The three feed fields that decide whether ChatGPT can sell your products

Since the Agentic Commerce Protocol launched, Shopify catalogs flow into ChatGPT shopping answers automatically. No merchant setup, no opt-in. Which means your feed is already being read. The only question is whether it parses.

The requirements split into two tiers, and the split matters: **required fields gate whether you can appear at all (fail one and you're invisible), while recommended fields decide how you rank** (present but outranked). Three fields do most of the damage we see:

1. Product identifiers (GTIN/MPN). The #1 cross-platform lever. Google reports GTIN-tagged products get up to 40% more clicks. A malformed GTIN (wrong length, dashes, spaces) is worse than a missing one, because it can get the product disapproved outright. Valid GTINs are 8, 12, 13, or 14 digits, no punctuation.

2. Hyper-specific product types. "Footwear" tells an engine nothing. "Men's insulated winter boots" answers the exact question a shopper asks. LLMs match your product type against buying-intent language, and generic types lose to specific ones every time. This is also the field small stores most often leave on the default.

3. Literal descriptions. Marketing copy ("Elevate your winter game!") is unreadable to a model deciding what the product is. Plain, functional descriptions covering materials, use cases, fit, and temperature range are what gets parsed and matched.

The uncomfortable part: you can't see any of this from the Shopify admin. Your store looks fine to you and unreadable to the machine reading it on a shopper's behalf.

We built a free check for exactly this. It reads your public product feed and flags what would block or bury you in AI shopping answers, with the exact fix for each finding. No install, no signup. → Run the free report