Should every blockquote on a page use Quotation schema?

No. Reserve the markup for quotes whose value depends on attribution—expert opinion, statements of record, definitions from a primary source. Decorative pull-quotes from your own copy or unattributed inspirational quotes add markup without adding citation surface area.

Does Quotation help with AI Overview or Perplexity citation share?

Indirectly, yes. AI engines like Perplexity build answers from cited quotes whenever possible, and structured Quotation markup gives the extractor a stable boundary plus a resolved speaker entity. Combined with isBasedOn, it improves both the likelihood of citation and the cleanliness of attribution in the final answer.

When should I use Quotation versus Claim or ClaimReview?

Use Quotation for any cited utterance regardless of factuality—opinion, color, dialogue, expert commentary. Use Claim for a discrete factual assertion that is or could be the subject of a ClaimReview verdict. Many fact-check pages benefit from both: a Quotation for the verbatim statement and a paired Claim plus ClaimReview for the verdict.

How do I attribute an organization rather than a person?

Set creator to an Organization node with name and sameAs pointing at the organization's canonical site, Wikipedia, or Wikidata entry. AI engines treat organizational quotes the same way as personal quotes for citation purposes, but resolution accuracy depends on the strength of the entity links.

Can I model a quote that has been translated?

Yes. Set text to the language version on the page, set inLanguage accordingly, and use isBasedOn to link to the original-language source. If you publish multiple translations, use one Quotation node per language and link them via translationOfWork and workTranslation on the parent CreativeWork.

Quotation Schema for AI Citation Extraction

Quotation is a schema.org CreativeWork subtype that represents a cited utterance attributed to a person or organization. By marking quotes explicitly with creator, spokenByCharacter, and isBasedOn, publishers give AI answer engines a structured handle to extract, attribute, and cite quoted content cleanly.

TL;DR

Quotation marks a quoted passage as a first-class entity in the page's structured data graph. Use text for the verbatim quote, creator or spokenByCharacter to attribute it to a Person or Organization, and isBasedOn to point at the source CreativeWork. AI answer engines like Perplexity and Google AI Overviews favor pages where quoted content is structurally distinct from surrounding prose, because it lets the extractor treat the quote as an indivisible citation unit.

Definition

Quotation is a schema.org type defined under Thing > CreativeWork > Quotation and canonically published at https://schema.org/Quotation. It represents any cited utterance—a quote from an interview, a passage from a book, a line of dialogue, a public statement, or any text reused verbatim from another source.

The core properties for AI citation are:

text: the verbatim quoted text.
creator / author: the Person or Organization who originated the quote (these properties are treated as equivalent on CreativeWork).
spokenByCharacter: the (possibly fictional) character, Person, or Organization to whom the quotation is attributed within the containing CreativeWork. This is useful for dialogue in fiction, dramatized statements, or when the in-text speaker differs from the document's author.
isBasedOn: a CreativeWork or URL the quote is derived from—typically the original interview, speech, paper, or article.
citation: a related work cited as a reference for this content.
inLanguage, datePublished, url: standard CreativeWork properties carrying language, date, and stable URL.

Unlike Claim, Quotation carries no implied truth value. It is the right type for opinion, color quotes, dialogue, and reused passages where attribution matters but the content is not under fact-checking review.

Why this matters

AI answer engines build their citations from quoted content disproportionately often, because quotes are easy to attribute, hard to paraphrase without distortion, and naturally answer-shaped. Perplexity's published pipeline (Perplexity Sonar features) returns inline citations for almost every claim in its synthesized answer; Google AI Overviews and ChatGPT Search behave similarly when the source has clean attribution markup.

Marking quoted passages with Quotation instead of leaving them as bare blockquotes does three useful things at once. It gives the extractor a stable boundary so the engine pulls the full quote rather than a truncated fragment. It binds the quote to a specific Person or Organization, which the engine can resolve against its entity graph for stronger attribution. And it links to the upstream source via isBasedOn, which lets the engine deduplicate when the same quote appears across multiple secondary articles.

The practical upside for publishers is asymmetric: the markup costs only a few JSON-LD nodes per page, adds no risk to traditional rankings, and makes every quote on the page eligible for direct citation by an AI engine that prefers structured citation surfaces.

How it works

A Quotation node typically lives inside the same