Perplexity groups citations by source type, making product pages vs. reviews explicit

Perplexity has been rolling out clearer citation panels that distinguish which sources an answer was built from, separating product pages from reviews, comparison content, and forum threads. The shift is small in interface terms but meaningful for merchants: the way a source gets labeled shapes whether shoppers click through to it — and which kind of content earns that click.

The change updates a long-running complaint about AI answers. Earlier citations were often a flat list, so a store that supplied a fact sat next to a forum post debating it, with no way to tell them apart. Grouping sources by role makes the underlying system more legible: a shopper can now see whether an answer leaned on a manufacturer listing, an independent review, or a seller's own product page.

If AI engines are beginning to tag sources by type, that changes the economics of building content. A merchant's own product page competing against a third-party review now does so in a visible category, not an anonymous list. That raises the bar for first-party content: the product page has to be good enough to be labeled a reliable primary source, because a reviewer who is explicitly separated from sellers still carries extra credibility.

It also makes the consequences of thin pages public. A page that offers little extractable detail is likely to be grouped under seller content regardless — and shoppers, seeing the label, may discount it before even reading. Clarity cuts both ways. In practice this nudges merchants toward the same fix as everything else in AI shopping: make the page carry verifiable, specific facts.

Citation labels won't rescue a product page that doesn't contain the facts an answer can use. Whether a source is tagged as a seller or as a reviewer, the engine still needs concrete detail — price, stock, shipping, and a stated purpose — before it can cite it at all. That part has not changed. What has changed is how visible the distinction is, which makes the marginal value of a well-structured product page higher than before. The free scan at geovisora.com/audit shows which of those fields a model can actually extract from your URLs today, so you can close the gap before a shopper's tab divides your page from a review.