Two outcomes that look the same in a screenshot
An answer that says "the Arizona felt pennant from Oxford Pennant is a good choice" has mentioned your product. An answer that says the same and links oxfordpennant.com/…/arizona-pennant has sent the buyer to it. The first is awareness; the second is a visit. Reporting both as "visibility" hides the only difference that reaches revenue.
We record the two separately, and treat only the second as a success: the right product named and the seller's official URL present in the answer.
Three grades of destination
Even when the seller's URL is present, it is not always the product page. Over 2,185 graded answers between 04.08.2026 and 16.09.2026:
- the product's own page — 643;
- another page of the same site — 1,445;
- the bare domain — 97.
The study has the per-model breakdown.
What makes a page a usable destination
- One stable URL per product, without variant parameters or session tokens,
that returns the same page to everyone.
- Identity in the HTML: brand, model name, GTIN and MPN, the variant, readable
without running JavaScript.
- Facts in sentences: what it is, what it fits, what is in the box, what it
does not do. A model cites what it can quote.
- A declared canonical, so the many copies of your product across the web
point back to one address.
Why the retailer so often wins
In answers about real products, 37% of citations pointed at the maker's or seller's own site (study). For a brand selling through retailers, the retailer's page is frequently the only page that states price, availability and a buy button in one place, and the model follows that. Whether that is acceptable is a commercial decision; it is worth making knowingly rather than discovering it in someone else's report.
What this does not promise
That a model will link your page. The rules that keep "we published it" apart from "the answer used it" are at /evidence.
Sources
- Do models agree about the same product?
- Whose pages AI cites when it answers about a product
- Evidence methodology, version 1.0