seo-geo-knowledge

Product Schema: Which JSON-LD Fields AI Needs

✍️ GEO Expert 📅 2026-03-13 ⏱️ 5 min

Product Schema for AI Visibility: Which Fields Your JSON-LD Really Needs

For e-commerce operators: Without a correctly implemented Product Schema according to the Schema.org standard, an online shop is almost invisible to AI search engines. In this article, you will learn which fields your JSON-LD absolutely must contain—and which optional fields make the crucial difference in AI visibility.

The Role of Structured Data in the AI Era

When an AI-based search engine processes a complex query—for example, "Best running shoes for injury-prone marathon runners with delivery by tomorrow"—it needs machine-readable facts. The Product Schema functions as the only reliable data source.

While Google traditionally also evaluates the visible text on a page, LLM crawlers like GPTBot or PerplexityBot primarily rely on structured data. The reason: an LLM can extract facts much more reliably from a JSON-LD block than from continuous text that mixes marketing language, ambiguous formulations, and decorative elements.

Be Standard, not creative

The motto is: "Be Standard, not creative". Use the official vocabulary of Schema.org. While the visual user interface can be creative, the technical description must follow strict standards.

Common errors in practice:

Invented property names like productCategory instead of category

Missing @type declarations in nested objects

Inconsistent currency formats (EUR vs. € vs. Euro)

Price as a string with a comma instead of a period ("159,90" instead of "159.90")

Each of these errors can lead to the AI discarding or misinterpreting the data.

The Indispensable Mandatory Fields

name: The full product name—clear, descriptive, and without marketing jargon. "Pro-Run Marathon Shoe X1" instead of "The ULTIMATE running shoe!!!".
offers: Price, currency, and availability. Without this trio, your product is irrelevant for commercial AI queries. Always use the Schema.org URL format for availability (e.g., https://schema.org/InStock).
review or aggregateRating: Social proof is indispensable for AI recommendations. A product with 124 reviews and 4.8 stars will be recommended significantly more often than one without any review data.

Crucial Fields for AI Visibility

Identifiers (SKU, GTIN, MPN)

Identifiers are the backbone of variant differentiation. The GTIN (Global Trade Item Number) allows the AI to match and merge data from different sources. If your product is identically identified via the GTIN on Amazon, in your shop, and on a price comparison portal, the AI's confidence in the data quality increases significantly.

The SKU (Stock Keeping Unit) additionally helps with internal product identification and the differentiation of variants (size, color). Without identifiers, the AI treats each listing as a potentially different product.

Brand

The brand field is often underestimated but essential. It links the product to the overarching brand entity in the Knowledge Graph. Always use the nested type: "brand": { "@type": "Brand", "name": "Pro-Run" }.

Price and Availability

This data must exactly match the information on the landing page in the JSON-LD. Any discrepancy between the displayed price and the schema price can be interpreted by crawlers as an attempt at manipulation and permanently damage trust.

Pay special attention to:

priceCurrency in ISO-4217 format (EUR, USD, GBP)

price as a decimal number with a period (not a comma)

priceValidUntil for time-limited offers

availability as a Schema.org URL

Shipping Details

Through shippingDetails, you can specify delivery times and shipping costs precisely. Only then can the AI answer questions like "What can be delivered by Friday?" or "Which shop delivers for free?". This field is increasingly becoming a distinguishing feature, as users often filter by delivery conditions in AI searches.

Aggregate Rating and Reviews

Reviews serve as social proof. AI systems prioritize products that have a positive history. Important: Use aggregateRating for the overall rating and additionally individual review entries for detailed customer opinions. The more verified reviews, the stronger the trust signal.

Practical Example: Correct JSON-LD

{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "Pro-Run Marathon Schuh X1",
  "description": "Leichter Laufschuh für Marathonläufer mit verstärkter Dämpfung und atmungsaktivem Obermaterial.",
  "sku": "PR-X1-2024",
  "gtin13": "4012345678901",
  "brand": { "@type": "Brand", "name": "Pro-Run" },
  "image": "https://example.com/images/pro-run-x1.jpg",
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.8",
    "reviewCount": "124"
  },
  "offers": {
    "@type": "Offer",
    "priceCurrency": "EUR",
    "price": "159.90",
    "availability": "https://schema.org/InStock",
    "priceValidUntil": "2025-12-31",
    "shippingDetails": {
      "@type": "OfferShippingDetails",
      "deliveryTime": {
        "@type": "ShippingDeliveryTime",
        "transitTime": {
          "@type": "QuantitativeValue",
          "minValue": 1,
          "maxValue": 2,
          "unitCode": "d"
        }
      }
    }
  }
}

Validation and Testing

A correctly structured Product Schema is of little use if it contains syntax errors. Use these tools for validation:

Google Rich Results Test: Checks if your schema qualifies for Rich Snippets
Schema.org Validator: Validates technical correctness against the standard
AI Trust Audit: Specifically checks your structured data for AI visibility and completeness

After every change to your shop template, test whether the structured data is still correctly output—especially after updates to your shop system.

Conclusion

Consistently use unique identifiers (GTIN, SKU), maintain shipping and delivery details directly in the schema, keep price and availability data meticulously up-to-date, and strictly adhere to the Schema.org standard. A complete Product Schema is not a nice-to-have—it is the admission ticket to your shop's AI visibility.