First off, what exactly is agentic commerce?
Agentic commerce is the shift from a customer manually searching and buying, to an AI agent handling parts of that journey on their behalf. That includes finding relevant products and, increasingly, completing the purchase itself.
Google frames this plainly in its own announcement: agentic commerce is AI completing tasks on people's behalf, already happening rather than theoretical.
Why this needed a new standard
Letting AI agents shop on someone's behalf creates a coordination problem. Every AI platform would otherwise need a custom integration with every retailer, and vice versa.
Google's answer is the Universal Commerce Protocol (UCP), an open standard covering the full shopping journey. Per coverage of its January 2026 launch, Google built it with retail partners including Shopify and Walmart.
For any ecommerce business, this sits between your product data and the customer. How well you're set up for it affects whether AI agents can find and check out your products at all.
What is Google's AI shopping experience?
Two things power this: AI Mode as the conversational interface, and the Shopping Graph as the data engine underneath it.
AI Mode moves away from ranked links toward a conversational experience. Per a breakdown of the system, it embeds product recommendations directly into AI-generated results, with images, prices, and reviews.
This also runs inside the Gemini app, letting people browse shoppable listings and comparison tables without switching context.
None of it works without a large, current product dataset. Google's own figures put the Shopping Graph at over 50 billion listings, refreshed more than 2 billion times an hour.
That dataset comes from Merchant Center. Per an analysis of the pipeline, data starts in a retailer's own systems, gets mapped into Google's structure, and enters Merchant Center before joining the Shopping Graph. Thin or stale data here means a visibility problem before anyone even searches.
What is agentic checkout like in Google?
Agentic checkout lets a shopper go from browsing to buying without leaving the AI surface they're using.
The core of this is Universal Cart. Per Google's announcement, it lets shoppers save products across retailers into one cart, then check out with Google Pay or transfer to the merchant's site. The brand stays merchant of record either way.
Google confirmed this is expanding to Nike, Sephora, Target, Walmart, Wayfair, and Shopify merchants like Fenty and Steve Madden, rolling out in the U.S. this summer.
A few specific UCP capabilities make this work, per Google's updates announcement:
Real-time catalog access to variants, inventory, and pricing.
Multi-item cart building in one action.
Identity Linking, so shoppers get the same loyalty benefits they'd get logged into a retailer's own site.
Payment interoperability. PayPal now supports UCP, alongside Affirm and Klarna inside Google Pay.
Rollout is expanding market by market - the U.S. comes first, with Canada, Australia, and the U.K. to follow.
What this means for merchants on Google
Being listed in Merchant Center puts you in this ecosystem technically. Getting recommended is a separate bar.
Attribute detail matters more than it used to. AI Mode matches detailed queries to specific product attributes, not just a title and a competitive bid.
Reviews factor in too. Merchant Center now scores your product data based on quality, which factors in completeness, image quality, level of detail, and pricing accuracy according to a breakdown of the system. The higher the quality score assigned to your product data by Merchant Center, the more likely your product is to be surfaced in AI-driven recommendations.
Treat Merchant Center as a feed you maintain regularly, not something you configure once. Gaps or stale data limit whether AI Mode surfaces your products at all.
'Ranking' in Google's AI shopping assistant
Traditional Shopping ranking leaned on relevance and bid. Recommendation inside AI Mode runs on a different logic.
Attribute completeness carries real weight.
A product with a handful of filled-in fields loses out to a competitor with dozens whenever a shopper asks something specific.
Conversational attributes are the newest lever.
At Google Marketing Live 2026, Google introduced new optional fields built for AI shopping, per detailed coverage of the announcement. These include Q&A pairs, related products, and document links, built to answer the kind of questions people ask an AI assistant directly.
Reviews feed the recommendation logic itself.
Detailed, genuine reviews give the AI concrete signals to match against nuanced queries.
Spend doesn't move your position here.
AI Mode's logic runs on data completeness and query match, not ad budget.
Google's UCP & how your product data reaches Google's AI shopper
It's worth tracing the actual path your data takes. Appearing on Google and being usable by Google's AI shopping agent are two different things.
Data typically starts in your own systems (a PIM, ERP, CMS, or ecommerce platform), holding attributes like ID, brand, title, price, availability, variants, and GTINs. It gets mapped into Google's structure and submitted to Merchant Center, where Google checks for missing data and flags issues before approval.
Once approved, your products enter the Shopping Graph, joining attributes, reviews, and pricing into the same dataset AI Mode and Gemini pull from. UCP sits on top of that layer, giving an agent the ability to pull real-time details, build a cart, and recognize loyalty benefits through Identity Linking.
Getting into the Shopping Graph makes you discoverable. Supporting UCP is a separate, higher bar: it's what makes you transactable by an AI agent directly.
How e-commerce businesses can get ahead of the curve in agentic commerce
A few concrete places to start, roughly in order of effort versus impact:
Audit your Merchant Center feed completeness. Filling gaps in material, size, color, use case, condition, and category costs nothing and has the biggest immediate impact.
Write titles the way people actually ask questions. A generic title competes against thousands of near-identical listings. Specific, attribute-rich titles give the AI something distinct to match.
Build up genuine product reviews actively. These now carry real weight in AI-driven recommendations.
Try the new conversational attribute fields as they roll out. These are built to help products surface in natural-language queries.
Track your UCP readiness, especially on Shopify or any platform with a direct integration path.
Treat this as ongoing work. The Shopping Graph refreshes billions of listings hourly, so an accurate feed today can go stale fast without upkeep.
How this changes the customer journey & shopping experience
The customer journey has traditionally moved in linear steps: search, browse, compare, land on a product page, check out, often across several tabs. Agentic commerce compresses those steps into something closer to a single conversation.
A shopper describes what they want, gets a curated set of products pulled from tens of billions of listings, refines that conversationally, and increasingly completes the purchase directly, without leaving Search or the Gemini app.
For shoppers, this means less friction and less tab-switching. For merchants, a growing share of the customer relationship (discovery, comparison, even loyalty recognition through Identity Linking) now happens on Google's surfaces before a customer lands on your own site, if they land there at all.
Your brand and product experience still matter. But the moment a customer forms their first impression is moving earlier in the journey, onto a surface you don't fully control. The businesses that adapt well here will be the ones that put real care into their product data, since for a growing share of customers, that data is the first storefront they see.
