Structured product data for agentic shopping
Stuart Kearney, co-founder of Vetted, on AI agents in shopping
Machine readable product pages are becoming a new distribution surface, not just a technical cleanup task. In agentic shopping, the merchant that exposes clean facts like exact SKU, price, stock, delivery window, and return terms gives the model something it can trust and compare. That shifts page design from persuading a human browser to feeding a recommendation engine that is deciding whether to route demand at all.
-
This is the same bottleneck Vetted describes in the handoff from research to purchase. The hard part is not finding a blender, it is knowing which merchant has the right variant, in stock, with fast shipping and acceptable returns. If those fields are missing or stale, an agent has to fall back to weaker guesses or avoid recommending the page.
-
The plumbing already exists, but adoption is uneven. Schema.org includes product offer, shipping, and return policy objects, and Google Merchant Center explicitly asks merchants for shipping costs, speeds, and return information because those details influence buying decisions. The change is that AI shopping systems now depend on the same data for ranking and execution, not just search snippets.
-
The closest comparable is product information software like Salsify, which helps brands keep product data synced across channels, and newer commerce infrastructure from players like Klarna and Forter that are building machine readable paths for agents. The winning stack increasingly looks like organized catalog data upstream, trusted recommendation in the middle, and checkout or fraud systems downstream.
Over the next few years, ecommerce teams will treat structured product data the way they once treated mobile design and SEO. The brands that keep merchant data accurate and continuously synced will be easier for assistants to rank, compare, and buy from, which means more traffic will flow to the cleanest catalogs rather than the loudest advertisers.