Shopify /products.json — the AI bulk-ingest feed

Every Shopify store has a public, paginated, fully-structured feed of its catalog at /products.json. It's the single most-ingested surface on a Shopify store and the path AI shopping agents — ChatGPT Shopping, Perplexity, Google AI Mode, Shopify's Global Catalog — use to discover what you sell at scale. In our 100-store launch scan, 40% of headless DTC brands had accidentally killed it during their Hydrogen or Next.js cutover. Their PDPs still rendered fine; their AI-readable surface had silently disappeared.

Last updated 2026-04-30 · Floor signal · 25 pts · Second-largest signal weight

What it is

The Shopify /products.json endpoint is part of Shopify's storefront API surface. It's been stable since the early 2010s, requires no authentication on a public storefront, and returns up to 250 products per call with standard pagination via ?page=N&limit=N. The shape:

{
  "products": [
    {
      "id": 4582134823219,
      "title": "Wool Runner",
      "handle": "wool-runner",
      "body_html": "<p>Merino-wool runner with cushioned EVA midsole...</p>",
      "vendor": "Allbirds",
      "product_type": "Footwear",
      "tags": ["sustainable", "running", "merino"],
      "variants": [
        { "id": 31..., "title": "M 9", "sku": "AB-WR-M-09",
          "barcode": "0012345678905", "price": "98.00",
          "available": true, "option1": "9", "option2": "Natural Black" }
      ],
      "options": [...],
      "images": [
        { "src": ".../runner.jpg", "alt": "Wool Runner — Natural Black" }
      ]
    },
    ... up to 249 more
  ]
}

It is not the same as the Shopify Admin API or the Storefront GraphQL API. /products.json is the legacy "open feed" — no auth, no token, public-internet readable.

Why AI shopping agents depend on it

Three reasons this surface gets weighted at 25 points (second only to Product JSON-LD):

• Bulk-ingest path. Indexing 10,000 product PDPs at one HTML page per request is hours of crawl work. Reading the same catalog from /products.json in 40 paginated 250-product responses takes seconds. Agents that need fresh inventory data — for "in stock?" queries, for sale-aware ranking — depend on the bulk path.
• Structured fields. The HTML PDP is a render target; /products.json is a data source. Variant structure, parent-product groupings, inventory flags, vendor + product_type — these are clean fields, not parsed-out-of-DOM guesses. A store with a clean feed is a store an agent trusts more.
• Discovery anchor. Even agents that index PDPs deeply (for the JSON-LD on each one) often discover those PDP URLs from /products.json first. Sitemap is the alternate discovery path; both should work.

How to test it on your store

One curl command:

curl -s https://yourstore.com/products.json?limit=1

Expected response: a JSON body with a products array containing exactly one product (or zero if your store is empty). What a failed feed looks like instead:

• 404 Not Found. Headless front-end didn't proxy this path. Most common.
• HTML, not JSON. Storefront-password gate is on; or the headless front-end is rendering a 404 page with a 200 status code.
• 302 redirect to /password. Storefront password is set in Admin → Online Store → Preferences. Turn it off to go live.
• Status 200 with {"products":[]}. Empty catalog. If you have products in the Shopify admin, this is a sales-channel publishing issue — products aren't published to the Online Store channel.

To verify pagination works (some headless proxies hardcode the first page only), try ?page=2 and confirm you get the next 250.

How to fix it

4 mistakes we keep finding