You have probably seen yellow stars and review counts appearing beneath a site title in search results. This effect is produced by Review and AggregateRating types from the schema.org vocabulary. When a user sees such a rating, trust in the link grows instantly and the click goes to that result instead of neighbouring ones. Because of this clear advantage many site owners rush to deploy the markup, but without understanding the rules these efforts often backfire and trigger penalties.
The difference between Review and AggregateRating
The Review type describes a single specific opinion and includes fields such as ratingValue, author, datePublished and reviewBody. It records the view of one customer along with who left it and when. AggregateRating represents a generalised assessment: ratingValue stores the average score while reviewCount or ratingCount shows how many people evaluated the object. These two types complement each other and are usually embedded inside a primary object such as Product, LocalBusiness or Service as a nested field.
Customers of Sayt.uz frequently confuse the two. Some add only AggregateRating without any Review element, and Google refuses to display the rich result. Others list individual Review entries but forget the aggregated rating. The correct approach is to provide both aggregateRating and a review array inside the main object simultaneously.
The 2023 Google update and the self-serving review ban
In August 2023 Google substantially tightened its rich result guidelines. Previously many site owners wrote glowing texts about their own products and embedded them inside schema, earning stars in search. This is now officially forbidden and labelled as self-serving review. In other words, a business owner or their employee may not place inside schema markup any ratings they have written themselves about their own company.
In 2026 the rule operates even more strictly. Google does not merely ignore such constructions but completely revokes the entire site's rich result eligibility when a violation is detected. Only authentic ratings from external customers may appear in schema. Moreover, these reviews must be visible to users directly on the page, not hidden inside the code alone.
JSON-LD example and correct deployment
The example below shows how AggregateRating and several Review entries are attached to a Product object. The code must match the actual reviews on the page and align with the rating the user sees.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Product",
"name": "Hosting Premium",
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.7",
"reviewCount": "238",
"bestRating": "5",
"worstRating": "1"
},
"review": [{
"@type": "Review",
"author": {"@type": "Person", "name": "Anvar Karimov"},
"datePublished": "2025-12-14",
"reviewRating": {"@type": "Rating", "ratingValue": "5"}
}]
}
</script>The fake review mistake and its consequences
The most damaging error is injecting fabricated reviews into schema. Many owners invent author names to give themselves top scores in pursuit of a competitive edge. Google detects such cases through a patented algorithm that analyses name patterns, date sequences, IP addresses, linguistic features and dozens of other signals. A flagged site receives a manual penalty and practically disappears from organic search.
One e-commerce company exposed more than 200 authentic customer reviews through schema and increased organic CTR by 28 percent, which had a noticeable impact on sales volume. Another company chasing the same competitor added fabricated ratings and within six months received a manual penalty that nearly wiped out its organic traffic. This contrast shows how thin the line is between success and disaster: only genuine customer opinions produce long-term gains.