Skip to content
All AI readability rules
WarningExtractability4 of 30 ptsAnswer Format

How to fix: No explicit definitional sentences

A definitional sentence is a plain sentence that states what a term is, in the form X is a Y that Z — the sentence you are reading is one. AccessKnight scores these as aeo.definitions, worth 4 of the 30 Extractability points, because a definitional sentence is what an answer engine lifts when a reader asks what something means.

The check

What is this?

This check asks whether the page states, in plain sentences, what its key terms are. AccessKnight sends the page's content sample to a model with one instruction for this criterion: award 0 to 4 points for explicit definitional sentences of the form X is a Y, covering the terms the page is actually about. Implied definitions earn nothing.

The highest-value place for one is the first paragraph after the h1. That sentence carries disproportionate weight when any system summarises the page, and it is the first thing a reader landing cold from a search result or an AI answer needs. The pattern is dull on purpose: name the entity, name its category, then name what distinguishes it. "AccessKnight is a WCAG audit tool that scans a site and returns a remediation report" beats any amount of atmosphere.

What fails is a page that uses a term forty times and never says what it is. Close behind are the substitutes that feel like definitions and are not: defining by example ("think of it like a spellchecker for your markup"), defining by contrast ("unlike the other tools, we…"), and defining by benefit ("the fastest way to find out what is wrong"). Each assumes the reader already knows the category. The grader is looking for the sentence that supplies it.

The impact

Who does this affect, and why does it matter?

Answer engines are affected first, for the most literal reason of any check in this pillar. "What is X" is one of the commonest shapes a question takes, and ChatGPT search, Copilot, Perplexity, Google AI Overviews and AI Mode, and Claude answering that shape are looking for a sentence stating what X is. A page without one supplies no candidate.

Readers gain from the same sentence, and some readers depend on it. Anyone arriving mid-topic from a search result, a shared link, or an AI answer has none of the context your homepage would have given them. Non-native speakers and readers with cognitive disabilities benefit specifically from terms being defined at first use — a WCAG AAA-level practice under Unusual Words, above the AA floor, and cheap to adopt.

Extractability is the 30-point pillar that measures how easily an engine can lift a clean answer from a page, and this is its smallest check at 4 points — but the threshold is tight. The row passes only at 3 or 4, warns at 2, and fails below 2. Partial credit reads as a warn, so a page with one definition amid a dozen undefined terms usually lands at 2.

The mechanic is straightforward. A definitional query has an obvious extractable unit, and if the page does not contain that unit there is nothing to lift no matter how thoroughly the rest of the page covers the subject. Writing the sentence is a precondition for being usable as a definitional source; it does not promise that any engine will use the page.

This is the smallest check with the largest transfer, which is why it is worth more attention than four points suggests. The same sentence works as your meta description, as the subhead under your h1, as the opening line of a pitch, and as the answer when someone asks what you do. Most pages that fail this check fail it because nobody ever wrote that sentence down, not because writing it was hard.

The fix

How do I fix it?

Write one sentence saying what the page's subject is, in the form X is a Y that Z, and place it as the first paragraph after the h1 as ordinary text. Then define each remaining term at its first use. A definition set inside a hero graphic or an SVG is never read.

  1. List the terms this page assumes the reader already knows — the product name, the category it belongs to, and any jargon sitting in the h1 or the section headings.
  2. Write one sentence for the page's main term in the form "X is a Y that Z", and place it as the first paragraph after the h1, ahead of any narrative or brand copy.
  3. Keep that sentence in the document as ordinary paragraph text. A definition living inside a hero image, an SVG, or a canvas is not read, because svg elements are stripped before grading and image content is never parsed.
  4. Define each remaining term in one sentence at its first use in the body, rather than letting the surrounding context imply it.
  5. Replace definitions by example, by contrast, or by benefit with a plain statement of what the thing is — the grader is looking for an explicit "is a" statement, not a comparison.
  6. Re-scan with "Also grade AI extractability" ticked and confirm the "Explicit definitional sentences" row reports 3 or 4 of 4.
Article body (HTML) — Before
<h1>Machine-readability scanning</h1>
<p>We built this because too many sites are invisible to the
  tools people now use to find things. Our scanner has been
  helping teams since 2025, it takes about 30 seconds, and it
  gives you something your developers can actually action.</p>
Article body (HTML) — After
<h1>Machine-readability scanning</h1>
<p>Machine-readability scanning is an automated check of how
  easily an AI crawler can reach, parse, and quote a web page.
  It reads robots.txt, the server-rendered HTML, and the page
  structure, then scores each of the three out of 100.</p>
<p>We built this because too many sites are invisible to the
  tools people now use to find things. The scan takes about 30
  seconds and returns something your developers can action.</p>
Detection

How does AccessKnight detect this?

This check is graded by a model, not matched by a rule — there is no regular expression hunting for the words "is a". AccessKnight strips script, style, noscript, template, svg, nav, footer, and header from the page, prefers the contents of main over body, keeps the first 6,000 characters, and sends that sample to Claude with all four Extractability prose criteria in one call. The criterion here: explicit definitional sentences of the form X is a Y, covering the page's key terms, scored 0 to 4. Because svg elements are stripped and image content is never read, a definition living only inside a hero graphic scores nothing. Temperature is 0, so the same page should score the same, but a model grade is a judgment and can move between runs. This row passes only at 3 of 4.

On screen right now

What strings mean this?

If one of these strings is in front of you right now — in a config file, an HTTP header, an AccessKnight report, or another checker’s output — it is describing the failure explained on this page.

AccessKnight report
Explicit definitional sentences
AccessKnight report
Graded from presence of "X is a ..." statements.
AccessKnight report
Extractability (AEO)
AccessKnight dashboard
Also grade AI extractability
FAQ

Frequently asked questions

What does a definitional sentence look like?

It names the thing, names its category, then names what distinguishes it: "A definition lead is an opening sentence that states what the page's subject is before any other copy." The shape is deliberately plain. If the sentence would still parse with the brand adjectives removed, it is doing the job.

Where on the page should the definition go?

The first paragraph after the h1, as plain text in the document. That position carries the most weight when a system summarises the page, and it is what a reader arriving cold from a search result or an AI answer needs first. Define the remaining terms at their first use in the body.

Why is my definitions row a warn when the page does contain definitions?

This row passes only at 3 or 4 out of 4, so a 2 is a warn even though it is half marks. The grader looks for definitions of the page's key terms in the plural, not one sentence. A page that defines its headline product but leaves every supporting term undefined commonly sits at 2.

Does a definition inside a hero image or SVG count?

No, and this is verifiable rather than a rule of thumb. The content sampler removes svg elements before the page text is sent for grading, and it never reads the contents of an image. A definition rendered as a graphic is invisible to this check, and to any retrieval crawler reading the document as text.

Is this the same sentence as a featured-snippet definition?

In practice it is the same sentence doing two jobs. Classic search has long drawn definitional snippets from an explicit "X is a Y" statement near the top of a page, and this check grades the same construction. Writing it once serves both surfaces, which is the argument for doing it even though the check itself is only worth 4 points.
Sources

Where does this come from?

Every external claim on this page traces to one of the documents below — the vendors’ and standards bodies’ own documentation. What AccessKnight adds is the scan, not the standard.

Is this check
failing on your site?

AccessKnight scans any page against all 18 AI-readability checks — including this one — and shows every issue with a fix. Free, no credit card.

✓ Free — no credit card✓ Nothing installed on your site✓ WCAG + AI readability