How to Set Up Google UCP: Complete Step-by-Step Guide
Reading time: 22 min Table of Content What You Need Before Starting Step 1: Set Up Your Google Merchant Center Account Step 2: Build an AI-Ready Product Feed Step 3: Add Conversational Commerce Attributes Step 4: Implement Product Structured Data on Your Site Step 5: Connect Your Ecommerce Platform to Merchant Center Step 6: Configure UCP Settings in Merchant Center Step 7: Test and Validate Your Setup Common Mistakes to Avoid What to Expect After Setup How to Monitor AI Visibility Next Steps Instead of buyers scrolling through search results and clicking links, AI agents in Google’s AI Mode and Gemini now evaluate product data directly, compare options, and recommend products to buyers in real time. If your products aren’t set up for UCP, they’re invisible to these AI agents. It doesn’t matter how good your Google Ads are or how well your site ranks organically. AI-powered shopping is a separate channel, and it requires its own setup. This guide walks you through the complete Google UCP setup, step by step. It’s platform-agnostic. Whether you’re on Shopify, WooCommerce, BigCommerce, Magento, or a custom platform, the Google side of UCP works the same way. What You Need Before Starting Before diving into UCP setup, make sure you have: A Google Merchant Center account (free to create) An active product feed connected to Merchant Center Products listed and available for purchase on your website Access to your website’s code or CMS for structured data implementation A Google Business Profile (recommended but not required) Step 1: Set Up Your Google Merchant Center Account Google Merchant Center is the foundation of everything UCP-related. It’s where Google stores, validates, and serves your product data to AI agents. Without it, UCP doesn’t work. Creating Your Account If you already have a Merchant Center account, skip to the optimization section below. If not: Go to merchants.google.com Sign in with your Google account (use your business Google account, not personal) Enter your business name and website URL Select your country and time zone Accept the terms of service Verify and Claim Your Website Google needs to confirm you own the website you’re listing products from: In Merchant Center, go to Settings → Website Enter your website URL Choose a verification method (HTML tag, Google Analytics, Google Tag Manager, or DNS record) Complete the verification Claim the URL so only your account can use it HTML tag verification is the fastest for most merchants. Just add the meta tag to your site’s head section. Enable Free Product Listings This is often overlooked. Google offers free product listings across Search, Shopping, Images, Maps, and YouTube. These listings also feed into AI Mode recommendations. In Merchant Center, go to Growth → Manage programs Enable “Free product listings” if not already active Ensure your products are opted in to “Surfaces across Google” This gives your products maximum visibility across Google’s ecosystem, including the AI agents that power UCP. Connect to Google Ads (Optional but Recommended) If you run Google Ads, linking your Merchant Center account gives AI agents additional performance signals about your products: In Merchant Center, go to Settings → Linked accounts Click “Google Ads” and link your account This enables Performance Max campaigns and gives Google more context about your products You don’t need Google Ads for UCP to work, but the additional data helps Google’s AI agents understand which of your products perform well and deserve recommendations. Step 2: Build an AI-Ready Product Feed Your product feed is the data AI agents use to evaluate your products. If the data is incomplete, vague, or inconsistent, AI agents will skip your products and recommend a competitor’s instead. This is the most important step in the entire process. Required Attributes Every product in your feed must have: id: Unique product identifier title: Descriptive, attribute-rich product title description: Functional product description (not marketing copy) link: URL to the product page on your website image_link: URL to the primary product image availability: in_stock, out_of_stock, or preorder price: Current price with currency code brand: Product brand name condition: new, refurbished, or used gtin or mpn: Global Trade Item Number or Manufacturer Part Number Missing any of these will either get your product disapproved or severely limit its visibility to AI agents. Recommended Attributes (Critical for UCP) These aren’t technically required by Merchant Center, but they’re essential for UCP. AI agents use these to compare and recommend products: product_type: Your own product categorization google_product_category: Google’s standard taxonomy additional_image_link: Multiple product images from different angles color: Standardized color names (“Navy Blue” not “Ocean Mist”) size: Standard sizing (“XL” not “Extra Large” on some and “XL” on others) material: Specific materials (“100% Organic Cotton” not “Cotton Blend”) weight: Product weight with units shipping_weight: Weight for shipping calculation shipping_dimensions: Length, width, height age_group: For apparel and accessories gender: For apparel products pattern: If applicable Optimizing Titles for AI Agents This is where most merchants get it wrong. Titles optimized for Google Shopping ads are not the same as titles optimized for AI agents. Shopping ad titles front-load keywords for click-through rate. AI agent titles need complete product information because the agent is evaluating whether to recommend your product, not whether a human will click on it. Bad title: “Blue Jacket XL – Free Shipping – Best Price” Good title: “Men’s Waterproof Hiking Jacket, Navy Blue, XL, Gore-Tex Shell, Rated to -10C, 340g” Bad title: “Industrial Air Compressor – SALE” Good title: “DeWalt 60-Gallon Two-Stage Industrial Air Compressor, 175 PSI Max, 6.5 HP, 230V Single Phase” Bad title: “Premium Dog Food” Good title: “Orijen Original Grain-Free Dry Dog Food, 25 lb Bag, Fresh Free-Run Chicken and Turkey” AI agents parse every word. Include: brand, product type, key specs, size/weight, distinguishing features. Leave out: promotional language, “best price,” “free shipping,” urgency words. Writing Descriptions That AI Agents Can Use Your descriptions need to be functional, not persuasive. AI agents don’t respond to emotional language. They need facts. Bad description: “Experience the ultimate in comfort with our premium jacket. You’ll love the way it feels. Perfect for any adventure!” Good description: “Waterproof hiking jacket with Gore-Tex shell membrane. Rated to -10C. Sealed
Agentic Commerce: What Ecommerce Brands Need to Know in 2026
Reading time: 22 min Table of Content What Is Agentic Commerce? Key Players in Agentic Commerce Agentic Commerce Adoption: Who Is Shopping This Way? Why Product Data Is the Competitive Moat in Agentic Commerce How Agentic Commerce Changes SEO Strategy How to Prepare for Agentic Commerce in 2026 In 2026, AI agents are starting to shop on behalf of customers. A person tells Gemini, ChatGPT, or Copilot what they need. The agent finds matching products, compares options, checks pricing and availability, and in some cases completes the purchase inside the conversation. No website visit required. This is agentic commerce. And if you sell products online, you need to understand what it means for your business. What Is Agentic Commerce? Agentic commerce describes a model where AI agents act as autonomous shopping assistants. Instead of a customer manually searching, browsing, and buying, they delegate part or all of that process to an AI agent. Here is what that looks like in practice: A customer opens Google Gemini and says, “Find me a waterproof hiking jacket under $200 that works for Pacific Northwest rain.” The agent searches across merchant catalogs, evaluates product attributes (waterproof rating, material, price, reviews), compares three to five options, and presents them with relevant details. The customer picks one, and the agent can complete the checkout without the customer ever visiting the brand’s website. This is not a concept. Google AI Mode already supports in-conversation checkout for select merchants. Shopify has launched Agentic Storefronts that enable this flow natively. Microsoft is building Copilot Checkout. The infrastructure is live. The critical shift is this: your product data is now your storefront. When an AI agent evaluates your products, it does not see your website design, your brand photography, or your homepage hero banner. It sees your product titles, descriptions, attributes, pricing, and structured data. That is all it has to work with. Key Players in Agentic Commerce Google Google is leading the push with the Universal Commerce Protocol (UCP), launched in January 2026 with Shopify. Google AI Mode now supports product discovery and in-conversation checkout. Gemini serves as the consumer-facing agent. Google Business Agent lets merchants configure how their products appear in AI-powered interactions. Direct Offers enable merchants to surface deals and promotions through AI channels. For a deep dive on UCP, read our guide: What Is the Universal Commerce Protocol (UCP)? Shopify Shopify co-developed UCP and is building native support through Agentic Storefronts. This lets Shopify merchants expose their product catalogs to AI agents with built-in support for discovery, cart, checkout, and post-purchase flows. Shopify has also launched an Agentic Plan that allows non-Shopify brands to participate in UCP through Shopify’s infrastructure, signaling how seriously they are investing in this channel. OpenAI OpenAI and Stripe built the Agentic Commerce Protocol (ACP), a competing standard. ChatGPT already supports product browsing and comparison, and 23-35% of adults have used it to