AI agents are no longer a concept demo at tech conferences. They are shopping. Twenty-three percent of Americans made a purchase through an AI agent in the past month, according to Morgan Stanley's AlphaWise survey. Adobe measured an 805% year-over-year increase in AI-driven traffic to US retail sites on Black Friday 2025. Salesforce reports that 39% of consumers already use AI for product discovery — and over half of Gen Z does.
The agentic commerce wave is not arriving. It has arrived. The question every online business owner needs to answer is whether their digital presence is built for this new buyer — one that does not browse, does not scroll, and does not click ads.
Key Takeaways
- 23% of Americans made a purchase through an AI agent in the past month (Morgan Stanley), and Adobe measured an 805% year-over-year increase in AI-driven traffic to US retail sites on Black Friday 2025.
- McKinsey forecasts agentic commerce will reach $5 trillion in global volume by 2030, with 15-25% of all ecommerce flowing through agentic channels.
- AI-generated product recommendations convert at 4.4 times the rate of traditional search, yet ChatGPT referrals still convert 86% worse than affiliate links — indicating merchant infrastructure is not keeping pace with demand.
- 42% of customers abandon purchases due to insufficient product information, and poor data quality costs businesses an average of $15 million annually — problems that AI agents amplify by moving to competitors with cleaner data.
- The practical readiness checklist includes auditing structured data, making content AI-parseable, publishing an llms.txt file, ensuring real-time data accuracy, and building agent-compatible checkout flows.
What Agentic Commerce Actually Means for Your Business
Traditional ecommerce assumes a human in the loop. Someone types a query, scans a results page, clicks through to your site, reads your product descriptions, compares prices, and eventually checks out. Every element of your online presence — from page design to promotional banners — was built to influence that human decision-making process.
Agentic commerce removes the human from most of that journey. An AI agent receives a request ("find me a reliable web hosting provider under $30/month with 24/7 support"), then autonomously researches options, evaluates providers, compares pricing and policies, and either makes a recommendation or completes the purchase directly. The agent does not see your carefully designed hero section. It does not notice your trust badges. It reads your structured data, parses your product attributes, and evaluates whether your offering matches the criteria it was given.
McKinsey forecasts agentic commerce will reach $5 trillion in global volume by 2030. Bain projects 15–25% of all ecommerce will flow through agentic channels by the same date. These are not projections about a distant technology shift. They describe a transition that is already underway.
The Readiness Gap Is Real — and Widening
The uncomfortable reality is that most online businesses are not ready. A PYMNTS Intelligence and Visa study found that while 80% of payment acquirers say their infrastructure supports agent-led transactions, merchants lag far behind in practical readiness. Integration costs, legacy systems, and the sheer effort of connecting AI-compatible tooling to existing operations remain significant barriers.
The numbers tell the story of a market caught between demand and delivery. AI-generated product recommendations convert at 4.4 times the rate of traditional search — yet ChatGPT referrals still convert 86% worse than affiliate links. The demand signal is strong. The merchant infrastructure to capture it is not.
Mirakl's research puts a price on the problem: 42% of customers abandon purchases due to insufficient product information, and poor data quality costs businesses an average of $15 million annually. When an AI agent encounters incomplete or contradictory product data, it does not ask for clarification. It moves on to a competitor whose data is clean.
Where Most Online Businesses Fall Short
The readiness gap is not about whether you have a website. It is about whether your website speaks the language that AI agents understand. Here are the areas where businesses most commonly fail.
Structured Data Is Missing or Incomplete
AI agents rely on structured data — Schema.org markup, clean product feeds, machine-readable pricing and availability — to evaluate your offering. Most product data was built for search filters and human browsing, not for AI parsing. If your pages lack proper markup for price, availability, shipping timelines, return policies, and specifications, agents cannot reliably include you in their evaluations.
Content Is Built for Humans Only
Your marketing copy might convert beautifully when a human reads it. But AI agents parse content differently. They look for clear, factual, structured information — not emotional appeals or clever wordplay. Optimising your content for AI search means providing direct answers to the questions agents are programmed to ask: What does this product do? What does it cost? How quickly does it ship? What is the return policy?
