The Storefront Isn't Disappearing. It's Just Stopped Being the Point of Entry.

Why drop-model brands with clean structured data will own the consideration set before their campaign goes live

Something I've been explaining to brands more frequently over the past few months is the difference between a storefront and an entry point.

Most fashion brands treat them as the same thing. They aren't. And Google's Universal Cart announcement makes that distinction more commercially urgent than it's ever been.

I want to explain what was actually announced, what it means in practice for drop-model brands, and why I think most ecommerce teams are thinking about the wrong problem right now.

What Universal Cart actually is

Before I get into the implications I want to be clear about what was confirmed at Google I/O and what's still ahead of the facts. I've seen a lot of articles this week describing Universal Cart in terms that go significantly beyond what Google announced. Some of the details being presented as confirmed functionality are speculation.

What's confirmed: Universal Cart is a persistent, cross-merchant cart that follows a user across Google Search, the Gemini app, YouTube, and Gmail. A user adds a product while browsing Search or watching a YouTube video and it sits in a single cart rather than a merchant-specific one. In the background the cart monitors price drops, price history, and flags restocks. It runs on Gemini models. Checkout works through Google Pay. Confirmed early partners include Nike, Target, Walmart, Wayfair, Sephora, and Shopify stores. It's rolling out across Search and Gemini in the US this summer with YouTube and Gmail following later.

This is not speculation. This is happening.

Two infrastructure pieces sit underneath it.

The Universal Commerce Protocol is an open standard Google co-developed with retail partners. It keeps brands as the Merchant of Record. You retain control of fulfilment, the unboxing experience, and the post-purchase customer relationship. The customer is buying from you. They're just not visiting your site to do it.

The Agent Payments Protocol is designed to let AI agents execute purchases on a user's behalf within limits the user defines. Specific brands, specific products, a spending cap. When conditions are met, the agent buys. This is important to qualify clearly: AP2 hasn't shipped in any consumer product yet. Google confirmed it's coming to their products in the coming months. The autonomous checkout scenario, where an agent executes a drop purchase the moment stock goes live, is the direction of travel, not the current state.

Where the horse and car analogy breaks down

I've been using an analogy in client conversations recently that I think is worth unpacking here because it frames the structural shift better than most of the commentary I've read this week.

The comparison I keep reaching for is Henry Ford. Not because it's original but because it's accurate. Before the car, horses worked. People built entire industries around them. Stables, feed merchants, carriage makers. They used horses to get from A to B. Then the car arrived and the question wasn't whether horses were good at what they did. It was whether they were the right tool for what came next.

The web in its current form was built for humans. Every design decision, every CRO framework, every UX principle was built on the assumption that a person was doing the navigating. That assumption held for about twenty-five years. AI shopping agents don't browse. They don't click through to a product page or read your brand story in the way a human does. They resolve product identity from structured data and decide which source to recommend based on the quality and coherence of that data.

The horse isn't going away. Human traffic isn't going away. But it's no longer the only mode of transport that matters.

The more precise analogy for what's actually happening inside the infrastructure is shipping containers and cargo manifests.

Before containerisation, moving goods internationally was slow and expensive not because the goods weren't real but because every port, carrier, and customs system had its own way of handling and describing cargo. The physical product existed. The systems that needed to move it couldn't read it consistently.

The shipping container standardised the physical format. But what made global trade function at scale was the electronic manifest. That was the structured data document that travels with every container and tells every system along the route what's inside, where it came from, who owns it, what it's worth, and where it's going. It's the equivalent of the passport.

Customs don't open the container. They read the manifest.

Map that directly to ecommerce architecture:

• The container is the product. It's real. The quality is there. The brand story is there.
• The ship is the ecommerce platform. It moves the product from warehouse to customer.
• The storefront such as the headless frontend, product pages, and mobile experience is the port. It's where humans come to collect things. It matters enormously for human conversion. But the agent never visits the port.
• The manifest is the structured data. The GTIN registered to your brand entity at GS1. The Organisation schema on your domain including tax details and registered addresses. The product schema on every PDP with brand field, material, sizing, and availability populated. The sameAs links that allow an agent to verify your identity against independent platforms. The PIM output feeding every downstream system with consistent, machine-readable product information.

Every AI shopping agent, every Universal Cart, every agentic checkout protocol reads the manifest. A container with no manifest doesn't get through customs. It gets held. It gets excluded. It's invisible.

What most fashion brands have actually built

I want to be direct about this because I think the gap between where most brands are and where they need to be is wider than most ecommerce teams realise.

Most of the brands I work with have invested heavily in the port. Headless frontends. Improved PDP layouts. Faster checkout flows. Strong creative direction.

All of it built for the human visitor arriving from a paid channel or an organic search.

The manifest is another matter.

