Product Schema and Google Shopping for AI Search (ChatGPT Shopping)
A buyer tells ChatGPT they want a running shoe under 4,000 rupees, size nine, in stock. It returns three specific products with prices. To be one of those three, your product data has to be clean, complete, and machine-readable. If your schema is thin or your Shopping feed is a mess, your product is simply not in the room when the answer gets built.
By Saurabh Garg. I have built a few D2C brands and I am still learning how AI reads a catalogue. This is the technical guide: the must-have schema fields, a Rich Results Test exercise you can run today, and a checklist to get your products into AI shopping answers.
- Your product pages have partial or missing schema and you are not sure it matters.
- Your Google Shopping feed throws warnings you have been ignoring for months.
- Buyers ask AI for product recommendations in your category and yours never appears.
- Meta CPMs are up 40 to 60 percent since 2023, and you need your catalogue found without paying for every click.
Then you have a product data problem. The AI shopping layer reads structured data, and yours is unreadable or incomplete.
AI shopping answers are built mostly from structured product data, not your marketing copy. Around 83 percent of the products ChatGPT surfaces in shopping answers come from Google Shopping data. So the work is clean, complete Product schema in JSON-LD on every product page, plus a Google Shopping feed above 95 percent attribute completion, kept accurate on price and stock. Below are the exact fields, a test you can run in five minutes, and the technical checklist.
This guide sits under the larger playbook for building a D2C brand in the age of AI. It is the plumbing that turns AI visibility into AI sales.
Why AI shopping reads data, not copy
When an engine answers a shopping question, it does not scrape your hero image and product story. It reads structured facts: name, price, availability, brand, rating, GTIN. Those facts come from two places that must agree, your on-page Product schema and your Google Shopping feed. When both are complete and consistent, the engine can confidently place your product in an answer. When they conflict or a field is missing, the safe move for the engine is to skip you.
That is the whole reason this technical work pays off. About 83 percent of the products ChatGPT surfaces in shopping answers come from Google Shopping data. Your feed is not just a Google Ads asset any more. It is the raw material for AI shopping answers. Treat it that way.
The must-have schema fields
Not all fields carry equal weight. Here are the ones that decide whether your product is eligible to appear, and what each one does. Mark up every product with these in JSON-LD Product schema.
| Field | What it is | Why it matters |
|---|---|---|
| name | Exact product name | The engine matches this to the buyer’s query. Be specific, not clever. |
| brand | Your brand name | Ties the product to your entity so brand-level trust carries over. |
| offers / price | Price and currency | Price filters are common in AI shopping queries. Wrong or missing price drops you. |
| availability | InStock or OutOfStock | Engines avoid recommending out-of-stock items. Stale stock data burns trust. |
| gtin / mpn | Global trade or part number | The unique ID that lets engines match your item across sources with confidence. |
| aggregateRating | Average rating and review count | Third-party proof the engine weighs heavily when choosing what to name. |
| review | Individual reviews | Feeds the “what do people say” part of a shopping answer. |
| image | Clear product image URL | Required for rich shopping results and visual answer formats. |
| description | Factual product description | Where key attributes like size, material, and use case get read. |
Open Google’s Rich Results Test and paste in one of your live product page URLs. Then do three things:
1. Check whether Product schema is detected at all. If nothing shows, your markup is missing or broken.
2. Read the warnings. Google will list missing recommended fields like aggregateRating, gtin, or availability. Write them down.
3. Fix the template, not the single page. Whatever field is missing here is missing on every product. Correct it once in your product template so all products inherit the fix.
Then re-run the test on two more product pages to confirm the template fix carried through. That is your schema baseline done in one sitting.
The Google Shopping feed side
Schema gets your on-page data right. The feed is the other half, and it is where most brands lose points. Google Merchant Center scores your feed on attribute completion, and you want it above 95 percent. Every missing attribute is a reason for the engine to prefer a competitor whose data is complete.
The two fields that quietly kill shopping visibility are price and availability, because they go stale. A buyer asking for something in stock under a price point gets your product only if your feed says, truthfully and right now, that it is in stock and under that price. Sync your feed often enough that it never lies. A feed that recommends an out-of-stock item once teaches the engine to trust you less.
In AI shopping, your product data is your salesperson. If the data is incomplete, the salesperson never gets to speak.
