Shopify Product JSON-LD

Product JSON-LD is the single largest signal in the CatalogScan score: 30 of 170 possible points, more than any other test. Every AI shopping agent — ChatGPT Shopping, Perplexity Shopping, Google AI Mode, Shopify's own Global Catalog — parses your product detail pages' JSON-LD first and only falls back to scraping the page body if the structured block is missing or malformed. This page explains the exact shape agents need, how to test what your theme is actually emitting, and how to fix it in product.liquid.

Last updated 2026-04-30 · Floor signal · 30 pts · Most points of any signal

What it is

JSON-LD ("JSON for Linked Data") is the structured-data format Google, OpenAI, Anthropic, Perplexity, and every other major AI agent agree on for reading product pages without needing to parse raw HTML. On a Shopify product detail page, it lives in a <script type="application/ld+json"> block near the bottom of the page, and the load-bearing field is the top-level @type: only Product and ProductGroup count for AI-shopping discovery. Organization, WebSite, BreadcrumbList, WebPage can be present alongside but do not score this signal.

The minimum viable shape an AI agent expects on a single-variant PDP:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Product",
  "name": "Wool Runner — Natural Black",
  "image": ["https://yourstore.com/cdn/shop/.../runner.jpg"],
  "description": "Merino-wool runner with cushioned EVA midsole...",
  "sku": "AB-WR-NB-09",
  "gtin13": "0012345678905",
  "brand": { "@type": "Brand", "name": "Allbirds" },
  "offers": {
    "@type": "Offer",
    "url": "https://yourstore.com/products/wool-runner",
    "priceCurrency": "USD",
    "price": "98.00",
    "availability": "https://schema.org/InStock"
  },
  "aggregateRating": {
    "@type": "AggregateRating",
    "ratingValue": "4.7",
    "reviewCount": "3284"
  }
}
</script>

For a multi-variant PDP (one parent with size/color options) a single Product node is the wrong shape — see our long-form on ProductGroup JSON-LD for the correct ProductGroup + hasVariant shape.

Why AI shopping agents care

Three reasons this signal weighs more than any other:

• It's the source of truth for downstream signals. Brand entity, AggregateRating, Offers availability, canonical URL hygiene, GTIN, SKU — every one of those is read out of the same Product JSON-LD block. If the block is missing, those signals cannot score either; you lose the floor and every dependent ranking-spread signal at once.
• Agents read this first, scrape body second. Crawling the rendered HTML body, walking the DOM, and inferring price + availability + reviews is orders of magnitude slower and orders of magnitude noisier than reading one JSON object. Stores that emit clean JSON-LD get crawled at higher frequency and lower error rate.
• It's the canonical citation surface. When ChatGPT Shopping cites your product in a response, the name, image, and offers.price fields are what the user sees in the rich card. Missing JSON-LD = degraded card = lower click-through even when you do show up.

How to test it on your store

The fastest hand-test, on any one PDP, is two steps:

1. View the page source (Cmd-U / Ctrl-U), search for application/ld+json.
2. Inside the matching <script> blocks, find the one whose top-level @type equals "Product" or "ProductGroup". If neither exists, you score zero.

To validate the JSON parses correctly and matches the schema vocabulary AI agents expect, paste your full PDP URL into Google's Rich Results Test — any red X under "Detected items" means a malformed block, which silently disqualifies you even if the data looks right at a glance.

The fastest way to test all your PDPs at once is the free CatalogScan scan: it pulls a product from your /products.json feed, parses the live PDP, and reports whether the JSON-LD on that page is valid + which sub-fields (price, brand, GTIN, AggregateRating) it found.

How to fix it

Fixes ordered cheapest to most-involved:

5 mistakes we keep finding