What Is Product Schema? Offer and Rating Markup for AI Shopping

Product schema is the schema.org markup that describes a product and its commercial terms: the Product entity with name, description, image, and brand, wrapping an Offer carrying price, priceCurrency, and availability, plus aggregateRating and identifiers like gtin, mpn, and sku. It is the densest commercial signal a page can emit — the difference between an engine inferring your price from prose and being told it.

What do the core fields actually do?

Each field group serves a distinct consumer need. Offers make commercial terms machine-readable: price, currency, availability state (InStock, OutOfStock, PreOrder), and priceValidUntil for time-bound pricing. Ratings — an aggregateRating with ratingValue and reviewCount — feed the stars in rich results and the confidence engines place in recommendations. Identifiers do entity resolution: a GTIN ties your page to the exact same product on every other retailer, review site, and price aggregator, letting engines consolidate signals instead of treating your listing as an island. Google's Product documentation splits the requirements between product-snippet and merchant-listing experiences.

Why does this markup matter more in AI shopping?

Commerce queries are moving into conversational interfaces. ChatGPT rolled out a dedicated shopping experience in 2025 that presents product cards assembled from structured metadata and feeds; Perplexity ships shopping results with buy buttons; Google's AI answers lean on the Shopping Graph, which ingests Product markup alongside Merchant Center feeds. In each pipeline, a product's eligibility depends on the system confidently knowing what it is, what it costs, and whether it is available — precisely the fields Product schema encodes. Pages without markup force inference, and inference-heavy sources lose to structured competitors in recommendation contexts.

How should ecommerce and SaaS teams implement it?

Generate markup from the same inventory system that renders the page, so price and availability can never disagree with visible content — mismatches violate guidelines and erode engine trust. Include every identifier you legitimately hold. Keep aggregateRating tied to genuinely collected reviews. For SaaS, the sibling type SoftwareApplication usually fits better than Product. Then close the loop: track which engines actually surface your products in shopping-flavored answers and where competitors displace you, the visibility feedback that tells you whether the structured layer is paying off.