Review and AggregateRating schema: social proof for AI

Why reviews are crucial for AI visibility

Reviews and ratings are one of the most powerful forms of social proof on the web. They influence purchasing decisions, build trust and give potential customers an objective picture of your product or service. But reviews are more than a conversion tool: they are a crucial signal for search engines and AI models to assess the quality and reliability of a business.

With Review and AggregateRating schema you make this social proof machine-readable. Search engines display the yellow stars in search results, and AI models use rating data to substantiate answers. When a user asks ChatGPT "What is the best-rated coffee shop in Utrecht?", the model searches for businesses with high ratings and readable reviews. Without schema markup, that information is invisible to machines.

Review schema works together with other Schema.org types to provide a complete picture. A Product with AggregateRating, a LocalBusiness with reviews or a Course with student ratings: the combination of the content schema type with review data is always stronger than either one alone.

Review schema: marking up individual ratings

The Review schema type describes an individual review of an item. It contains the author, the rating value, the date and optionally the written review text. Reviews are always linked to an itemReviewed, the object being reviewed.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Review",
  "itemReviewed": {
    "@type": "LocalBusiness",
    "name": "Coffee Bar The Bean",
    "address": {
      "@type": "PostalAddress",
      "addressLocality": "Utrecht"
    }
  },
  "reviewRating": {
    "@type": "Rating",
    "ratingValue": "5",
    "bestRating": "5",
    "worstRating": "1"
  },
  "author": {
    "@type": "Person",
    "name": "Lisa van den Berg"
  },
  "datePublished": "2026-04-10",
  "reviewBody": "The best specialty coffee in Utrecht. The team knows their beans inside and out, and the homemade cakes are irresistible. Every week a different single-origin espresso, always surprising and always great.",
  "name": "Best coffee in the city"
}
</script>

The reviewRating contains the numerical score. Always specify bestRating and worstRating to make the scale clear. Most platforms use a scale from 1 to 5, but Schema.org supports any scale. The author field is required and must refer to a real person or organization. Anonymous reviews without an author are ignored by Google.

The itemReviewed field

The itemReviewed field links the review to the reviewed item. This is crucial for Google to understand what exactly is being reviewed. Use the correct schema type: Product for product reviews, LocalBusiness for business reviews, Course for course reviews. The itemReviewed may be a shortened version with only the name and possibly key identifiers.

AggregateRating: the overall picture

Where Review describes an individual rating, AggregateRating provides the statistical overview: the average score, the number of ratings and the number of written reviews. This is the schema type that generates the yellow stars in Google.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "AEO Strategy Course",
  "description": "Complete online course on Answer Engine Optimization for marketers and SEO specialists.",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.8",
    "bestRating": "5",
    "worstRating": "1",
    "ratingCount": "456",
    "reviewCount": "312"
  },
  "review": [
    {
      "@type": "Review",
      "reviewRating": {
        "@type": "Rating",
        "ratingValue": "5",
        "bestRating": "5"
      },
      "author": {
        "@type": "Person",
        "name": "Pieter Jansen"
      },
      "datePublished": "2026-03-20",
      "reviewBody": "Finally a course that explains AEO practically and understandably. The Schema.org modules were particularly valuable.",
      "name": "Practical and in-depth"
    },
    {
      "@type": "Review",
      "reviewRating": {
        "@type": "Rating",
        "ratingValue": "4",
        "bestRating": "5"
      },
      "author": {
        "@type": "Person",
        "name": "Anouk de Groot"
      },
      "datePublished": "2026-04-05",
      "reviewBody": "Good course with many practical examples. Would like to see more about AI Overviews and the latest Google updates.",
      "name": "Good, but could be more current"
    }
  ]
}
</script>

Note the difference between ratingCount and reviewCount. ratingCount is the total number of people who gave a score (including those who did not write text). reviewCount is the number of written reviews. Google displays ratingCount next to the stars in search results. If you only have reviewCount (no distinction between ratings and reviews), use ratingCount.

Review schema and E-E-A-T signals

Reviews are a direct expression of the "Trust" component in E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness). A business with hundreds of positive reviews has demonstrably earned customer trust. AI models use these signals to determine which sources they consider reliable.

