Agentic Commerce Is Coming: Is Your Marketing Organization Ready?
Agentic commerce is the practice of AI agents shopping, comparing, and completing purchases on behalf of a person or a business, often without a human clicking through every step.
The agent searches, compares prices, checks reviews, and in some cases finishes the checkout. Analysts disagree on how big this gets by 2030, and the range runs from $190 billion to $5 trillion depending on what gets counted. But every major research firm agrees on one point: agents are already shaping how people discover and buy products, and marketing teams built for human clickstreams are not set up for machine buyers.
A shopper used to type a query into Google, click a link, and land on a product page your team designed. Now that shopper might ask ChatGPT to find "a waterproof jacket under $150," and the agent pulls product data straight from a feed, compares three options, and never sees your homepage at all. That shift changes what marketing has to build, measure, and staff for.
The Market Numbers Are All Over the Map, and That's the Point
Ask five research firms how big agentic commerce will be by 2030 and you get five different answers. That's not sloppy research. Each firm counts something different, and the gap tells its own story about how fast the ground is moving.
McKinsey puts the US opportunity at $900 billion to $1 trillion by 2030, with a global range of $3 trillion to $5 trillion, counting orchestrated revenue across the full commerce chain. Morgan Stanley takes a narrower view and requires real autonomous action from the agent, landing on $190 billion to $385 billion, or 10 to 20 percent of US ecommerce. Bain & Company includes agent-influenced purchases, not just agent-completed ones, and projects $300 billion to $500 billion, or 15 to 25 percent of US ecommerce. eMarketer counts only checkout that happens inside an AI platform itself, the strictest definition of all, and forecasts $20.9 billion for 2026, close to 1.5 percent of total ecommerce.
On the B2B side, the numbers jump. Gartner expects 90 percent of B2B buying to run through AI agents by 2028, pushing more than $15 trillion through agent-to-agent exchanges.
| Research Firm | 2030 Estimate | What It Measures |
|---|---|---|
| McKinsey | $900B–$1T US / $3T–$5T global | Orchestrated revenue across the full commerce chain |
| Bain & Company | $300B–$500B (15–25% of US ecommerce) | Agent-completed and agent-influenced purchases |
| Morgan Stanley | $190B–$385B (10–20% of US ecommerce) | Purchases requiring meaningful autonomous agent action |
| eMarketer | $20.9B in 2026 (~1.5% of total ecommerce) | Checkout completed entirely inside an AI platform |
| Gartner (B2B) | $15T+ through agent exchanges by 2028 | B2B procurement volume moved through AI agents |
Consumers Are Not Waiting for the Perfect Definition
People started using AI agents to shop before most brands finished debating what "agentic commerce" means. Adobe tracked AI-driven traffic to US retail sites climbing 693 percent year over year during the 2025 holiday season, and the pace held into 2026, up 393 percent in the first quarter and peaking at 1,151 percent in December.
The conversion story flipped fast, too. In March 2025, AI-referred traffic converted 38 percent worse than regular traffic. By March 2026, it converted 42 percent better, a record in Adobe's data set. Shoppers who arrive through an AI agent spend 48 percent longer on site and generate 37 percent more revenue per visit than shoppers who arrive through traditional search.
Salesforce's review of more than 1.5 billion shoppers during the 2025 holidays found AI and agents drove $262 billion in global online sales. Over Cyber Week alone, agents influenced 20 percent of all purchases. And per Checkout.com, 47 percent of consumers planned to use an AI agent for their holiday shopping.
None of this means checkout has gone fully autonomous. Forrester's research through mid-2026 found that most "agentic" shopping experiences are still conversational: a chatbot recommends and compares, but a person still clicks buy. True hands-off purchasing stays rare, and B2B agent-to-agent negotiation is early but growing. The behavior is real. The full autonomy narrative is running ahead of it.
The Real Problem: Most Marketing Organizations Are Not Built for This
Here is the gap that matters more than any market-size debate. According to the Martech for 2026 report from Scott Brinker and Frans Riemersma, 90.3 percent of marketing organizations now use AI agents somewhere in their stack. Only 23.3 percent have moved an agent into full production. The rest sit in pilots, experiments, or workflows too narrow to matter.
