Originally published on The Searchless Journal
Google spent the last decade trying and failing to build a commerce business that could rival Amazon. Google Shopping, Google Express, Google Pay as a checkout layer — none of it stuck. Amazon owned the purchase moment, and Google owned everything upstream: the research, the reviews, the product discovery. Google was the billboard. Amazon was the register.
That division of labor is collapsing. At Google Marketing Live 2026 and the preceding Brandcast event, Google laid out a commerce infrastructure play that is its most credible challenge to Amazon's checkout dominance to date. Not because Google built a better shopping app. Because Google is embedding purchase capability directly into the AI surfaces where people are already asking questions, watching videos, and searching for products.
The architecture is deceptively simple: Gemini shopping agents that understand what you want, a Universal Commerce Protocol that connects merchants to Google's AI surfaces, shoppable YouTube ads that let you buy without leaving the video, and Google Pay as the transaction layer underneath it all. Sundar Pichai, speaking at NRF's Big Show in January and again through his post-I/O press tour, framed it as removing friction from commerce. Ashish Gupta, Google's VP and GM of merchant shopping, put it more bluntly during a GML press briefing: the goal is to "remove friction and close the entire loop."
Closing the loop is the operative phrase. Google does not want to recommend products anymore. It wants to sell them.
What Google Actually Announced
The announcements from GML and Brandcast were dense with product names and feature updates. Here is what matters for anyone tracking the commerce infrastructure shift.
Shoppable YouTube with native checkout
Google claims users watch 110 million hours of shopping-related content on YouTube every day. That is a staggering distribution surface. At Brandcast, Google announced that YouTube ads connected to its Universal Commerce Protocol can now process a purchase without the user ever leaving the video. A viewer watching a product review or unboxing on their TV can complete a purchase with two clicks on their remote while the video continues playing.
The mechanics rely on Google Pay and the UCP integration. Merchants who participate in UCP can pipe their inventory, pricing, and checkout flow directly into the YouTube ad unit. The user sees the product, prompts the Gemini agent embedded in the ad for details, and completes the transaction. No redirect to a product page. No cart abandonment at the merchant site. The purchase happens inside Google's surface.
This is not theoretical. Google is already running these units with select merchants. The TV remote as a checkout device is an unusual commerce surface, but the math is compelling: 110 million daily hours of shopping content times a checkout-embedded ad format equals a commerce channel that Amazon cannot replicate because Amazon does not own the content surface.
Universal Commerce Protocol expansion
The Universal Commerce Protocol, which Google co-developed with Shopify and announced at NRF in January, is Google's answer to the Agentic Commerce Protocol backed by OpenAI, Stripe, and PayPal. At GML, Google announced that UCP is expanding beyond Google's primary shopping surfaces into YouTube ads and Google Maps ads. A Google Maps ad for a hotel can now convert directly into a booking. A local restaurant ad can trigger a food delivery order through an agentic conversation.
Google also added merchant incentives to pull more retailers onto UCP. Merchants who integrate with the protocol can now export their loyalty points and member-exclusive discounts directly into agentic ads. This is a meaningful upgrade. One of the reasons Amazon has dominated commerce is its loyalty infrastructure: Prime members get faster shipping and better prices, which creates lock-in. Google is attempting to replicate that dynamic by letting merchants bring their own loyalty programs into the agentic ad experience.
Direct Offers and Business Agents
Two new ad products deserve attention. Direct Offers, currently in beta, allows merchants to upload discounts, coupons, and incentives that Google's AI then "matches or combines on the fly to present the most compelling offer" to each user. During the GML press briefing, Google VP of Global Ads Dan Taylor used the example of a grill manufacturer offering a 10 percent discount plus a free accessory if the AI determines the customer needs an extra nudge to convert.
Business Agents for Leads takes the agentic ad concept further. Gemini AI agents are embedded directly into ad units, trained on the advertiser's own website content. A user researching a general query might encounter an ad containing a conversational agent that can answer detailed questions about a product, a university program, or a service offering. The agent collects contact information and passes it to the advertiser as a lead.
Taylor called Business Agents his favorite new product. That is revealing. Google is not just selling ad impressions anymore. It is selling AI-mediated conversations that convert directly into leads and purchases. The line between informational AI and commercial AI is dissolving inside Google's ad stack.
The Protocol War: UCP vs ACP
Beneath the product announcements lies the real battleground. The protocol layer will determine which AI surfaces can actually close transactions, and right now there are two competing standards.
The Agentic Commerce Protocol was first. OpenAI developed it with Stripe, later adding PayPal, to give ChatGPT and other AI surfaces a standardized way to process purchases on behalf of users. ACP handles the payment mechanics: tokenized wallets, spending limits, transaction authorization. It is backed by the payment processors that already handle a significant share of online transactions.
The Universal Commerce Protocol is Google's counter. Co-developed with Shopify, it is backed by Walmart, Target, Etsy, and Wayfair. Google describes UCP as "agnostic" and compatible with existing industry protocols, including Google's own Agent2Agent protocol and the Model Context Protocol developed by Anthropic.
