Home / Blog / Agentic commerce by the numbers

Agentic commerce by the numbers (2026)

6 min read · Market data · Updated 2026

Agentic commerce stopped being a forecast and became a line on the dashboard. Over the 2025 holiday season, Adobe measured a 693% year-over-year jump in traffic to US retail sites from generative-AI sources, and Salesforce estimated that AI and agents influenced 262 billion dollars of online holiday sales. AI assistants now send real visitors, shape real spend, and crawl real catalogues every day. Here is the picture in numbers.

A note on the figures below. Every number here is attributed to the firm that reported it, and this space moves fast, so methodologies and figures shift from one release to the next. Treat them as the shape of a steep trend, and check the latest primary source, all linked at the end, before you drop a single figure into a deck. The third decimal place is not the point. The direction is, and the direction is not in doubt.

The traffic signal: AI referrals grew in triple digits

The clearest early signal is referral traffic. Adobe Analytics, whose retail figures are drawn from more than one trillion visits to US retail sites, reported that traffic from generative-AI sources grew 693% year over year across the 2025 holiday season, and that retail saw the biggest gains of any industry, up 769% in November alone. Earlier in the year the same channel was up 393% year over year in the first quarter. The absolute share is still small next to search and direct, but a channel compounding at that rate does not stay small for long. This is the hockey-stick portion of the curve.

What those visitors do once they arrive matters just as much. Adobe found AI-referred shoppers convert 31% better than other sources over the holiday, and by March 2026 that reached a record 42% better than the site average, a full reversal from a year earlier when AI traffic actually converted worse. They also engage more: sessions run 48% longer, visitors browse 13% more pages, and they are 33% less likely to bounce. Revenue per visit from AI traffic was up 254% for the holiday. The agent is not sending you tyre-kickers. It is pre-qualifying intent and handing you a visitor who already decided your store is a plausible answer.

The spend signal: AI influenced 262 billion dollars

Salesforce put a number on the money. Across the 2025 holiday season it estimated that AI and agents influenced 262 billion dollars of online sales and touched roughly 20% of all retail purchases, out of 1.29 trillion dollars in global online holiday spend. Traffic from third-party AI search such as ChatGPT and Perplexity doubled versus the year before, and shoppers arriving from those channels converted 9 times more often than shoppers referred from social media. Retailers that ran their own shopper agents grew sales 59% faster than those that stayed on the sidelines.

Influence is not the same as a checkout completed inside a chat window, and the two should not be conflated. But it is the leading indicator. Spend that an assistant shapes today is spend an assistant will complete tomorrow, as agentic checkout protocols mature and more shoppers trust an agent to press buy. The takeaway for a merchant is not the exact dollar figure. It is that a large and growing share of buying decisions now passes through a model before it reaches your store, and the model is deciding whether you are even in the consideration set.

The crawler signal: agents are already reading your store

You do not need a forecast for this one. It is in your server logs right now. AI crawlers such as GPTBot, PerplexityBot and Google's AI crawlers are already visiting e-commerce sites, pulling product data, prices and policies into the systems that answer shopper questions. AI assistants already recommend specific products and specific stores by name in response to plain-language requests.

This changes the nature of the risk. Being absent from agentic commerce is not a future decision you get to make later. The agents are visiting today, and they are forming a view of your catalogue today. If your structured data is thin, your availability values are missing, or your policies live only in a PDF, the agent forms that view anyway, and it forms it against you. Silence in your logs is not safety. It usually means the crawler came, could not read enough, and moved on.

The projection signal: analysts see zero-click buying

Looking further out, Gartner projects that AI agents will intermediate more than 15 trillion dollars in B2B spending by 2028, with a growing share of those transactions involving no human in the loop. Gartner also expects 33% of enterprise software to include agentic AI by 2028, up from under 1% in 2024, and 60% of brands to use agentic AI to run one-to-one customer interactions by the same year. Analyst projections are directional by nature and should be read as trajectory rather than promise. But the direction is consistent across firms and consistent with the traffic and spend data already on the ground. When the leading indicators and the long-range forecasts point the same way, the trend line is not a guess.

You do not have to believe the forecasts. The crawlers are already in your logs.

Where the figures come from

To keep this honest, here is who reported what. Verify against the primary source before you quote any of it.

• Adobe Analytics, 2025 holiday shopping data: AI referral traffic to US retail up 693% year over year, plus the conversion, engagement and revenue-per-visit lift for AI-referred visitors. Adobe report.
• Salesforce, 2025 holiday shopping data: 262 billion dollars of AI-influenced online sales, roughly 20% of purchases, AI-search traffic doubling, and the 9x conversion versus social. Salesforce report.
• Gartner: 15 trillion dollars in agent-intermediated B2B spend by 2028, 33% of enterprise software agentic by 2028, and 60% of brands using agentic AI for one-to-one interactions. Gartner newsroom.
• Your own crawler and server logs for GPTBot, PerplexityBot and Google AI crawler activity against your specific catalogue.

What the numbers add up to

Put the four signals together and the story is coherent. Referral traffic from AI is growing in triple digits. The visitors it sends convert better than your other channels and stay longer once they arrive. AI already influenced 262 billion dollars of holiday spend, and the analysts expect agent-led buying to keep compounding for years. None of these depends on a single vendor's methodology being exactly right, which is precisely why the trend survives the uncertainty in any one number.

That is the real cost curve. Every month a store stays unreadable to agents, it forfeits a slice of a channel that is compounding, and it does so silently, with no bounce-rate spike or error page to warn anyone. The work of making a store readable, trusted and buyable by agents, and keeping it that way as the standards move, is exactly what a Agentic Commerce Orchestrator™ (ACO™) does. Goofre™ is the platform that turns e-commerce developers into ACO™s and gives merchants a measured read on where they stand before the curve gets any steeper.