Do customer reviews help your products show up in AI shopping answers?

Why reviews matter more in a chat

On a search results page, stars sit next to a link, and the shopper decides. In a chat, the assistant does part of the deciding. A shopper asks "which of these is the quietest?" or "do people say it runs small?" and the answer often comes from what reviewers wrote.

OpenAI is clear about this. It says ChatGPT's labels such as "Budget-friendly" are generated from information available to the model, and gives an example: a "Budget-friendly" label "might mean that reviewers frequently mention good value". In other words, what customers write about your product can become the label a new shopper sees.

So reviews now do two jobs: they persuade the human, and they give the AI words to describe you.

What ChatGPT and Google say they read

Two things follow. First, ChatGPT reads reviews on public websites, so reviews hidden behind logins or loaded in ways crawlers can't see may not count. Second, Google leans on product identifiers to match reviews to the right product, so missing or wrong GTINs can cost you stars.

How to make your reviews readable by AI

1. Put the review text in the page HTML

Many review apps inject reviews with JavaScript after the page loads. Google can render that, but in Vercel's analysis of AI crawler traffic (December 2024), the crawlers from OpenAI, Anthropic, Meta, ByteDance and Perplexity did not execute JavaScript. If your reviews only appear after scripts run, those crawlers may see a product page with no reviews at all.

Check it: open a product page, view the page source (not the inspector), and search for a sentence from one of your reviews. If it isn't there, ask your review app or developer whether it can render reviews on the server, or at least output the rating and the latest reviews in the HTML.

2. Add Product markup with the rating

Google's Product snippet documentation lets you add aggregateRating and review to your Product structured data. A minimal example:

{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Merino trail sock, crew height",
  "sku": "MTS-CREW-42",
  "gtin13": "5601234567890",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.6",
    "reviewCount": "128"
  }
}

Google's review snippet guidelines set the rules: the reviews must be visible to users on the marked-up page, the review must be about a specific product rather than a category, and you must not aggregate reviews or ratings from other websites. Google also says there is no special schema needed for its AI features, and that structured data should match the visible text, per its guide to AI features and your website.

3. Send the same numbers in your feeds

If you use Google Merchant Center, the Product ratings add-on shows star ratings in ads and free listings. You need at least 50 reviews across all products. Most stores send reviews through a supported review aggregator; you can also upload them yourself, or start collecting with Google's customer reviews add-on.

If you have access to OpenAI's product feed, its products feed reference has review_count and star_rating. OpenAI's rules are worth copying everywhere: the count and the rating must describe the same group of reviews, and you should omit the rating when there are no reviews rather than send zero. Store ratings go in separate fields and must not be mixed with product reviews.

Whatever you send, the rating on the page, in the markup and in each feed should be the same number.

4. Use real product identifiers

Google says ratings are matched to products mainly by GTIN, and that matching on SKU, brand plus MPN or URL "often result[s] in fewer reviews being correctly associated with your products". Put the GTIN in your feed and in your Product markup, and keep brand and MPN spelled the same in your product data and your review data.

5. Ask for reviews on more than one site

ChatGPT's review summaries are based on reviews from public websites, not only yours. Ask buyers for a review a few days after delivery, and link to the places that matter in your market: your own product page, a review platform you use, or the marketplace you sell on. Reply to negative reviews politely and fix what they point to. A visible, sensible reply is part of what the next shopper, and the AI, reads.

Review rules you can't skip

Reviews are regulated, and breaking the rules can cost you more than your stars. This is a summary, not legal advice.

In practice: invite every buyer, not only happy ones; publish negative reviews as well as positive ones; if you give anything in return for a review, say so next to it; and add a short "How we collect reviews" note near your reviews. That note is exactly what the EU rule asks for, and it costs you nothing.

Mistakes that hide your reviews

1. Reviews only in a JavaScript widget. Many AI crawlers won't see them.
2. Stars in the markup that aren't on the page. Google requires the rating to be visible to users.
3. One store-wide rating on every product. Google wants reviews about a specific item, and OpenAI keeps store and product ratings in separate fields.
4. Imported ratings from another site. Google says not to aggregate reviews or ratings from other websites in your markup.
5. Different numbers in different places. 4.6 on the page, 4.8 in the feed and 4.4 in the markup makes your data look unreliable.
6. Missing GTINs. Fewer of your reviews get matched to the right product in Google.

Where to start this week

1. Pick your 10 best sellers and view their page source. Is the rating there? Is review text there?
2. Validate their Product markup with Google's Rich Results Test and fix any rating errors.
3. Count your reviews. If you have 50 or more across all products and use Merchant Center, look at the Product ratings add-on.
4. Check your review invite. Does it go to every buyer, and is any incentive disclosed?
5. Add a "How we collect reviews" line under your reviews.
6. Ask an assistant. Type a real shopper's question about one of your products into ChatGPT or Google AI Mode and read what it says reviewers think.

What this won't do

Reviews don't buy you a place in an AI answer. OpenAI says product results are selected independently and "not all available products will necessarily be shown", and it ranks merchants on factors like availability, price and quality. Google says reviews and stars are filtered and may not always show. Good, visible, honest reviews remove a reason to skip you, and give the assistant accurate words to describe your product. That's the realistic goal.

For the rest of the product page, see our guides to product descriptions AI shopping assistants can use and showing shipping and return policies to Google and AI.

Check your site

The free Web2Agents AI check scores your homepage on the basics AI tools look for, with results on screen. The €19 AI-Ready Kit runs all 36 checks on up to 25 pages, including product pages and their schema, and ranks the fixes.