Review ratings in AI answers

01Do review stars in AI answers still matter?

Review stars in AI answers matter in a narrower way than stars in classic search results did. A star rating in an AI answer is a summary of reviews shown next to a product, place or service inside a generated response. When one appears, it usually comes from structured product or business data rather than from a single article that happens to mention a score.

This guide covers where AI answers get ratings, how on-page markup relates to feeds and review platforms, and what to mark up. It does not cover how to write review schema itself; for that, read the guide to review snippet markup rules. It belongs to the AI answers section of the handbook.

Two definitions keep the rest clear. A review is one person's assessment of one item. An aggregate rating is the average of many reviews, usually given with a count, such as 4.4 from 89 reviews. AI answers are most likely to show the second kind, because it compresses many opinions into one number.

02Where do AI answers get ratings?

Google's guide to optimizing for generative AI features says generative AI responses can include product listings, product information and information about local businesses. It points merchants to Merchant Center feeds and Google Business Profiles to keep that information visible in AI responses and other results. In other words, the documented route for product and local data into Google's AI features runs through those systems as well as through the web page.

ChatGPT documents a similar path for shopping. Its merchant page for product discovery in ChatGPT invites stores to share product feeds and describes richer feeds as a way to improve how products appear in results.

That page does not say how reviews in a feed are weighted, and it does not mention schema.org markup at all. For other answers, ChatGPT search fetches pages through its search crawler, as the guide to how ChatGPT and Perplexity read your pages explains. A rating printed on a fetched page is text the answer can quote, but no documentation describes ratings being lifted from markup into a star display.

03On-page markup vs review platforms

On-page review markup is still the only route to a Google review snippet, the stars under a blue link. Google's review snippet documentation lists the types that can show them, including products, recipes, software apps, books and local businesses when the site reviews other businesses. The markup must describe a specific item, and the reviews must be visible on the page.

Review platforms and marketplaces are a separate source. A business profile, a marketplace listing or a third-party review site collects ratings that answer engines can read wherever those platforms publish them. You do not control those pages, and Google's review snippet guidelines say not to aggregate reviews or ratings from other websites in your own markup.

So the two sources do different jobs. Your markup states your own reviews precisely and keeps a page eligible for rich results. Platforms carry ratings that other people collected. An AI answer that compares products may draw on either, and neither engine publishes how it balances them.

Google's review snippet documentation also makes a recommendation that suits AI answers as well as stars: accept only ratings that come with a review comment and the author's name. A bare number tells a reader little. A rating with a written reason and a named reviewer is something an answer can summarize with context.

Ratings as text

Text answers in ChatGPT and Perplexity work from the words on a fetched page. If a rating matters to your page, state it in a sentence that survives quotation: the score, the scale, the number of reviews and the date, as in "4.4 out of 5 from 89 verified buyers, as of November 2026". That sentence is useful to a reader whether or not any engine reads the markup underneath it.

04Product and local results

For products, the strongest position is consistency. The rating in the product page's AggregateRating, the rating shown to visitors and the rating in any merchant feed should match. A mismatch between feed and page is the kind of data conflict that makes any system less certain which number to trust.

For local businesses, the rules are stricter. Google's review snippet guidelines say that when the entity being reviewed controls the reviews about itself, its LocalBusiness or Organization pages are ineligible for star review features, even when the reviews come through an embedded third-party widget. Local ratings that appear in Google's results are therefore normally drawn from Business Profile reviews, not from a business marking up praise on its own site. The guide to schema and AI Overviews covers where markup does help local and product pages.

05Why do honest reviews matter more than stars?

Because the rules are moving toward authenticity, and AI answers make fake signals easier to spot at scale. Google's review snippet documentation now carries a specific guideline on fake and undisclosed incentivized reviews.

The examples in that guideline are reviews not based on a genuine experience, and reviews written in exchange for money, discounts, vouchers or free products without clear and prominent disclosure. Google's general structured data guidelines add that reviews or ratings not by actual users may result in a manual action, which removes rich result eligibility for the affected markup.

An honest rating is also a better trust signal for any reader, human or machine. For example, a product with 4.3 stars from 212 named reviews, some of them critical, reads as real. A wall of identical five-star reviews reads as manufactured. The home page of the wprichsnippets.com handbook explains why review markup sits at the centre of this site's subject.

06What should you mark up?

• Mark up reviews and ratings that are visible on the page, collected from real customers, about the specific item the page describes.
• Use AggregateRating with a real count, and keep it in sync with the number shown to visitors.
• Do not mark up testimonials about your own business on your own site and expect stars; local and organization self-reviews are ineligible.
• Disclose any incentive for a review next to the review itself, not in a footer policy.
• If you sell products, keep the rating in your merchant feeds equal to the rating on the page.
• Leave the rating out entirely if you do not have genuine reviews. An empty field is better than an invented one.

None of this guarantees stars in an AI answer, and no engine promises them. It does keep your data accurate in every place an answer engine might look.