Learn · AEO Index
How AI shopping agents decide what to show
When an agent shops, it isn’t browsing your website. It is calling a structured catalog endpoint, reading product data, and ranking it against the shopper’s phrasing. Understanding that pipeline is the difference between guessing at AEO and engineering it.
1. The agent queries a catalog, not a webpage
Modern Shopify stores expose a machine-readable catalog (the Universal Commerce Protocol endpoint) that agents can search directly. The agent sends the shopper’s request as a query and gets back a ranked list of products with titles, prices and attributes. If a store doesn’t expose that endpoint, it is simply not in the running — the agent has nothing to read. Check whether your store exposes one →
2. Lexical match does the heavy lifting
For specific queries ("merino tank," "cutting mat," "ankle socks"), the strongest signal is whether the words a shopper used appear in your product titles. Stores that win a query overwhelmingly converge on the same vocabulary — we can watch it happen. On every category page the "language that wins" section shows the exact terms that ranking stores share, ranked by how many independent shops use each.
3. Intent and values are the blind spot
For abstract requests ("something sustainable," "a gift for a new dad," "non-toxic"), pure word-matching breaks down — a product can be perfectly sustainable and never use the word. This is the frontier of AEO: agents increasingly weigh meaning, not just tokens, and most catalogs aren’t written for it. Stores that encode intent and values into their product data win queries their competitors can’t even see.
4. Then it ranks the field
Once the agent has candidates, it orders them — by relevance, and increasingly by price, availability and rating. We track that ordering per query across the catalog: who appears, who wins the #1 slot, and how the field shifts day to day. Browse who’s rising in AI search, or open any query to see the live agent result for it.