Google UCP Expands: Cart, Catalog & Identity Linking Explained

On March 19, 2026, Google announced the biggest expansion to the Universal Commerce Protocol (UCP) since its initial launch. Three new capabilities — Cart, Catalog, and Identity Linking — transform UCP from a single-product checkout protocol into a full-featured commerce layer that AI agents can use to browse inventory, build multi-item orders, and recognize returning customers. As Google's official announcement described it, this update makes UCP the standard interface between AI agents and retailers.

For context: UCP is Google's open standard that lets AI agents — Gemini, third-party assistants, and agentic surfaces across the web — interact with retailer systems to complete purchases on behalf of shoppers. When UCP launched, it supported a single action: checking out one product. That was useful but limited. A shopper telling Gemini "I need running shoes, socks, and a water bottle" would hit a wall because the agent could only process one item at a time. The March 19 update removes that limitation and adds capabilities that bring UCP much closer to replicating a full online shopping experience.

What Changed: Three New Capabilities

The expansion adds three distinct capabilities to the UCP specification, each addressing a specific gap in how AI agents interact with retailer commerce systems. Together, they move UCP from a checkout protocol to a comprehensive commerce protocol.

Cart: Multi-Item Transactions

Before: UCP only supported single-product checkout. An AI agent could help a shopper buy one pair of shoes, but if the shopper also wanted the matching belt, the agent would need to start a separate transaction. That friction killed cross-sell opportunities and made the experience feel clunky compared to a regular shopping cart.

Now: AI agents can build baskets with multiple items from the same retailer and process them as a single transaction. This enables the kinds of natural shopping conversations that humans actually have — "I need the shoes, the belt, and throw in a pair of those no-show socks." The agent adds all three to a cart and checks out once.

Cart also opens up restock scenarios. A shopper who buys the same set of skincare products every month can tell an AI agent to "reorder my usual routine," and the agent can rebuild the entire basket and complete the purchase. For retailers, this means higher average order values and repeat purchase automation without building proprietary subscription systems.

Catalog: Real-Time Inventory Queries

Before: AI agents relied on product feed snapshots from Merchant Center, which could be hours old by the time an agent referenced them. A shopper asking "Is this jacket available in size L?" might get a yes based on stale feed data, only to find out it was out of stock when they reached checkout. As Search Engine Land reported, this disconnect was a major source of friction in early agentic commerce experiences.

Now: AI agents can query a retailer's live inventory and pricing through the Catalog capability. Instead of reading from a cached feed snapshot, the agent calls the retailer's catalog API directly to get current stock levels, real-time pricing, and up-to-the-minute product availability. This makes recommendations significantly more accurate and reduces the abandoned-checkout scenarios caused by stale data.

For retailers with rapidly changing inventory — flash sales, limited editions, perishable goods — Catalog is a game-changer. The AI agent can confidently tell a shopper "Yes, there are 3 left in your size" because it just checked, not because a feed file said so six hours ago.

Identity Linking: Loyalty and Membership Integration

Before: Every interaction through UCP was anonymous. The AI agent had no way to know that the shopper was a loyalty member, had earned rewards points, or was eligible for member-only pricing. This meant that loyal, high-value customers got the same generic experience as first-time visitors — a missed opportunity for retailers who invest heavily in loyalty programs.

Now: Identity Linking allows retailers to connect their loyalty and membership systems to UCP. When a recognized customer shops through an AI agent, the agent can surface member pricing, available rewards, and exclusive deals. A shopper might hear "As a Gold member, you get 15% off this item plus free shipping" during an AI-assisted shopping conversation.

This capability is significant because loyalty programs are one of the strongest competitive moats in retail. Letting AI agents access and promote those benefits means the loyalty advantage extends into agentic commerce surfaces, not just your own website and app.

Simplified Onboarding via Merchant Center

Alongside the three new capabilities, Google announced a dramatically simplified onboarding path for UCP. Previously, implementing UCP required developer resources to build and expose the necessary API endpoints. As Search Engine Journal covered, Google is now offering streamlined setup directly through Merchant Center, reducing the technical barrier significantly.

Google named six launch partners for the simplified onboarding: Shopify, Salesforce, Stripe, BigCommerce, Adobe, and Commerce Inc. If you're on one of these platforms, UCP integration will be available through pre-built connectors rather than custom development. The Merchant Center interface will guide you through connecting your existing commerce infrastructure to the UCP specification.

