Guide
Does DefinedTerm Schema Help Glossary Pages Get Cited?
The short answer
Does DefinedTerm schema help glossary pages get cited?
No evidence isolates the markup as a cause. Our glossary page — shipped on day 3 of a greenfield build with a 1-sentence extractable definition and DefinedTerm schema — was observed cited in Google's AI Overview within days [our data]. But Ahrefs' 1,885-page controlled test found no citation lift from adding schema, so we credit the definition text as the likely mechanism and ship the markup anyway: cheap, standard, unproven-as-causal.
DefinedTerm schema sits inside the best origin story we can offer: a glossary page on a brand-new domain, shipped on day 3, observed cited in Google's AI Overview within days — with DefinedTerm markup on it. It would be easy, and profitable, to sell that as "the schema that gets you cited." The controlled evidence says we can't, so this page separates what the markup is, what the page shape actually was, and which of the two the evidence lets us credit.
What is DefinedTerm schema?
DefinedTerm is schema.org's type for "a word, name, acronym, phrase, etc. with a formal definition" — the natural vocabulary for a glossary entry. The working properties are few: name carries the term, description carries the definition, and inDefinedTermSet points at a DefinedTermSet node that groups the site's entries into 1 named glossary. Nothing about the type is exotic; it is a machine-readable mirror of a definition the visible page must already state.
That mirroring rule is the constraint worth respecting. Google's structured-data guidelines require markup to reflect visible content, so the description should be the same definition the reader sees in the page's opening line — not a second, markup-only version that can drift from the visible one.
What did the glossary page that won actually look like?
A page built citation-shaped from its first commit, on which the markup was one element among several. The full field report is our first AI Overview citation; the shape that shipped:
| Element | What shipped |
|---|---|
| Opening line | A 1-sentence, ≤50-word definition, quotable whole with no context-dependent clauses |
| Schema | DefinedTerm markup naming the term and mirroring that definition |
| Figures | Primary-source-cited numbers in the quotable sentences |
| Structure | Question-shaped H2 sections, each answering in its first sentence |
| Target | The definitional query for the build's core term |
The observation attached to it: shipped on day 3 of a greenfield insurance build, and recorded as cited in the AI Overview for its term within days — including an inline, hyperlinked brand mention in the answer body [our data]. The record is dated manual checks of the live SERP, because Google Search Console has no AI Overview dimension; no report would have told us.
None of that shape is proprietary. It is the extractable-passage pattern from Google's May 2025 guidance on performing well in AI experiences, applied to a page whose entire job is stating one definition cleanly.
Does the markup cause the citation?
We cannot say it does, and the best available evidence says don't. The win is an n of 1 in which definition text, page structure, sourcing, and markup all shipped together — nothing isolates any single variable. The controlled evidence that does isolate markup is Ahrefs' test of 1,885 pages that added JSON-LD against roughly 4,000 controls (August 2025–March 2026): no meaningful citation lift on any platform tested. The whole markup-and-citations question is accounted for in what the evidence shows, and DefinedTerm earns no exception to it.
So the honest attribution is asymmetric. The definition text is the likely mechanism: engines demonstrably quote visible passages, definitional queries reward a liftable definition, and the sentence that was quoted was the visible opening line. The markup is unproven-as-causal: standard, harmless, and — on the only controlled test available — worth roughly 0 citations by itself.
Definitional queries are also the easiest class to win, which caps what the win can prove. A clean definition of a term nobody else answers competes with almost nothing; that is a reason to own your niche's definitions, not evidence that a schema type moves engines.
How do you implement DefinedTerm in JSON-LD?
One node per glossary page, one set node for the glossary, templated once. The shape our builds generate, with placeholder values [our data]:
{
"@context": "https://schema.org",
"@type": "DefinedTerm",
"@id": "https://example.com/glossary/some-term/#term",
"name": "Some term",
"description": "The same ≤50-word definition the page opens with, verbatim.",
"inDefinedTermSet": {
"@type": "DefinedTermSet",
"@id": "https://example.com/glossary/#set",
"name": "Example Co Glossary"
}
}
Across our fleet this is generated from each page's frontmatter at build time — the writer supplies the definition once and the markup mirrors it, which is what keeps the visible text and the description from ever disagreeing [our data]. On any templated stack the marginal cost per page is effectively 0, and that number is the entire economic case for shipping it.
Should you invest in DefinedTerm markup?
Invest in the definitions; let the markup ride along free. Every hour available for a glossary page does more work in the visible text — a definition that survives being lifted alone, real questions answered underneath it, figures with sources — than in any JSON-LD block. That ordering is what our own win supports, and it is what the controlled evidence demands.
Said against interest, since glossary builds are work we sell: you do not need anyone to add DefinedTerm schema for you. If your stack templates JSON-LD, it is a few lines; if it doesn't, skip the markup before you skip the definition text. The same cost logic governs its sibling type in our FAQ schema verdict — mirror real visible content for free, or don't bother. What nobody can sell you is the citation itself: nobody controls answer-engine output, and the glossary formula — definitions first, markup second, promises never — is documented in full in our generative engine optimization guide.
Frequently asked questions
What is DefinedTerm schema?
schema.org's type for 'a word, name, acronym, phrase, etc. with a formal definition.' On a glossary page it names the term and carries its definition machine-readably, usually with name, description, and an inDefinedTermSet reference grouping the site's glossary into 1 set.
Did DefinedTerm schema cause your AI Overview citation?
We cannot say, and we refuse to claim it. The page shipped its definition text, structure, and markup together on day 3 of the build, so nothing isolates one variable — and Ahrefs' 1,885-page controlled test found no causal link between schema and AI citations generally.
What should a glossary page look like to be quotable?
Open with a definition of 50 words or fewer that survives being lifted alone, then answer the term's real questions in question-shaped sections with sourced figures. Google's May 2025 guidance points the same direction: easy-to-extract passages that directly answer the query.
How do I add DefinedTerm markup?
A JSON-LD node with @type DefinedTerm, the term as name, the same definition the visible page opens with as description, and an inDefinedTermSet pointing at 1 DefinedTermSet node for the glossary. Templated once, it costs roughly 0 per additional page.
Should every glossary page have DefinedTerm schema?
If your stack templates it, yes — it is standard vocabulary, costs nothing at the margin, and mirrors visible content. If adding it costs real engineering time, spend that time on the definition text instead: the evidence locates the quoting behavior in the visible words, not the markup.