Universal Cart and UCP in Europe: what changes and what to prepare now

More of the customer journey can now happen outside your website. That changes what ecommerce teams need to prepare.

Google’s Universal Commerce Protocol, or UCP, creates a common way for AI shopping experiences, merchants and payment systems to exchange the information needed for product discovery, cart creation, checkout and order handling. Universal Cart makes that more visible to the customer by keeping shopping activity connected across Google products.

For European ecommerce leaders, the useful question is not simply what UCP is. It is what needs to be ready before these flows become more widely available in Europe: product data, Merchant Center, checkout rules, customer-data controls, internal ownership and governance.

The useful distinction is between being ready and being live. A feature does not need to be available to every European consumer before it matters to a European retailer.

What Google Universal Cart is

Universal Cart is Google’s shopping cart for use across merchants and across Google services. Google announced it on 19 May 2026 and is rolling it out on Search and the Gemini app in the US, with YouTube and Gmail to follow. Google, 19 May 2026

A shopper can add products from several retailers while searching, chatting with Gemini, watching YouTube or reading Gmail. The cart then works in the background: it tracks price drops and price history, flags when items are back in stock and warns about incompatible products. Because it is built on Google Wallet, it can also take payment perks and loyalty information into account.

At checkout, the shopper can pay with Google Pay in a few taps or transfer the cart to the merchant’s own site. Google says the brand remains the merchant of record either way. The checkout runs on UCP, which is why the two are easy to confuse: Universal Cart is what the shopper sees, and UCP is the protocol underneath.

The merchant may still own the transaction, fulfil the order, handle returns, maintain the customer relationship and remain Merchant of Record. But more of the journey connecting discovery, comparison and purchase may happen outside the merchant’s own website.

What UCP actually does

UCP is easy to overcomplicate. It is not a website feature, a checkout button or a switch that makes a company ‘agentic commerce ready’.

It is a commerce protocol: a shared set of rules that allows different systems to exchange the information needed to support shopping and checkout. Google describes it as an open standard for agentic commerce that enables agents and systems to work together across the commerce ecosystem. Google UCP overview

Commercially, UCP helps systems answer questions such as:

• Is this product available?
• Is the correct variant available?
• What is the current price?
• Can this product participate in a particular checkout flow?
• Which shipping and fulfilment options apply?
• Which policies or restrictions need to be shown?
• How should the cart be handed to the merchant?
• How should an order be completed?
• How does the merchant remain responsible for the transaction?

That is why UCP matters beyond the technical team. Another system needs to be able to read your commerce information and use it correctly.

In business language, four functions matter most.

The practical change is that more of checkout can now happen through structured system-to-system interactions.

UCP, ACP and AP2: which rails matter

UCP is Google’s rail for agentic commerce, but it is not the only one. For a European leader, the useful question is not which protocol to build first. It is which customer journeys matter to the business, and which rails those journeys run on.

ACP: OpenAI’s route to checkout

OpenAI’s Agentic Commerce Protocol (ACP), developed with Stripe, powers checkout inside ChatGPT. It launched first in the US. If your customers research in ChatGPT as much as in Google, ACP is the parallel rail to watch.

AP2 and payment trust

UCP covers how commerce systems interact. Google’s Agent Payments Protocol, AP2, addresses a different but related problem: how an AI-assisted payment can be authorised and recorded so it can be checked later. Google describes AP2 as an open protocol designed to create a common language for secure, agent-led payments. Google Cloud AP2 announcement

The important point is not the protocol name. AI-assisted payments need more than convenience. Customers, merchants, payment providers and regulators need clear evidence of who authorised the payment and who is responsible.

For European organisations, this matters when payment, identity and customer data begin to move through AI-assisted commerce systems.

The preparation is the same for every rail: product data, claims and policies that machines can read and trust; checkout rules that still work when the journey starts elsewhere; and clear ownership of which channels are switched on. That work carries over, whichever protocol reaches your market first.

From click to buy to intent to order

Traditional ecommerce was built around a familiar sequence:

AI-assisted commerce begins earlier, with intent.

A customer may ask an AI system to compare options, find a product that meets specific requirements, check compatibility, remember preferences, monitor prices or prepare a purchase before they ever visit the merchant’s website.

The emerging journey looks more like customer intent → AI identifies products → products are compared → price and availability are checked → cart or checkout is prepared → customer confirms → merchant fulfils the order.

The change is not simply from one interface to another. More of the journey now depends on systems exchanging structured product, cart and checkout information.

