Product Schema Markup

Product schema markup is structured data embedded in a product page using the Schema.org Product vocabulary, usually as JSON-LD. It gives search engines and AI systems a machine-readable view of the product (name, description, brand, SKU, GTIN, price, availability, rating, shipping and return policy) that does not depend on parsing the rendered HTML correctly.

Schema.org was founded by Google, Microsoft, Yahoo and Yandex in 2011 as a shared vocabulary for structured web data. The Product type has grown steadily since, adding offers, shipping details, merchant return policies and additionalProperty for arbitrary attributes.

JSON-LD is the preferred format because it sits in a self-contained script block rather than woven through the HTML. Google recommends it over Microdata and RDFa, and the AI engines parse it, which makes product schema the highest-leverage technical fix for AI visibility on a product page.

A complete Product schema block for AI commerce includes:

Core identity fields (required):

• name: the product title
• image: one or more high-resolution images, several angles preferred
• description: a complete, specific product description
• brand: the brand as a structured Brand object
• sku: your internal SKU

Commerce fields (required for shopping results):

• offers: with price, priceCurrency, availability (InStock, OutOfStock, PreOrder), priceValidUntil and url
• gtin, gtin13 or mpn: global identifiers, required for Merchant Center and increasingly expected by AI engines

Trust and enrichment fields (strongly recommended):

• aggregateRating: with ratingValue, reviewCount, bestRating
• review: individual reviews with author, ratingValue and reviewBody
• shippingDetails: shipping cost and delivery time
• hasMerchantReturnPolicy: return policy per Schema.org's MerchantReturnPolicy type

Attribute depth (the differentiator):

• material, color, size, audience, gender, ageGroup: structured attributes where they apply
• additionalProperty: an array of PropertyValue objects for anything else (waterproof rating, fit, use case, certifications). This is where the attributes that win comparisons usually live.

AI engines reason about entities, not strings. Product schema makes each product an entity with attributes the engine can query directly. An assistant answering "waterproof hiking boots under 200 dollars for flat feet" can filter on offers.price, additionalProperty (waterproof) and audience (wide fit), the exact fields schema exposes. Without schema it has to infer those from free text, which is slower and less reliable.

BrightEdge reported pages with complete product schema were 2.5 times more likely to be cited in AI Overviews (2025). The lift is larger on commerce queries than on general informational ones.

Schema is also how products enter the knowledge graphs engines consult when composing shopping answers. Google's Knowledge Graph in particular underlies AI Overviews and AI Mode product answers. A product with rich schema becomes an entity in that graph. A product without it is largely invisible to that layer even when its page ranks well in classic search.

That makes schema the technical baseline for any AI visibility work. Content and attribute improvements only pay off once the schema carries them.

The recurring mistakes across ecommerce schema audits:

1. Mismatched data. Schema says 99 dollars, the page shows 89. Google treats this as a spam signal and can apply a manual action. Schema must match the visible page, always.
2. Missing gtin / mpn. Without a global product identifier, your product cannot be confidently merged with the same product from other sources in the knowledge graph. This is one of the most common reasons a product page ranks well but does not appear as a product card in AI answers.
3. Stale priceValidUntil. priceValidUntil must be set to a future date. A past date tells Google the offer expired and the product drops from shopping results.
4. Thin description field. A 20-word description gives the retrieval layer almost nothing to match against. Invest in 150-300 word descriptions that cover material, fit, use case, and differentiators.
5. Variant inconsistency. Product pages with size/color variants must keep the JSON-LD schema, rendered HTML, and Google Merchant Center feed aligned on the selected variant. Variants that exist in the feed but not in the schema (or vice versa) cause the product to disappear or appear incorrectly.
6. Missing aggregateRating when you have reviews. If reviews exist on the page, aggregateRating must be populated. Missing it is a significant negative signal for AI product cards.
7. Stale availability. If schema says InStock and checkout fails, every engine that cached it gets a bad signal. Render schema from live inventory.

Google's Rich Results Test and the Schema.org validator catch structural errors. They do not catch schema that disagrees with the page or the feed, or a description field that is technically valid and says nothing. Those take a person looking at the product. A Paz expert checks schema, page and feed together for the products that matter most, fixes the gaps from your sources, and our experts publish within agreed guidelines. See the product page SEO service, or run the free AI readiness report on one product page to see its schema gaps.