Michael Fan · Writing

How to optimize a Shopify product feed for AI shopping agents

· 2 min read

Agentic CommerceDTC

Listen

Make every product machine-readable. Fill in all structured attributes (no blanks), add valid Product + Offer schema to every product page, keep price and inventory accurate in real time, surface real review data, and expose a clean product feed agents can read. Agents rank products on data completeness and accuracy, not visual design. A complete feed is the highest-return move you can make in 2026.

I run an 8-figure DTC CPG store on this stack. Here's the checklist, in priority order.

The checklist

  1. Fill every product attribute. Title, brand, GTIN/UPC, category, price, availability, condition, size or variant, material or ingredients. Blanks get excluded from filtered queries.
  2. Add Product + Offer JSON-LD to every product page. Shopify themes often ship partial schema, so validate and complete it.
  3. Sync inventory and price in real time. Stale data gets you dropped.
  4. Expose structured reviews (review count and rating in schema). Volume and recency are tie-breakers.
  5. Write descriptions an agent can read. Lead with the facts a buyer filters on, not brand poetry.
  6. Keep a clean product feed (Google Merchant Center or Shopify feed). Many agents read these.
  7. Make checkout protocol-ready. See the ACP vs UCP explainer.

Which attributes matter most to agents?

Attribute Why agents need it Common gap
GTIN/UPC Matches your product across the web Often blank on DTC SKUs
Availability Agents drop out-of-stock items Not synced in real time
Price Core ranking and filter input Sale price mismatches
Structured reviews Trust, and a tie-breaker Locked in an app, not in schema
Specific attributes (ingredients, size, ship speed) Query filters Buried in prose, not structured

How do I check if my feed is agent-ready?

Run your product pages through Google's Rich Results Test and a schema validator. Audit your Merchant Center feed for disapprovals and blank fields. Test real buyer queries in ChatGPT and Perplexity to see if your products show up. If they don't, it's almost always a data-completeness problem, not a budget problem.

Frequently asked questions

Do I need an app or custom code? Start with what's free. Complete your attributes, validate your theme's schema, fix Merchant Center errors. Custom work comes later.

Does this help with Google Shopping too? Yes. The same clean data powers Google Shopping, AI Overviews, and agent discovery. One effort, several channels.

How often should I audit the feed? Monthly at least, and after any theme or catalog change. Agents punish stale data fast.

Note: seeded draft. Add screenshots from your own store and your real before-and-after attribute-completeness numbers before publishing.

← All writing