At NRF in January, Pichai addressed the inevitable protocol fatigue directly. "You might wonder why we are introducing another protocol," he said during his keynote. He argued that "the industry needs a protocol that works at global scale and takes into account the nuances of commerce journeys." Translation: OpenAI's ACP works fine for a chatbot processing a single purchase, but Google believes commerce at scale — across YouTube, Maps, Search, and Shopping — requires something more robust.
The framing is characteristic Google. ACP is positioned as a startup's protocol. UCP is positioned as an enterprise-grade infrastructure layer. The retailers and merchants I have spoken with through AdExchanger's reporting express a different concern: they worry about walled gardens. Google and Amazon both have what industry observers call a "content fortress mentality." The merchants' fear is that integrating with UCP means becoming dependent on Google's ecosystem, just as integrating with ACP means becoming dependent on OpenAI's.
That concern is valid. But it also misses the point. The protocol war is not about ideology. It is about distribution. Whichever protocol reaches more consumers through more AI surfaces will win merchant adoption by default, because merchants go where the transactions happen.
Why ChatGPT's Checkout Failure Is a Warning, Not a Dismissal
A common narrative in the coverage of Google's commerce push is that ChatGPT already tried native checkout and failed, therefore agentic commerce is a mirage. This is the wrong lesson.
ChatGPT did launch a native checkout feature. It did fold it. The reason matters. ChatGPT's checkout failed not because users do not want to buy through AI surfaces, but because ChatGPT lacked the commerce infrastructure that makes checkout feel natural. ChatGPT has no video content surface like YouTube where users are already in a shopping mindset. It has no maps product where local commerce transactions are a logical next step. It has no loyalty program integration. It has no merchant network at the scale of Google's Shopping Graph.
Google has all of those things. The 110 million daily hours of shopping content on YouTube is a pre-existing demand signal. The Google Maps integration creates a local commerce surface. The Shopify partnership gives UCP access to millions of merchants from day one. Google Pay, despite its historical struggles, is already installed on every Android device globally.
ChatGPT's checkout failure was a product failure, not a category failure. Google is building the infrastructure that ChatGPT lacked. Whether Google can execute where ChatGPT stumbled depends on one critical variable: Google Pay adoption.
The Google Pay Problem
Google's entire commerce stack rests on a fragile foundation. The AI agents, the shoppable ads, the UCP protocol, the universal cart — all of it funnels toward a single transaction moment where the user pays with Google Pay.
Google Pay has not been the success Google hoped for. In markets where Google competes with Apple Pay, contactless payment adoption on Android remains uneven. In markets where Google competes with local payment systems — WeChat Pay in China, UPI in India, iDEAL in the Netherlands — Google Pay is often a secondary option or absent entirely.
If the checkout friction at the point of sale is higher than Amazon's one-click purchase, the entire AI-to-purchase pipeline loses its advantage. Amazon's checkout experience is the benchmark because Amazon spent two decades reducing purchase friction to a single button. Google's agentic checkout needs to be at least as seamless, or the consumer will abandon the AI surface and go directly to Amazon, which is exactly what they do today.
This is the strategic risk that Google's commerce executives are aware of but rarely discuss publicly. Gupta's comment that "more and more people are coming to AI surfaces to shop" is the aspiration. The data on actual completed transactions through AI surfaces is still thin. The $2.3 billion in agentic commerce transactions in Q1 2026 that payment processors reported is growing fast, but it is still a rounding error against $6 trillion in annual global ecommerce.
Google's bet is that the transaction volume will follow the infrastructure. Build the checkout rails into every AI surface, make the purchase flow seamless, and the volume will come. It is a reasonable bet. But it requires Google Pay to work flawlessly at scale, and Google Pay has never worked flawlessly at scale.
What This Means for Brand Visibility
Here is the part that should worry every brand that is not tracking this shift.
If Google succeeds in building an AI-to-purchase pipeline, brand visibility inside AI answers stops being a branding exercise and starts being a revenue metric. When a Gemini shopping agent recommends your product and the user can buy it with two clicks without leaving the AI surface, your visibility inside that AI answer is your point of sale.
This is a fundamentally different dynamic than traditional search. On a SERP, the brand that ranks first gets the most clicks, but the user still navigates to the brand's website to complete the purchase. The brand controls the checkout experience. The brand controls the upsell. The brand owns the customer relationship.
In an agentic commerce world, the AI surface owns the checkout experience. The brand provides the product, the inventory, the pricing, and the loyalty data through a protocol like UCP or ACP. But the consumer interaction — the recommendation, the product details, the transaction — happens inside Google's or OpenAI's interface. The brand is a supplier to the AI surface, not the owner of the customer relationship.
This is why the agentic commerce tipping point matters beyond the raw transaction numbers. The $2.3 billion in Q1 agentic purchases is significant because of the growth rate, not the absolute value. The infrastructure being built now — the protocols, the payment rails, the AI agent integrations — will determine how trillions of dollars in commerce flow over the next decade.
