// THE 2027 CMO / MARKETING TO MACHINES
Marketing to Machines: When AI Agents Become the Customer
Agent commerce has disclosed transactions, pilots, and large forecasts. What is still hard to establish is comparable retail scale and the share of purchases an agent actually completes.
Gartner's dated forecasts anticipate agent-intermediated buying and a change in channel-based marketing. The 2027 CMO series treats those as planning questions, separate from observed sales.
The available disclosures use different measures: assistant referrals, merchant checkout, completed agent-initiated transactions, and machine-to-machine payments. They need separate counts and denominators before a marketer can compare adoption.
The stat everyone is going to misquote
The 85 percent figure is already circulating as "85 percent of customer data will come from AI agents." That is not what Gartner published. The actual subheading, from a January 2025 release, reads: "By 2027, 85% of Customer Data will be Collected from Automated Interactions or Those Led by AI Agents." The word "or" is not decoration. It folds in ordinary automation, form fills, app telemetry, tracking pixels, alongside anything an agent actually does. No methodology or sample is attached to the number either. It is an analyst forecast, and the version worth repeating is the one Gartner actually wrote, not the tighter, scarier one that spreads faster.
The same Gartner release forecasts lower mobile-app usage by 2027, attributing it to both assistants and consolidation of branded apps. These are forecast explanations, not measured evidence that assistants alone caused the decline.
Gartner also forecasts that more than 40 percent of agentic AI projects will be cancelled by the end of 2027 because of cost, unclear value, or inadequate controls. That is a planning risk, not evidence that a particular commerce implementation will fail.
What the machine actually buys, right now
OpenAI's March 24, 2026 product-discovery update says Instant Checkout did not provide the intended flexibility and describes merchant checkout in an in-app browser. Discovery in an assistant and checkout on a merchant site remain different from an agent autonomously placing an order.
Etsy's cited earnings remarks put agentic traffic below one percent of total traffic, a visit measure rather than an order count. Shopify reports AI-referred order-growth multiples without an absolute base in the cited disclosure. Visa disclosed hundreds of completed agent-initiated transactions in December 2025, without dollar volume. That is a transaction count. The x402 counter measures another market, machine payments, and cannot establish retail purchasing scale.
The cited Adobe retail reports describe AI-sourced visits and growth rates; missing absolute shares limit comparisons. Salesforce's Cyber Week definition includes recommendations and conversational service, and its holiday traffic forecast includes competitor scrapers. Those categories cannot be relabelled as agent-completed checkout. Define the counted event before carrying any of these figures into a plan.
The stats that do not survive a source check
Ahrefs' 23-times conversion claim comes from ahrefs.com, and Semrush's 4.4-times value claim does not supply a comparable sample, period, and conversion definition in the cited passage. They cannot establish universal visitor value. The per-engine measurement article separates conversion events from revenue per visitor and checks the benchmark arithmetic.
The cited Semrush and Ahrefs traffic-share figures come from vendor site panels with their own periods and definitions. They measure referrals, not all agent traffic or autonomous purchases. Academic agent experiments below address behavior under controlled settings, a different kind of evidence. Neither type substitutes for completed-transaction measurement at a merchant.
What actually changes an agent's decision
Researchers at Bayes Business School and King's Business School ran real shopping agents against the standard toolkit of persuasion: scarcity cues, countdown timers, strike-through discounts, bundles. Across four models and four product categories, none of it reliably moved the agent, and some of it backfired. Two things did work with consistency: star ratings swayed the agent's choice, and price worked in the predictable direction, a higher price reliably hurt selection. Treat those results as hypotheses for the agents your customers use. Test the effect before committing campaign spend.
The What Is Your AI Agent Buying? study reports sponsored-tag penalties and endorsement effects in its tested agents. One Polluted Page Is Enough reports manipulation under its experimental setup, and Incumbent Advantage examines brand and optimization effects. These experiments identify hypotheses and vulnerabilities to test; they do not establish that every production shopper responds in the same way.
What the llms.txt evidence establishes
llms.txt was proposed in September 2024 as a way to provide readable site context to AI tools. Providing that context and improving search citations are separate claims.
Google's guidance says Search does not use these proposed files. Ahrefs' May 2026 sample of 137,210 domains found 97 percent of published llms.txt files received no requests that month; some agents and training/retrieval crawlers did fetch the remainder. A request does not prove citation impact. Shopify's documentation describes llms.txt and llms-full.txt as alternate URLs mirroring agents.md by default. Availability does not establish usage or benefit.
