What is ChatGPT Shopping?
ChatGPT Shopping is a feature that lets ChatGPT recommend specific products directly in its answer when it detects a buying intent, including a product image, price and a link to the retailer. In Germany the feature has been live since July 2025. Instead of a list of blue links, users receive a curated selection of matching products, often with a short reason why a product fits the query.
For online shops, this shifts part of product discovery from the classic search engine into the dialogue with the AI. Whoever appears in these recommendations gains qualified customers without paying for every click. Whoever is missing simply does not get mentioned by the AI.
Industry estimates for 2026 paint a clear picture. AI overviews in Google search cause a loss of around 38 percent of organic clicks, and about 30 percent of B2B research already runs through ChatGPT and Perplexity. At the same time, the German-speaking market has a first-mover window of 12 to 18 months in which structured product data has an outsized effect.
Where ChatGPT gets its product data
To optimise effectively, you need an honest understanding of what feeds ChatGPT's product recommendations. There are two halves:
- The Google Shopping feed. According to industry estimates, around 83 percent of the product data comes from the structured feed that merchants maintain through Google Merchant Center. This half is the basis for many price and availability details.
- Structured data on your own product page. The second half is Schema.org markup directly on the shops' product pages, which AI crawlers read and evaluate. It provides context, brand, ratings and unique product identifiers.
Feed-AI delivers the second half: structured, Schema.org-compliant product data plus your own Google-indexed product page that AI systems can use as a source. Feed-AI explicitly does not create the Google Shopping feed itself. You continue to maintain that through Google Merchant Center. This clear separation is intentional, because it is honest and you know exactly which building block comes from where.
Required fields in the Product schema
For an AI to reliably match and recommend a product, the product page needs a clean Product schema following Schema.org. These fields are decisive:
| Field | Description | AI relevance |
|---|---|---|
name |
Exact product name | Very high |
offers |
Offer object with price and availability | Very high |
price |
Price as a number, without currency symbol | Very high |
priceCurrency |
Currency in ISO format, for example EUR | High |
availability |
Availability, for example InStock or OutOfStock | High |
brand |
Product brand | High |
aggregateRating |
Average rating and count | Very high |
gtin / mpn |
Unique product identifier (GTIN or manufacturer number) | High, for unique matching |
Important: the values in the schema must match the visible page content. If the price in the schema differs from the price on the page, Google treats it as a violation, and the AI loses trust in the source.
Common mistakes: why products stay invisible to ChatGPT
Many shops have well-maintained product pages and still do not appear in AI recommendations. Usually it comes down to one of these points:
- AI crawlers are blocked. The robots.txt is missing the release for GPTBot, OAI-SearchBot, PerplexityBot or Google-Extended. Without that release, the AI does not see the page at all.
- No Product schema. The product data sits only as body text or an image on the page. Without machine-readable
Productschema, the AI lacks the structure. - Price and availability only via JavaScript. If price or stock status load only later in the browser, they are missing from the crawled HTML.
- No unique product identifier. Without GTIN or MPN, the AI cannot cleanly distinguish variants and identical products.
- Contradictory data. The price in the schema and the price on the page do not match, or the availability is outdated.
- No own indexed product page. Selling only through a marketplace hands visibility to the marketplace and leaves no own source for the AI to cite.
In practice, a blocked or forgotten crawler release is by far the most common reason for invisibility. Check your robots.txt first, before you fine-tune schema and content.
Step-by-step checklist for product data optimisation
Release AI crawlers in your robots.txt
Explicitly allow GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended. Without that release, every further optimisation stays ineffective.
Add Product schema to every product page
A JSON-LD block with name, offers (price, priceCurrency, availability), brand and ideally aggregateRating. Add GTIN or MPN for unique matching.
Serve price and availability server-side
Make sure price and stock status already appear in the delivered HTML and do not load only via JavaScript, so crawlers can see them.
Write concrete, factual product descriptions
State material, dimensions, intended use and distinguishing features clearly. The AI only recommends what it clearly understands. Marketing phrases do not help.
Maintain Google Merchant Center
Keep the Google Shopping feed current and complete. This is the first of the two data halves and is fully in your hands.
Measure visibility and adjust
Regularly check whether and how products appear in ChatGPT, Perplexity and Gemini, and add fields as soon as AI search expects new details.
A sporting goods shop had well-maintained product pages, but no Product schema and an outdated robots.txt entry that blocked GPTBot. After releasing the crawlers, adding the Product schema with GTIN and serving prices server-side, the first models appeared in ChatGPT recommendations within a few weeks, for example on the query for a well-rated trail running shoe under 150 euros.
How Feed-AI helps with ChatGPT Shopping optimisation
The checklist above can be worked through by hand. Feed-AI takes the technical part off your plate, so you do not have to deal with JSON-LD, markup errors or crawl details. You enter your product data, and the platform handles the rest.
Specifically, Feed-AI takes care of:
- Product schema automatically. You enter your product data no-code, and Feed-AI generates valid Schema.org markup with all required fields (step 2 of the checklist).
- Your own indexed product page. Each listing gets a Google-indexed page on feed-ai.de, served server-side and therefore readable for AI crawlers (steps 3 and 4).
- AI visibility check. Feed-AI measures whether and how your product appears in ChatGPT, Perplexity and Gemini, and proves the progress before and after (step 6).
- Self-learning field audit. Feed-AI regularly checks the fields, adds new requirements, verifies existing entries and rates them by importance for AI search.
What stays in your hands, we say plainly: releasing the AI crawlers in your robots.txt (step 1) and the Google Shopping feed in Merchant Center (step 5). Feed-AI does not replace the feed, it adds the second, structured half of the data.
AI systems constantly change which details they expect for a recommendation. Anyone who enters their product data once and never adjusts it loses visibility over time without noticing. This is exactly where the self-learning field audit works: Feed-AI checks the fields regularly, adds new requirements automatically, verifies existing entries and prioritises them by importance. That way your visibility stays high instead of slowly eroding because no one keeps it up to date.
Get started with Feed-AI and create your first listing in a few minutes, or first see how Feed-AI works.
Frequently asked questions
What is ChatGPT Shopping? +
ChatGPT Shopping is a feature that lets ChatGPT recommend specific products directly in its answer, including a product image, price and a link to the retailer. In Germany the feature has been live since July 2025. Instead of a list of blue links, ChatGPT returns a curated selection of matching products.
Where does ChatGPT get its product data? +
From two sources. The larger part comes, according to industry estimates, from the Google Shopping feed, about 83 percent of the product data. The second half is structured data following Schema.org directly on the shops' product pages. Both sources complement each other.
Can Feed-AI add my products to the Google Shopping feed? +
No, and we say that plainly. Feed-AI delivers the second half: structured, Schema.org-compliant product data plus your own Google-indexed product page that AI systems use as a source. You maintain the Google Shopping feed itself through Google Merchant Center.
Which Schema.org fields are required for ChatGPT Shopping? +
The Product type with name, an offers object (price, priceCurrency, availability), brand and ideally aggregateRating. For unique product identification, gtin or mpn belong with it.
How long does it take for products to appear in ChatGPT? +
It depends on the crawl and update cycle and often takes a few weeks. What matters is that the product data is structured, crawlable and released for AI crawlers at all. Without that foundation a product does not appear.
Is ChatGPT Shopping optimisation worth it for small shops? +
Especially now. Industry estimates point to a first-mover window of 12 to 18 months in the German-speaking market, during which structured product data has an outsized effect. Small shops that optimise early secure visibility before competitors catch up.