Review

The itemReviewed property specifies the entity that is the subject of the review. This is a required property that connects the evaluation to the specific Thing being critiqued.

• Expected type: Thing (typically Product, CreativeWork, Organization, or Service)
• Without this property, the review is semantically orphaned and cannot be linked to a specific entity in a knowledge graph
• For local businesses, use LocalBusiness or its subtypes like Restaurant
• Example: A review of a book would set itemReviewed to a Book type with its own name and author properties

The reviewRating property captures the qualitative or quantitative assessment given by the reviewer. It typically uses the Rating type to express a numerical score.

• Expected type: Rating
• The Rating type itself uses ratingValue (e.g., "4.5") and bestRating (e.g., "5") to normalize the score
• For star ratings in search results, pair this with AggregateRating on the parent product page
• A review without a rating can still be valid if it provides a textual reviewBody
• Example: "reviewRating": { "@type": "Rating", "ratingValue": "4", "bestRating": "5" }

The author property identifies the person or organization that created the review. This establishes provenance and trustworthiness for AI-driven citation systems.

• Expected type: Person or Organization
• Use Person for individual reviewers with properties like name and url
• Use Organization for editorial bodies or publications with properties like name, logo, and sameAs
• For anonymous reviews, this property can be omitted, though it weakens the ClaimReview credibility signal
• Example: "author": { "@type": "Person", "name": "Jane Doe", "url": "https://example.com/author/jane-doe" }

The reviewBody property contains the full textual content of the review. This is the actual prose evaluation that provides context beyond a numerical score.

• Expected type: Text
• This field should contain the complete review text, not a truncated summary
• AI models use this field for semantic analysis and sentiment extraction during Retrieval-Augmented Generation
• For multi-paragraph reviews, include all content within a single string value
• Example: "reviewBody": "The noise cancellation is exceptional, outperforming competitors in the same price bracket. Battery life exceeded the advertised 30 hours."

The datePublished property indicates when the review was first made publicly available. Freshness is a critical signal for AI-generated overviews and search rankings.

• Expected type: Date or DateTime in ISO 8601 format
• Use YYYY-MM-DD for date-only precision or YYYY-MM-DDThh:mm:ss+00:00 for full timestamp
• Pair with dateModified if the review has been updated to reflect new product versions
• Stale reviews without dates may be deprioritized by generative engines seeking current information
• Example: "datePublished": "2024-11-15"

The publisher property identifies the organization ultimately responsible for making the review available. This is distinct from the author and represents the publishing entity.

• Expected type: Organization or Person
• Use when the publishing platform differs from the individual reviewer (e.g., a magazine publishing a freelancer's review)
• Include name, url, and logo properties to strengthen the publisher's Brand Entity Optimization
• For self-published reviews on a personal blog, the publisher and author may be the same entity
• Example: "publisher": { "@type": "Organization", "name": "TechReview Weekly", "url": "https://techreviewweekly.com" }

A specialized subtype used to evaluate the veracity of a specific statement. Essential for publishers and fact-checking organizations.

• @type: ClaimReview
• claimReviewed: The exact statement being evaluated
• reviewRating: Uses a defined scale (e.g., True, False, Mostly True)
• author: Must be a recognized authority
• itemReviewed: A CreativeWork containing the claim

Google uses this markup to display fact-check labels in search results, directly influencing how AI overviews assess the credibility of contested information.

Google's official documentation on implementing Review structured data. Covers eligibility requirements, required properties, and technical guidelines.

• Required properties: itemReviewed, reviewRating, author
• Rating scale: Must use a defined bestRating and worstRating
• Prohibited practices: Self-serving reviews, generic placeholder reviews
• Monitoring: Use Rich Results Test for validation

This is the canonical reference for ensuring your Review markup qualifies for rich result display and is correctly parsed by AI crawlers.

A Schema.org type representing the average rating calculated from multiple individual reviews. It uses properties like ratingValue (the numerical score) and reviewCount (total number of ratings).

• Critical distinction: AggregateRating summarizes collective opinion, while Review captures an individual's evaluation.
• Search display: Powers star ratings in rich results, directly influencing click-through rates.
• Best practice: Always nest AggregateRating within the Product or Service entity it describes, not as a standalone object.

A fact-checking schema used to evaluate the veracity of a specific claim. It connects a claimReviewed (the statement being checked) with a reviewRating that outputs a verdict like True, False, or Misleading.

• AI citation signal: Google explicitly uses ClaimReview to surface fact-checks in generative overviews, making it a powerful trust signal.
• Structure: Requires an author (the fact-checking organization) and a url linking to the full analysis.
• Difference from Review: Review assesses a product or creative work; ClaimReview assesses a factual assertion.

The Schema.org type representing any tangible item or intangible service offered for sale. Review and AggregateRating are typically nested as properties of a Product to establish the subject of evaluation.

• Entity linking: Use sameAs to connect a Product to its Wikidata or Wikipedia entry, strengthening entity identity for knowledge graphs.
• Rich results: Combining Product with Review, Offer, and AggregateRating enables product knowledge panels with star ratings and pricing.
• Properties: Includes name, description, image, sku, brand, and offers.

Defines an entity such as a company, corporation, or institution. In the context of reviews, the Organization type is used to identify the author of a Review or the entity being reviewed.

• Brand authority: A well-structured Organization with sameAs links to Wikidata, Crunchbase, and social profiles establishes a definitive entity identity.
• Review authorship: When an organization publishes reviews, linking the Review.author to the Organization entity builds a verifiable provenance chain.
• Nested types: LocalBusiness, Restaurant, and Corporation are sub-types that inherit all Organization properties.

A more granular alternative to AggregateRating, used to specify a single qualitative or quantitative assessment within a review. It defines the bestRating and worstRating scale boundaries.

• Use case: When a reviewer scores multiple sub-aspects (e.g., Service, Ambiance, Value) of a restaurant, each can be a separate Rating node.
• Property: reviewAspect names the specific dimension being rated, providing structured granularity that AI models can parse.
• Relationship: Multiple Rating objects can be nested within a single Review to decompose the evaluation.

Represents an individual human being. In review markup, Person is used to identify the author of a Review, establishing the credibility and identity of the reviewer.

• Author authority: Linking a Person to their professional profiles via sameAs helps search engines verify reviewer authenticity.
• Properties: Includes name, jobTitle, affiliation (linking to an Organization), and image.
• Entity disambiguation: A fully defined Person entity prevents confusion between individuals with similar names in knowledge graphs.