In the old model, the brand gained more control once the shopper reached its website. In the new model, a larger part of the decision context may be created before that happens.

What this changes for marketing across the buyer journey

For a marketing or ecommerce leader, the practical question is not whether the journey is changing. It is what changes at each stage, and what the business needs to do differently.

Awareness

Traditional awareness work is built around paid media, organic search, social, PR and content.

In AI-assisted product discovery, some of that awareness now happens inside AI answers and shopping experiences. A customer asks a question and the system decides which brands or products are relevant enough to show.

That makes a new question important: Can the system understand what you sell, who it is for and why it is relevant?

For Google specifically, Merchant Center is becoming more important because it supplies product information to Google’s AI shopping experiences. Shopify’s current Google documentation says product discovery in Google AI Mode and Gemini is powered through the Google & YouTube sales channel, which syncs product and store information to Merchant Center. Shopify documentation

Consideration

In a traditional journey, the brand can devote substantial effort to product pages, comparison guides, reviews and persuasive content.

In an AI-assisted journey, the system may perform much of that comparison for the customer. It can summarise options, surface competitors, explain trade-offs and recommend products.

The brand’s content still matters, but it now has two jobs. It must persuade the customer directly and give AI systems enough accurate, consistent and credible information to describe the product correctly elsewhere.

What that means for positioning is covered in how to keep brand differentiation in AI-mediated commerce.

Conversion

Conversion no longer starts only on the checkout page.

Google’s Merchant Center UCP hub distinguishes capabilities such as cart transfer, native checkout and identity linking. Shopify makes a similar commercial distinction: eligible merchants can keep Google AI Mode and Gemini direct checkout active or turn it off while products remain discoverable and customers are redirected to the merchant site. Merchant Center UCP hub Shopify documentation

That means product identifiers, price, availability, shipping, policies and checkout rules become part of checkout readiness.

Retention

Retention introduces a different question.

Traditionally, loyalty and repeat purchase live mostly inside the brand-customer relationship: email, account, app, CRM and direct channels. AI-mediated commerce adds another interface between the two.

The question is not only whether an AI system can help generate another sale. It is also what customer, loyalty and account information should be shared, under what conditions, and which parts of the post-purchase relationship should remain under the merchant’s control.

The existing marketing funnel still matters. The difference is that an AI system can now influence several stages of it.

Merchant of Record does not mean control of the decision

Merchants still play a central commercial role, but legal ownership of the sale and control of the customer journey are not the same thing.

A brand can remain Merchant of Record while an AI system influences which products make the shortlist, how they are described, which competitors appear, which attributes are emphasised and whether the customer continues to the merchant’s website or another interface.

The customer still confirms the purchase

Agentic commerce does not mean customers disappear from the transaction. Current implementations still keep people involved in sensitive fulfilment and payment steps. Shopify’s Google AI Mode and Gemini flow, for example, lets customers enter shipping and payment information and confirm the purchase inside a Shopify-powered direct checkout. Shopify documentation

The immediate change is that AI systems are shortening the distance between discovery, comparison, cart creation and checkout. That matters because brands have traditionally used this part of the journey to influence customers through merchandising, content, offers, product detail and experience design.

The customer may still confirm the purchase. But more of the information and comparison leading to that decision may be shaped elsewhere.

Why product data becomes a leadership issue

In traditional ecommerce, product data quality was often treated mainly as a feed-management problem.

In AI-assisted commerce, it becomes part of whether products can be understood and used correctly.

Google’s Merchant Center UCP documentation says Merchant Center remains the central hub for preparing product data and can alert participating merchants when feed attributes, returns or account configuration need attention. Merchant Center UCP hub

Google has also introduced optional conversational attributes intended to help AI systems and conversational agents understand product nuances. These include question-and-answer data, document links, related products, item-group titles, variant options and popularity rank. Conversational attributes

This does not mean adding more fields automatically makes a product more visible or more likely to be recommended.

It means AI commerce increasingly depends on product information that is structured, consistent and detailed enough to use.

A mismatch between the website, Merchant Center, checkout rules and retailer data can become more than a feed problem. It can affect whether a product is understood, shown correctly or able to move towards purchase.

The Merchant Center foundation

This is where AI visibility and agentic commerce begin to overlap. For Google, Merchant Center is no longer only a Shopping feed destination. It increasingly supplies structured product information to Google’s AI shopping experiences.

