Google's Universal Commerce Protocol and Merchant Center: The new infrastructure for AI shopping

AI shopping is shifting from "new feature" to infrastructure. Google's latest commerce update introduces a set of tools that move commerce deeper into conversational experiences, including a new open standard called the Universal Commerce Protocol (UCP), branded chat experiences via Business Agent, and incentive-led Direct Offers inside AI Mode. Together, they signal a shift from AI helping users discover products to AI actively guiding them through evaluation and checkout.

This goes beyond ads appearing in chat. The opportunity is to build the transactional rails for agent-led commerce, with ChannelEngine acting as the control panel that powers these commerce flows, and Merchant Center playing a central role to feed Google’s agentic shopping experience.

Before stepping into the strategic implications, here is what Google has introduced and what is already live.

Google's announcement, explained simply

Announced on 11 January 2026, the update introduced 3 elements that map to different layers of the shopping journey:

1) Universal Commerce Protocol (UCP)

UCP is Google’s new open-source standard designed to help AI agents communicate with commerce systems across the buying journey. Importantly, UCP is not conceptual. It is already active in powering native checkout flows inside Google AI Mode and the Gemini app for a select number of eligible US merchants.
At the time of writing, the feature is available only in the United States and only for a limited set of participating merchants and partners through an early access rollout. Google is releasing the protocol in phases, and merchants can express interest through a participation program.

Google positions UCP as ecosystem infrastructure, not a proprietary format. It was developed in collaboration with major commerce and payments partners, including Shopify, Walmart, Wayfair, Target, Etsy, and others. More than 20 commerce and payments organizations have endorsed or integrated with the standard.

In practical terms, UCP enables smoother “ask, decide, buy” journeys inside conversational interfaces using payment methods and wallets shoppers already trust.

2) Business Agent

Business Agent brings brand-led conversational help into Search, positioned like a “virtual sales associate” that can answer product questions in a brand’s voice.

It is rolling out through Merchant Center for eligible US retailers and is already live in pilot environments. It connects structured product data with conversational responses so that brands can guide decision-making without redirecting users off the platform.

3) Direct Offers

Direct Offers allows retailers to surface personalized incentives inside AI Mode when signals indicate purchase intent.

This is not traditional keyword bidding. It is context-aware promotion inside a conversational decision flow. Retailers can test discounts and other value-led incentives with tighter margin guardrails. Direct Offers is being piloted in supported categories and markets.

Why this is bigger than "another ads update"

For years, Google Shopping visibility was largely a function of feeds, bids, and landing pages. This announcement signals a structural change: shoppers can discover, evaluate, and potentially purchase products inside conversational flows across Google surfaces, including AI Mode and Gemini.

That changes the competitive game in two important ways:

• Selection replaces ranking. In conversational commerce, the agent narrows choices down to a small set of products. That means more pressure on data confidence, not just keywords.
• Eligibility matters. Protocols and platform integrations increasingly determine what can be bought inside the flow, not just what can be shown.
  
  

UCP: The infrastructure layer for in-chat commerce

Universal Commerce Protocol (UCP) is best understood as a connective layer for agent-driven commerce. It is not a “new ad format.” It is a standard intended to make agent-driven shopping interoperable across merchants, platforms, and payment networks.

Google is aligning UCP with other emerging standards around agent communication (such as Agent2Agent (A2A), Agent Payments Protocol (AP2), and Model Context Protocol (MCP)), and payments, creating an interoperable framework for agentic commerce. When implemented, UCP can enable checkout directly on Google surfaces such as AI Mode in Search or the Gemini app.

For brands and retailers, the implication is clear. Participation in these agent-led shopping flows increasingly depends on structured, reliable commerce data and integrations that allow systems to exchange information programmatically. Clean product feeds and well-governed data pipelines are prerequisites for being eligible to transact within conversational experiences.

Merchant Center is becoming the Google AI shopping control plane

Google also reinforced Merchant Center’s expanded role in AI shopping.

The practical reality: more attributes, more complexity

What changes with AI shopping is not just how products are shown, but also new attributes that your product data needs to have. Google is rolling out new Merchant Center attributes designed specifically for conversational experiences, where an agent needs more context than a standard product listing can provide.

