Product discovery is moving into chat. Shoppers ask an assistant for "the best waterproof trail runners under £150 that ship by Friday" and get a shortlist, not ten blue links. In 2026 the plumbing behind this, agentic commerce, has matured quickly.
This guide is for Shopify and custom-store merchants who want to show up in those shortlists and convert the traffic.
The protocols in plain English
| Standard | Who's behind it | What it does |
|---|---|---|
| Universal Commerce Protocol (UCP) | Google and Shopify, with retail partners | A common language for agents to discover products, build carts and complete checkout across merchants |
| Shopify agent tools (Catalog, Checkout MCP) | Shopify | Let AI agents search Shopify's catalog and run checkout through MCP; see shopify.dev/docs/agents |
| Agentic Commerce Protocol (ACP) | OpenAI and Stripe | An open spec for agent-initiated purchases; see OpenAI's commerce docs |
The landscape is still moving. OpenAI paused Instant Checkout inside ChatGPT in March 2026, as CNBC reported, and most purchases now hand off to the merchant's own checkout. Discovery is where the volume is today, and it rewards the same things regardless of protocol.
What agents look at when choosing products
Agents can't be charmed by a hero video. They read data:
- Complete, structured product data: title, brand, GTIN, variants, materials, dimensions, compatibility
- Accurate price and stock, in real time. Agents penalise "out of stock" surprises
- Shipping speed and cost by region
- Return and warranty policies in plain, machine-readable language
- Reviews and ratings, plus third-party mentions
- Clear answers to common questions on the product page
The merchant checklist
Product data
- Fill every relevant attribute in Shopify (or your PIM), not just the title and description
- Add GTINs and MPNs where they exist
- Write descriptions that answer buying questions ("fits wide feet", "machine washable at 30°C")
- Keep Google Merchant Center feeds clean. Many agents use them as a source of truth
Structured data on the storefront
Product,Offer,AggregateRatingandMerchantReturnPolicyschema on product pages- Prices and availability in structured data that match what's visible
- See Google's product structured data docs
Inventory and performance
- Real-time inventory sync between your warehouse system and the store
- Fast product and cart APIs. Agents time out just like humans, only faster
- Load test the APIs, not just the pages (our Black Friday checklist covers how)
Policies and trust
- Dedicated, crawlable pages for shipping, returns and warranty
- Consistent business details across your site, Google Business Profile and marketplaces
Should you build your own agent integration?
For most Shopify merchants, no: the platform handles the protocol layer. Custom work pays off when you:
- Run a headless or custom stack (see headless vs Liquid)
- Sell configurable or B2B products that need quote or compatibility logic
- Want your own assistant on-site that uses the same catalog tools
- Have a SaaS-like commerce product that would benefit from an MCP server
How to track agentic traffic
- Referral sources such as chatgpt.com, perplexity.ai, copilot.microsoft.com and gemini.google.com
- UTM parameters on feed links where platforms support them
- Orders with agent attribution in Shopify, where available
- Conversion rate by source. Agent-referred shoppers often arrive with high intent
Bottom line
Agentic commerce rewards boring excellence: complete data, honest stock, clear policies and fast APIs. Fix those and you'll be ready whichever protocol wins.
Our e-commerce team audits product data, structured markup and storefront performance in a single sprint. Book a call.
Frequently asked questions
What is agentic commerce?
Agentic commerce is when an AI assistant discovers, compares and helps purchase products on a shopper's behalf, using structured catalog data and commerce protocols instead of a human browsing your storefront.
Do Shopify merchants need to do anything for agentic commerce?
Shopify exposes catalog and checkout capabilities to AI agents at platform level, but the quality of your product data, inventory accuracy, policies and reviews decides whether agents recommend you.
Did OpenAI stop Instant Checkout?
OpenAI paused its in-chat Instant Checkout in March 2026, and merchants now generally complete purchases in their own checkout. Product discovery in ChatGPT continues, so feed and data quality still matter.