Universal Commerce Protocol: What UCP means for your online store

The Universal Commerce Protocol (UCP) is an open standard that lets AI agents shop on their users’ behalf: from finding a product to building the cart, checking out and handling post-purchase support. Google introduced it in January 2026, co-developed with Shopify, Etsy, Wayfair, Target and Walmart.

UCP isn’t a closed Google tool. Any store that implements the standard can be reached by any agent that supports it too. In this article, we explain how UCP works, what’s in it for you as a retailer and how to get your store ready.

Agentic commerce shortens the customer journey to a single request

The classic customer journey has many steps: A user searches on Google, clicks an ad, lands on a landing page, accepts cookies, looks for the product, adds it to the cart, enters an address and pays. Customers drop off at every one of these steps.

With language models like Gemini, this process is shifting toward agentic commerce. The customer states an intent, such as “I need a waterproof two-person tent under €300 that will arrive by Friday,” and the AI handles the search, comparison and purchase preparation.

UCP makes it technically possible for an agent to initiate a purchase in your store. Without a shared standard, every AI platform would have to build a custom integration for every store. UCP solves that problem for both sides.

UCP is an open standard backed jointly by retailers and payment providers

UCP was initiated by Google and released as open source under the Apache 2.0 license. Shopify, Etsy, Wayfair, Target and Walmart co-developed it. More than 20 other companies back it, including Adyen, American Express, Best Buy, Mastercard, Stripe, Visa and Zalando.

According to its project page, the standard covers five core functions:

• Catalog: Agents search products and retrieve details.
• Cart: Agents build carts.
• Account linking: Users connect their retailer account to the agent, secured via OAuth 2.0.
• Checkout: The purchase is completed, including shipping options and discounts.
• Order management: status and fulfillment after the purchase

Shopping is the core. There’s a draft for lodging, and further areas such as food ordering have been announced. UCP works alongside existing protocols: Agent2Agent (A2A) for agent-to-agent communication, Agent Payments Protocol (AP2) for payments and Model Context Protocol (MCP) for connecting to language models.

The key difference from earlier solutions: UCP is decentralized. Implement the specification, and you’re reachable by every agent that supports the protocol, not just a single platform.

You remain the seller and keep the customer relationship

UCP changes the path to purchase, but not who does the selling. That gives retailers three advantages.

Fewer steps to purchase

Every step removed means fewer drop-offs. Google has announced UCP as the foundation for checkout on eligible product listings in AI Mode in Search and in the Gemini app. Payment runs through Google Pay, with payment and shipping details pulled from Google Wallet. So the customer no longer has to fill out a form.

Customer data and brand stay with you

On marketplaces, retailers often lose their direct line to the customer. With UCP, according to Google and the project page, you remain the merchant of record, meaning the seller in the legal sense. You receive the order data, ship under your own brand and manage post-purchase service yourself.

Higher profit per sale from buyers with clear intent

In many markets, click costs are rising. An agent that comes to your store arrives with a specific purchase task, not out of curiosity. How much this affects your acquisition costs and margin can only be measured once agent-driven sales are available in your market.

Your store describes its capabilities in a public UCP profile

For the technical implementation, you need to understand how an agent communicates with your store. Much like robots.txt, a file sits at a fixed location: At /.well-known/ucp your server publishes a UCP profile in JSON format. According to Google, it must be publicly accessible without a login.

The profile contains:

• the UCP capabilities your store supports, such as checkout, shipping, discounts and order management
• the specification versions you use
• your API endpoints
• payment settings, for example for Google Pay
• public keys that can be used to verify messages from your server

An agent reads this profile first and then knows whether and how it can buy from you. Your server determines which version and capabilities to use based on what both sides support. The specification is continuously expanded and versioned by date. To connect with Google, you enter your profile’s address in Merchant Center.

Payments rely on verifiable customer authorizations

The biggest hurdle in agentic commerce is trust: How does a customer approve a payment that an AI carries out?

UCP relies on the Agent Payments Protocol (AP2) for this. Payments are based on customer payment mandates and verifiable credentials. According to the project page, every authorization is cryptographically tied to the user’s consent. Payment providers plug in through a modular architecture.

For you as a retailer, this means sensitive payment data doesn’t have to travel openly through the protocol, and every payment can be traced back to a customer authorization. That reduces the risk of fraud and disputes.

Complete product data determines whether an agent recommends your product

With UCP, the goal is no longer to rank first for a search term like “buy running shoes.” An agent has to recognize your product as the right choice for a specific request. Read more in our article on Generative Engine Optimization.

Agents need decision criteria

The foundation is strong product data in Google Merchant Center. But agents need more than the standard fields. One example: If a user asks for a laptop for professional video editing, processor and RAM aren’t enough. Details on the display’s color gamut coverage and performance on typical tasks help. Google has announced new Merchant Center attributes for this, designed to make products discoverable in AI Mode, Gemini and Business Agent. They’ll initially be rolled out to a small group of retailers.

Accurate real-time data builds trust

Agents are meant to complete purchase tasks reliably. If your system shows three units in stock and checkout then fails, the sale is lost, and the agent has a reason to pick a different retailer next time. Accurate inventory, pricing and delivery times are therefore non-negotiable.

How to prepare your store now

UCP requires marketing and IT to work together. This sequence has proven effective:

1. Complete your product data: UCP draws on your product data . Fill gaps in your feed and use every relevant attribute, from material to pattern to energy_efficiency_class. The more an agent knows about a product, the more likely it is to recommend it.
2. Check your interfaces: Find out whether your e-commerce platform, such as Shopify, Shopware or Adobe Commerce, can expose carts and shipping costs through an API for external requests. Shopify is one of UCP’s co-developers.
3. Track availability: At launch in January 2026, UCP checkout was limited to US retailers; Google has announced a global rollout. Check Merchant Center regularly to see whether the integration is available in your market, and watch for errors in your structured product data.

Our agentic commerce page shows how we prepare retailers to sell through AI agents. For a comparison with OpenAI’s Agentic Commerce Protocol, see our article on ACP and Shopify.

Conclusion: In agentic commerce, the winners control their data and interfaces

UCP is meant to let retailers handle purchases through AI agents while remaining the seller themselves.

What makes the difference is well-structured product data, reliable real-time information and a technical integration that lets agents shop without friction. Retailers who lay these foundations now will be ready the moment agent-driven sales launch in their market.