Google’s Universal Commerce Protocol (UCP) is not a future roadmap item anymore. On February 11, 2026, Vidhya Srinivasan, Google’s VP/GM of Advertising & Commerce, confirmed that Wayfair and Etsy already support UCP, letting shoppers buy directly inside AI Mode. Waitlists are still the entry point for most merchants, but the capabilities behind them are shipping in production right now.

That timeline matters because UCP is a bigger undertaking than its sibling protocol, the Agentic Commerce Protocol (ACP). ACP is mostly scoped to checkout, fulfillment, and payment. UCP goes further, covering the entire commerce lifecycle through six distinct capabilities. For SEOs and ecommerce teams, this means the job is shifting from optimizing for clicks to optimizing for selection — making sure an AI agent chooses your product, not a competitor’s, through data integrity, complete product signals, and commerce data that is genuinely machine-readable. Call it Agentic Commerce Optimization (ACO): a discipline that did not exist eighteen months ago and now needs a checklist.

The six capabilities UCP is built on

Google laid out its methodology the day UCP launched, and it breaks down into six core capabilities: product discovery, which governs how agents find and surface inventory during research; cart management, covering multi-item baskets, dynamic pricing, and complex basket rules; identity linking, using OAuth 2.0 authorization for personalized experiences and loyalty recognition; checkout, handling session creation, tax calculation, and payment; order management, run through webhook-based lifecycle and logistics updates; and vertical capabilities, extensible modules for specialized needs like travel booking windows or subscription schedules.

UCP’s schema authoring guide shows these capabilities are defined through versioned JSON schemas that form the protocol’s foundation. For anyone coming at this from an SEO background, that means familiar structured data properties — offers, aggregateRating, shippingDetails — stop being just rich-snippet fuel. They become the exact fields an agent queries throughout the entire discovery-to-purchase process.

Schema.org still matters, even though UCP doesn’t build on it directly

UCP’s technical specification uses its own JSON schema-based vocabulary rather than extending schema.org. That does not make schema.org optional. As Pascal Fleury put it at Google Search Central Live in December, “schema is the glue that binds all these ontologies together.” The practical division of labor is this: UCP handles the transaction itself, while schema.org is what helps an agent decide who to transact with in the first place.

That makes a full product schema audit table stakes, not a nice-to-have. Product schema needs complete coverage of the core fields — name, description, SKU, GTIN, brand, images, and offers — and each offer needs price, priceCurrency, availability, URL, and seller populated, plus aggregateRating and review data to establish a positive third-party signal. Every product variant should output correctly, shippingDetails should include delivery estimates, and Organization or Brand markup should be in place to support “merchant of record” verification (fall back to Person markup if Organization does not apply). A dedicated FAQPage is also worth setting up, since these get incorporated alongside product-level FAQs as part of an agent’s decision-making process. None of this is new advice on its own — it is standard SEO hygiene — but it needs auditing now, before UCP rollout accelerates and gaps become expensive.

Your Merchant Center feed needs its own audit

UCP uses an existing Merchant Center feed as its discovery layer, which means the requirements go beyond on-site schema into fields that live only inside Merchant Center itself. Return policies need to be complete, including return costs, return windows, and policy links, since these get used both at checkout and as a factor in whether a product gets selected at all — advanced accounts need this set at each sub-account level. Customer support information matters too, partly because some entry-level support queries can be handled by an agent directly, reducing load on human support teams.

Three fields are easy to miss but functionally required. Products need the native_commerce attribute set to even be eligible for agentic checkout. Every product needs a stable identifier that correlates with the product ID used in the checkout API. And any product carrying a safety or usage warning needs the consumer_notice attribute applied. Google’s own recommendation is to populate these through a supplemental data source in Merchant Center rather than editing the primary feed directly, which avoids formatting errors that could invalidate the whole feed. It is also worth checking a given product category against UCP’s product restrictions list before assuming eligibility.

Conversational commerce attributes reduce hallucination risk

In the UCP announcement, Srinivasan described a new set of data attributes built specifically for the conversational commerce era: “we’re announcing dozens of new data attributes in Merchant Center designed for easy discovery in the conversational commerce era, on surfaces like AI Mode, Gemini and Business Agent. These new attributes complement retailers’ existing data feeds and go beyond traditional keywords to include things like answers to common product questions, compatible accessories or substitutes.” The purpose is straightforward: the more precisely a product is described, the less an agent has to guess, and guessing is where hallucination creeps in during the selection process.

