R3 — One indexable page per record (distinct URL per item / variant) — per Google's merchant-listing guidance

Rule: Every record in a structured catalogue gets its own distinct, crawlable URL — one page per item, or per variant where variants differ on a queryable attribute. Lumping multiple records onto a single "products page" disqualifies them from product rich results and forfeits the per-record findability and AI-citation surface.

Why: Direct primary-source warrant from Google: "Product rich results only support pages that focus on a single product (or multiple variants of the same product)... This includes product variants where each product variant has a distinct URL" (Google Search Central — product rich results require "a distinct URL" per product (or per variant); confirms one findable page per record). schema.org ItemList / ListItem / Product types (schema.org — ItemList / ListItem / Product / Offer types exist precisely to mark up individual records and lists as machine-readable structured data) assume each list item carries its own URL. The mechanism generalises beyond products — any record worth filtering on is a record worth its own findable URL (Mechanism summary — structured catalogues expose attributes as data; prose/PDF/images lock them in a format no filter can reach).

How to apply:

• Default catalogue architecture: list page + detail page per record. Do not crash records together into "category pages with everything inline."
• Where variants differ on a queryable attribute, each variant gets its own URL.
• The same discipline supports AI-citation eligibility (AI-citation eligibility via structured records — DIRECTIONAL; vendor-blog driven, Google's own position is "not a direct ranking factor") — extractable records are more cite-able than locked PDFs/images (USPTO Patent 8,700,594 — structured data is "searchable by data type"; unstructured data (bitmaps, audio, text docs) is not).