No Machine-Readable Trust Signals
Trust is a prerequisite for agentic transactions. Seventy-eight percent of financial institutions expect fraud to spike from AI shopping agents, according to Salesforce. Agents need verifiable trust signals — transparent policies, clear identity information, third-party certifications, and consistent data across platforms. A beautiful "About Us" page does not substitute for structured authority signals that machines can verify.
Technical Infrastructure Cannot Handle Agent Interactions
AI agents interact with your site differently from human visitors. They may send requests at unusual rates, access API endpoints or product feeds directly, and expect real-time data on inventory and pricing. Sites built solely for browser-based human interaction often break under agent-style access patterns — returning CAPTCHAs, blocking automated requests, or serving outdated cached data.
A Practical Readiness Checklist
Preparing for agentic commerce does not require rebuilding your entire digital presence. It requires targeted improvements to how your business communicates with machines.
Audit your structured data. Every product or service page should have complete Schema.org markup — Product, Offer, Organization, FAQ at minimum. Price, availability, shipping, and return policies must be machine-readable, not buried in paragraph text.
Make your content AI-parseable. Review your key pages through the lens of an AI agent. Can a machine extract your value proposition, pricing structure, and differentiators without interpreting marketing language? Direct, factual content wins in agentic commerce.
Publish an llms.txt file. This emerging protocol tells AI agents what your business does, what you offer, and where to find key information. It is the equivalent of robots.txt for the agentic era.
Ensure real-time data accuracy. Inventory levels, pricing, and availability must be current. An agent that finds a discrepancy between your listed price and checkout price will not complete the purchase — and may deprioritise your business for future queries.
Test your AI visibility. You cannot improve what you cannot measure. Run an AI readiness scan to understand how AI agents currently perceive your business — which signals they can read, which they cannot, and where the gaps are.
Build for agent-compatible checkout. As standards like Shopify's Universal Commerce Protocol (UCP) mature, businesses will need checkout flows that AI agents can navigate programmatically. Start by ensuring your checkout does not rely on visual cues or JavaScript interactions that agents cannot process.
The Window Is Open — but Closing
The current moment offers an asymmetric advantage. Most businesses are not yet optimised for agentic commerce. Those that move now — structuring their data, cleaning their product information, making their content machine-readable — will be the ones that AI agents learn to trust and recommend first.
AI agents build preference models. Once an agent successfully completes a transaction with a business that has clean data, transparent policies, and reliable fulfilment, that business earns priority in future agent recommendations. Early movers do not just win the first transaction. They build compounding visibility as agents learn which sources are reliable.
The businesses that wait for agentic commerce to "mature" before acting will find themselves in a market where agent preferences are already locked in — and the cost of breaking into those preference models is significantly higher than the cost of preparing now.
Frequently Asked Questions
How do I know if my business is ready for agentic commerce?
Check whether your product or service pages have complete Schema.org markup (Product, Offer, Organization, FAQ at minimum), whether your key content is machine-parseable (direct answers rather than marketing language), whether your data is accurate in real time (pricing, availability, inventory), and whether your checkout flow can be navigated programmatically. A free AI scan reveals how AI agents currently perceive your business in under 30 seconds.
What is an llms.txt file?
An llms.txt file is an emerging protocol that tells AI agents what your business does, what you offer, and where to find key information. It functions as the equivalent of robots.txt for the agentic era — a machine-readable summary that helps AI agents quickly understand your business without parsing your entire website.
How quickly is agentic commerce growing?
Adobe measured an 805% year-over-year increase in AI-driven traffic to US retail sites on Black Friday 2025. Morgan Stanley's AlphaWise survey found 23% of Americans made a purchase through an AI agent in the past month. Salesforce reports 39% of consumers already use AI for product discovery, with over half of Gen Z doing so.
What happens if my business isn't optimised for AI agents?
AI agents that encounter incomplete or contradictory product data move on to a competitor whose data is clean — they do not ask for clarification. Businesses that wait for agentic commerce to "mature" will find agent preferences already locked in, with the cost of breaking into those preference models significantly higher than the cost of preparing now.
Your customers are already delegating purchase decisions to AI. The only question is whether your business is part of the answer those agents return. You can see a preview of how AI-ready your website is with a free AI scan — 30 seconds, no signup. For the complete picture, SwingIntel's AI Readiness Audit delivers expert research across 9 AI platforms.