The same report found 52.4 percent of marketing leaders name organizational and process readiness as their biggest struggle, ahead of the technology itself. Gartner has separately reported that martech buyers use only 49 percent of the capability they already pay for. Teams keep buying tools before they fix the operating model underneath them.
Forrester's take on this is direct: the biggest barrier to agentic commerce readiness is not the AI. It's internal silos and a culture that resists change. Marketing, IT, customer service, and legal all have to coordinate on shared data and shared goals before an agent can act on a brand's behalf with any confidence.
Data structure sits at the center of the problem. AI agents pull from product feeds, structured content, and APIs, not from a page a designer laid out for a human eye. A retail strategist quoted in MarTech put it plainly: if a retailer's product data isn't structured and machine-readable, agents skip it and move to a competitor that got the data right. Content that hasn't been refreshed in 30 to 60 days loses visibility in AI-powered search, too, so stale content becomes a real cost, not just an SEO nuisance.
What "Ready" Actually Looks Like
A marketing organization ready for agentic commerce has a few things in place before the agents show up:
Machine-readable product and content data. Pricing, inventory, specs, and reviews need to live in structured, API-accessible formats, not buried in a PDF or a page template. If an agent can't parse it in milliseconds, the product doesn't exist to that agent.
Cross-functional teams built around the customer journey, not the channel. Channel-specific teams (email, social, paid) worked when a person clicked through each channel in sequence. Agents skip that sequence, so teams need to organize around outcomes and hand-offs instead.
A single source of truth for customer and product data. Composable customer data platforms and data warehouses now serve as the backbone most teams need before AI can act on their behalf with any reliability. Fragmented data produces fragmented, wrong decisions from an agent, fast.
Governance and guardrails, written down. Suppression rules, spend limits, and eligibility logic have to exist before an agent operates on a brand's behalf, not after something goes wrong.
A measurement plan that accounts for a channel you can't fully see. Only 13.6 percent of marketing teams currently measure AI inclusion rate or agent-referred conversion, according to the Martech for 2026 report. Attribution breaks the moment discovery moves inside a chat window, so teams need a plan for that blind spot now.
Where This Meets Demir Digital's Work
This is Design Ops and Marketing Ops territory, and it's the work Demir Digital builds for clients every day. Structured data, clean workflows, and teams that know why they're doing what they're doing don't happen by accident. They come from system audits, component libraries, playbooks, and embedded guidance that outlast a single project.
Getting ready for agentic commerce doesn't start with picking an AI vendor. It starts with the same groundwork that makes any digital system work: map the gaps, put the right frameworks in play, and build teams that can run the system after the consultants leave. That's the difference between chasing a trend and building capability that holds up when the next one arrives.
FAQ
-
Agentic commerce is the use of AI agents to search, compare, and purchase products or services on behalf of a person or business, sometimes with a human approving each step and sometimes without.
-
Estimates range widely by definition. McKinsey projects $3 trillion to $5 trillion globally, Bain projects $300 billion to $500 billion for the US, and Morgan Stanley projects $190 billion to $385 billion for the US, depending on how much autonomy each firm requires to count a purchase as "agentic."
-
Both are moving, but B2B is moving faster in relative terms. Gartner expects 90 percent of B2B buying to run through AI agents by 2028, with more than $15 trillion in transaction volume passing through agent-to-agent exchanges.
-
Research from Forrester and the Martech for 2026 report points to the same answer: organizational readiness, not technology. Silos between marketing, IT, and customer service, plus unstructured product and content data, hold most teams back more than any missing tool.
-
Start with a data and workflow audit. Structure product and content data for machine readability, break down channel silos in favor of journey-based teams, and put governance rules in place before agents start acting on the brand's behalf.
-
Trust is growing but uneven. Checkout.com found 47 percent of consumers planned to use an AI agent for holiday shopping, and Forrester's research found most current experiences are still conversational rather than fully autonomous, with a person approving the final purchase.
Related Reading
