Use Review markup for one genuine review of one specific item, with the review visible on the page. Use AggregateRating markup for an average of several ratings or reviews that the page also shows.
Do not use either type on your own site to qualify reviews of your own business for Google review stars. Decide which type matches what the page shows, write the code, then test it.
Valid markup can make a page eligible for a review snippet, but Google decides whether and when to show a rich result. No markup guarantees stars, rankings, AI citations or traffic, so the useful goal is markup that is accurate, matches the page and meets the rules of each system that reads it.
The same rules apply on WordPress, Shopify, Wix, Squarespace, Webflow and hand-built sites, because the markup describes the page and not the platform.
Choose Review or AggregateRating for the page
A Review is one person's or one organization's evaluation of one item. It carries an author, a reviewRating (a Rating object with a ratingValue) and, usually, the reviewBody text and a datePublished. A customer who writes three paragraphs and gives four stars out of five has written one Review, and the page that shows that opinion is the page to mark up with it.
An AggregateRating is the combined result of many ratings. It carries a ratingValue that is the average, plus a count, plus the ends of the scale. A product page that shows "4.4 out of 5 from 89 reviews" is describing an AggregateRating. If a page shows both a few individual opinions and the average of all ratings, Google's review snippet documentation says to include an aggregate rating based on those reviews when that is appropriate.
Two properties in the aggregate type cause most of the confusion. Use reviewCount for the number of people who submitted written reviews, and ratingCount for the total number of ratings, including ratings with no written review. Google requires at least one of the two. Use both when both numbers are accurate and visible, and keep each equal to what the page displays.
The scale matters too. The default is 1 to 5. If your page uses another scale, such as 1 to 10, provide bestRating and worstRating so the numbers mean what visitors see. Write the average with a decimal point in JSON-LD, as 4.4, even if the visible page uses a decimal comma.
Pages where Google documents review snippets
Google documents review snippets for a fixed list of item types: Book, Course list, Event, Movie, Product, Recipe, Software App, CreativeWorkSeason, CreativeWorkSeries, Episode, Game, MediaObject, MusicPlaylist and MusicRecording. Local business and Organization are on the list only when the site captures reviews about other local businesses or organizations. Google documents EmployerAggregateRating separately for employer reviews.
Every review must concern one specific item, not a category or a list. A page that reviews "the best lanterns" as a group is not a review of one item, while a page that reviews the Blue Lantern 4000 is. The reviewed item needs a name, and the ratings should come directly from users. Do not collect ratings from other websites and total them on your page.
If your item type is not on Google's list, Review and AggregateRating are still valid Schema.org vocabulary, and other systems may read them. A generic object does not create a Google feature for an unsupported type, so do not expect a Google review snippet from it. Schema.org defines Review and AggregateRating independently of any search engine's display.
Reviews of your own business and reviews from other sites
Google's September 18, 2019 guidance on review rich results defines a self-serving review as a review about entity A placed on entity A's own website, either written into the markup or shown through a third-party widget. Google says pages with those reviews are not eligible for the star review feature for LocalBusiness and Organization. The same guidance says the rule alone does not cause a manual action and does not affect a Google Business Profile.
Suppose a kayak rental shop collects customer comments and posts them on its homepage. Marking those up as reviews of the shop will not make the homepage eligible for Google review stars, whether the comments are typed in or embedded from a widget. The comments are still worth showing to visitors. Markup for a business's reviews of itself is the part Google excludes.

The rule has a clear counterpart. A review site, a directory or a publication that reviews other businesses can mark up those reviews, because the reviewed business is a different entity from the site. Reviews of products that you sell are a separate case again: a product page can carry reviews of that product, which is why the examples below use a product.
Copying is the other limit. Do not mark up reviews or ratings that you copied from Google Business Profile or another website. Google's guidelines also rule out fake reviews, and they rule out incentivized reviews unless the incentive is clearly and prominently disclosed. Google's documentation update log records that guideline on July 24, 2026, and the blog post on Google's fake and incentivized review rule covers what changed.
Properties Google requires for each type
Google splits its review snippet properties into required and recommended. For a standalone Review, Google requires author, itemReviewed and reviewRating, and the item needs a name. The Product snippet requires name plus one of review, aggregateRating or offers. Everything else, such as datePublished, reviewBody, bestRating and worstRating, is recommended for Google, not required.
