The OpenAI product feed spec, explained for Shopify merchants
Last reviewed
Short answer
The OpenAI product feed spec is the published format for product data entering ChatGPT shopping. Required fields: item_id, title, description, brand, url, image_url, price, availability and seller_name, plus eligibility flags. Titles cap at 150 characters, descriptions at 5,000. Shopify merchants do not submit this feed themselves, Shopify Catalog does, which makes the spec your quality rubric rather than an integration task.
Why a Shopify merchant should read a feed spec at all
Because it is the closest thing AI shopping has to published ranking rules, and your store is already being judged against it. OpenAI states that for merchants on Shopify, product data is integrated into ChatGPT through Shopify Catalog with no additional work required. The pipe is automatic. The quality of what flows through it is yours.
One honesty note before the fields: the spec page publishes no version date, so everything below is as read on 29 August 2026, and the source link at the foot of this page outranks this page if they ever disagree.
The required fields, mapped to your admin
Nine fields are required, and every one maps to something you already have in Shopify. The mapping is the useful part, because it turns the spec into a checklist you can act on without leaving the product page.
| Spec field | Rule | Where it lives in Shopify |
|---|---|---|
| item_id | Unique, stable identifier | Handled by Shopify, variant ID |
| title | Max 150 characters, avoid all-caps | Product title |
| description | Max 5,000 characters, plain text | Product description |
| brand | Max 70 characters | Vendor field |
| url | Working product URL | Handled by Shopify |
| image_url | JPEG or PNG, HTTPS preferred | First product image |
| price | Price with ISO currency | Price field |
| availability | Current stock state | Inventory tracking |
| seller_name | Max 70 characters | Your store name |
The flags that decide where you can appear
Three boolean flags govern eligibility, and the names matter because older write-ups circulate wrong ones. The current spec uses is_eligible_search for search visibility, is_eligible_checkout for direct purchase, and is_ads_eligible for ads. The spec states is_eligible_search must be true for is_eligible_checkout to be enabled: no search presence, no checkout.
Checkout eligibility also drags in trust fields. With is_eligible_checkout set, the spec requires seller_privacy_policy and seller_tos URLs. That is the spec-level echo of what Shopify's own eligibility list asks for: completed policy pages are not bureaucracy, they are a checkout gate.
The optional fields that do the ranking work
OpenAI's key-concepts guidance says recommended attributes, like rich media, reviews, and performance signals, improve ranking, relevance, and user trust. The optional field list is where those live: gtin and mpn identifiers, sale_price, variant attributes, reviews, Q&A content, return_policy and target_countries.
Third-party research points the same way. Profound's June 2026 analysis of roughly a million offers found feed-sourced offers dominating the top position, with reviews and Q&A on product pages correlating with higher ranking. Correlation, not spec, and labeled as such. But when the spec says reviews improve ranking and the research finds reviews correlating with rank, betting against reviews is the strange position.
What to actually do with this
Read your own catalog the way the feed does. Titles over 150 characters get cut, all-caps titles violate the stated guidance, empty descriptions ship as empty, a missing vendor field ships as a missing brand, and untracked inventory ships as unknown availability.
That audit is mechanical, which is the point: it is checkable by rules, not vibes. It is also exactly what Entitled's free score automates across the whole catalog, one grade per product, against these fields. Do it by hand for ten products or let the app do all of them, but either way the spec is the rubric.
Frequently asked questions
Do I need to submit a feed to OpenAI myself?
No. OpenAI states Shopify product data is integrated automatically through Shopify Catalog, with no additional work from individual merchants. Direct feed submission exists for platforms and large retailers and requires applying for access.
What are the exact eligibility field names?
is_eligible_search, is_eligible_checkout and is_ads_eligible, as read from the current spec. Posts referencing enable_search or enable_checkout are describing an older shape of the spec, which is a reason to check the source link rather than any secondary write-up, including this one.
Does meeting the spec guarantee my products appear?
No. The spec governs what can be ingested and ranked, and OpenAI states not all available products will be shown. Meeting it removes the failures you control. It cannot buy placement, and neither can anything else.
My titles are longer than 150 characters. Does it matter?
The spec sets 150 as the maximum and warns against all-caps. A longer title gets truncated somewhere you do not control, and ChatGPT may rewrite titles regardless. Front-load brand, product type and the attributes a buyer would filter on.
Sources
Keep reading
Score your own catalog
Entitled grades every product against the rules on this page, explains each failure, and fixes them with AI you approve. The score is free forever.