Shopify Product Facts for AI Systems

A recommendation system cannot compare a product using facts that exist only in a merchant’s head. It also cannot reliably resolve five conflicting versions of the same fact.

The central technical job for Shopify AEO and GEO is to create a product fact layer that is explicit, consistent, and attached to a stable product identity.

The fact model

Divide product information into six groups:

Group	Examples	Recommended owner
Identity	name, brand, SKU, GTIN, canonical URL	Shopify product or variant
Classification	product type, taxonomy category, audience	taxonomy and metafields
Physical	material, size, weight, dimensions, ingredients	typed metafields
Eligibility	age, skin type, skill level, fit, use case	typed metafields plus visible copy
Commercial	price, currency, availability, subscription	variant and market data
Policy	shipping, returns, warranty, safety exclusions	policy pages or metaobjects

Use typed metafields for facts that must be reused. A free-form description can explain why a material matters, but a material metafield should hold the normalized value.

Build a source-of-truth map

Before editing the storefront, complete this table for the selected product:

Fact	Shopify source	Visible location	Structured source	Update trigger
Price	variant price	buy box	Offer.price	price change
Availability	inventory policy	buy box	Offer.availability	inventory change
Material	product metafield	specifications	Product.material	merchandising edit
Warranty	warranty metaobject	policy panel	referenced text/URL	policy edit
Fit	product metafield	size panel	description or property	product edit

If a fact has two owners, choose one. If a value is time-sensitive, avoid a manually copied value.

Write facts at the level buyers compare

“High quality” is not a comparable fact. The following are:

• 1440 × 1440 video resolution
• compatible with 2.4 GHz Wi-Fi
• 4.2 oz and 1.7 oz size options
• designed for beginners age 12 and over
• 100% cotton, regular fit
• 30-day return window

For each product, list five questions a customer asks immediately before purchase. Make every answer visible in plain text on the canonical page.

Keep identity stable across variants

Variants create ambiguity when names, SKUs, URLs, and offers are disconnected.

Use:

• one stable parent product identity
• a stable SKU for each commercially distinct variant
• variant URLs that resolve to the correct selected option
• explicit labels such as Size: 4.2 oz, not Big
• market-specific currency and availability

For a product group, a simplified pattern is:

{
  "@context": "https://schema.org",
  "@type": "ProductGroup",
  "@id": "https://example.com/products/glazing-milk#product",
  "name": "Glazing Milk",
  "brand": {"@type": "Brand", "name": "Example Brand"},
  "variesBy": ["https://schema.org/size"],
  "hasVariant": [
    {
      "@type": "Product",
      "sku": "GM-42",
      "size": "4.2 oz",
      "offers": {
        "@type": "Offer",
        "price": "32.00",
        "priceCurrency": "USD",
        "availability": "https://schema.org/InStock"
      }
    }
  ]
}

Treat this as a model, not copy-paste production code. Shopify’s native structured_data output, theme version, apps, and market setup determine the correct implementation.

Handle claims as evidence

Claims need qualifiers:

• “water resistant” needs a rating or test condition
• “works with Alexa” needs the supported feature and region
• “for sensitive skin” needs product guidance and appropriate limitations
• “lifetime warranty” needs a link to the terms and exclusions

Store the claim, its source URL, owner, applicable products, and last review date. Avoid converting marketing superlatives into numeric facts.

Audit consistency

For ten priority products, compare:

1. visible title, price, availability, and selected variant
2. canonical URL
3. all Product, ProductGroup, and Offer objects
4. Shopify product feed or Merchant Center data
5. collection cards and internal search
6. policy pages referenced by the product

Classify each difference:

• critical: wrong price, currency, stock, SKU, or canonical
• material: wrong size, compatibility, ingredients, or warranty
• descriptive: wording differs but meaning agrees
• intentional: market or variant context explains the difference

Acceptance checklist

A product is fact-ready when:

• the canonical URL identifies one product
• variant identity is recoverable
• visible and structured commercial facts agree
• decisive specifications are text, not image-only
• policies and limitations are linked
• claims have an owner and review date
• unsupported facts are omitted rather than guessed

This does not force an answer engine to recommend the product. It gives the engine fewer reasons to ignore it, misclassify it, or describe it incorrectly.

Primary references

• Schema.org Product
• Schema.org ProductGroup
• Shopify product structured data
• Google product structured data