How does Perplexity choose which stores to cite?

The mechanism is straightforward. A question comes in, the system issues one or more searches, fetches a set of candidate pages, and composes an answer with references to the sources it drew on. If your page was not fetched, or was fetched and returned something unusable, it cannot be cited no matter how good it is.

That puts a set of unglamorous technical conditions first. The page has to be reachable by the fetchers, which means not blocked in robots rules and not challenged by bot management at the edge. It has to return content quickly, because a retrieval step does not wait indefinitely. And the substance has to be present in the response rather than assembled by scripts afterwards.

Stores fail this more often than publishers do, because ecommerce pages carry heavy client-side rendering, region selectors, consent walls and price components that populate late. A product page whose price and availability only appear after JavaScript runs may be retrieved as a shell.

So the first audit is a fetch test: request your key pages the way a simple client would, and read what comes back. If the important facts are missing, that is the finding.

Specificity. An answer needs statements it can carry across: a measurement, a material, a compatibility, a price, a policy, a date. Pages built from adjectives give a summariser nothing to quote, while pages built from facts are easy to draw on.

Directness helps too. A page that answers the exact question near the top, in plain language, is easier to use than one that arrives at the answer after four sections of preamble. Headings that state conclusions rather than tease them do the same work.

Self-containment matters more than on a normal web page. A section that depends on the reader having read the previous one is harder to lift, so each section should stand alone.

Freshness is a factor for anything with a price or a policy attached. A clearly stated last reviewed date, and content that is genuinely current, both help a system decide the page is safe to rely on.

And the page should be the right type for the question. A product page answers what is this and can I buy it. A guide answers which one should I choose. Asking a product page to serve a comparison question is a mismatch that shows up as a competitor's roundup being cited instead.

For product questions, the sources are rarely only retailer pages. Roundups, review sites, forums, marketplaces and manufacturer pages all get retrieved, and they may be describing your products with information you did not supply.

That has two consequences. The first is accuracy: if a directory lists a discontinued price or a review site has your shipping policy wrong, that inaccuracy can end up in an answer. Auditing what the major third-party sources say about you, and correcting them at source, is unglamorous work with a direct effect.

The second is inclusion. Being present and correctly described in the places that get cited for your category matters as much as your own pages, which is an argument for the supplier listings, association directories and specialist coverage that also serve ordinary search.

Perplexity has also published arrangements for merchants to supply product information for its shopping experiences, and the shape of those programs has changed more than once. Check the current merchant documentation directly rather than acting on a description written some months ago, and evaluate it as you would any other channel: what data it requires, what control it gives you, and what it reports back.

Four things are within your control. Allow the fetchers you want to be read by, remembering that the crawler and the user-triggered agent are separate identities and that blocking one does not imply the other. Confirm at the CDN or firewall as well as in robots rules, since bot management defaults often block them silently.

Make the facts retrievable in the HTML, including price, availability, specifications and policies. Keep your product data consistent between your pages, your feeds and the marketplaces that carry you, so whatever source is used says the same thing.

Publish the comparison and guidance content that answers the question shapes people actually ask, rather than expecting a product page to serve every purpose.

Then check, on a schedule. Run your category's real questions from your selling market, record which sources are cited and what is said about your products, and keep the log. Pair that with server log filtering on the fetcher user agents to confirm your pages are being requested at all. Those two together are as close to measurement as this gets, and neither supports promising a citation to anyone.