This is a deliberate move to accelerate adoption. When UCP launched, the implementation requirements effectively limited it to large retailers with dedicated engineering teams. By pushing onboarding into Merchant Center and partnering with major ecommerce platforms, Google is bringing UCP within reach of mid-market and smaller retailers — the same audience that already manages product feeds through Merchant Center.

What This Means for Retailers

The UCP expansion signals that Google is serious about building agentic commerce into the core shopping experience. Here's what that means practically for retailers running Shopping campaigns.

Early adopters get preferential AI agent treatment. Retailers who implement UCP early will be the ones AI agents can actually transact with. If a shopper asks Gemini to buy something and your competitor supports UCP checkout while you don't, the agent will complete the purchase with your competitor. This is a new form of competitive advantage that's entirely separate from ad bidding and SEO.

Feed quality matters even more. The new Catalog capability queries your product data in real time, but it still relies on the structured data you maintain in Merchant Center. Incomplete product attributes, missing size data, and vague descriptions will limit what AI agents can do with your catalog. Use a feed intelligence tool to identify and fix data gaps before they become conversion gaps in agentic channels.

Loyalty programs become a competitive advantage in AI shopping. Identity Linking means your investment in loyalty infrastructure now pays off beyond your owned channels. Retailers without loyalty programs are at a structural disadvantage in personalized AI shopping conversations. If you have a loyalty program, prioritize exposing it through UCP. If you don't, this is another reason to build one.

Multi-item cart changes your average order value equation. With Cart support, AI agents can drive larger basket sizes by naturally cross-selling and bundling during conversations. This changes how you think about competitive pricing — winning the initial product recommendation opens the door to upsells that weren't possible in single-product checkout. Monitor how agentic transactions differ from direct web purchases using product-level analytics.

The UCP expansion also connects to other recent Google announcements. Direct Offers in AI Mode lets retailers share personalized deals during AI conversations, and Merchant Center Business Agents provide branded AI chat on Search. UCP is the transaction layer that makes all of these surfaces actionable. Without UCP, AI agents can recommend your products but can't close the sale.

How to Prepare Your Product Feed

Regardless of when you plan to implement the full UCP specification, there are concrete steps you can take now to ensure your product data and commerce infrastructure are ready. The better your foundation, the faster your onboarding when you're ready to enable UCP.

1. Maximize feed attribute completeness. AI agents need rich, structured product data to serve shoppers effectively. Audit your Merchant Center feed for missing attributes — sizes, colors, materials, GTIN/MPNs, product types, and custom labels. Every empty field is a question the AI agent can't answer. Our UCP product feed readiness guide covers exactly which attributes matter most for agentic commerce.

2. Build or expose real-time inventory APIs. The Catalog capability needs live stock data. If you're on Shopify, BigCommerce, or another supported platform, your inventory API likely already exists — you just need to connect it. If you're on a custom platform, you'll need an API endpoint that returns current stock levels and pricing by SKU. Even if you're not implementing UCP today, having this API ready positions you for fast onboarding later.

3. Ensure your cart accepts external SKU additions. The Cart capability means AI agents will be adding products to your cart programmatically. Your cart system needs to accept additions via SKU identifier from external sources, not just from your own product pages. Test whether your platform supports headless cart operations — most modern platforms do, but older custom systems may need updates.

4. Prepare your loyalty program APIs. If you have a loyalty program, you'll need API endpoints that can authenticate members, retrieve reward balances, and apply member pricing. Identity Linking connects to these APIs so AI agents can surface personalized offers. If your loyalty program lives in a third-party system (Yotpo, Smile.io, LoyaltyLion), check whether they've announced UCP integration plans.

5. Review your product feed optimization strategy. Titles, descriptions, and images that work well for standard Shopping ads also form the foundation of agentic commerce. But the bar is higher: AI agents use product data conversationally, so descriptions need to read naturally and titles need to be genuinely descriptive. Use custom labels to segment products by UCP readiness level so you can track which parts of your catalog are prepared.

6. Monitor the competitive landscape. Track which of your competitors are implementing UCP. When their products become available for agentic checkout and yours aren't, you'll lose transactions you might otherwise win. The UCP vs ACP comparison explains how different commerce protocols compete in the agentic space.

Frequently Asked Questions