MCP connects applications with tools and context. UCP defines a business profile at /.well-known/ucp declaring supported commerce capabilities. A context document and a working checkout capability have different jobs. Publishing the profile alone does not establish that transactions work or that an agent has used it.
What I actually do, and what I do not claim it does
This site keeps llms.txt and llms-full.txt as optional context files. Their presence does not establish why a page gets cited. Measure actual retrieval and reader outcomes before making that claim.
Useful site work remains accurate product information, accessible pages, clear authorship, and structured data that reflects visible content. Valid schema alone does not establish a causal citation gain. For commercial outcomes, follow observed referrals into pipeline and count eligible transactions under the same definition across periods.
Frequently asked questions
Will 85 percent of customer data come from AI agents by 2027?
Gartner's January 2025 forecast includes automated interactions OR those led by AI agents. Ordinary automation is included. The forecast is not a measured pure-agent share.
Are AI agents completing purchases?
Visa disclosed hundreds of completed agent-initiated transactions in December 2025 without a dollar volume. That establishes a disclosed count, but not broad retail scale. Assistant discovery, merchant-site checkout, and an agent-initiated transaction are different measures.
Does llms.txt establish a citation advantage?
Google says Search does not use the proposed file format. Ahrefs' May 2026 domain sample found most published files received no requests, while some agents and crawlers did fetch them. Neither source establishes a causal citation benefit across all assistants.
What changes an AI shopping agent's decision?
The cited shopping-agent experiments found effects from price, ratings, endorsements, and product-description changes under their tested models and settings. Results and manipulation risks need validation in the system a merchant actually serves.
How should a marketer prepare for agent commerce?
Make product information accurate and accessible, define which transactions an agent may perform, and test the relevant checkout integration. For UCP implementations, the business profile declares supported capabilities at /.well-known/ucp. Measure completed eligible transactions separately from discovery traffic.
Gartner mobile app usage prediction, including the 85 percent customer-data subheading and the one-third gen-AI content subheading: Gartner newsroom, January 15, 2025, analyst Emily Weiss. Gartner's channel-based-marketing prediction: Gartner newsroom, January 15, 2026. The 90 percent B2B agent-intermediated buying and $15 trillion figure: Gartner newsroom, October 21, 2025. The 40-plus percent agentic-project cancellation prediction and the roughly 130 real vendors estimate: Gartner newsroom, June 25, 2025, analyst Anushree Verma. OpenAI's Instant Checkout retreat: OpenAI, "Powering Product Discovery in ChatGPT," March 24, 2026. Etsy's under-1-percent agentic traffic disclosure: Etsy Q4 2025 and Q1 2026 earnings calls, CEO Kruti Patel Goyal. Visa's hundreds-of-transactions disclosure: Visa, December 18, 2025. The x402 micropayment protocol figures: x402.org live counter, read August 4, 2026. Adobe's retail AI traffic-share chart and the conversion reversal: Adobe Digital Insights, AI-Sourced Traffic Insights, January 12, 2026, and follow-on reports in April and June 2026. Salesforce's Cyber Week and 2026 holiday-prediction wording: Salesforce newsroom, December 5, 2025 and July 20, 2026. Ahrefs' 23x conversion claim and its own caveat on click-through: Ahrefs, June 2025. Semrush's 4.4x claim: Semrush, July 21, 2025. Semrush's 50,000-site traffic-share study: Semrush, April 27, 2026. Ahrefs' 74,752-site fixed-cohort tracker, climbing from 0.28 percent in March 2026 to 0.39 percent by June: Ahrefs, May and June 2026 updates. The shopping-agent persuasion research: Jafar Sabbah and Oguz A. Acar, reported in Harvard Business Review, May 12, 2026. The sponsored-tag penalty and market-share findings: "What Is Your AI Agent Buying?", arXiv 2508.02630, August 2025. The single-polluted-page manipulation rates: "One Polluted Page Is Enough," arXiv 2606.13610, June 2026. The incumbent-advantage and GEO-arms-race collapse findings: "Incumbent Advantage," arXiv 2606.17443, June 2026. Google's statement that Search ignores llms.txt: Google Search Central, "Guide to Optimizing for Generative AI Features," updated July 10, 2026. John Mueller's llms.txt statement: Bluesky, June 17, 2025. The 97-percent-zero-requests finding: Ahrefs, June 15, 2026, 137,210 domains. Shopify's llms.txt auto-deployment as an agents.md alias: Shopify changelog, May 2026. The Model Context Protocol's server count and foundation transfer: Anthropic and the Agentic AI Foundation, December 2025.
About the author
Jeff Brokaw is a CMO and Certified Chief AI Officer who builds AI workflows, measurement systems, and the commercial operations around them.