Google’s AI Performance Insights report provides data on conversational shopping visibility, including share of voice, competitors, products showing, search intent and shopping stages. Since 16 September 2026 it has been generally available in Australia, Canada, India, New Zealand and the US. No EU country is included, so European brands cannot yet use it as a cross-market measurement layer. AI Performance Insights Google, 16 September 2026

That changes the order of the work. Inside Google, AI visibility and Merchant Center are increasingly connected, so the question is not simply which one to work on first.

But it is important not to overstate the relationship. Strong Merchant Center data does not guarantee recommendation or visibility across every AI system. It does mean that, inside Google’s own commerce environment, product data quality is part of whether Google can interpret and act on the catalogue correctly.

For how products gain or lose recommendations across assistants, see how ChatGPT, Gemini and Claude recommend products.

From visible to selected

Visibility used to mean appearing in search results, shopping listings, ads or marketplaces. In AI-assisted commerce, that is only the first stage.

A product can appear and still be described incorrectly. It can be represented correctly and still lose to a competitor. It can be recommended and still be unable to progress into checkout.

This is why product data, claims, evidence and checkout readiness need to be considered together.

For UCP specifically, visibility is not only about being found. The product and checkout information also need to be reliable enough to support the next step.

Lex Agentica’s AI commerce visibility audit tests the first stages of this sequence, whether products are visible, represented correctly, recommended and selected, for a brand’s priority products and markets.

Why EU beauty brands should watch this

Beauty is visible in Google’s first wave of agentic checkout. When Google announced Universal Cart, it named Sephora, Ulta Beauty and Fenty among the merchants for its new checkout features in the US. Google, 19 May 2026

For European beauty brands, that is an early signal rather than a deadline. It shows the kind of retailers in Google’s first wave, and what they must supply: accurate product data, clear policies and checkout rules that still work when the journey starts somewhere else.

Claims make beauty harder. Efficacy, ingredient and sustainability claims travel with the product into AI answers and carts, and new EU rules on environmental claims have applied since 27 September 2026. Claims that are consistent across the brand’s own site, Merchant Center and retailers give AI systems less contradictory information to work from.

For evidence on how AI systems find, describe and recommend European products, see Lex Agentica’s research on AI product recommendations.

What is different for European brands

Europe is not simply a later market. Three things make preparation different here: large retailers are moving before consumers can buy, governance obligations apply whatever the protocol, and measurement gets harder as the journey leaves your website. Google has not given a reason for the EU timing. Explanations that point to the Digital Markets Act or data-protection rules are interpretation, not something Google has stated.

European retailers are already preparing

For Europe, two facts need to stay separate. Retailers can prepare for UCP before UCP checkout is widely available to consumers. Carrefour and Zalando are examples of preparation, not proof of broad consumer availability.

Carrefour’s strategic plan describes a Google partnership to co-create a shopping experience directly from Gemini using UCP. Zalando has described itself as one of two European launch partners for Google’s Universal Commerce Protocol. Carrefour 2030 Zalando results

Neither example proves that UCP checkout has become a mainstream European commerce channel.

What they show is that large European retailers are willing to prepare before widespread consumer rollout.

Governance, consent and accountability

UCP can make commerce flows more structured, but it does not resolve GDPR, consumer-law, Digital Markets Act, AI governance or accountability questions by itself.

Those obligations depend on how the company uses the technology, what data moves between systems, how recommendations and customer information are handled, and who is responsible when something goes wrong.

This is one of the reasons a US implementation cannot simply be treated as a European operating model.

Account-linked and persistent shopping journeys make this concrete. Before enabling them, decide which customer, loyalty and consent data may move between an AI surface and your systems, on what basis, and who approves it.

Measurement

As more of the journey happens before a website visit, traditional analytics becomes less complete.

Shopify notes that Google Analytics and custom client-side pixels do not fire inside Google AI Mode and Gemini direct checkout; only server-to-server checkout events are supported there. Shopify documentation

That means someone needs to own measurement beyond normal website analytics.

Consumer behaviour

It is not yet clear how quickly consumers will move from using AI for research and comparison to trusting it with more of the purchase journey.

That will vary by category. Delegating a low-risk repeat purchase is different from buying skincare with sensitive claims, choosing a high-value product or making a decision with safety implications.

Is this relevant to your brand right now?

Not every European brand needs to build a UCP integration in the next 90 days. The answer depends on three questions.

1. Is your category already active in AI-mediated shopping?

Run a small, repeatable set of customer-style buying queries across the categories that matter most commercially.

Do not only record whether your brand appears. Check whether the product is represented accurately, which attributes are used, which competitors appear, whether your product is recommended or substituted, and whether results vary between repeated runs.