When schema and feed disagree, you lose
Here is the failure mode almost nobody checks for. Your on-page schema says the price is 3,499. Your Google Shopping feed, last synced a week ago, says 3,999. Now the engine has two prices from the same brand and no way to know which is true. The safe move is to drop you from the answer, because recommending a wrong price burns the engine’s own trust with its user. You just lost the sale to a conflict you never saw.
These conflicts creep in because schema and feed are usually owned by different people. The developer sets up schema once. The performance marketer manages the feed for Google Ads. Neither treats the other’s data as their problem, so the two drift apart on price, title, and stock. The fix is not clever. It is ownership. One person or one audit has to confirm that for every SKU, the schema and the feed agree on price, availability, and title. Run that check monthly, and after any pricing change or sale.
The same discipline applies to title. If your schema calls it “Acme Vitamin C Serum 20%” and your feed calls it “Face Serum for Glow,” the engine cannot confidently match them as the same product. Pick one exact product name and use it identically in both places.
The technical checklist
Run this end to end. Each item is a common point of failure that keeps products out of AI answers.
- Product schema in JSON-LD on every product page, applied through the template, not page by page.
- All must-have fields present: name, brand, price, availability, gtin, image, description.
- aggregateRating and review markup live, pulling real review data.
- Every product page passes Google’s Rich Results Test with no errors.
- Google Shopping feed above 95 percent attribute completion in Merchant Center.
- Price and availability synced frequently so the feed never shows stale data.
- Schema data and feed data agree on price, title, and availability for every product.
- GTIN or MPN present on every item so engines can match it across sources.
- AI crawlers allowed in robots.txt: OAI-SearchBot, PerplexityBot, Google-Extended.
This product work sits on top of a clear brand. If the engine does not understand your brand as an entity, product-level trust has nothing to attach to. Get that right first with brand entity SEO, then fit this into the wider program in the GEO playbook for D2C brands.
Three brands, three lessons
Each of these treats product data as an asset, not an afterthought. That discipline is why their catalogues show up when an engine builds a shopping answer.
Allbirds
Clean product pages with complete data and a clear sustainability story per item. The structured detail per product is exactly what an engine needs to place a specific shoe in a specific answer.
Mumzworld
A deep mother-and-baby catalogue with structured product data and heavy review volume. That completeness is what lets an engine match a specific product to a specific parent query.
Sugar Cosmetics
Detailed product pages with shade-level specifics and strong ratings. The granularity in the data is what makes individual SKUs eligible for AI shopping answers, not just the brand name.
Where brands get stuck
The schema itself is not the hard part. Keeping it right at scale is. Product templates change, developers ship updates that quietly break markup, and the feed drifts out of sync on price and stock. Most teams fix schema once, celebrate, and never check again, so it decays. The other gap is ownership: schema lives with the developer, the feed lives with the performance marketer, and nobody owns the fact that they must agree. This is where an outside partner earns its fee, treating product data as a living system with the audits to catch drift. That is the work we do at C4E.
Frequently asked questions
What is product schema for AI search?
Product schema for AI search is structured data in JSON-LD on your product pages that states the facts an AI shopping engine reads: name, brand, price, availability, GTIN, rating, and image. It makes your products machine-readable so engines can place them in shopping answers.
Does ChatGPT use Google Shopping data?
Yes. Around 83 percent of the products ChatGPT surfaces in shopping answers come from Google Shopping data. A complete, accurate Google Shopping feed above 95 percent attribute completion directly affects whether your products appear in AI shopping answers.
Which schema fields matter most for AI shopping?
Name, brand, price, availability, and a unique GTIN or MPN are essential because they let the engine match and filter your product. aggregateRating and review add the third-party proof engines weigh heavily, and image is required for rich shopping results.
How do I test my product schema?
Paste a live product page URL into Google’s Rich Results Test. It shows whether Product schema is detected and lists missing recommended fields. Fix the missing fields in your product template so every product inherits the correction, then re-test.
Why do price and availability matter so much?
AI shopping queries often filter by price and stock. If your feed shows a stale price or recommends an out-of-stock item, the engine either drops you or learns to trust your data less. Sync price and availability frequently so the feed never lies.
Want your products in AI shopping answers?
We treat product data as a living system: complete schema, a clean feed above 95 percent, and audits that catch drift before it costs you a sale. If your catalogue is invisible when buyers ask AI for a recommendation, the fix is technical and we do it.
Write to hello@c4e.in or use the form below, and tell us your category. We will run the Rich Results Test on your products and send you what we find.