A live tracking dashboard published today by Originality.ai reveals that just 26 websites out of more than 3 million scanned have publicly implemented the Universal Commerce Protocol - the open standard Google launched in January 2026 to enable AI agents to conduct purchases on behalf of consumers.

The numbers are stark. Four months after Google announced UCP with considerable fanfare at the National Retail Federation conference, the infrastructure meant to underpin the next generation of AI-powered shopping remains nearly invisible on the public web. Originality.ai's dashboard, which continuously scans websites for the presence of the /.well-known/ucp file path, found only 26 detectable implementations as of May 2026 - and none of them belong to the major companies that co-developed or endorsed the standard.

A protocol without a public footprint

Google announced UCP on January 11, 2026, co-developing the standard alongside Shopify, Etsy, Wayfair, Target, and Walmart, with endorsements from more than 20 additional companies including American Express, Best Buy, Stripe, Mastercard, Visa, and Macy's Inc. The ambition was significant: a single open standard that would allow AI agents to discover products, negotiate checkout parameters, link customer identities, and manage the full post-purchase lifecycle without requiring custom integrations for each merchant-platform combination.

According to Originality.ai, no public /.well-known/ucp implementation was detected on the primary domains of any of those named companies. That absence does not mean they are not experimenting with UCP privately or through non-public integrations. The scan can only detect what is publicly visible. Still, the gap between the January announcement and the current state of publicly detectable adoption is notable.

The methodology is straightforward. Originality.ai used PublicWWW to scan more than 3 million websites for the UCP file path, tracking adoption as a trend over time from February 2026 through May 2026. The trend chart shows minimal movement across that period, with adoption essentially flat at low single-digit figures before reaching 26 sites in total.

What UCP is and how it works

Universal Commerce Protocol is an open-source standard designed to standardise communication between AI agents, merchants, and payment providers across the entire shopping journey. Before UCP, if a merchant wanted its products to be purchasable by every AI platform, it would need a separate custom integration for each one. An AI agent trying to shop across multiple retailers would similarly need to understand each retailer's unique checkout requirements, authentication methods, and product data formats.

UCP solves this with four defined participant roles: Platforms (the AI surfaces such as Gemini or AI Mode), Businesses(the merchants), Credential Providers, and Payment Service Providers. Each role has defined interfaces and data formats, so implementing UCP once is sufficient for a merchant to be accessible to any UCP-compliant agent.

The protocol is compatible with existing web standards. According to Originality.ai, UCP is designed to work with REST, JSON-RPC, and the Agent2Agent Protocol (A2A). PPC Land's coverage of the UCP help page published in March 2026noted that the standard is also compatible with the Agent Payments Protocol (AP2) and the Model Context Protocol (MCP), giving merchants considerable flexibility in how they connect their existing infrastructure.

Loyalty data is part of the standard too. According to Google, UCP will retrieve member benefits so that consumers "get the best value inclusive of your member benefits" - meaning an AI agent completing a purchase on behalf of a user would factor in that user's loyalty points, membership discounts, or rewards status rather than simply comparing listed prices.

Implementation is not straightforward

The technical process for adopting UCP on Google's surfaces involves several sequential steps. According to Originality.ai, merchants need a Merchant Center account and must join Google's UCP waitlist for approval to appear on AI surfaces such as Gemini and Google AI Mode. They then need to set up Google Pay, create and publish a UCP profile, integrate a required checkout flow, and connect Google's order lifecycle update system. Merchants must also decide whether customers check out as guests or via account-linked sessions.

That is a non-trivial sequence of steps. For smaller retailers without dedicated engineering resources, each stage represents a potential friction point. Originality.ai notes that this implementation complexity may partly explain why early adoption has been limited, with smaller businesses in particular potentially relying on platforms like Shopify to handle UCP integration on their behalf - introducing a layer of platform dependency that may concern merchants who value direct control over their commerce relationships.

The protocol's technical architecture involves standardised methods for discovering merchant capabilities, negotiating checkout parameters, exchanging payment credentials, linking customer identities through OAuth 2.0, and managing order lifecycle webhooks. That breadth is part of what makes UCP potentially significant - and part of what makes it more complex than a simple file drop.

The agentic commerce context

The adoption numbers land against a backdrop of rapidly growing interest in AI-powered shopping. According to Originality.ai's study, AI and agents drove approximately $67 billion in sales during Cyber Week 2025. Retail sites have already seen 4,700% year-over-year growth in generative AI traffic as of 2025. McKinsey projections cited in the study estimate that agentic commerce could orchestrate between $3 trillion and $5 trillion in global revenue by 2030.

The conversion gap between AI traffic and traditional traffic is narrowing. According to Originality.ai, in January 2025 AI traffic sources were 49% less likely to convert than non-AI sources. By July 2025, that gap had fallen to 23%. Over the same period, AI-driven revenue per visit increased 84% compared with non-AI sources.

Those numbers help explain why PPC Land has tracked UCP-related developments closely since January 2026, covering Target and Walmart's early checkout integrations within Google's Gemini app, Mastercard's Agent Pay infrastructuredesigned to authenticate agentic transactions, and more recently Google's expansion of UCP to hotels, food delivery, and new countries announced at Google Marketing Live on May 20, 2026.

