Agentic Commerce by the Numbers Your Board Will Actually Believe
Sourced agentic commerce statistics for 2026: merchant adoption, consumer delegation intent, AI-influenced holiday sales, and early conversion data from agent checkouts.
On this page
AI and agents influenced $262 billion in online revenue during the 2025 holiday season, 20% of all orders, according to Salesforce. That one number settles the "is agentic commerce real" argument. What it does not settle is how fast delegation scales, which merchants are ready, and whether agent traffic actually converts. The numbers below answer those questions, each with a named source, grouped so you can lift them straight into a deck.
The headline numbers
- Global online holiday sales hit $1.29 trillion in 2025, up 7% year over year, and AI plus agents touched 20% of it. Source: Salesforce Shopping Index, January 2026. Read: agent influence is already a fifth of peak-season commerce, before most brands have shipped any agent-facing infrastructure.
- AI-referred traffic to retail sites grew 693% year over year during the 2025 holiday season, the largest jump of any industry Adobe tracks. Source: Adobe Digital Insights, January 2026. Read: assistants are now a referral channel with its own growth curve, and it compounds quarterly.
- Gartner expects AI agents to handle 90% of B2B purchases by 2028, intermediating more than $15 trillion in spend. Source: Digital Commerce 360, November 2025. Read: the B2B forecast is more aggressive than any consumer projection; procurement teams will meet agents before shoppers do at scale.
Consumer delegation intent
The question budget owners ask first: will people actually let an agent spend their money? Survey data from payments providers, who see transaction intent rather than sentiment, says yes with conditions.
- 44% of UK shoppers are open to AI handling the entire shopping process, final purchase included, once criteria like budget, features, and brand are set. Source: Adyen Retail Report, January 2026. Read: delegation intent is mainstream before the checkout rails are.
- Adyen's US edition of the same report puts American willingness above 50%, and its Canadian data lands at 51%. Read: North American shoppers are ahead of the UK on full delegation, which is unusual for a payments behavior.
- 33% of consumers expect at least 10% of their purchases to be AI-driven within a year. Source: Checkout.com, 2026 agentic commerce research across six countries. Read: a third of your customers have already priced in a delegated share of wallet.
| Market | Share open to agent-completed checkout | Source |
|---|---|---|
| United States | 50%+ | Adyen Retail Report 2026 |
| Canada | 51% | Adyen Retail Report 2026 |
| United Kingdom | 44% | Adyen Retail Report 2026 |
The condition buried in every one of these figures: preferences must be set first. Consumers delegate execution, and they keep control of criteria. That makes structured preference data the scarce asset, which is why teams investing in AI audience segmentation for ecommerce ads are effectively pre-building the profiles agents will shop against.
Merchant adoption and readiness
- 42% of merchants are already testing agentic commerce in some form. Source: Checkout.com Agentic Commerce Report, 2026. Read: testing is the median position; doing nothing now signals a deliberate bet against the channel.
- 72% of merchants agree consumers will adopt agent-led shopping faster than most merchants can support it. Source: Checkout.com, same report. Read: the industry has publicly conceded a readiness gap; the deck-worthy question is whether yours is smaller than your category's.
- 91% of small businesses, 97% of mid-market merchants, and 99% of large enterprises report being at least somewhat familiar with agentic commerce. Source: PayPal Agentic Commerce Pulse Report, 2026. Read: awareness is saturated; differentiation now comes from execution, which is where agentic AI automation work actually gets funded.
Traffic and early conversion data
This is the section that unlocks budget, because it moves the conversation from intent to observed behavior.
- AI referrals to retail sites outperformed non-AI traffic sources across nearly every metric Adobe measures this holiday season, from conversion to bounce rate to revenue. Source: Adobe Digital Insights quarterly AI traffic report, January 2026. Read: AI-referred visitors arrive later in the funnel, pre-qualified by the conversation that sent them.
- Agent-assisted sessions show roughly 4.4x higher conversion potential, yet actual agent conversion runs about 86% worse on sites whose infrastructure blocks or breaks automated buyers. Source: MetaRouter, 2026 trends analysis. Read: the demand exists and the plumbing loses it; bot mitigation tuned for scrapers is silently refusing revenue.
That second bullet is the single most useful chart-in-waiting for an agent-readiness pitch. High-intent traffic, low completed conversion, and the gap is an engineering problem rather than a demand problem. If you want to know how visible your catalog already is to assistants, run the AI visibility checker before the meeting; a concrete score beats a hypothetical every time. The measurement discipline is the same one covered in ecommerce campaign optimization with AI customer insights: instrument first, then argue.
The rails that shipped
Projections mattered less once real checkout infrastructure went live.
- ChatGPT's Instant Checkout lets US users complete purchases inside the chat, with merchants paying a fee only on completed transactions and product ranking unaffected by checkout participation. Source: OpenAI, with Stripe as launch payments partner. Read: the Agentic Commerce Protocol it runs on is an open standard, so the integration you build for one surface is designed to travel.
- The protocol launched with Etsy sellers and opened to Shopify merchants, meaning long-tail catalogs get agent distribution on day one. Source: OpenAI, same announcement. Read: agent surfaces will not wait for enterprise release cycles; smaller competitors can transact there first.
Projections to treat carefully
Two habits keep a statistics slide credible. First, separate observed data (Salesforce, Adobe) from stated intent (Adyen, Checkout.com) from analyst projection (Gartner); mixing the three is how decks get challenged. Second, convert every stat into a company-specific number before presenting. "20% of holiday orders were AI-influenced" is trivia; "20% of our December orders would flow through surfaces we currently can't transact on" is a budget line. The same conversion discipline applies to paid channels, where the arithmetic in the AI audience targeting for ecommerce ads ROAS playbook shows how influenced-revenue claims should be discounted before they hit a forecast. Pressure-test the resulting numbers in a ROAS calculator rather than presenting influence as incrementality.
One more caveat worth saying out loud in the room: "AI-influenced" is a generous metric. Salesforce counts orders where AI touched the journey at any point, which includes recommendations and chat support alongside true agent checkouts. The genuinely agentic share, where software selected and bought autonomously, is still small. The growth rates are the story, and they are steep enough that the caveat cuts against complacency rather than against investment.
Sources
All external links appear inline above, once per source: Salesforce 2025 Holiday Shopping Data; Adobe Digital Insights quarterly AI traffic report; Adyen Retail Report 2026 (UK edition); Checkout.com agentic commerce consumer and merchant research; PayPal Agentic Commerce Pulse Report; Digital Commerce 360 coverage of Gartner's B2B agent forecast; OpenAI's Instant Checkout announcement; MetaRouter's 2026 agentic commerce trends analysis.