What a "ChatGPT Product Feed" Actually Means
There is no official ChatGPT product feed format in the way Google Shopping has a Merchant Centre feed. When people search for how to set up a ChatGPT product feed, they are usually asking one of two related but distinct things: how do they get ChatGPT to surface their products when someone asks for a buying recommendation, and how do they make their product data readable by AI systems in general. Both questions have practical answers, and they overlap more than you might think.
ChatGPT, Perplexity, and similar AI tools build their product answers from a combination of sources: their training data, live web browsing (where available), and structured signals embedded in web pages. A "product feed" in the AI sense is really about making your product information structured, crawlable, and trustworthy enough that AI systems choose to pull from your store rather than a competitor's.
So the goal here is not to upload a spreadsheet somewhere. It is to configure your store so that every product page communicates its data clearly, in a format AI can read and reproduce accurately.
Structured Data: The Foundation of AI-Readable Products
If you do one thing to make your products visible to ChatGPT and other AI search tools, it should be implementing Product schema markup using JSON-LD. This is the structured data format recommended by Google and read by virtually every major AI system that browses the web.
A proper Product schema block tells AI systems the name, description, price, availability, brand, SKU, reviews, and more, all in a machine-readable format that does not depend on the AI correctly parsing your page layout. Without it, an AI browsing your site has to guess what your product name is, what it costs, and whether it is in stock. With it, those answers are explicit.
The core fields that actually matter
A lot of schema guides tell you to include every possible property. In practice, the fields that drive AI citations and rich results are a shorter list:
- name: The exact product name, matching what appears on your page.
- description: A genuine, detailed description. Not a keyword-stuffed blurb. AI systems read this and quote from it.
- image: A high-quality, publicly accessible image URL.
- offers: Price, currency, availability, and URL. This is what tells AI whether the product is buyable right now.
- brand: Your brand name as an
OrganizationorBrandentity. - aggregateRating: If you have reviews, include average rating and review count. AI tools treat rated products as more credible.
- sku and gtin: These identifiers help AI systems match your product to known entities, which is significant for disambiguation.
The offers.availability field is one people forget. Use the full schema.org URL format: https://schema.org/InStock rather than just writing "In Stock". AI parsers handle the full URL reliably.
JSON-LD placement on product pages
Place your JSON-LD block inside a <script type="application/ld+json"> tag in the <head> or <body> of each product page. On Shopify, this goes into your product template (usually product.json or product.liquid). On WooCommerce, plugins like Schema Pro or Rank Math can generate this automatically, but verify the output with Google's Rich Results Test because auto-generated schema is often incomplete.
The key thing to check: is the schema dynamic? It should pull live data from your store's product database so that price and availability update automatically. Hard-coded schema that goes stale is worse than no schema, because AI systems may surface incorrect pricing and damage your credibility with potential buyers.
Building a Crawlable Product Catalogue AI Can Actually Read
Schema markup handles the structured layer, but AI systems also browse your pages and read prose. Your product descriptions need to be genuinely informative, specific, and written in natural language that answers real buying questions.
Think about what someone would ask ChatGPT before buying your type of product. "What is the best waterproof jacket under £150?" or "Which standing desk is good for tall people?" Your product descriptions and category pages should contain the kind of language that answers those questions, not just a list of bullet points with feature names.
Sitemaps and crawlability
AI tools that browse the web, including ChatGPT's browsing feature and Perplexity, rely on being able to find and index your pages. A clean XML sitemap that includes all product URLs, submitted to Bing Webmaster Tools and Google Search Console, gives these systems a reliable map of your catalogue.
Bing powers a significant portion of ChatGPT's browsing results, so Bing Webmaster Tools is not optional. If you want to know which of your pages are actually being cited by AI answers, Bing Webmaster Tools now has an AI performance report that shows exactly that.
Also check that your robots.txt is not blocking the crawlers that AI tools use. GPTBot (OpenAI's crawler) and PerplexityBot are the two to watch. Many stores unintentionally block them by using aggressive bot-blocking rules that were set up for spam prevention years ago.
Product page content depth
Thin product pages, just a title, a photo, and three bullet points, are unlikely to be cited by AI. AI systems favour pages that provide enough context to be useful in an answer. That means:
- At least 200 to 300 words of original product description per page.
- Clear answers to common questions (dimensions, materials, compatibility, care instructions).
- Genuine customer reviews displayed on-page and marked up with
Reviewschema. - Comparison information where relevant (how this product differs from your other variants or from common alternatives).
This is not about stuffing pages with text. It is about giving AI enough material to accurately represent your product in an answer.
