On this page
- The short answer: export, remap, deliver, refresh
- What OpenAI documents today (and what it does not) for WooCommerce and BigCommerce
- Map WooCommerce and BigCommerce fields to the required feed columns
- Variants, prices and availability: the common mapping mistakes
- Delivery: file upload, hosted URL refresh or a feed provider
- Check the feed, then check whether ChatGPT shows your products
- FAQ: WooCommerce and BigCommerce catalogues in ChatGPT
- Build the feed once, then measure what ChatGPT does with it
- Related Business Hub guides
To get a WooCommerce or BigCommerce catalogue into ChatGPT, export your products, remap them to OpenAI's product feed schema, deliver the file by a documented method and refresh it whenever price or stock changes. A WooCommerce product feed for ChatGPT is not a button on your platform. It is a mapping job. The platform export is raw material, not a finished feed.
This guide covers that job in order: what OpenAI documents, which columns you must produce, where WooCommerce and BigCommerce exports go wrong, how to deliver the file and how to check the result. Facts about OpenAI's programme are as checked on 18 November 2026. Commerce in ChatGPT is still changing, so recheck the specs before you build.
The short answer: export, remap, deliver, refresh
Treat your WooCommerce or BigCommerce export as a source file, remap every row to OpenAI's feed schema, deliver it by a documented method and keep it fresh. That is the whole loop.
Export. Pull products and variations from the platform, one row per purchasable item.
Remap. Rename and reshape columns to the required fields:
item_id,title,description,url,brand,seller_name,image_url,availabilityandprice, plus the geo fieldstarget_countriesandstore_country.Deliver. Upload the file, host it at a URL that refreshes, or use a feed provider.
Refresh. Re-run the export and remap whenever prices or stock change.
OpenAI's own pointer is short: if you sell retail products, submit a merchant feed so ChatGPT can use your catalogue for product discovery (OpenAI Commerce). The rest of this guide is the detail behind those four steps. If you already have a CSV and want the column-level walkthrough, the ChatGPT Business Hub product feeds CSV setup guide covers it.
What OpenAI documents today (and what it does not) for WooCommerce and BigCommerce
OpenAI documents a product feed schema and file-based delivery. It does not document a native WooCommerce or BigCommerce connector.
What is documented: a feed reference with required fields, variant handling, price format, availability, merchant info, returns and geo tagging (OpenAI product feed schema). The same schema is published for file uploads (OpenAI file upload spec). Feeds can arrive by file upload, hosted URL or SFTP, and OpenAI also mentions integrations through commerce platforms and feed providers.
What is not documented: a first-party WooCommerce plugin or a BigCommerce channel that submits your catalogue. If you search for a WooCommerce ChatGPT Instant Checkout plugin or a BigCommerce agentic commerce integration, treat any claim of a native connector as unverified until OpenAI documents it. Third-party feed providers such as Feedonomics say they support both platforms, but that is a vendor service, not an OpenAI-built connector.
Two more points trip people up. First, the feed is for product discovery. Second, the standard upload currently targets the US.
Map WooCommerce and BigCommerce fields to the required feed columns
Build one mapping table and apply it to every row. The WooCommerce CSV and the BigCommerce export both need reshaping, because neither uses OpenAI's column names.
| Feed field | WooCommerce source | BigCommerce source |
|---|---|---|
item_id |
SKU or variation ID, stable and unique | SKU or variant ID, stable and unique |
title |
Product name | Product name |
description |
Cleaned plain-text description | Cleaned plain-text description |
url |
Product permalink | Product URL |
image_url |
Main image | Primary product image |
brand |
Brand attribute or taxonomy | Brand field |
seller_name |
Your store name | Your store name |
availability |
Stock status mapped to in_stock, out_of_stock or backorder |
Inventory level mapped the same way |
price |
Regular price plus currency, e.g. 79.99 USD |
Price plus currency |
group_id |
Parent variable product's stable ID | Parent product's stable ID |
target_countries, store_country |
Set per store | Set per store |
Two rules from the schema matter most. Submit one row per purchasable item or variant. And write money as an amount, a space and an uppercase ISO 4217 currency code, for example 79.99 USD, a decimal amount in major units (OpenAI product feed schema).
