The Era of AI Buying for You: 57% Would Switch Brands for a Better Deal—How EC Should Prepare

The way people shop online is changing quietly but fundamentally. Until now, finding a product, comparing options, adding to cart and pressing "buy" was the human's job. But in 2026, "agentic commerce"—where an AI agent handles that entire sequence on your behalf—is moving from proof-of-concept to practical use.

According to a survey published June 9, 2026 by the payments firm Checkout.com (across six markets including the UK and US), 33% of consumers expect at least 10% of their purchases to go through AI within a year. At the same time, 27% said they trust no organization to operate an AI shopping agent. Rising demand and stubborn distrust are advancing side by side.

This shift shakes the premises of marketing and e-commerce. Ranking high in search results, catching the eye with ads, nurturing brand affection—these playbook moves assume a human buyer, and they work differently the moment the buyer becomes an "agent." Below, we trace the reality and the numbers of agentic commerce, then dig concretely into what EC and brands should prepare.

The agent does the "search, compare, buy"

What decisively separates agentic commerce from conventional recommendations or chat assistance is that it transacts at the end. When a user states a goal in natural language (e.g., "a wrinkle-free shirt under ¥10,000 for next week's business trip"), the agent plans the steps, queries multiple stores and marketplaces, evaluates options against the constraints, executes payment within preset limits, and triggers the order and fulfillment. At the end of the conversation, money actually moves and an order is placed.

A range of consumer-side agents has emerged—OpenAI's ChatGPT, Perplexity's Comet, Amazon's Rufus, Google's AI shopping. Buyers will increasingly choose "which agent to delegate to" rather than "which site to buy from." The time a human spends looking at the shelf (the search results page) shrinks, while upstream the agent narrows the candidates—meaning the point of sale moves one stage further up the funnel.

The numbers show demand and distrust advancing together

First, the market's temperature, via Checkout.com's figures.

What stands out: while only 3% of transactions currently involve AI agents, 89% of merchants are already preparing. Adoption lies ahead, but the supply side is acting on the assumption it is coming. At the same time, consumers will not delegate unconditionally. There is a ceiling on the per-transaction amount they will allow without approval, and spending caps (30%), instant permission revocation (29%) and easy cancellation (28%) ranked as non-negotiables. Willing to delegate, but keeping the reins—that is the buyer's honest stance.

Brand choice shifts from the human to the agent

The heaviest number for marketing is that 57% said they would switch to another brand for a better deal. For groceries, 41% are open to delegating to AI, with household supplies close behind. In other words, the more an item is bought repeatedly and easily compared, the more likely an agent is to coolly swap it for the optimal choice.

With a human buyer, the impression of the packaging, a chance encounter on the shelf and past affection swayed the choice. But an agent compares dispassionately on "structured facts"—price, stock, delivery terms, review ratings, ease of returns. The emotional edge a brand built over years risks being passed over unless it registers in the agent's criteria. Conversely, elements that are clearly superior as data will be evaluated more fairly than ever.

The payment rails are already being laid

To the worry of "is it safe for AI to pay on its own," the payments industry has already begun building dedicated machinery. Visa is advancing a Trusted Agent Protocol (a transaction standard for verified agents); Mastercard's Agent Pay lets permitted agents transact on a person's behalf using "agentic tokens." OpenAI and Stripe co-developed the Agentic Commerce Protocol (ACP), and Google, with Shopify and others, launched the Universal Commerce Protocol (UCP).

The crucial part is that these begin to define technically "who, with what authority, up to how much, and revocably." Agentic commerce is not a one-off experiment but is becoming formal, cross-industry infrastructure built atop the foundations of payment, identity and permission management. For merchants, how they connect their inventory, pricing and product information to these protocols will determine whether they can win this acquisition channel in the near future.

What EC and brands should prepare

So where should sellers start? The starting point is improving the "machine readability" of product information. Price, stock, size, material, delivery time, return terms—organize these as structured data an agent can reliably retrieve, not as decoration on a page meant for human eyes. If the feed is low-quality, the brand won't even enter the agent's comparison set. This is the same shift as search SEO moving its center of gravity from "articles people read" to "information machines understand."

Next, sharpen the competitiveness of price and terms as "facts." In a world where 57% will switch, a small price gap or easier delivery and returns becomes the deciding factor. Emotional branding does not become meaningless, but it works only once you are "on the agent's playing field." Finally, design for trust. Make the spending caps, revocation and easy cancellation that consumers want explicit in your own purchase flow, and show that you are a seller buyers can safely delegate to even via an agent. It is surely coming but hard to time—which is why starting with product-data hygiene, a preparation that never goes to waste, is the sensible move.

Key takeaways

• "Agentic commerce," where AI agents handle everything from search to payment and fulfillment, is reaching practical use. Buyers will increasingly choose "which agent to delegate to" over "which site to buy from."
• In Checkout.com's survey (June 2026; six markets incl. UK/US), 33% expect 10%+ of purchases to be AI-driven within a year, 27% distrust all operators, and 57% would switch brands for better value. AI-involved transactions are 3% today, but 89% of merchants are preparing.
• Brand choice shifts from emotion to "structured facts." Because agents compare coolly on price, stock, delivery and returns, machine-readable product data is the top priority.
• Visa, Mastercard, OpenAI×Stripe and Google×Shopify are building payment and permission protocols. Merchants should hurry on three fronts: structured product data, competitive price/terms, and trust design such as spending caps.