ChatGPT Shopping Optimization for Shopify: 5-Step Setup Guide

This is the most common and most silent failure mode. If your robots.txt blocks OpenAI's crawlers, ChatGPT Shopping has never seen a single product page from your store.

Check your current state:

curl -s https://yourstore.com/robots.txt | grep -A2 -i "gptbot\|oai-searchbot"

If you see Disallow: / under either agent, remove it. The correct configuration:

User-agent: GPTBot
Allow: /products
Allow: /collections
Allow: /

User-agent: OAI-SearchBot
Allow: /

• GPTBot is not blocked in robots.txt
• OAI-SearchBot is not blocked in robots.txt
• Store is not password-protected (dev mode off)
• curl -A "GPTBot" https://yourstore.com/ returns 200

ChatGPT Shopping uses your /products.json endpoint for bulk catalog discovery — it's faster than crawling individual PDPs. It must be publicly accessible, paginated correctly, and include variant data.

curl -s "https://yourstore.com/products.json?limit=1" | jq '.products[0] | {title, variants: .variants | length}'
# Should return product title and variant count, not an error

Common failures:

• Headless stores: If you're running Hydrogen or a custom storefront, /products.json may be disabled. You need to proxy it or use the Storefront API.
• Password protection: Any store password blocks all crawlers. This is the #1 reason a new merchant isn't in ChatGPT Shopping.
• Pagination: The default limit is 250 products per page. Stores with 250+ products must be paginated. ChatGPT respects ?limit=250&page=N.

• /products.json returns HTTP 200 without authentication
• Variant data is included (check variants[].barcode)
• All pages accessible (test ?limit=250&page=2 if you have 250+ products)

GTINs are the product identifier ChatGPT Shopping uses for exact-match variant lookup. When a user asks "where can I buy [Brand] [Model] in size M?" ChatGPT resolves it to a specific GTIN. Variants without GTINs can't appear in those results.

Fill the Barcode (ISBN, UPC, GTIN) field on every variant in Shopify admin. For bulk entry, export your product CSV, fill the Variant Barcode column from your supplier or GS1 account, and re-import. See the full GTIN guide for format requirements and check-digit validation.

• Every variant has a non-empty barcode field
• Barcodes are valid GTINs (12-digit UPC or 13-digit EAN)
• No internal SKUs or ASINs used in the barcode field

ChatGPT Shopping uses structured data from your PDPs to confirm product details, validate GTINs, and look up variant-specific pricing. Without it, it's working from the feed alone — less accurate, lower confidence, fewer recommendations.

// Minimal Product JSON-LD for a single-variant product
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Classic T-Shirt",
  "gtin": "012345678901",
  "brand": { "@type": "Brand", "name": "YourBrand" },
  "offers": {
    "@type": "Offer",
    "price": "42.00",
    "priceCurrency": "USD",
    "availability": "https://schema.org/InStock",
    "url": "https://yourstore.com/products/classic-tshirt"
  }
}

For products with size/color variants, use ProductGroup instead — see the Shopify schema markup guide for the full Liquid template implementation.

• Every PDP has valid Product or ProductGroup JSON-LD
• gtin field populated with the variant's barcode value
• offers.price matches the actual current price
• offers.availability reflects actual stock status
• Google Rich Results Test returns no errors for a sample PDP

ChatGPT Shopping cites product descriptions when recommending products in response to informational queries ("best hiking boots for wide feet"). Short descriptions (under 100 words, or just a size chart) get cited less. Descriptions that cover use case, materials, fit notes, and distinguishing features score higher.

You don't need to rewrite your entire catalog at once. Prioritize by revenue: start with your top-20 best-sellers and expand from there. Each description should answer: what is it made of, who is it for, what problem does it solve, and what makes it different?

• Top-20 products have 150+ word descriptions
• Descriptions cover material, use case, fit/sizing notes
• No duplicate descriptions across variants of the same product
• Descriptions are in the body_html field (visible in /products.json)