Treat this as a first indication, not as a full visibility benchmark.

2. Are your competitors preparing?

Look beyond announcements. Check whether competitors are improving product data, joining agentic-commerce partnerships, testing AI shopping or changing how they structure commerce information.

One competitor announcement is not enough to create urgency. Look for a pattern before deciding that your category is moving faster than your internal plans.

3. Is your product data ready anyway?

This is the safest place to start. Strong product data, clear claims, consistent attributes, reliable identifiers, accurate availability and clearly structured policies are valuable whether or not UCP becomes a near-term priority.

If those areas are weak, fixing them will still improve the business even if UCP takes longer to reach Europe.

What EU commerce leaders should do now

Do not start with a large technical project. Start with the products and categories that matter most commercially.

1. Test how your products appear in AI. Check whether Google’s AI experiences understand your products, use cases and differentiators correctly. Record missing products, incorrect descriptions, weak matches, competitor substitutions, unsupported claims and important differences between repeated runs.
2. Check Merchant Center and product data. Review titles, descriptions, identifiers, variants, availability, price, shipping, returns, imagery, category information and conversational attributes where relevant. Check for differences between the website, Merchant Center, structured data and retailer listings.
3. Check product claims and evidence. This is particularly important for European consumer brands where sustainability, efficacy, safety, origin, certification or technical performance influence selection. Check whether important claims are precise, supported, consistent and easy to verify.
4. Check checkout readiness. Review cart transfer, native checkout, fulfilment rules, product-ID consistency, policies and customer-data handling. Not every brand needs all of this now, but leadership should know where the gaps are.
5. Define ownership and controls. Decide who owns agentic-commerce channels, who can turn them on or off, who approves checkout participation, which products can participate, how customer data may be used, how analytics are handled and who is responsible when the AI journey behaves differently from the website.

Before any AI commerce channel is switched on, the leadership team should be able to answer six questions:

• Who knows the channel is active?
• Who approved it?
• Which products are eligible?
• What customer data is involved?
• What happens to analytics, consent and checkout customisations?
• Who owns the decision to opt out?

Those are not abstract future questions. Some merchants already need to answer them.

Four areas to check

The work can be organised into the four areas of Lex Agentica’s agentic commerce readiness model.

That is the work now. UCP does not need to be fully available across Europe for product data, checkout, ownership and controls to matter.

The commercial bottom line

The website, brand, SEO, Merchant Center, retail media, content and performance marketing still matter. The change is that more of the customer’s decision can happen somewhere else before the website visit.

For European ecommerce leaders, UCP is neither a reason to ignore the change nor a reason to rush into a large implementation. It is a reason to check whether product data, checkout, ownership and controls are ready for a different buying journey.

The useful questions are simple: Can AI systems understand our products correctly? Are our claims and product data reliable enough to support recommendation? Can our commerce systems support new checkout paths? Do we know who controls those paths when they become available?

Those are the gaps European ecommerce teams can start checking now.

Status tracker: Universal Cart and UCP by market

This tracker records what Google and Shopify have announced for each market, taken from their own posts and documentation. A market changes status here only when an official source changes. Agency and press claims appear separately in the status log.

Last verified 28 September 2026. Google’s Wallet Help says Universal Cart is currently available only in the United States. AI performance insights in Merchant Center is also available in India and New Zealand. On Shopify, eligible stores are in the US, UK, Australia and Canada, and direct checkout in AI Mode and Gemini is shown only to US customers. Google Wallet Help Google, 19 May 2026 Google, 16 September 2026 Shopify documentation

Merchant eligibility, product discovery and customer checkout availability are three different things. A brand can be eligible to sell and its products can be discoverable while its customers still cannot check out inside an AI experience.

What has changed since May 2026

When this guide was first published in May 2026, UCP could still be described mainly as an emerging protocol. Google now runs a Merchant Center UCP integration hub where merchants can validate their UCP profile, cart transfer and native checkout, and in September it added cart transfer to the merchant’s site and enhanced checkout-flow testing. Shopify has moved Google AI Mode and Gemini into its Agentic Storefronts, with direct checkout active by default for eligible stores and shown only to US customers. Each change is dated in the status log below. Merchant Center UCP hub Google, 16 September 2026 Shopify documentation

Status log

Every change to the status above, newest first. Dates are the source’s publication date. Third-party claims are listed as reported and do not change a market’s status until an official source confirms them.

Last verified 28 September 2026.