← Back to blog
Review and AggregateRating schema after Google's 2024 changes
April 15, 2026 6 min read

Review and AggregateRating schema after Google's 2024 changes

Google narrowed which review markup earns organic-search rich results. The schema itself is unchanged — still valid Schema.org, still parsed by Google Merchant Center and Bing. What's documented, what's plausible, and what to keep vs. drop.

schemagooglereviewsai-search

Reference reading: For the underlying product schema base, see product schema for Shopify. The forthcoming validation guide will cover the Schema.org Validator and Rich Results Test workflow referenced below.

The short version: between 2023 and 2024 Google narrowed the set of review markup that earns organic-search rich results. The schema itself didn’t change — Review and AggregateRating are still in Schema.org, still valid, still parsed by other commerce surfaces. What changed was Google’s organic-search rich-result eligibility, which is one specific surface among several that consume the markup.

A common 2024 reaction was “strip the review schema, it doesn’t do anything anymore.” That conflates Google’s organic-search policy with the markup’s broader role. The structural facts below are what the schema continues to do.

What Google actually changed

Google restricted FAQ and HowTo rich results in August 2023, limiting FAQ rich results to authoritative health and government sites and narrowing HowTo. Reviews followed a separate trajectory: Google has documented stricter requirements for product review rich results, particularly around third-party review providers and the underlying review HTML being indexable rather than JavaScript-only.

For the precise current eligibility criteria, refer to Google’s product review snippet documentation and merchant listings documentation — the specific requirements evolve and the live docs are the source of truth.

The narrowing affected Google organic-search rich-result eligibility. It did not invalidate the schema, did not change Schema.org’s definitions, and did not affect what other surfaces read from the markup.

Where review schema continues to be read

Three surfaces read review structured data and were not affected by Google’s organic-search narrowing:

Google Merchant Center

GMC accepts product reviews via a separate Product Reviews feed, distinct from the standard product feed. Reviews submitted through the feed populate the star ratings on Shopping ads and free listings. The product feed’s aggregateRating is also a documented input.

The narrowing affects organic Search rich results; it does not affect Shopping. Star ratings continue to appear in Shopping cards.

Bing and Microsoft Copilot

Bing Webmaster Tools’ structured data documentation covers Review and AggregateRating as supported types. Microsoft has not announced a Google-equivalent narrowing for review markup, and Microsoft’s October 2025 ads guidance for AI search explicitly recommends structured data as an input for inclusion in AI-generated answers.

AI agent retrieval layers

Less is publicly documented here, and what’s true is more carefully hedged.

Searchviu’s October 2025 testing suggested that AI agents (ChatGPT, Claude, Perplexity, Gemini) may not consistently extract JSON-LD on direct page fetch. If that pattern holds, schema’s value to AI surfaces flows mostly through the underlying retrieval layer (Bing’s index, Google’s index, ChatGPT’s product index) — agents read the index; the index reads the schema.

What this means for review schema specifically: stripping AggregateRating removes the catalog from whichever index inputs the markup was feeding, including the indexes that AI agents query. Whether and how those agents weight the rating value when generating recommendations is not publicly documented. Treat the surfacing benefit as plausible-via-the-index-path rather than as a measured mechanism.

For the longer treatment of this finding, see AI agents and JSON-LD: the honest picture.

What to ship in 2026

A pattern that respects what’s documented:

Keep aggregateRating on every product page that has real reviews. The block is small and feeds GMC and Bing directly. The Schema.org definition is a documented type the major search engines read.

"aggregateRating": {
  "@type": "AggregateRating",
  "ratingValue": "{{ product.metafields.reviews.rating }}",
  "reviewCount": "{{ product.metafields.reviews.count }}",
  "bestRating": "5",
  "worstRating": "1"
}

The example uses Shopify Liquid; the same JSON-LD shape applies on WooCommerce, BigCommerce, Adobe Commerce, Squarespace, or any ecommerce platform that can render server-side templates.

Render individual Review blocks only when the third-party provider’s integration documents their use. Yotpo, Bazaarvoice, Reviews.io, Okendo, and Stamped each have documentation on how their schema integrations work; follow the provider’s documented pattern rather than a generic one.

Render the underlying review HTML, not just a JavaScript widget. This is the documented Google requirement for review eligibility and the Searchviu testing reinforces the broader point: server-rendered HTML is what indexers read; JavaScript-only widgets are not reliably indexed.

Use the GMC Product Reviews feed where reviews matter for Shopping. For catalogs running paid Shopping ads or aiming for free listing visibility with star ratings, the dedicated review feed is a documented input alongside the schema layer.

What to drop

Two patterns worth removing:

  • Review blocks for first-party-only reviews when targeting Google rich results. They don’t earn rich results post-narrowing and they bulk up the JSON-LD. Keep aggregateRating (which has separate eligibility); drop the individual Review enumeration if it’s not earning a surface.
  • Fake or aspirational ratings. Populating aggregateRating for products with no real reviews is a Google Merchant Center policy violation and against Bing’s guidelines. Enforcement is uneven; the policy risk is real.

The honest framing

The 2024 reaction to Google’s narrowing was often overcorrection. The narrowing was a specific Google organic-search policy change, not a schema deprecation. The schema is still defined by Schema.org, still valid, still read by GMC and Bing, and still feeds the indexes that AI surfaces query.

A useful frame: review schema in 2026 is for GMC, Bing, and the broader index path first; for Google organic rich results conditionally and only where eligibility criteria are met.

What to ship this week

  1. Audit current state. Validate product pages through the Rich Results Test and the Schema.org Validator. Note which surfaces show errors vs. warnings vs. valid.
  2. Restore aggregateRating where it was stripped if real reviews exist. The block is a small template change.
  3. Add the GMC Product Reviews feed if running paid Shopping and not already submitting it.
  4. Verify the review HTML renders server-side. A JavaScript-only widget is not reliably indexed regardless of what schema is shipped.

The narrowing of Google rich results was a real change. The conclusion that review schema doesn’t matter anymore confuses one surface’s policy with the markup’s broader role. The schema continues to be read by other surfaces; treating it as deprecated removes the input that was carrying that surfacing.