Only 2.8% of ChatGPT's Product Recommendations Show Shoppers a Star Rating

A shopper asks ChatGPT for the best robot vacuum and gets a row of product cards, each with a photo, a price and a store. On most of those cards there is no star rating, so the shopper has to take ChatGPT's word that the product is worth buying, or click through to find out.

Star ratings are one of the first things shoppers check, and on a retailer's own page they sit right next to the price. On ChatGPT they are rare, which makes the handful that do appear more visible, and shifts the weight of the decision onto ChatGPT's own choice of product.

ChatGPT shows a star rating on 2.8% of its product cards. When it does, the product is almost always well rated: the typical rated recommendation has 4.6 stars from 1,300 reviews, and 4.6% of rated recommendations fall below 4 stars.

On September 23, 2026 we asked ChatGPT 2,000 "best product or service" questions, such as "best coffee maker" and "best automatic cat litter box", twice each, and collected the 7,048 product cards in its answers. A product card is the shopping tile ChatGPT shows with a product's name, photo, price and store, and sometimes a star rating and review count.

195 cards, spread across 55 questions, showed a star rating. For each we recorded the rating and the number of reviews behind it, and counted how many fell under 4 stars. The median is the middle value when every rated product is lined up from lowest to highest.

195 of 7,048 cards carry a rating, about one card in 36. For every other product, the shopper sees a name, a price and a store, with nothing on the card to say how other buyers rated it. ChatGPT's answer text may describe a product as highly rated, but the card itself rarely backs that with a score.

Local business recommendations work differently. When ChatGPT recommends local businesses, the typical business card shows 4.8 stars, higher than the 4.6 median of rated products. Local shoppers see a score next to the name; product shoppers mostly do not.

Of the 195 rated products, 9 have a rating under 4 stars, 4.6%. Most sit well above that line: 56 are rated between 4.5 and 4.69 stars, 48 between 4.7 and 4.89, and 15 at 4.9 or 5.0. In all, 119 of the 195, more than half, have 4.5 stars or better.

The typical rated product has 4.6 stars and 1,300 reviews. ChatGPT's picks lean on products with a long track record, where a high average rests on many buyers rather than a handful.

The nine products under 4 stars come from everyday categories: skincare, robot vacuums, cat litter boxes, coffee makers, running shoes, strollers and grills. The lowest is a gel moisturizer rated 3.2 stars from 8,100 reviews. A robot vacuum rated 3.6 stars has 13 reviews, so a few ratings can move its score a long way.

Some of the low-rated picks carry thousands of reviews, 8,100 for the moisturizer and 4,482 for the coffee maker. A below-4 score on a popular product did not keep it out of ChatGPT's answers, which shows that ChatGPT picks products on reputation and coverage as much as on the star average.

Because most product cards show no rating, ChatGPT's choice rests on other signals: the expert reviews, comparison articles and retailer listings that name your product. Get your product into the roundups and review sites that cover your category, and make sure its retailer listings have complete names, descriptions and prices that ChatGPT can match to a card.

When a rating does appear, it sits right next to your product and price, so keep your ratings strong on the retailers ChatGPT links to, such as Walmart, Best Buy and Target, the most common stores on its product cards. Reply to critical reviews and fix the issues they raise, since a sub-4 score is visible to every shopper who sees the card. Our ecommerce seo services track how your products appear in AI shopping answers.

Embed this research

Cite this study