Is your store ready for AI shoppers?
When a customer asks an assistant to find or buy a product, the assistant reads the structured data on the page. We checked 240 product pages for what an agent needs to recommend a product with confidence, what each store offers agents directly, and what it tells AI crawlers in robots.txt.
7IT SolutionsPublished
- 50.8%give a product page complete machine-readable data
- 40.8%publish a barcode or manufacturer number
- 69.7%expose a commerce endpoint for AI agents (UCP)
What a product page tells an agent
Share of product pages that publish each field in structured data.
| Product described in structured data | 83.8% | |
| Price | 81.3% | |
| Currency | 80.4% | |
| In stock or not | 78.8% | |
| Brand | 59.6% | |
| SKU | 69.2% | |
| Barcode or manufacturer number (GTIN/MPN) | 40.8% | |
| Ratings or reviews | 14.6% | |
| Shipping or return policy | 14.2% |
What a store offers agents directly
Share of stores that answered. Shopify now generates these for its stores, so most of the coverage comes from the platform, not from the store owner.
| Commerce endpoint for agents (UCP) | 69.7% | |
| Instructions for agents (agents.md) | 63.5% | |
| llms.txt | 71.7% | |
| Business described in structured data | 68.4% |
What stores tell AI search crawlers
Share of stores (with a robots.txt) that block each assistant’s search crawler. Blocking it keeps the store out of that assistant’s answers.
| OAI-SearchBot | 0% | |
| ChatGPT-User | 0% | |
| PerplexityBot | 0% | |
| Claude-SearchBot | 0% | |
| Claude-User | 0% |
By platform
Share of product pages with complete data, by platform.
| Shopify (219) | 53.4% |
Method. Public pages only, fetched from the US by an identified crawler (7IT-Research) that respects robots.txt and does not work around bot protection: robots.txt, llms.txt, agents.md, the UCP discovery file, the home page and one product page per store (the first product a Shopify store lists, otherwise the first product link on the home page). Structured data read from JSON-LD. Aggregates only; no store or app is named. Related research: all studies.
An assistant will recommend the product it can describe accurately: name, price, currency, stock, brand and an identifier. A page missing those is passed over for one that has them, even when the product is better. None of this is visible to a shopper, and all of it is a one-time fix in the page template.