browse products. However, OpenAI pulled back native checkout functionality on March 6, 2026, making ACP’s transaction layer less mature than UCP’s for now. Microsoft Microsoft is building Copilot Checkout, integrating shopping capabilities into its AI assistant. With Copilot embedded across Windows, Edge, and Microsoft 365, this gives Microsoft a distribution advantage for AI-powered commerce. Payment Networks Visa, Mastercard, Stripe, and Adyen are all backing these protocols. Their involvement is critical because it means the payment infrastructure for AI-completed transactions is being built by the companies that process the majority of global digital payments. This is not an experiment. These are infrastructure companies making long-term bets. Agentic Commerce Adoption: Who Is Shopping This Way? Let’s be honest about where things stand. Agentic commerce is real, but it is early. What the data shows: Shopify reported a 15x increase in AI-originated orders. That sounds dramatic, but it is growing from a small base. A 15x increase on a tiny number is still a relatively small number. Forrester’s March 2026 research found that completing purchases inside AI conversations is the least adopted use case for AI shopping tools. Most consumers use AI for product discovery and comparison, then complete checkout on the merchant’s website. 23-35% of adults have browsed products in ChatGPT, showing significant awareness and experimentation. LLM traffic still accounts for roughly 2% of total referral traffic to ecommerce sites. Growing fast, but not dominant yet. The honest assessment: Agentic commerce is in the “early adopter” phase. The technology works. The infrastructure is being built. Consumers are experimenting. But mainstream adoption of AI-completed checkout is still ahead of us. The comparison that matters: Mobile commerce followed a similar trajectory. Early mobile shopping was clunky and low-volume. Skeptics said people would never buy on their phones. Today, mobile accounts for over 70% of ecommerce traffic. The brands that optimized for mobile early captured an outsized advantage. Agentic commerce is following the same curve. The question is not whether it will matter. It is whether you will be ready when it does. Why Product Data Is the Competitive Moat in Agentic Commerce In traditional ecommerce, brands compete on website experience, brand perception, advertising spend, and pricing. In agentic commerce, the primary competitive factor is product data quality. AI agents cannot squint past bad product titles. They cannot look at a photo and figure out what “Summer Vibes Collection #3” actually is. They cannot infer that your product is waterproof if you did not include a waterproof rating in your attributes. What the data shows: Specific, descriptive product titles that include key attributes (material, size, use case, brand) Complete product attributes covering every relevant specification Accurate inventory and pricing updated in real time Structured variant information (sizes, colors, configurations) with consistent naming Functional product descriptions that describe what the product does, not just how it looks The Inference Advantage Products that are easier for AI to understand sell more in agentic commerce. This is the “inference advantage.” If two hiking jackets are functionally identical but one has detailed attributes (waterproof rating: 20,000mm, weight: 12oz, material: 3-layer Gore-Tex, fit: regular) and the other says “great
What Is the Universal Commerce Protocol (UCP)? Everything You Need to Know
Reading time: 22 min Table of Content Who Built the Universal Commerce Protocol and Why How the Universal Commerce Protocol Works UCP vs ACP: What Is the Difference? What UCP Means for Ecommerce Brands The Work Overlaps With Traditional SEO UCP Rollout Status: What Is Live in 2026 How to Prepare Your Ecommerce Store for UCP This guide explains what UCP is, how it works, who built it, and what you need to do about it. Who Built the Universal Commerce Protocol and Why UCP was developed by Google and Shopify as a joint initiative, but the coalition behind it is much larger. Walmart, Target, Etsy, Wayfair, Visa, Mastercard, Stripe, and Adyen have all backed the protocol. Why? Because AI agents need a common language to interact with online stores. Without a standard, every AI platform would need custom integrations with every ecommerce platform. That does not scale. UCP solves this by creating one protocol that any AI agent can use to interact with any merchant. Think of it like how HTTPS standardized secure web communication. UCP is doing the same thing for AI-powered commerce. The backing from major payment networks (Visa, Mastercard, Stripe, Adyen) is significant. It signals that the infrastructure for AI-completed transactions is being built at every layer, from discovery to payment processing. How the Universal Commerce Protocol Works UCP creates a structured