too if you trade from a real address. Every other signal on every page attaches to this one.
- 2
Make the title and the H1 agree
They should name the same thing. This is a one line change per page and it settles the brand consistency check immediately.
- 3
Add the schema type that matches the page
Service on service pages, Article on editorial pages, HowTo wherever you genuinely teach a sequence. Structured data is 30 per cent of the score and most sites leave three of its five checks unclaimed.
- 4
Rewrite three headings as questions and answer them immediately
Use the wording customers use. Put the answer in the first sentence or two beneath the heading, then expand. That single habit feeds the direct answers check and gives models something to quote.
- 5
Mark up the FAQ you already answer
FAQPage schema over questions that are visible and genuinely answered on the page. Do not create an FAQ block purely for the markup.
- 6
Convert one comparison and one sequence
One prose comparison becomes a table, one described process becomes an ordered list. Two edits usually clear the structured content check.
- 7
Put a named author and an honest date on editorial pages
A real person, their credentials, Person schema, and a datePublished you do not touch unless the content changed. Authorship is worth 25 points in the authority category and is the most common zero we see.
What this checker does not do
Worth knowing before you act on the number, because a check that is deliberately narrow can read as a failure when it is not.
- It reads the HTML your server returns for one URL. Content that only appears after JavaScript runs in the browser is not counted. That is deliberate, because it is close to what most answer engine crawlers see.
- The citation check recognises .edu and .gov domains, Wikipedia and Google Scholar. A regulator, a trade journal or a manufacturer's documentation will not be counted, so a well sourced page can score lower than it deserves on that one check.
- It does not ask ChatGPT, Perplexity or Google AI Overviews whether they currently cite you. It measures whether the page is built to be citable, which is the part you control.
- It scores a page, not a site. Run it on your homepage, your strongest service page and one article, because the three usually fail in different places.
Keep going
The rest of the AEO material on this site, in the order most people need it.
How to run an AEO audit
The full manual method behind this tool, step by step.
AI search optimisation
What we do when a site needs the work done rather than diagnosed.
AI visibility tracking
Track whether AI assistants are actually mentioning you over time.
Free SEO audit tool
The classic technical and on-page side, checked separately.
SEO and AEO glossary
Plain definitions for the terms used on this page.
AEO checker FAQs
How the checker works, what it looks at, and what the score means.
An AEO checker is a tool that analyses whether a page is built to be quoted by AI answer engines such as ChatGPT, Perplexity, Google AI Overviews and Copilot. It looks at the same HTML those crawlers see and reports what is missing.
18 checks in four categories: entity recognition (Organisation, LocalBusiness and WebSite schema, plus title and H1 consistency), structured data (FAQ, HowTo, Service, BreadcrumbList and Article schema), content quality for AI (question style headings, visible word count, lists and tables, heading hierarchy) and authority signals (author, publication date, external citations, internal links, metadata).
Each category is scored out of 100 from its own checks, then combined as a weighted average: entity recognition 25 per cent, structured data 30 per cent, content quality 25 per cent and authority signals 20 per cent. The total is rounded to the nearest whole number.
70 and above is AI-Ready, 45 to 69 needs improvement, and below 45 is not AI-Ready. Most business sites we check land in the middle band, and structured data is usually the category carrying the loss.
No. Enter a URL and the full score, every category and every failed check appear on the page. Your name and email are only needed if you want the written report emailed to you afterwards.
SEO works towards ranking a page in a results list. AEO works towards being quoted inside an answer that cites a handful of sources. The fields overlap, but AEO asks the title to identify an entity, the headings to carry questions, and the page to make its author and sources explicit.
A few seconds in most cases. The checker fetches the URL with a 15 second timeout and scores the HTML it receives. If the page cannot be fetched in that window the result comes back as zero, which is a fetch failure rather than a judgement on the page.
After any structural change, and otherwise once a quarter. The checks measure how a page is built rather than how it performs on a given day, so the score only moves when the page does.
Want to improve your AEO score?
Our AEO specialists can optimise content, schema, and entity signals to improve AI visibility.
Get AEO Consultation