Merchant feeds were built for Shopping ads. They are increasingly being read by something else entirely — AI assistants answering product questions directly, without the user ever reaching a product page.
Structured data is what machines trust
When an AI assistant recommends a product, it favours information it can verify. Structured product data — clear titles, specific attributes, accurate availability, real pricing — is easier to trust than marketing prose, so it gets weighted accordingly.
Feeds optimised purely for Shopping tend to be terse. Titles are keyword-stuffed, descriptions are truncated, and attributes are left blank because they were never required for the ad to serve. That is now a visibility problem.
Write attributes for a reader who cannot see the photo
Material, dimensions, finish, compatibility, coverage per pack, whether fitting is included — these are exactly the details that decide whether an assistant can confidently recommend a product for a stated need. If a shopper asks for matte porcelain suitable for a wet room, the model needs those attributes present as data, not implied by a photograph.
Consistency between feed and page matters
Where the feed and the product page disagree on price, availability or specification, models tend to discount both. Keeping them synchronised is basic hygiene that has quietly become a ranking factor of sorts.
The practical upside
Feed quality work pays twice. It improves Shopping and Performance Max relevance immediately, and it improves how well your catalogue can be surfaced in AI answers. Few optimisations serve both channels as directly.