GTINs missing on new drop SKUs. Brand fields empty on product schema. Organisation schema sameAs links pointing to Instagram handles changed during a rebrand eighteen months ago that nobody thought to update. PIM data that's clean in the ERP and degraded by the time it reaches the product feed. None of this shows up in GA4. It doesn't trigger a Search Console error. There's no dashboard that surfaces it as a problem.

It just sits there making the manifest unreadable to agents that are now doing meaningful consideration work before any human visits the site.

The co-ordination problem underneath this is the same one I encounter in every structured data conversation. It lives in the gap between technical SEO, trading, and operations. Nobody feels fully responsible for it. So it doesn't get prioritised and the port keeps getting more beautiful while the manifest stays broken.

Universal Cart makes this a more urgent problem than it was six months ago. A customer who adds your product to their Universal Cart while watching a YouTube video three weeks before your drop is a customer in your consideration set who never visited your site. The agent is monitoring. When the drop goes live and the conditions are met, it acts.

A customer who tried to add your product but couldn't because the inventory data wasn't structured in a way the cart could read is a customer who moved to a competitor. There's no session recording of that. No drop-off point to diagnose. The consideration happened in Google's infrastructure and you weren't in it.

The funnel now has a stage before the funnel you're currently measuring.

What the Merchant of Record detail means in practice

I want to spend a moment on this because I think it changes something specific about how drop-model brands should be thinking about post-purchase work.

The UCP design keeps you as the Merchant of Record. You control fulfilment. You control the unboxing experience. You own the post-purchase email sequence. The customer is buying from you.

What you don't control is the discovery and consideration stage. The customer may not visit your site until the delivery confirmation email lands.

For brands where the site experience is part of the brand perception, and for premium streetwear that's almost always the case, this shifts where the first real brand impression happens. The PDP is no longer the entry point. The product card in the Universal Cart is.

The downstream implication is that the post-purchase sequence needs to do different work than it used to. If a meaningful proportion of buyers are arriving without the PDP experience that used to warm them up to the brand, the first delivery, the packaging, the first email after purchase then all of that is carrying more weight. The foundation for repeat purchase is being built in a different place.

That's worth rebuilding the sequence around.

What to have in place before the next drop

I want to be clear that some of what follows is preparation for infrastructure that isn't fully live yet. AP2 isn't shipping this week. The autonomous checkout scenario for drops is coming months away at minimum. But Universal Cart itself is rolling out this summer in the USA and the structured data requirements for appearing in it are the same requirements I've been writing about across this series.

Four weeks out from a drop is the minimum runway. You can't fix the manifest in a 48-hour window.

New SKU GTINs need to be assigned and verified at source before the product page goes live. Not present in the feed. Verified through GS1's Verified by GS1 tool to confirm the identifier resolves to your brand entity. A GTIN resolving to a factory's GS1 prefix is a manifest pointing to the wrong owner. Not on launch day. Not the week before.

Product schema needs to be complete on every new PDP before launch. Brand field populated. Material and sizing attributes present. Availability status accurate. Price matching the live feed. Incomplete schema on a new SKU means the agent can't resolve it as a verified product. It enters the drop window as an unknown entity and unknown entities don't get recommended.

Organisation schema sameAs links need to be verified. Every URL opened and confirmed as a current, active, brand-owned profile. I've audited brands where sameAs links pointed to Instagram handles changed two rebrands ago. That's not a minor gap. It's a contradictory identity signal the agent has to discount.

At least one piece of authored content published before the drop with proper Author schema markup. It doesn't need to be long. It needs to exist, be marked up correctly, and be live before the window opens. Early access content, a piece on the manufacturing story, a design breakdown. Whatever is on brand. This refreshes authorship signals in the entity graph in the weeks before the window opens.

The part that isn't speculation

I want to close on something I think gets lost in most of the commentary on agentic commerce. It tends toward either breathless excitement or dismissive scepticism. The honest position is somewhere more boring than either.

Universal Cart is real and it's rolling out this summer. UCP is real. AP2 is confirmed infrastructure that hasn't shipped yet in any consumer product. The autonomous drop checkout scenario is real as a direction and probably further away than the announcement implies. OpenAI attempted something comparable in 2025 and retreated when the implementation complexity became clear. Google has more infrastructure and more merchant relationships. But this is still hard.

What isn't speculative is the structured data requirement underneath all of it.

Whether autonomous checkout arrives this summer or eighteen months from now, the manifest has to be readable before the agents can act on it. Every drop where you build it properly makes the next one better. The GTIN coverage grows. The entity graph gets richer. The gap between your manifest and a counterfeit operation's manifest widens every time.

If you're focusing on just making the port look prettier for th evisitors. I'd spend some of that attention on the manifest instead.