Those labels describe Google's feature only. A property can be valid Schema.org vocabulary without being required by Google, and a complete property set does not make a page eligible if the content breaks a guideline.
Nesting changes one rule. If a review sits inside a parent item through the review property, omit itemReviewed, because Google treats the parent as the reviewed item. The same applies to an aggregateRating placed inside a parent through the aggregateRating property. A review author also needs a valid name shorter than 100 characters for an author-based snippet, so use the reviewer's name and not promotional text.
Google's documentation also allows a standalone review to leave out an individual rating when the markup includes both an author and a review date. Include the rating anyway when the page shows one, so the markup matches the page.
Add the product review JSON-LD
This example marks up one review of one product. The product, the reviewer and the publisher are fictional, and every URL uses example.com. Replace the @id values, the product name, the rating, the author name, the date and the review text with what appears on your own page, and keep the bestRating and worstRating values in line with your visible scale.
{
"@context": "https://schema.org",
"@type": "Review",
"@id": "https://example.com/reviews/blue-lantern-4000-review#review",
"name": "Bright output in a compact body",
"itemReviewed": {
"@type": "Product",
"@id": "https://example.com/products/blue-lantern-4000#product",
"name": "Blue Lantern 4000"
},
"reviewRating": {
"@type": "Rating",
"ratingValue": 4,
"bestRating": 5,
"worstRating": 1
},
"author": {
"@type": "Person",
"name": "Jordan Lee"
},
"datePublished": "2026-07-20",
"reviewBody": "The Blue Lantern 4000 is bright enough for a small room and easy to carry. The battery lasts through an evening, although the charging cable is shorter than expected.",
"publisher": {
"@type": "Organization",
"name": "Example Review Desk"
}
}The review title, the body text and the star score should all appear in the visible page, because Google's guidelines require review content to be visible and immediately apparent on the marked-up page. This example is also written for a page that belongs to a reviewer or review publication. It is not suitable for a business to review itself when the business controls the review process.
When the review lives on the product page itself, nest it inside the Product and drop itemReviewed. The next block shows that shape with a shorter review.
{
"@context": "https://schema.org",
"@type": "Product",
"@id": "https://example.com/products/blue-lantern-4000#product",
"name": "Blue Lantern 4000",
"review": {
"@type": "Review",
"name": "Easy to carry",
"reviewRating": {
"@type": "Rating",
"ratingValue": 5,
"bestRating": 5,
"worstRating": 1
},
"author": {
"@type": "Person",
"name": "Sam Rivera"
},
"datePublished": "2026-07-22",
"reviewBody": "Light enough to carry from room to room."
}
}Add the product aggregate rating JSON-LD
This example puts an average of 4.4 on the same fictional product, with 89 written reviews and 104 total ratings on a scale of 1 to 5. The numbers are placeholders. Each count must match what your page shows to visitors, so if the page says 89 reviews, reviewCount is 89.
{
"@context": "https://schema.org",
"@type": "Product",
"@id": "https://example.com/products/blue-lantern-4000#product",
"name": "Blue Lantern 4000",
"description": "A fictional rechargeable lantern for indoor emergency lighting.",
"image": "https://example.com/images/blue-lantern-4000.jpg",
"url": "https://example.com/products/blue-lantern-4000",
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": 4.4,
"reviewCount": 89,
"ratingCount": 104,
"bestRating": 5,
"worstRating": 1
}
}Here name is required for the Google feature, together with aggregateRating, its ratingValue and at least one of the two counts. The description, image and url properties are optional for the review snippet, though they describe the same product that the page shows. For product identity, price and availability, follow the guide to Product schema markup so the rating sits inside a complete Product entity.
Keep the markup in step with the page. When a new review arrives and the visible average changes, update ratingValue and the counts as well. An aggregate that no longer matches the visible rating data is one of the violations listed in Google's guidelines. If you would rather build the code from a form, the Review schema generator produces the JSON-LD for you.
Place the markup on your platform
Check what your platform already outputs before adding anything. Shopify says its themes automatically include structured data and product microdata, and that product pages may already carry price, availability and review information, so look at the theme's output on the Shopify SEO help page before pasting a second description of the same product.
Squarespace says it automatically generates structured data, with Product schema among the generated types, and that some generated schemas cannot be edited or removed (see Squarespace's search optimization article). Confirm what a template emits before adding custom review code.
Webflow offers a page-level schema field for manual JSON-LD, supports binding Collection fields into the code, recommends the Schema Markup Validator and warns about duplicate markup. Wix publishes JSON-LD troubleshooting guidance on punctuation, visible page content and its script-tag limits.