Stores most often leave brand blank, because neither platform guarantees it is filled in. Decide where your brand value lives (an attribute, a taxonomy or a custom field) before you export, or the column will arrive empty.
Variants, prices and availability: the common mapping mistakes
Most rejected rows come from variants, formatting and stale stock, not from missing titles. Check these first.
Exporting only the variable parent. A WooCommerce variable product exports a parent row plus one row per variation. ChatGPT needs one row per purchasable variation, each with its own item_id, price and availability. Exporting just the parent hides every size and colour.
Reusing one item_id across variants. The schema defines item_id as a stable ID unique per purchasable item or variant (OpenAI product feed schema). Give each variation a unique item_id and share the parent's group_id.
Passing WooCommerce's Parent column through as group_id. That column is a WooCommerce relationship field, not a feed field. Convert it to a stable ID and put it in group_id.
Dropping the variant options. Each variation's attribute values (for example colour and size) belong in the variant fields, and top-level attributes should stay consistent with them. Flag the listing as having variations.
Bad price formatting. Commas as decimal separators, thousands separators or a missing currency code all break the format. Output 1299.00 USD, not 1.299,00 or $1,299.
Sale prices in the wrong place. Submit the current regular price as price and an active sale price in sale_price, rather than overwriting the regular price.
Stale availability. Availability and price change daily. OpenAI asks for current stock status on each row, so a feed built once and forgotten drifts out of date fast.
Delivery: file upload, hosted URL refresh or a feed provider
Pick the delivery method by how often your catalogue changes, not by how easy the first upload feels. All three routes below are documented.
| Method | Best for | Catch |
|---|---|---|
| File upload | A first test, or a small catalogue that rarely changes | You must re-upload by hand after every price or stock change |
| Hosted URL refresh | Most WooCommerce and BigCommerce stores | You need a place to host a regularly regenerated file |
| Feed provider | Larger catalogues, or teams without developer time | It is a paid third-party service, and you still own the mapping decisions |
For a store with changing stock, hosted URL refresh is the practical default. A scheduled job exports from the platform, runs your remap, writes the file and serves it at a stable address. SFTP is the other documented transport if you would rather push files than host them.
Whichever you choose, keep item_id values stable between runs. If IDs change every export, ChatGPT sees new products each time instead of updates to the old ones. If you manage client stores, AI search reporting for agencies covers what to show them once the feed is live.
Check the feed, then check whether ChatGPT shows your products
Validate the file against the schema before you submit, then watch for your products in ChatGPT answers over the following days. A feed that uploads is not the same as a catalogue that appears.
Before submitting, run this checklist:
All nine required fields plus
target_countriesandstore_countryare present on every row.Every variation has its own
item_id, price and availability.Every price matches the
amount CURRENCYformat.Every
urlandimage_urlresolves and points to the right product.No row is a variable parent standing in for its variations.
After submitting, test with shopping-style prompts a buyer would use: a product type plus a use case, a brand plus a category, a comparison between two items you stock. Note which ones return your products, and rerun them after each feed refresh. Appearing is never guaranteed, so treat the results as a measurement, not a pass or fail.
This is where Revnu fits. Revnu is an agent that finds customers and grows revenue across every channel, and it tracks how a business appears in ChatGPT, Claude, Perplexity and Gemini answers and in Google. What it adds is tracking how your business appears, with the results reported in Slack. Pro is $99 a month, self-serve. For the reporting side, see ChatGPT Business Hub metrics and weekly reporting.
FAQ: WooCommerce and BigCommerce catalogues in ChatGPT
The FAQ below answers the questions merchants ask most. For the wider picture of who can use these tools, read ChatGPT Business Hub: what it is and who can use it.
Build the feed once, then measure what ChatGPT does with it
The catalogue work is the same on both platforms: export, remap, deliver, refresh. The stores that do it well treat the feed as a living file, not a one-off upload. Once yours is live, the open question is whether ChatGPT actually surfaces your products. Connect your website to Revnu on the Pro plan to track how your business appears in AI answers, and use that to decide what to fix next.
Related Business Hub guides
Written by
Co-founder and CEO, Revnu
