
ChatGPT product feed for Shopify
When ChatGPT shows products, part of the catalog data comes from feeds that merchants submit to OpenAI. Feedflock turns your Shopify catalog into an OpenAI-format feed: every required field, structured variants, prices with currency. You paste the feed URL into ChatGPT Business Hub and OpenAI fetches it from there.

Two yes/no settings decide where ChatGPT shows your products.
Every row carries is_eligible_search and is_ads_eligible as lowercase true/false: the first puts the product into ChatGPT's shopping answers, the second into sponsored placements. The flags are set per product in Feedflock. The feed simply reports them.
Catch feed issues before ChatGPT does.
Before every generation Feedflock checks that every field the channel requires is filled, and, on templates verified against the channel's own spec, that the values fit its rules. Most problems are fixed with one click, without editing products in Shopify.
Setup
ChatGPT feed live in five steps
From a bare Shopify store to a listed offer: the whole setup fits into one coffee break.
Claim your business in ChatGPT
Start in ChatGPT Business Hub (chatgpt.com/business-hub), claim your business and verify your identity in the products section. Then choose the countries and currencies you sell in. Standard uploads target the US for now.
Install Feedflock
Add Feedflock from the Shopify App Store. It reads your products, variants, stock and images directly through the Shopify API.
Pick the ChatGPT Shopping template
Create a new feed and choose ChatGPT Shopping. Every required field is mapped for you, seller_name included: it comes from your Shopify store name, and you can override it if your trading name differs.
Check variants and identifiers
Variants go out as one row each, with a shared group_id and a variant_dict of the selected options. gtin comes from the Shopify barcode, and a code that isn't a valid GTIN length is left out. mpn reads a custom.mpn metafield, so it stays empty unless you keep the manufacturer's part numbers.
Connect the feed URL
In Business Hub, choose a hosted URL as the delivery method and paste the feed's address. Feedflock regenerates the file every hour at that same URL. To put only some products into ChatGPT ads, tag them chatgpt-ads in Shopify.
https://feedflock.app/f/demo/chatgpt/chatgpt.csv
The feed
Sample feed: what ChatGPT receives
"is_eligible_search" "is_ads_eligible" "item_id" "group_id" "listing_has_variations" "title" "description" "url" "image_url" "price" "availability" "gtin" "brand" "variant_dict" "seller_name" "target_countries" "true" "false" "SKU-2048-M" "8812345678901" "true" "Linen shirt Sand - M" "Breathable linen shirt with a relaxed regular fit." "https://shop.example/products/linen-shirt-sand?variant=46012345678901" "https://cdn.shop.example/img/linen-shirt-sand.jpg" "59.00 USD" "in_stock" "8594000123453" "Nordlin" "{ ""Size"": ""M"" }" "Shop Example" "US"

Nordlin
Linen shirt Sand
$59
Feed fields
is_eligible_search
true
is_ads_eligible
false
item_id
SKU-2048-M
group_id
8812345678901
listing_has_variations
true
title
Linen shirt Sand - M
description
Breathable linen shirt with a relaxed regular fit.
url
https://shop.example/products/linen-shirt-sand?variant=46012345678901
price
59.00 USD
availability
in_stock
gtin
8594000123453
brand
Nordlin
variant_dict
{ "Size": "M" }
seller_name
Shop Example
target_countries
US
The channel
Shopping questions, answered
Products surface inside the conversation, and what ChatGPT can say about them comes from your feed: titles, descriptions, variants, prices and stock. Factual descriptions that answer a shopper's questions are what the model reads.
Discovery and ads from one feed
The same product data powers organic discovery and ChatGPT ads. is_ads_eligible opts a product into Ads processing independently of search, so you can advertise some products and keep the rest organic.
No category tree to fight
OpenAI names no taxonomy: product_category is your own path, broad to specific. Nothing gets rejected for being off-tree, and Feedflock can still send Google's path if you have it mapped.
The feed
- The ChatGPT Shopping template maps every required field: the nine basics (item_id, title, description, url, brand, seller_name, image_url, availability, price) plus target_countries and store_country. seller_name comes from your Shopify store name
- OpenAI's canonical column names, one row per variant, UTF-8, at one stable URL that OpenAI fetches
- Prices go out as amount + space + ISO 4217, always with two decimals (59.00 USD). A compare-at price becomes price plus sale_price, and a sale that isn't strictly lower is left out
- Availability from Shopify's stock status as in_stock or out_of_stock, never a Google spelling: the one field whose bad value rejects the row
- Variants ship as one row each, with a shared group_id, listing_has_variations and a variant_dict of the selected options; url opens the product page with that variant selected
- Ads eligibility per product: is_ads_eligible is false by default and switches on for the whole feed or for products tagged chatgpt-ads
- Every generation is checked against OpenAI's spec (enum values, money format, GTIN lengths, title and description limits), and empty required columns are listed before OpenAI reads the file
- Regenerated every hour at a stable URL, so every fetch by OpenAI reads a current full snapshot
Deep dive
Required fields, one row per variant
- Required: item_id, title, description, url, brand, seller_name, image_url, availability, price, target_countries, store_country
- Titles at most 150 characters, descriptions plain text up to 5,000
CSV,TSVorJSONLinUTF-8, with publicHTTPSlinks- No placeholder strings: an optional field you don't know is left empty
Technical details
The OpenAI product feed is one row per purchasable item or variant. Nine fields describe the product: item_id, title, description, url, brand, seller_name, image_url, availability and price. Two more, target_countries and store_country, say where it is sold. Titles stay at or under 150 characters and descriptions are plain text up to 5,000.
