Two stores sell the same kettle at the same price. In Google's results one listing shows a star rating under the title and the other shows nothing, and an assistant asked for a reliable kettle names the first store and adds that "reviewers praise the fast boil". Neither store has a better kettle. One has evidence around it that Google and the assistant can read. Reviews and the content customers create are the trust signals both systems use, and they count only when they're real, visible and marked up correctly.
Stars in results come from markup about real reviews
Google's review snippet documentation is precise about where the stars come from: an aggregate rating or individual reviews inside structured data for a supported type, and Product is one of them. The markup sits inside the Product object on the product page and states the average rating and the number of ratings or reviews. Google's guidelines add conditions that stores break constantly. The reviews must genuinely come from customers, they must be visible on the page the markup sits on, and they must be about the product on that page.
The commonest violation is the self-serving rating. A review app injects the store's overall rating into every page, or the homepage carries an aggregate rating for the business itself. Google's review snippet guidelines rule that ineligible: an organisation can't rate itself, and a store-wide score isn't a rating of the kettle. At best the markup is ignored. At worst the page earns a manual action for structured data that misrepresents what's on it.
Stars on free listings and Shopping ads come from a different place. Google's Merchant Center help documents a product ratings programme, fed by a reviews file you submit or by an approved review aggregator. Page markup doesn't feed it and it doesn't feed page markup, so a store that wants stars in both places needs both. Product schema and rich results covers the page side in detail.
Collect reviews the way Google and the law allow
The mechanics are simple and most stores skip them. Ask every buyer, after delivery rather than after purchase, with a link that lands on that product's review form. Keep the form short. Show the reviews on the product page itself, in the page's HTML, so Google and an assistant's crawler read them as part of the page rather than as a widget that loads after a click.
Two lines you don't cross. Google's review policies prohibit fake reviews and reviews that are paid for, and in the US the Federal Trade Commission's rule on fake reviews makes buying or faking them illegal. Keep the negative ones too. A page showing only perfect scores reads as curated to a shopper and, in observed behaviour, to an assistant summarising sentiment, while a page with a spread of honest scores and a reply from the store under the critical ones reads as trustworthy.
Reply to the critical reviews in public. The reply is content on your page that answers the exact objection the next buyer has.
Customer content gives pages the words shoppers use
User-generated content (UGC) is anything a customer adds: reviews, questions and answers, photos, fit comments. Its search value is in the language. You'd write "stainless steel body". A customer writes "survived a week of camping in the rain and still looks new". That sentence matches searches and sub-questions no product copy would, and it's the kind of line an assistant quotes when it says what reviewers think.
Three things make UGC count. It lives on the product page, not on a review platform's own domain, where the content builds their site rather than yours. It's rendered in the HTML Google receives, which is worth checking on a review widget because some render only after scrolling or clicking. And it's moderated for spam without being edited for substance, because a curated set of customer voices is marketing in a different font.
Questions and answers deserve their own section on a product page. A shopper asking "does it fit a standard UK plug" and a staff answer is exactly the question-and-answer pair that Comparison and FAQ content teaches you to write, except a customer wrote it for you.
Brand signals are the web agreeing on who you are
For a Google rich result, the signal is markup on your page. For an assistant naming you, the observed pattern is different: the brands that get named are described consistently by several independent sources. A brand that exists only on its own website has nobody vouching for it.
The parts you control directly are few and worth doing. Google documents Organization structured data for stating your name, your logo and the other profiles that belong to you through the sameAs property. Put it on the homepage and keep the name identical everywhere. Use the same brand and product names in your feed, on your pages, in marketplace listings and on the manufacturer's site, down to the model number. Write an About page that says plainly what you make, for whom, and where you're based, because that's the page an assistant reads when it describes you.
The parts you earn are the ones that decide whether you get named: reviews on retailer sites, a mention in a roundup, a recommendation in a forum thread, a comparison article that includes you. Forum threads in particular show up often among the sources assistants cite for "which should I buy" questions. That's observed rather than documented, and it's consistent enough to act on. The next lesson puts these in priority order.
What to take away
- Google shows review stars from aggregate rating markup inside Product, and only for real, visible reviews of the product on that page; a store rating itself is ineligible.
- Stars on free listings and Shopping ads come from Merchant Center's product ratings programme, which is separate from page markup.
- Customer reviews, questions and photos add the language shoppers search with and assistants quote, provided they sit on your page and in its HTML.
- Being named by an assistant follows from independent sources describing your brand consistently, which you earn rather than mark up.
Next
Put the page, the consistency and the mentions in the right order in Being the product AI recommends.