Amazon's posture toward third-party AI shopping agents has been notably different. As PPC Land documented, the company spent much of 2025 blocking AI crawlers while simultaneously assembling a 40-person engineering team dedicated to agentic commerce experiences - a tension that reflects the structural complexity facing large platforms invested in their own retail media businesses.

What the data does and does not show

Originality.ai is careful about what its scan measures. The dashboard detects only publicly accessible /.well-known/ucp files. A company could be running UCP-based transactions through internal staging environments, private integrations, or behind authentication layers without that activity appearing in the public scan at all.

This matters for interpreting the absence of major brands. Shopify, for instance, participated as a co-developer and is likely involved in UCP implementations through merchant accounts rather than its own primary domain. Target and Walmart both announced integrations at the January launch, enabling checkout within Gemini and AI Mode. That those integrations do not surface in a public file scan does not mean they are inactive - it means they may be structured differently from the /.well-known/ucp path that the dashboard tracks.

What the data does show is that the open, publicly detectable layer of UCP adoption - the layer that would allow any UCP-compliant AI agent, not just Google's own surfaces, to discover and interact with a merchant - is still minimal. UCP's value as an open standard rather than a proprietary Google checkout button depends on broad implementation. Twenty-six sites, however notable those sites may be, is not broad implementation.

The llms.txt parallel

The slow uptake is not without precedent in the world of AI-related web standards. PPC Land covered the llms.txt adoption stall in July 2025, when analysis found that no major LLM provider - including OpenAI, Anthropic, or Google - actively supported the protocol eight months after its launch. A March 2026 study found only 7.4% of Fortune 500 companies had implemented llms.txt, despite widespread promotion by SEO tools.

Originality.ai draws a direct comparison to this pattern, noting that its llms.txt tracking study demonstrated how implementation of new web standards can take significant time to grow as organisations discover and learn about them. The difference with UCP is that the commercial stakes are considerably higher. llms.txt affects content discoverability. UCP affects whether transactions happen at all.

What it means for merchants and marketing professionals

The marketing community has reason to watch UCP adoption closely, even at this early stage. PPC Land has documented the structural shift underway: if product discovery, comparison, and checkout increasingly occur inside AI interfaces rather than on brand websites, the metrics and levers that digital marketers have relied on shift substantially.

According to Originality.ai, direct website traffic may decline as agents handle more of the shopping journey on users' behalf. A 25% drop in search engine volume due to AI chatbots and agents is already projected by 2026. Clear, current, and structured product data becomes more important for discovery, since agents rely on machine-readable signals - price, availability, return policy, specifications - rather than visual merchandising or narrative brand content. Ratings and reviews may also carry greater weight: one study cited by Originality.ai found that doubling review counts would allow a seller to raise prices by 37% with GPT-4.1.

The profit margin question is real. AI agents do not browse impulsively, respond to promotional framing, or get upsold. They match user constraints to available products. A merchant outside a user's specified price range is simply excluded. That mechanical precision changes the economics of conversion in ways that have not yet fully worked through marketing planning cycles.

The possibility of what Originality.ai calls "agentic fees" remains speculative but structurally plausible. If multiple UCP-compliant merchants offer identical products at the same price, the question of which merchant an agent selects is not yet formally resolved. The study notes that prioritisation could become a monetised layer, analogous to app store commissions, where merchants pay to be considered first. Whether that layer emerges - and how Google and other platforms might structure it - has significant implications for the economics of AI-mediated commerce.

For now, the main takeaway from Originality.ai's dashboard is that the infrastructure layer of agentic commerce remains in its earliest stages despite over four months of headline activity around UCP. The Universal Cart announced by Google at Google Marketing Live on May 19, 2026, adds a consumer-facing persistence layer that lets users save products across retailers within Google's surface - which may drive further UCP implementation among merchants wanting to participate in that shopping mode.

Twenty-six sites is a starting point. Whether it becomes a floor or a ceiling for the standard's broader adoption will depend on how quickly implementation tooling matures, how consistently major platforms enforce or incentivise compliance, and whether consumer behaviour shifts toward agent-mediated purchasing at the pace projections suggest.

Timeline

Summary

Who: Originality.ai, a content authenticity platform, conducted and published the study. Google announced and co-developed UCP with Shopify, Etsy, Wayfair, Target, and Walmart, alongside more than 20 endorsing companies including Mastercard, Visa, Stripe, and American Express.

What: A live tracking dashboard and study scanning more than 3 million websites for public implementation of the Universal Commerce Protocol found only 26 detectable deployments. No public /.well-known/ucp file was found on the primary domains of any of the major companies named in Google's original January 2026 announcement.

When: The dashboard published today, May 21, 2026, with adoption tracking running from February 2026 through the present. UCP itself was announced on January 11, 2026.

Where: The scan covered more than 3 million public websites indexed by PublicWWW. UCP is a global open-source standard, with initial commercial implementations centred on US markets through Google Search, AI Mode, Gemini, and Merchant Center.

Why: The study matters because UCP's potential value as an open standard for agentic commerce depends on broad, public adoption by merchants. With only 26 detected implementations four months after launch, the infrastructure layer enabling AI agents from any platform to execute purchases across a wide range of merchants remains far less developed than the volume of industry announcements might suggest.

Share this article
The link has been copied!