Beyond standard product and brand FAQs, three attribute categories are worth prioritizing. Compatibility data creates up-sell opportunities by telling an agent what pairs with what. Substitution data gives an agent something to recommend when an item is out of stock rather than losing the sale entirely. Related-product data supports cross-sell. It also pays to get specific rather than generic — a product described only as “purple” gives an agent less to work with than one described as “dark purple,” or even a named shade like “Wolf,” which still falls under purple but carries more precise signal. The same logic applies to size, material, and any other variant attribute.

Understanding query fan-out with a real example

Optimizing conversational attributes well increases the odds of surfacing across fan-out query results — the branching set of sub-queries an AI system runs to fully answer a broader question. Tools like WordLift’s Visual Fan-Out simulator make this visible by showing how a single product image decomposes into multiple search intents.

Running On’s Cloudsurfer Max shoe through that simulator across three horizons shows the pattern clearly. At the surface level, the tool correctly identifies the product as an On shoe from the Cloudsurfer range. One horizon deeper, it expands into the broader category of performance running gear. A horizon beyond that, fan-out queries surface related gear across the category more generally. That progression is useful diagnostic information: it shows how widely an LLM is willing to range from a single product image, and it suggests where a brand should be reinforcing attributes if it wants to stay visible as the query fans further out.

UCP’s roadmap goes well beyond retail

UCP is not staying scoped to single-item retail purchases. Its published roadmap points toward multi-item carts and complex baskets — bundling, promotions, and realistic tax, shipping, and fulfillment logic beyond a single-item checkout. Loyalty and account linking are planned too, standardizing how agents apply points, member pricing, and cross-merchant benefits. Post-purchase support is on the list as well, covering order tracking, returns, and handoff to customer service. Personalization signals — cross-sell and upsell logic, wishlists, purchase history, context-based recommendations — are also planned, alongside expansion into entirely new verticals: travel, services, digital goods, and food and restaurant use cases.

Any brand operating in those adjacent verticals should treat this roadmap as a reason to move early rather than wait for a finished spec, since eligibility requirements are likely to mirror the retail rollout closely.

Third-party trust signals still decide selection

No amount of on-site data integrity fully replaces external validation. Third-party review platforms such as Trustpilot and G2 are frequently cited and trusted across most major LLMs, which means collecting and maintaining positive brand and product reviews remains directly relevant to whether a product gets selected during agentic discovery — it is not a separate workstream from technical readiness, it is part of it.

Frequently asked questions

What’s the difference between ACP and UCP?

ACP focuses on the checkout-to-fulfillment-to-payment journey. UCP covers the entire commerce lifecycle across six capabilities, from product discovery through post-purchase support, and is designed so merchants integrate once rather than building separate connections for every agent platform.

Do I need to rebuild my schema markup for UCP?

Not from scratch. UCP uses its own JSON schema vocabulary rather than extending schema.org, but schema.org markup remains essential for helping agents decide which merchant to transact with. A full audit of existing product schema is the right first step.

What is the native_commerce attribute?

It is a Merchant Center feed attribute that determines whether a product is even eligible for agentic checkout under UCP. Products without it set will not be considered, regardless of how complete the rest of the feed is.

Is UCP only relevant to retail businesses?

Not for long. UCP’s roadmap already extends into travel, services, digital goods, and food and restaurant use cases, so businesses in those verticals should prepare alongside traditional ecommerce merchants.

Where to start

Join the UCP waitlist, and treat the wait as prep time rather than downtime. Get Merchant Center in order — return policies complete, native_commerce attribute set. Make sure developers have actually read the UCP documentation rather than skimming a summary. Populate conversational commerce attributes, especially question-and-answer content, compatibility, and substitution data. And run a full schema audit now, before rollout accelerates further. Brands treating UCP as a distant announcement rather than an active shift, similar to the gap already showing up in why most top retailers are currently invisible to agentic commerce, are the ones most likely to be caught flat-footed. For a broader view of the space, see our complete guide to agentic commerce, and for merchants deciding which spec to prioritize first, our breakdown for small merchants is a useful next read. Feed ownership is also worth settling internally now — see why your product feed is an SEO asset and who should own it.