Examples include Product_faq (retailer-provided Q&A), Product_use_cases (best-fit scenarios), Product_substitutes (alternatives when something is unavailable), Checkout_eligibility (signalling whether an item is eligible for agent-led purchase), and more.

From "hygiene attributes" to "strategy attributes"

Historically, many feed fields were hygienic. You filled them to comply and improve relevance marginally. In an AI shopping environment, a new class of attributes becomes strategic because it helps an agent understand intent and relationships, not only product facts.

A useful way to think about it is: Can the agent answer questions, guide a decision, and keep the sale moving? Here are three practical jobs your product data now needs to do:

1) Help the agent answer buyer questions

Conversational shopping increases the number of “micro-questions” a shopper asks. Capturing the answers in structured product data reduces ambiguity, improves consistency, and increases the agent’s confidence in what it recommends.

2) Help the agent recommend the right product for the situation

Agents need signals that connect products to scenarios, not just specifications.

3) Help the agent keep the sale alive

In conversational commerce, the agent can prevent drop-off by presenting a better alternative when something i(ps unavailable or by recommending what completes the purchase.

The operational catch:

When the number of attributes grows, two things happen fast:

• It becomes easier for errors to creep in (conflicts, stale values, mismatched variants, incomplete coverage).
• Product data has a more direct impact on performance because agents use it to decide what to show, what to recommend, and what to exclude.
  
  

Knowing which attributes matter is the easy part. The harder part is rolling them out across thousands of SKUs, multiple teams, and constant catalog changes.

How to operationalize strategy attributes at scale

You do not need to wait for "full agentic checkout maturity" to start preparing. The goal is to make your catalog more machine-decidable inside Google’s ecosystem.

Put governance in place before you scale

• Assign owners: core commerce data, content, compatibility, substitution, offers.
• Set refresh cadences: fast for stock and price, slower for FAQs and use cases.
• Define QA rules to catch conflicts and missing values.
  
  

Build an attribute strategy by category, not by SKU

Start with your top 1 to 3 revenue categories and define:

• Top intents (use cases)
• Top objections (FAQ themes)
• Top attach opportunities (accessories)
• Top "save the sale" options (substitutes)
  
  

Then expand.

Enforce formatting discipline

Google is explicit that missing or inaccurate product information can cause issues and reduce serving across Google surfaces. Treat formatting like an engineering problem:

• controlled units and conversions
• standardized values for common fields
• consistent variant rules
  
  

Use rules and enrichment to avoid constant source catalog rewrites

If you rely on manual edits in multiple systems, you will fall behind as attributes expand.

Aim for a layered approach:

• Source catalog = truth
• Enrichment layer = transforms and derives fields
• Channel mapping = outputs the format each destination needs
  
  

Test like a growth team

Do not try to perfect everything at once. Run a controlled rollout:

• Phase 1: coverage (what % of SKUs have complete strategy attributes)
• Phase 2: visibility (impressions, eligibility, click quality)
• Phase 3: conversion signals (assisted conversions, return rate changes where measurable)
  
  

Treat Direct Offers as a separate track with guardrails

If Direct Offers is available in your market and category, approach it as experimentation with margin protection:

• define margin floors
• exclude volatile inventory
• test offer types carefully (discounts, free shipping, bundles over time)
  
  
  
  

Where ChannelEngine fits (without adding complexity)

As Google adds more fields and more AI-driven surfaces, the risk for multichannel sellers is fragmentation: different titles, different attributes, different availability signals across channels.

The good news is you do not need a separate "AI Mode feed." Google’s AI Mode reads from the same Merchant Center product feed used for Google Shopping. That means the work you do to improve feed quality and enrich attributes can influence both traditional Shopping results and AI-driven recommendations.

As a multichannel marketplace integration platform, ChannelEngine helps you manage that Merchant Center feed from one place, centralize product enrichment, and distribute consistent, channel-specific outputs at scale. When updates flow automatically, you reduce the inconsistencies that AI shopping experiences penalize and make product data easier for conversational systems to interpret.

The retailers who win will be the ones who can operationalize richer attributes, govern them at scale, and continuously test how their catalog is interpreted in AI-driven discovery.