On WordPress or a hand-built site, add the JSON-LD through the theme's or the SEO plugin's supported head or page-output mechanism, then validate the rendered page. The WordPress schema markup guide covers existing plugin output, and the guide to adding schema markup covers the general routes.
Test the markup in a fixed order
Test in the order below, because each step answers a different question: whether the JSON is valid, whether the vocabulary is valid, whether a named Google feature accepts the page, and whether Google shows anything. Run the code through a JSON syntax check first, then the Schema Markup Validator for vocabulary, then Google's Rich Results Test for review snippet eligibility. The Schema Nova schema validator and schema checker help with the first steps.

A passing result is not a verdict on policy. Google says syntax validation cannot catch every spam or content-policy problem, so open the live page and compare it with the markup yourself: the review text, the average, the counts, the scale and the item name should all appear on the page. JSON-LD is a JSON-based format for linked data, defined in the W3C JSON-LD 1.1 Recommendation of July 16, 2020, and its syntax rules are strict, so one stray comma stops the whole block from parsing.
After publishing, watch Search Console for the page. Google provides Search Console reports for structured data, which list warnings such as missing fields. When one appears, the guide to fixing missing field warnings works through them one property at a time.
Make review markup readable for Bing and AI search
Bing reads Schema.org in JSON-LD, including review and aggregated-review scenarios, so the same markup serves it (Bing's structured data documentation). Its URL Inspection tool has a Markup card that shows the JSON-LD on a page, and its webmaster guidelines say markup should reflect the visible content. Yandex offers a structured data validator, and recognized markup can take about two weeks to appear (semantic markup FAQ).
To make a page eligible for ChatGPT search, allow OAI-SearchBot to crawl it (OpenAI's help article). Keep accurate markup and visible reviews on the page. See whether AI systems use schema markup.
Fix the problems that stop review stars
When stars do not appear after a clean test, start with eligibility. Confirm the item type is on Google's list, that the page is about one item and that the reviews are not about the site's own business. Then compare the markup with the visible page, since a mismatch is the most common cause of a rejected or ignored snippet.
Google says structured data violations can lead to a structured data manual action, which removes eligibility for the affected rich result but does not itself remove the page from ordinary web search. The violations include markup for content visitors cannot see, inaccurate or misleading information, a different item from the one on the page, reviews copied from another site, fake reviews, undisclosed incentivized reviews, invalid reviewer names, an aggregate that does not match visible data, a scale that conflicts with the page, and reviews of a category instead of one item.
Even a page that follows every rule may show no stars. Google says it decides whether and when to display a rich result, and Google's reviews system says product structured data may help it identify product reviews but is not the sole signal. Keep the page accurate, keep it crawlable and indexable, and recheck Google's documentation updates page when you publish, since the review guidelines have changed before.
Google stopped showing FAQ rich results in May 2026; the FAQ schema guide explains where FAQPage stands.
Questions people ask
Can I add review schema to my own business website?
You can add it, but Google treats reviews about your own LocalBusiness or Organization that you control as self-serving, and those pages are not eligible for the star review feature. Reviews of products you sell are a different case, and a page that reviews other businesses can qualify. Markup on your own business reviews does not change what Google Business Profile shows.
Can I copy reviews from Google Business Profile into my markup?
No. Google's guidelines do not allow reviews or ratings copied from another website, and an average built from other sites' ratings is not allowed either. Ratings should come directly from users of your own page.
Does an aggregate rating need individual reviews?
An AggregateRating needs a ratingValue and at least one of reviewCount or ratingCount, so it can describe ratings that have no written reviews. Use ratingCount for the total number of ratings and reviewCount for written reviews, and keep both equal to what visitors see.
Do I need bestRating and worstRating?
Google lists both as recommended. They matter whenever your scale is not the default 1 to 5, since without them a score of 8 would be read against the wrong range. Adding them to every example makes the scale explicit.
Why are stars not showing after the test passed?
A passing test confirms parsing and documented eligibility signals only. Google decides whether and when to show a rich result, and the item type, the review source and the visible content all have to meet its guidelines.
Can a product review use a 10-point scale?
Yes. Set bestRating to 10 and worstRating to 1 and show the same scale on the page. A visible score of 8 out of 10 with markup written for a 5-point scale is a mismatch that Google lists as a violation.
