Product Structured Data for AI Agents and Search: A schema.org Guide
A practical guide to schema.org Product and Offer markup in JSON-LD: which fields to include, how to avoid common errors, and how to test your pages for AI agents and search engines.
Product structured data is a block of JSON-LD code that describes a product in a standard format from schema.org. At a minimum, a product needs a name, an image and an Offer with price, currency and availability. Search engines and AI agents can read this block directly, so it gives them clean facts without guessing from your page layout.
What structured data is
Structured data is machine-readable information added to a page. The most common format is JSON-LD, a script block in the page head or body. It uses the schema.org vocabulary.
The block does not change how your page looks. It tells software what the page is about. For a product page, the two key types are:
- Product. The item: name, image, description, identifiers, brand.
- Offer. The sales terms: price, currency, availability, seller, link.
Google documents which fields it reads for product rich results and merchant listings. Other search engines and AI systems may read the same markup. Support differs by system, and no one can promise that markup leads to a mention or a sale.
A minimal, correct example
This example shows a product with one offer.
{
"@context": "https://schema.org",
"@type": "Product",
"name": "Merino beanie, grey, one size",
"image": ["https://example.com/images/merino-beanie-grey.jpg"],
"description": "Soft beanie made from 100% merino wool. One size fits most adults.",
"sku": "BEANIE-GREY",
"gtin13": "4006381333931",
"brand": { "@type": "Brand", "name": "Example Brand" },
"offers": {
"@type": "Offer",
"url": "https://example.com/products/merino-beanie-grey",
"priceCurrency": "EUR",
"price": "29.90",
"availability": "https://schema.org/InStock",
"itemCondition": "https://schema.org/NewCondition"
}
}
Use your own real values. The URLs, brand and numbers above are examples only.
Fields to include
Product fields
name: the product name, the same as on the page.image: one or more full image URLs.description: a plain text summary.sku: your internal product code.gtin,gtin8,gtin12,gtin13orgtin14: the barcode number, when you have one.mpn: the manufacturer part number.brand: the brand name.
Offer fields
price: a plain number with a dot as decimal separator. No currency symbol, no thousand separator.priceCurrency: the ISO 4217 code, such as EUR.availability: a schema.org URL, such ashttps://schema.org/InStock,OutOfStockorPreOrder.url: the page where the buyer can buy.itemCondition: new or used.priceValidUntil: the end date of a sale price, if you have one.
Optional fields that help
aggregateRatingandreview: only if they come from real reviews that buyers can see on the page.shippingDetails: shipping cost and delivery time, using the OfferShippingDetails type.hasMerchantReturnPolicy: the return window and terms, using the MerchantReturnPolicy type.
Check Google's current documentation for the exact fields it supports for each use.
Variants
Products with sizes or colors need care. Schema.org has a ProductGroup type with hasVariant and variesBy for this. Each variant is a Product with its own SKU, price and availability.
If your platform does not output ProductGroup, make sure at least that the selected variant on each URL has the right price and availability.
Common errors
- Price with a symbol or comma. Write 29.90, not "€29,90".
- Price mismatch. The markup says one price and the page shows another. This can lead to warnings or lost eligibility.
- Wrong availability. Sold-out items marked as InStock.
- Duplicate blocks. A theme, a plugin and the platform all output a Product block, with different values.
- Markup for invisible content. Rating or review data that visitors cannot see on the page.
- Missing image. Many features need at least one.
- Fake reviews. Do not add ratings you made up. It breaks search engine rules and it is misleading.
- Old data in caches. Cached pages keep an old price in the markup.
How to test
- Open the Schema Markup Validator at schema.org and check that the JSON-LD parses.
- Use Google's Rich Results Test on a live URL. Fix errors first, then warnings.
- In Google Search Console, check the product and merchant listing reports, if you use them.
- Compare the markup with the visible page: price, currency, stock, title.
- Re-test after theme, plugin or platform updates.
Shopify and WooCommerce notes
- Shopify. Most themes output Product and Offer markup. Check yours on live pages, and test variant URLs.
- WooCommerce. WooCommerce outputs Product markup. SEO plugins can output more. Check for duplicates and for sale price logic.
If you edit markup in theme code, work on a copy first.
Frequently asked questions
Is JSON-LD better than other formats?
JSON-LD is the format Google recommends. It is easier to maintain than Microdata, because it sits in one block and does not mix with your HTML.
Do AI agents use schema.org markup?
Many systems can read it, because it is a standard. But we cannot say that every agent does. Keep the same facts in your visible page text too.
Can structured data guarantee that I appear in AI answers?
No. It makes your data easier to read. Whether you appear depends on many other factors, such as relevance, trust and price.
How often should I check my markup?
Check after each theme, plugin or platform update, and test a few top products every month.
Clean product data, clean orders
Once the sale happens, the order needs a correct invoice. See what we build for EU stores.
See our products →