Connecting Your Store to AI Shopping Channels
Beyond your own site, there are emerging channels that bring products directly into AI interfaces.
Bing Shopping and Microsoft Merchant Centre
Microsoft Merchant Centre feeds power product carousels in Bing and, by extension, some product results in Microsoft Copilot. If you are running a traditional Google Shopping feed, adapting it for Microsoft Merchant Centre is relatively low-effort and extends your product visibility into the Copilot ecosystem. The feed format is similar to Google's: a TSV or XML file with required fields like title, price, link, image link, availability, and brand.
Copilot is increasingly pulling product data from this feed for shopping-related queries, so this is a meaningful channel for e-commerce stores to be present in.
OpenAI's shopping features
OpenAI has been rolling out native shopping results inside ChatGPT for certain product queries, pulling data from trusted retail sources and schema-marked-up pages. As of mid-2025, these results prioritise pages with complete Product schema, strong review signals, and clear pricing. There is no separate feed submission process for this yet, but ensuring your schema is complete and your pages are indexed by GPTBot is the right preparation.
At FlinnSchema, we see this pattern consistently across client stores: the ones with well-structured product schema get cited in AI shopping answers; the ones without it simply do not appear, regardless of how good their products are. You can run a free AI visibility audit to see exactly where your product pages stand right now.
Keeping Your Product Data Fresh
AI systems that browse the web in real time are sensitive to freshness. A product page that has not been updated in 18 months, with a cached schema block showing a price from last year, is a reliability risk. AI tools learn to deprioritise sources that have given them inaccurate information before.
A few practical steps to keep things current:
- Set up automated schema regeneration that fires whenever product data changes in your database. On Shopify, this is manageable through theme liquid files or a schema app. On WooCommerce, it depends on your plugin but most reputable ones handle this.
- Use the
priceValidUntilfield in your offers schema if you run time-limited promotions. This tells parsers the price expiry date rather than leaving them to guess. - Periodically validate your schema with Google's Rich Results Test and Schema.org's validator, especially after theme updates or platform migrations.
There is also an argument for republishing or refreshing your most important product pages on a regular cycle, even when the product itself has not changed. How often to republish content to stay cited by AI covers the frequency and approach in detail.
Measuring Whether It Is Working
This is where most stores fall short. They implement schema and then have no idea whether it is actually driving AI citations or traffic. The measurement toolkit for AI visibility is still developing, but there are things you can do now.
First, set up UTM tracking and referral monitoring in Google Analytics to capture visits from ChatGPT and Perplexity. These sessions show up under referral traffic from chat.openai.com and perplexity.ai. The volume may surprise you, either positively or negatively.
Second, run manual tests. Search for your product categories in ChatGPT and Perplexity and see whether your store appears. Try different phrasings: "best [product type] for [use case]", "where to buy [product name] in the UK", "reviews of [brand name]". Record your results and repeat them monthly to track improvement over time.
Third, check Bing Webmaster Tools. If ChatGPT is browsing Bing's index to find product answers, your Bing performance is a proxy for your ChatGPT product visibility. Page impressions, click-through rate, and the AI performance report give you real data to work with.
Frequently Asked Questions
Does ChatGPT have an official product feed submission process?
Not yet, at least not in the way Google Merchant Centre works. ChatGPT's product answers currently come from web browsing, training data, and schema-marked-up pages. The closest thing to a "submission" is ensuring GPTBot can crawl your site and that your product schema is complete. Microsoft Merchant Centre is the most direct feed-based route into AI shopping results via Copilot.
Will adding Product schema guarantee my products appear in ChatGPT answers?
No, schema is a necessary condition, not a sufficient one. AI systems also consider the trustworthiness of your domain, the quality and depth of your content, whether your products have reviews, and how recently your pages were crawled. Schema makes you eligible; the rest of your SEO and content work determines whether you are selected.
My store is on Shopify. Is Product schema added automatically?
Shopify's default themes include some basic Product schema, but it is often incomplete. Common gaps include missing aggregateRating data, incomplete offers blocks, and no GTIN or brand entity markup. It is worth validating your existing schema rather than assuming Shopify has covered everything. Many stores we audit have schema present but broken in ways that prevent AI tools from using it.
How long does it take to see results after implementing Product schema?
It depends on how frequently AI tools crawl your site. For Bing-indexed pages, you can accelerate crawling using IndexNow, which pushes updates to Bing almost immediately. For ChatGPT's browsing, the lag is less predictable, but most stores see measurable changes in AI referral traffic within four to eight weeks of a full schema implementation.