For a shopping assistant to consider a product for a particular request, it needs reliable product and offer information. Start with the product page and keep variants, price and stock consistent with the feed and cart. The merchant documentation for each channel explains what data it accepts and how to send it.
What should the product page answer?
Take a real customer service question. For “Will this child seat fit my car?”, compatible models and years, the type of mounting and any exceptions are helpful. Then show the purchase details for the selected variant, especially price, availability and delivery terms. Clear public information makes these answers easier for both people and systems to use.
What do feeds and structured data do?
A feed sends your catalogue in the format a particular channel expects. Structured product data makes the information on the page machine-readable. For Google, merchants can use product markup on their site and Merchant Center. For ChatGPT, OpenAI’s commerce documentation describes catalogue integration and product data submission. Access and supported features depend on the programme and market.
How do I tell whether it works?
Pick five variants and compare the page, structured data, feed and cart. Include an out-of-stock item and a price change. Then check whether the channel accepted the data, whether the offer appeared for a specific query and whether it led to a shop visit. Those are three separate measurements.
Book chapters, articles and sources
- Chapter 5 · Being found before anyone visits the shop
- Chapter 9 · When purchasing moves between interfaces
- When product data ages: A field contract for shopping agents
- Stefan Hamann · Agentic Commerce Is Not Dead. The Discovery Layer Just Moved.
- OpenAI · Commerce documentation
- Google Search Central · Product structured data