Write factual copy that answers the questions a shopper would ask, because that is what the model reads. Feedflock renders the tab-delimited form with OpenAI's canonical column names. Unknown optional values stay empty, since the spec rules out placeholders like n/a or unknown.
Identifiers and variants
- item_id is stable and unique: never reuse it for a different item
- Variants share a group_id, set listing_has_variations and name their options in variant_dict
GTINis exactly 8, 12, 13 or 14 digits with a valid check digit, leading zeros kept- mpn is the manufacturer's part number, sent together with brand
Technical details
An item_id is a promise: stable, unique per item or variant, and never reused for a different product. OpenAI matches products by item_id, not by file name, and keeps it through price, stock, title and image changes. Feedflock uses the variant SKU and falls back to the variant ID when a SKU is missing.
Variants of one product go out as separate rows sharing a group_id that differs from every item_id, with listing_has_variations set and a variant_dict naming the selected options, as serialized JSON in the cell, as the spec requires. Each row carries its own price, availability, URL and image. Where the manufacturer assigns a GTIN, send it: exactly 8, 12, 13 or 14 digits with a valid check digit, no spaces or dashes.
Ads and checkout in ChatGPT
- is_ads_eligible: true opts a product into Ads processing; false opts it out
- Ads eligibility is independent of search discovery: you control them separately
- Checkout inside ChatGPT is a separate integration; the template ships discovery-only
- Catalogs for Ads also arrive by
SFTP, set up in OpenAI's Ads Manager
Technical details
ChatGPT ads build on the same product feed. A row opts into Ads processing with is_ads_eligible: true, independently of search eligibility, so you can advertise a product without changing how it surfaces in organic answers, or keep it discoverable without advertising it. In Feedflock the flag is false by default; set it for the whole feed or per product, for example true only on products tagged chatgpt-ads in Shopify. You can set it in the app, or ask in a chat: the Feedflock MCP connector lets Claude or ChatGPT switch ads eligibility per product, always with a preview before anything changes.
Checkout inside ChatGPT needs its own integration with OpenAI; an eligibility flag alone does not complete it. The template therefore sends is_eligible_checkout as false, so buyers land on your store. If OpenAI enables checkout for you, it also needs seller_privacy_policy and seller_tos, which Feedflock can derive from your Shopify policy pages.
Delivery, prices and markets
- OpenAI fetches the feed from a hosted
URL; manual upload andSFTPalso work - Every delivery is a full snapshot of your catalog
- Retire an item right away with is_eligible_search: false
- Standard uploads target the US; Canada and Mexico need market setup
Technical details
In Business Hub you pick one of three delivery methods: a hosted HTTPS URL, a manual upload or SFTP. With a hosted URL, OpenAI fetches the file from Feedflock, and Feedflock regenerates it every hour, so each fetch reads current prices and stock. Every delivery is a full snapshot of the catalog.
To retire an item immediately, set is_eligible_search to false. Money is written as amount, space, ISO 4217 code (79.99 USD), and a sale_price counts only when it is positive, strictly below price and in the same currency. A standard upload targets the US; a currency or product URL on its own doesn't select another market.
ChatGPT feed: the takeaway
One stable URL that OpenAI fetches, an hourly refresh and a file that follows OpenAI's spec column by column. Your Shopify catalog stays the single source of truth, in discovery and in ads.
Claude · ChatGPT
Or set up your ChatGPT feed in chat
Connect Feedflock to Claude or ChatGPT and ask. It creates the feed as a draft, maps your categories to the tree ChatGPT uses and tells you what is still missing. You approve every change.
Create a feed for ChatGPT and map my categories.
I'll create it as a draft and show you each change before I make it.
FAQ
Prefer to set it up in a conversation? Feedflock's MCP connector lets Claude or ChatGPT create this feed as a draft and walks you through each change for approval. Set up your feeds in chat
Get on every marketplace that matters this afternoon.
Install free, connect your first feed in minutes, and let Feedflock keep it fresh and validated.