UCP Checker & Validator /.well-known/ucp

Why a UCP Manifest Matters: The Data Behind Agentic Commerce

AI-referred traffic to retail sites is growing faster than any other channel and now converts better than paid search. Here is the real data behind that shift, where the widely repeated "20% of e-commerce" claim actually comes from, and why a manifest is what lets an agent find you at all.

Quick answer

AI-referred traffic to U.S. retail sites grew 393% year over year in Q1 2026 and converted 42% better than non-AI traffic in March 2026, according to Adobe Analytics. It is still a small share of total e-commerce traffic overall, but growing faster than any other channel. AI shopping agents cannot reliably parse a website built for humans; a UCP manifest gives them a structured, direct answer instead.

How much shopping is actually happening through AI right now

The numbers vary a lot depending on exactly what is being measured, so it is worth being specific. Adobe Analytics tracks AI referral traffic across more than a trillion visits to U.S. retail sites, and its data shows consistent, fast growth:

393% Year-over-year growth in AI-referred traffic to U.S. retail sites in Q1 2026, peaking at 1,151% YoY during the 2025 holiday season. (Adobe Analytics)
42% How much better AI-referred traffic converted than non-AI traffic in March 2026, a reversal from converting 38% worse the same month a year earlier. (Adobe Analytics)
$3–5T McKinsey's moderate-scenario estimate of global consumer commerce that could be mediated by AI agents by 2030, including up to $1 trillion in the U.S. alone.

AI-referred shoppers also behave differently once they arrive. Adobe found they spend meaningfully more time on-site and generate more revenue per visit than shoppers arriving through other channels, a pattern that has held for several consecutive quarters rather than being a one-off spike.

Where the "20% of e-commerce" number actually comes from

If you have seen a claim that AI now accounts for roughly a fifth of online shopping, it most likely traces back to a real number, just not the one it is usually presented as.

What the number actually measures

Salesforce's Cyber Week 2025 report (Nov. 25 – Dec. 1) found that AI and agents were credited with influencing 20% of all global online orders that week, worth about $67 billion in sales. In the U.S. specifically, the figure was 17% of orders, about $13.5 billion.

That is a real, publicly reported figure from Salesforce. But it describes one peak shopping week, not typical year-round volume, and it measures orders influenced by AI, through personalized recommendations and conversational customer service, not orders where an agent independently completed the transaction end to end.

Set against that, Adobe's year-round tracking of direct AI-referred traffic to U.S. retail sites shows a much smaller share of total traffic, even with the triple-digit growth rates above. Both things are true at once: AI's role in shopping is still a modest slice of total volume most of the year, and it is growing faster than any other channel by a wide margin. The second fact, not an inflated headline number, is the actual reason a manifest matters now rather than later.

Why agents need a manifest, not just a good website

An AI shopping agent does not browse a storefront the way a person does. It cannot look at a page and infer where checkout lives, which payment methods are accepted, or what a "buy" button actually does. It needs that information exposed in a structured, predictable format it can query directly.

Adobe's own AI Content Visibility Checker, a tool it built to evaluate exactly this, assigns retail pages a score out of 100% for how much of their content is actually machine-readable to large language models. A page scoring 50%, per Adobe's own framing, means roughly half its content is effectively invisible to the AI systems now sending it traffic. That gap is not really a content problem, it is a structure problem, and it is precisely what a manifest is designed to close.

Two major standards have emerged to solve this in the last year. Google's Universal Commerce Protocol (UCP) launched January 11, 2026, co-developed with Shopify, Etsy, Wayfair, Target, and Walmart. OpenAI's Agentic Commerce Protocol (ACP), co-developed with Stripe, launched months earlier, in September 2025, and now powers Instant Checkout for ChatGPT's 700 million-plus weekly users. Both rely on the same basic idea: a structured, machine-readable manifest published at a known location, rather than an agent guessing at a website's layout.

What happens if a site doesn't have one

Without a manifest, an agent has no reliable way to confirm what a store sells, which payment methods it accepts, or how to complete a purchase there. In practice that means the store is simply skipped in favor of one an agent can parse deterministically, regardless of how good the underlying product or website actually is. As AI-referred traffic keeps compounding at the rates Adobe is reporting, that gap gets more expensive every quarter it goes unaddressed, not less.

Find out where your site stands

Check whether a domain publishes a valid UCP manifest, with a category breakdown and a specific fix for every issue found.

Frequently asked questions