ChatGPT Shopping: How to prepare your products for product search

When a question shows purchase intent, ChatGPT responds with product recommendations, complete with images, prices, reviews, and links to retailers. Ask “Which running shoes work for wide feet?” and you get a curated selection, not a list of links.

For retailers, the question is how to get into that selection. OpenAI has rebuilt the feature several times since its launch in spring 2025. This article covers the state of play as of October 2026: the available routes, the technical requirements, and what determines which products are shown.

Product results in ChatGPT are independent of ads

OpenAI is explicit: ChatGPT selects product results on its own. They are not ads and are not influenced by partnerships. No retailer can pay its way into these results.

Ads do exist now, but they are kept separate. Since August 2026, ChatGPT Ads have been running in 31 European markets, including Germany and Austria. They are shown only to users on the Free and Go plans, are clearly labeled, and, according to OpenAI, do not influence answers. For what this means for advertisers, see our article on advertising in ChatGPT and our ChatGPT Ads page.

Two ways to get your products into ChatGPT

Crawling your website

The OAI-SearchBot crawls publicly accessible websites for ChatGPT search. As long as it is not blocked in your robots.txt and your product pages are reachable without a login, your products can appear whenever they match a query. According to OpenAI, a feed is not strictly required.

Product feed

A feed lets you control how your products are presented and keep prices, availability, and details current. OpenAI cites broader catalog coverage, more complete product details, and better placement when shoppers compare options. Common formats and delivery methods are supported, including SFTP, APIs, leading e-commerce platforms, and feed providers.

If you sell on Shopify or Etsy, you are already connected: Shopify catalogs flow in through Shopify Catalog, with no separate application needed. All other retailers apply via OpenAI's merchant page and join a waitlist. OpenAI has announced a self-service onboarding platform for this year.

Apps for larger retailers

Retailers can also offer their own apps in ChatGPT. OpenAI describes these as optional and best suited to larger retailers that want more control over the shopping experience. Most retailers start with the feed.

Purchases happen on your website

In fall 2025, OpenAI introduced in-chat purchasing with the Agentic Commerce Protocol and Instant Checkout. It is now moving away from that model: rather than a standalone checkout inside ChatGPT, the company is focusing on better product search and the retailer's own checkout. Users discover and compare products in ChatGPT, then complete the purchase on the retailer's website or app. Purchases that start in ChatGPT carry no fees.

For you, this means the customer relationship, payment processing, and data stay with you. That makes it all the more important that your landing page delivers on the recommendation, and that you can identify ChatGPT traffic in your analytics. For more on the protocol, read our article on ACP, Shopify, and agentic commerce.

One limitation remains: the shopping feature is currently available only to users in the US. OpenAI plans to roll it out gradually to more retailers and regions. If you sell in German-speaking markets, use this time to get your website and feed ready.

ChatGPT combines product data from many sources

According to OpenAI, ChatGPT draws on the following when selecting products:

• structured metadata from retailers and third-party providers, such as price and product description, along with other third-party content,
• its own model responses generated before the new search,
• OpenAI's safety standards and product policies.

When several retailers sell the same product, ChatGPT shows a merchant list, ranked by availability, price, quality, and whether the retailer is the manufacturer or primary seller. The lowest-priced known option may carry a “Best price” label.

The takeaway: there is no single platform you must be listed on. But a product that appears in several places with consistent data and plenty of reviews gives ChatGPT more signals than one that lives on a single page.

Four technical foundations that make your products readable

Allow crawler access

Check your robots.txt. OpenAI uses separate crawlers for separate purposes: OAI-SearchBot for search, GPTBot for model training. To opt out of training but still appear in search, block GPTBot and allow OAI-SearchBot.

Add structured product data

Mark up your product pages with schema.org, especially Product, Offer and AggregateRating. This lets crawlers read title, price, availability, images, and reviews without ambiguity. Many stores already have this markup in place for Google, and it helps ChatGPT and its background data providers too.

Keep your feed up to date

According to OpenAI, price and shipping changes reach ChatGPT with a delay. A regularly refreshed feed shortens that lag. If you already maintain a clean feed for Google Shopping, most of the work is done: product ID, title, description, price, currency, availability, URL, image, and category. To see how we prepare product data, visit our feed management page.

Publish content around the product

Buying guides, comparisons, and how-to content on your website can be cited as sources in answers. They, too, need to be crawlable and clearly structured.

SEO, brand, and reviews drive the recommendation

ChatGPT does not follow a public ranking formula. It decides based on how well a product matches the user's intent, and three factors feed into that.

SEO lays the foundation

Only what gets found and understood can be recommended. A clearly structured website, descriptive URLs, meaningful titles, and structured product data all improve discoverability. A product with incomplete information struggles in ChatGPT just as it does in Google Search. For how language models choose their sources, see our article on Generative Engine Optimization.

Well-known brands offer more signals

ChatGPT draws on whatever information about a product is available online. Established brands with a good reputation have more tests, mentions, and reviews. Brand is not a formal criterion, but products that are discussed often and favorably stand a better chance. Personalization adds another layer: if users opt in, ChatGPT factors in their memories and preferences, such as a preference for certain brands.

Reviews shape the selection

Reviews play a central role. ChatGPT displays star ratings, summarizes reviews from public websites, and assigns labels such as “Budget-friendly” or “Most popular”. OpenAI does not verify reviews itself, which makes a broad base of genuine reviews all the more important. Actively ask satisfied customers for reviews, and respond to criticism.

Complete product attributes answer specific questions

Users ask ChatGPT more specific questions than they type into a search engine. If someone mentions a budget, price takes priority; if they rule something out, it drops from the list. So describe your products with every relevant attribute, such as “suitable for allergy sufferers” or “for large dogs”. The closer a product matches the need, the more likely it is to be selected.

Conclusion: Get your product data in order now and be ready for the European launch

ChatGPT's shopping feature is a channel for pre-selection; the purchase stays with you. Technically accessible pages and complete, up-to-date product data create an important foundation for product search in ChatGPT.

The feature is still limited to the US, but preparing now pays off: the same foundations work in Google Search, Google Shopping, and other providers' AI answers. To see how we support retailers, visit our ChatGPT Shopping page.