AI Growth Hub
← AI SEO

Rich Product Schema for AI Search: GTIN, Attributes, and Shipping That Get You Cited

schema-markup ai-search ecommerce-seo google-search-console

What the March Core Update Changed About Schema

Google deprecated FAQ schema in January 2026 and HowTo schema in February 2026. If your pages lost their rich snippets after those changes, that was expected. But the bigger shift happened in March.

The March 2026 core update repositioned structured data from “rich result trigger” to “AI citation confidence signal.” Adding schema is no longer primarily about generating star ratings or FAQ dropdowns in SERPs. It now influences whether Google AI Overviews, ChatGPT Shopping, and Perplexity include your products in AI-generated answers.

According to analysis from alhena.ai, products with complete structured data appear in AI-generated answers at 2.5x the rate of products with only basic schema. That gap is large enough to treat schema as infrastructure rather than an optimization extra.

For cross-border DTC sellers, schema now belongs in the same priority tier as site speed. It directly affects whether AI search sends you any traffic.

Why Basic Product Schema Falls Short

Most Shopify themes and WooCommerce plugins generate baseline Product schema automatically: name, price, image, description. Those fields were designed for traditional crawlers a decade ago. AI search engines need considerably more.

The gap by field:

FieldTypical basic schemaAI search importance
GTIN / MPNMissingHigh: AI product matching relies on unique identifiers
Attributes (color, material, size)MissingHigh: conversational long-tail query matching
shippingDetailsMissingMedium: AI shopping agents need delivery data for comparisons
hasMerchantReturnPolicyMissingMedium: trust signal during purchase consideration
aggregateRatingSometimes includedHigh: credibility signal in AI answers
brand.nameUsually includedMedium: brand entity consistency

AI search engines use schema to do two things: verify a product actually exists (GTIN/MPN), and understand what queries that product should answer (attribute fields). Without both layers, your products have low probability of appearing in AI-generated answers.

GTIN and MPN: The Foundation of AI Product Matching

GTIN (EAN, UPC, ISBN) is the most direct way for AI search engines to confirm a product is a specific, real item. ChatGPT Shopping prioritizes products it can cross-verify with a GTIN. Perplexity product cards use GTINs to aggregate data from multiple sources.

JSON-LD format:

{
  "@type": "Product",
  "name": "Organic Cotton T-Shirt",
  "gtin13": "0123456789012",
  "mpn": "TSHIRT-ORG-M-BLK",
  "brand": {
    "@type": "Brand",
    "name": "Your Brand Name"
  }
}

If your products don’t have EANs or UPCs, register through GS1 — it’s a one-time cost, not an ongoing expense. Cross-border sellers often already have FNSKUs on the Amazon side but have never added those product identifiers to their DTC site schema. That’s the easiest gap to close.

On Shopify, populate the Barcode field on the product page and reference product.selected_or_first_available_variant.barcode in your schema template. No need to hardcode each product.

Product Attributes: Matching Conversational Long-Tail Queries

When a user asks ChatGPT or Perplexity “recommend a lightweight waterproof jacket for hiking, under $150,” the AI needs to match against material, function, and price range. If those attributes aren’t in your schema, that query match goes to competitors who do have them.

The additionalProperty approach is the most flexible:

{
  "@type": "Product",
  "additionalProperty": [
    {
      "@type": "PropertyValue",
      "name": "Material",
      "value": "Gore-Tex waterproof fabric"
    },
    {
      "@type": "PropertyValue",
      "name": "Waterproof Rating",
      "value": "20000mm"
    },
    {
      "@type": "PropertyValue",
      "name": "Use Case",
      "value": "hiking camping outdoor"
    }
  ]
}

Prioritize: primary material, core functional differentiators, use cases, size range. Don’t duplicate generic description text into schema — include only the dimensions that matter for long-tail query matching. A product page already has the marketing copy; schema needs the machine-readable facts.

Shipping and Return Policy: Required for AI Shopping Agents

AI shopping agents in 2026 — Google AI Mode’s native checkout, ChatGPT Shopping’s direct purchase flow — filter products during the selection process. Missing shipping times or return policy data gets your products removed before they ever reach the recommendation stage.

Implementation:

{
  "@type": "Offer",
  "shippingDetails": {
    "@type": "OfferShippingDetails",
    "shippingRate": {
      "@type": "MonetaryAmount",
      "value": "0",
      "currency": "USD"
    },
    "deliveryTime": {
      "@type": "ShippingDeliveryTime",
      "handlingTime": {
        "@type": "QuantitativeValue",
        "minValue": 1,
        "maxValue": 2,
        "unitCode": "DAY"
      },
      "transitTime": {
        "@type": "QuantitativeValue",
        "minValue": 5,
        "maxValue": 8,
        "unitCode": "DAY"
      }
    }
  },
  "hasMerchantReturnPolicy": {
    "@type": "MerchantReturnPolicy",
    "returnPolicyCategory": "https://schema.org/MerchantReturnFiniteReturnWindow",
    "merchantReturnDays": 30,
    "returnMethod": "https://schema.org/ReturnByMail",
    "returnFees": "https://schema.org/FreeReturn"
  }
}

Before filling these fields, verify the data matches your actual policy page. AI engines cross-reference structured data against page content. Discrepancies lower your credibility score — publishing inaccurate schema is worse than publishing no schema at all.

Rollout and Validation

Run finished JSON-LD through Google Rich Results Test before deploying. Then confirm crawl status via the Structured Data report in Search Console.

For Shopify and WooCommerce: start with 3-5 high-traffic product pages, validate manually, then extend to the full catalog. Whole-site changes that break schema are painful to roll back — staged deployment costs a few days but prevents larger problems.

After deployment, request re-crawl in Search Console and monitor for 4-6 weeks. AI search indexing cycles run longer than traditional SERPs. Two weeks without visible change is not a signal that the work didn’t matter — it’s just not indexed yet.

阅读本文中文版: 产品 Schema 富属性：让 AI 搜索引擎真正引用你的商品

Related Articles