Perplexity Shopping: How Product Recommendations Get Built

Where the recommendations come from

Perplexity always searches, and it cites what it finds. For product questions, it tends to pull from three kinds of sources: retailer and brand product pages for specs and price, editorial review sites for the "best of" framing, and community threads, especially Reddit, for the "people actually like this" signal. The answer is a synthesis of all three. The product cards that appear are usually assembled from merchant data where Perplexity has it, and from the product page otherwise.

In our testing, a product that appears in two of those three source types tends to make the list. A product that only appears on its own site rarely does, because there is nothing independent confirming it belongs there.

The product page still matters

Perplexity needs to read your product page to get the facts right. That means the page has to be crawlable, the key specs have to be in the HTML rather than loaded by a script, and the price has to be visible. Product schema with price, availability, and rating gives it a clean set of fields to work from. A page that renders as an empty shell to a crawler gets described from whatever a third party said about it, which is sometimes wrong and often out of date.

Write the first paragraph as a plain statement: what the product is, who it is for, the one or two specs that matter most, and the price. The model tends to lift that paragraph nearly verbatim into the reason it gives for recommending you.

The independent sources do the convincing

The "best under $60" framing does not come from your site. It comes from review roundups and from people on forums saying they bought it and liked it. Perplexity tends to cite Reddit threads heavily for product questions, and it tends to cite established review sites for the comparison structure.

You cannot buy your way into those. What you can do is make sure the product exists in the places where the conversation happens. That means getting real reviews on the platforms that matter for your category, answering questions on community threads honestly, and giving reviewers accurate information to work with. Fake reviews and paid placements are not just against the rules, they tend to get discounted by models that have learned what astroturfing looks like.

Price and availability drift

A product answer is only as good as its price. Perplexity tends to show the price it found at crawl time, which may be stale. If your price changed last week and the cached page says otherwise, the answer is wrong and the user finds out at checkout. Keep schema and visible price in sync, and avoid pages where the real price only appears after a JavaScript call.

Availability matters the same way. A product listed as out of stock in schema tends to get dropped from recommendations.

What a shopping audit looks like

At AEO GEO Labs a shopping-focused audit starts with a prompt set of real buying questions in the category, run in Perplexity and the other engines, with the recommended products and cited sources recorded each week. The pattern that comes out is usually clear: the brands that win are present on the product page, on one or two review sites, and in at least one community thread, with the same facts in all three places. The brands that lose are often invisible to the crawler or only present on their own domain.

Three things to check now

Run your top five buying questions in Perplexity and note which products appear and which sources are cited. Check your product pages with a crawler checker and a JavaScript visibility checker to confirm the specs and price are readable. Then look at whether your product appears in any of the independent sources that got cited. Each gap is a concrete task. If you want the baseline measured and tracked, AEO GEO Labs does this weekly for ecommerce clients.