pathway for AI agents to interact with your store. Here is how it works at each stage: Discovery When a customer asks an AI agent to find a product, the agent searches across merchant catalogs semantically. This means it understands product attributes, not just keywords. If someone asks for “a lightweight running shoe for wide feet under $150,” the agent evaluates your product data against that request. Your product titles, descriptions, attributes, and structured data determine whether the agent finds you. Incomplete or vague data means the agent skips you. Cart Once the agent identifies a product, it can add items to a cart, apply available discounts, select variants (size, color, quantity), and calculate shipping. All of this happens through UCP without the customer visiting your website. Checkout The agent can complete the purchase inside the AI conversation. The customer confirms payment, and the transaction processes through your existing payment infrastructure. You stay the merchant of record. The order flows into your system like any other order. Post-Purchase UCP also handles post-purchase interactions. Customers (or their agents) can track orders, initiate returns, and get support, all through the AI interface while your existing fulfillment and support systems handle the backend. You Stay in Control This is a critical point: UCP does not disintermediate you. You remain the merchant of record. Orders go through your payment processor. You fulfill orders. You handle customer relationships. UCP just gives AI agents a standardized way to interact with your store. Need Help Setting Up UCP? We can audit your product data and get you UCP-ready in weeks, not months. Get Started UCP vs ACP: What Is the Difference? You may have heard about the Agentic Commerce Protocol (ACP), built by OpenAI and Stripe. ACP is a competing standard with similar goals. However, on March 6, 2026, OpenAI pulled back its native checkout functionality inside ChatGPT. This makes UCP the more stable and actively deployed protocol right now. Both protocols may coexist long-term. For now, UCP has stronger institutional backing and active deployment through Google AI Mode and Shopify Agentic Storefronts. What UCP Means for Ecommerce Brands UCP fundamentally changes what it means to have a “storefront.” Your website is still important, but it is no longer the only place customers interact with your products. AI agents are now browsing your catalog on behalf of shoppers, and your product data is the storefront they see. Your Product Data Is Your AI Storefront When an AI agent evaluates your products, it reads your titles, descriptions, attributes, pricing, inventory, and structured data. That is all it has. There is no product photography to fall back on. No clever homepage design. No brand story in the header. Just data. If your product title says “SKU-4892-BLK” instead of “Men’s Waterproof Hiking Boot, Black, Size 10,” the agent cannot recommend you. It does not know what you sell. AI Agents Are Literal Unlike human shoppers, AI agents cannot compensate for sloppy data. A person might look at a product image and figure out what “Summer Collection Item #7” actually is. An AI agent cannot. It takes your data at face value. Missing attributes, inconsistent naming, vague descriptions, and incomplete variant information all reduce your chances of being recommended. Clean Catalogs Win Brands with clean, structured, complete product catalogs will get recommended first. This is not speculation. It is how the technology works. AI agents rank products based on how well they match the customer’s request, and matching requires accurate, detailed product data. Want help getting your store ready? See our agentic commerce services. The Work Overlaps With Traditional SEO Here is the good news: the work required to make your products UCP-ready is largely the same work that improves your traditional SEO. Clean product data, rich structured markup, accurate Merchant Center feeds, and detailed product descriptions help you rank in Google Search AND get recommended by AI agents. You are not building two separate strategies. You are building one strategy that works across both channels. UCP Rollout Status: What Is Live in 2026 UCP is live, but still in early stages. Here is where things stand: Google UCP is active in AI Mode and Gemini. Select merchants can already receive orders through AI-powered conversations. Google Business Agent is available for eligible merchants, enabling automated product recommendations and checkout. Shopify Agentic Storefronts are rolling out in phases. Not all merchants have access yet, but Shopify is actively expanding the program. Shopify has also launched an Agentic Plan that allows non-Shopify brands to participate in UCP through Shopify’s infrastructure. Adoption Data Forrester’s March 2026 research shows that AI-assisted commerce adoption is growing, but completing purchases