How to get your products into ChatGPT Shopping

Shoppers now ask ChatGPT what to buy, and it answers with products and links to the stores that sell them. Here is how that works and what an online store needs to show up.

Ask ChatGPT "what is a good waterproof hiking boot under $150?" and it no longer just describes boots. It shows products, with images, prices and the shops that sell them, and links you to buy. Google does the same in AI Mode, and Perplexity and Microsoft Copilot have their own versions. The industry has a name for it, agentic commerce, and a small store owner has a simpler question: how do my products get into those answers?

The good news is that the route in is mostly the work a well-run store should already be doing. This guide sticks to what OpenAI and the platforms have published, because this is a fast-moving area full of confident vendor claims.

How ChatGPT finds the products it shows

ChatGPT gets product information in two ways, and a store can use either or both:

1. 1By reading your website. ChatGPT search uses OpenAI's crawler, OAI-SearchBot, to read public pages. If your product pages are readable, they can be found and cited, like any other page.
2. 2From a product feed. OpenAI publishes a commerce specification for merchants to send a structured, regularly refreshed feed (CSV or JSON) with identifiers, descriptions, prices, stock levels, images and delivery options. Its documentation says the required fields are what make price and availability display correctly, and that extras like reviews and rich media improve ranking and relevance.

If you sell on Shopify, OpenAI's help centre says your product data already reaches ChatGPT through the Shopify Catalog, with no extra work for individual merchants. Everyone else, including stores on WooCommerce, relies on the crawler or applies to OpenAI for direct feed access.

The checklist for a small online store

1. Let the AI crawlers in

Open yourstore.com/robots.txt. If OAI-SearchBot is blocked, or everything is blocked under User-agent: *, ChatGPT search cannot read your shop. Security plugins and CDN "bot protection" settings sometimes block AI crawlers without asking. GPTBot, OpenAI's training crawler, is separate, so you can allow search and refuse training. Our post on getting recommended by ChatGPT explains the difference.

2. Make every product page answer the obvious questions

An assistant matching a product to a question needs facts it can read as text: what it is, what it is made of, sizes, compatibility, price, delivery time and return policy. A product page that is a photo and a "Buy now" button gives it nothing to work with. Write the descriptions as if a careful shopper asked you the questions in person, because that is roughly what is happening.

3. Add product structured data

Schema.org Product markup states the name, price, currency, availability, brand and reviews in a format machines cannot misread. WooCommerce outputs basic product markup by default, and SEO plugins extend it. Check a product page with Google's Rich Results Test and fix anything it flags. It also powers rich results in ordinary Google search, so it pays twice.

4. Keep price and stock accurate

If ChatGPT quotes $49 and your page says $59, or shows a product you sold out of last week, the shopper leaves annoyed and the platform learns your data is unreliable. Whatever feeds you publish, to Google Merchant Center, Shopify or OpenAI, should update automatically from the store, not by hand.

5. Be in Google Merchant Center too

Google's AI Mode shopping results draw on its Shopping Graph, which is fed by Merchant Center. Free product listings cost nothing, and the product data you prepare there (identifiers, images, prices, shipping) is the same work an OpenAI feed needs. If you only do one feed this year, start with Google's.

6. Make the landing fast and the checkout simple

A shopper who arrives from ChatGPT has already decided they are interested. Lose them to a slow page or a nine-step checkout and you did all the work for nothing. Test the product page, basket and checkout on a phone. Our pre-launch checklist for online stores covers what to check.

Why reviews matter more than ever

When an assistant chooses between two similar products, it leans on what others have said: reviews on your site, on marketplaces and on independent sites. OpenAI's own documentation lists reviews among the attributes that improve ranking. Ask every customer for one, reply to the critical ones, and display them as text on the product page, not only inside an image or a widget a crawler cannot read.

How to tell if it is working

• Search for your own products. Ask ChatGPT, Google AI Mode and Perplexity the questions your customers would ask, and note which stores appear.
• Look for AI referrals. Visits from chatgpt.com and perplexity.ai show up as referral traffic in Google Analytics 4. Our GA4 guide shows where to find them.
• Watch for UTM tags. Links from ChatGPT often carry utm_source=chatgpt.com, so these sales can be attributed in your store's reports.

What to be wary of

Nobody outside OpenAI or Google controls which products their assistants pick. Be careful with anyone promising guaranteed placement, selling a secret "AI feed" format, or generating hundreds of near-identical product pages to game the results. The real levers are the boring ones above, and they also make your store better for the shoppers who arrive from everywhere else.

Where Serverlys can help

And when shoppers land with questions the product page does not answer, ConvoAI puts an AI agent on your site that answers from your own catalogue and policies, day and night. We explain how it works here.