AI Shopping Runs on Intent Clusters, Not Keywords. Your Product Data Decides If You Exist.

Practical Ecommerce's new piece on intent clusters lands a point most retailers have not absorbed yet: AI shopping surfaces do not match keywords, they map intent. Models embed products and shopper queries in the same mathematical space and cluster them by what the buyer is actually trying to do.

The mechanics matter. A query like a quiet coffee machine for a small office gets placed near products whose data expresses those attributes, even when no keyword matches. Discovery starts with an answer rather than a list of links. The model assembles a recommendation set, and visibility is curated: you are in the set or you do not exist. There is no page two of an AI answer.

The commercial scale is already visible. Salesforce's holiday analysis found AI influenced a fifth of global online sales in the 2025 peak season, and retailers running their own shopper agents grew sales meaningfully faster than those without.

Australian retailers with thin product data, meaning a title, a price and a one-line description, are increasingly invisible to intent-based retrieval. The model cannot place a product in the right cluster if the data never describes the use case, the setting, the compatibility or the buyer it suits. The optimisation surface has moved from the search results page into your own catalogue.

That is uncomfortable news for anyone who outsourced product copy to the intern years ago, and good news for retailers willing to do the unglamorous data work their competitors will not.

Assistants will assemble more of the shortlist every quarter. The retailers who win will be the ones whose products are describable by a machine, because describable is the new discoverable.