The combination of Review schema with other E-E-A-T signals is particularly powerful. A LocalBusiness with AggregateRating, author information and sameAs links to verified profiles gives AI models a strong trust signal. Each individual signal reinforces the whole.

• High AggregateRating (4.0+) with a significant number of reviews (50+) positions you as a trusted source.
• Recent reviews (last 3 to 6 months) show that your business is active and current.
• Detailed reviewBody texts give AI models specific, user-verified claims about products and services.
• Identified authors (with name and possibly profile) increase credibility above anonymous reviews.

Implementation patterns for different scenarios

How you implement Review schema depends on your platform and the type of content being reviewed.

E-commerce product reviews

For webshops, the combination of Product schema with AggregateRating and individual Review items is the standard. Generate the schema markup dynamically based on your review database, so new reviews are automatically included.

Local business reviews

For local businesses you can combine reviews from your own website with a reference to external review platforms. Use AggregateRating on your LocalBusiness schema and consider including a few representative reviews as examples.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "Kobalt",
  "@id": "https://kobaltdigital.com/#organization",
  "url": "https://kobaltdigital.com",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.9",
    "bestRating": "5",
    "ratingCount": "47",
    "reviewCount": "38"
  },
  "review": [
    {
      "@type": "Review",
      "reviewRating": {
        "@type": "Rating",
        "ratingValue": "5",
        "bestRating": "5"
      },
      "author": {
        "@type": "Person",
        "name": "Thomas Mulder"
      },
      "datePublished": "2026-03-12",
      "reviewBody": "Kobalt built our AEO strategy from the ground up. Within three months we were being cited by Perplexity and ChatGPT. Professional team with deep knowledge."
    }
  ]
}
</script>

Service reviews and testimonials

For service providers, testimonials are a form of reviews. Mark them with Review schema and link them to your Organization or Service type. Ensure the testimonials are actually visible on your website: Google requires that schema markup matches visible content.

Common mistakes and Google guidelines

Google follows strict guidelines for Review and AggregateRating schema. Violations can lead to the loss of rich results or even a manual action.

1. Publishing self-reviews: rating your own business on your own website is not allowed. Reviews must come from third parties.
2. Fabricating or manipulating reviews: the schema markup must reflect authentic, actually user-written reviews.
3. Showing reviews in schema that are not visible on the page: every review in your markup must also be visible to visitors.
4. Placing AggregateRating on the homepage without it belonging to a specific item: AggregateRating must always be nested in a supported schema type.
5. Leaving outdated review data in place: keep your review counts and average score current. An AggregateRating with data from two years ago is unreliable.

Reviews as an AI source: how models use ratings

AI models use review data in multiple ways when generating answers based on web content.

First, they use AggregateRating as a quality signal. A product with a 4.8 average based on 500 reviews is considered more reliable than a product with the same score but only 5 reviews. Second, AI models extract information from reviewBody texts to answer specific questions. If someone asks "Is the battery life of product X good?", an AI model can pull that information from customer reviews. Third, reviews strengthen your overall AEO position through the trust signal they provide.

• AggregateRating functions as a quantitative quality signal for AI models when ranking recommendations.
• ReviewBody texts provide AI models with specific, user-verified claims about products and services.
• Recent reviews give AI models confidence that the information is current and relevant.
• The combination of high scores and detailed reviews makes your content a reliable source for AI citations.

Summary: key takeaways

• Review schema marks individual ratings with author, score, date and text. AggregateRating provides the statistical overview with average score and number of ratings.
• Google allows Review rich results on specific schema types: Product, LocalBusiness, Book, Course, Event and more. Self-reviews are not allowed.
• AggregateRating generates the yellow stars in Google search results, which demonstrably increases click-through rate.
• AI models use review data as a quality signal, as a source for specific product claims and as a trust indicator.
• Keep reviews current, authentic and synchronized between schema markup and visible page content to prevent manual actions.

Frequently asked questions

In a world where AI models summarize your reputation in a single sentence, structured reviews are your most powerful evidence.