Brands that optimize their product data, entity clarity, and AI discoverability now will be the ones that AI agents recommend when the transaction volumes become meaningful. Brands that wait for agentic commerce to be proven before investing in AI visibility will be invisible to the agents doing the selling. Google's own AI Performance Insights tool, announced at GML, is designed to help brands measure exactly this: how visible they are inside AI search results. The fact that Google built this tool signals that the company understands the connection between AI visibility and commerce outcomes.
The Infrastructure Layer Matters More Than the Chatbot Layer
Most coverage of the Google-vs-Amazon commerce war focuses on the consumer experience: which AI assistant gives better product recommendations, which chatbot has a more natural shopping flow, which voice assistant handles "order me paper towels" more reliably.
That is the wrong layer to watch.
The competition that will determine the outcome is happening at the infrastructure level. UCP vs ACP. Google Pay vs Stripe vs PayPal. Google's Shopping Graph vs Amazon's marketplace. YouTube's shopping content vs Amazon's product videos. Shopify's merchant network integrated into Google's protocol vs Amazon's third-party seller platform.
Amazon's advantage is its installed base of paying customers with stored payment methods and fast shipping expectations. Google's advantage is its distribution surface: Search, YouTube, Maps, Shopping, and every Android device on the planet. Amazon owns the purchase habit. Google owns the discovery moment. The question is which is more valuable in a world where AI agents collapse the distance between discovering a product and buying it.
The honest answer is that neither company has solved this yet. Google's commerce push is its most credible attempt in a decade, but it still requires Google Pay to work at scale and consumers to trust AI agents with their purchasing decisions. ChatGPT's advertising revenue reaching $100M ARR in six weeks proves that advertisers see the potential. But advertising and commerce are different businesses. Convincing a brand to buy an ad on ChatGPT is easier than convincing a consumer to complete a purchase through an AI agent.
Google is ahead because it has the commerce infrastructure layer — Shopping Graph, Google Pay, YouTube content, merchant partnerships — that no other AI company can match. Amazon is ahead because it has the consumer trust and the checkout muscle memory that no other commerce company can match. The agentic commerce war will be won at the intersection of those two advantages, by whichever company can combine AI-mediated discovery with frictionless checkout at a scale that matters.
What Smart Brands Should Do Now
The protocol war is not waiting for brands to get comfortable. Google is signing merchants onto UCP. OpenAI and Stripe are pushing ACP. Payment processors are choosing sides. Shopify merchants are being asked which protocol to integrate with. The infrastructure decisions being made now will shape the AI commerce landscape for years.
For brands, the immediate priorities are straightforward.
First, ensure your product data is AI-readable. Schema markup, structured product feeds, entity clarity, and clean attribute data are the minimum requirement for being visible to Gemini shopping agents, ChatGPT product recommendations, or any other AI commerce surface. If an AI agent cannot parse your product catalog, it cannot recommend your products.
Second, integrate with at least one agentic commerce protocol. If you are a Shopify merchant, UCP integration is the path of least resistance. If you are already working with Stripe, ACP is the natural choice. Being on both is ideal. Being on neither is a strategic risk.
Third, track your AI visibility as a commerce metric, not just a branding metric. When a Gemini agent recommends your product in a shoppable YouTube ad, that recommendation is the digital equivalent of shelf placement in a retail store. You would not ignore your retail placement strategy. You should not ignore your AI placement strategy.
Fourth, watch the Google Pay adoption data. If Google Pay adoption accelerates as UCP expands into YouTube and Maps, Google's commerce stack has a real chance of challenging Amazon. If Google Pay remains a weak link, the entire pipeline stalls at the moment that matters most.
Check your brand's AI visibility score for free at audit.searchless.ai. Find out which AI engines recommend your products and which ones leave you invisible.
Sources
- AdExchanger: "Google Wants To Do Your Shopping For You At GML This Year" (May 22, 2026) — detailed GML commerce breakdown including UCP expansion, Direct Offers, Business Agents for Leads, and Google VP quotes
- AdExchanger: "Getting Lost In The Protocols At NRF's Big Show 2026" (January 2026) — UCP vs ACP protocol comparison, Pichai NRF keynote, merchant concerns about walled gardens, Shopify partnership details
- Google GML press briefing (May 2026) — official product announcements for UCP expansion into YouTube and Maps, Direct Offers beta, Business Agents for Leads, Ashish Gupta and Dan Taylor quotes
- Google/NRF: Sundar Pichai NRF keynote address (January 2026) — UCP introduction rationale, "global scale" framing
- Searchless internal data: Q1 2026 agentic commerce transaction volumes ($2.3B), AI Overviews coverage data
- Stripe: Agentic Commerce Protocol documentation and Q1 2026 merchant transaction data ($340M in agent-initiated transactions)
- PayPal/OpenAI: ACP partnership announcement (October 2025) — PayPal integration with ChatGPT agentic checkout
- NYT: "Google Search Bar AI Gemini Overhaul" (May 19, 2026) — search bar redesign for AI prompting
Explore Searchless pricing for full-service GEO that makes your products visible to every AI agent, every commerce surface, every time.
