Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Liability Is the Real Dealbreaker
The report highlights a fundamental question: when an AI‑driven purchase goes wrong, who’s accountable? Merchants point to payment providers (35%), AI agent platforms (28%), then themselves (22%). Consumers blame the AI platform first (38%), then themselves (18%), merchants (10%), and payment providers (9%).
This confusion about accountability is why agentic commerce remains stuck in the experimental phase. Nobody wants to own the risk.
Consumer Demands Before Trusting an AI Agent
Before letting an agent buy on their behalf, consumers want concrete safeguards. Checkout.com’s data lists the essentials:
- A spending cap (30%)
- The ability to revoke the authorization instantly (29%)
- Simplified cancelation or returns (28%)
- Required validation above a certain amount (27%)
- Display of available options before finalizing the purchase (26%)
Here’s where merchants are missing the mark. A quarter of consumers say they’d stop using an AI agent if a purchase was difficult to return or dispute, yet only 16% of merchants see returns and customer service as a barrier to their preparedness.
Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Merchants and Consumers Disagree on What to Delegate
Merchants expect delegation to start with complex, high-value categories: travel planning (35%), subscriptions and provider switching (33%), financial products (31%). But consumers want to hand over low-risk, everyday purchases first: groceries (41%), household items (31%), fashion (26%), beauty (25%). The logic is simple—you trust an AI to buy groceries before it alters your investment portfolio.
Liability Is the Real Dealbreaker
The report highlights a fundamental question: when an AI‑driven purchase goes wrong, who’s accountable? Merchants point to payment providers (35%), AI agent platforms (28%), then themselves (22%). Consumers blame the AI platform first (38%), then themselves (18%), merchants (10%), and payment providers (9%).
This confusion about accountability is why agentic commerce remains stuck in the experimental phase. Nobody wants to own the risk.
Consumer Demands Before Trusting an AI Agent
Before letting an agent buy on their behalf, consumers want concrete safeguards. Checkout.com’s data lists the essentials:
- A spending cap (30%)
- The ability to revoke the authorization instantly (29%)
- Simplified cancelation or returns (28%)
- Required validation above a certain amount (27%)
- Display of available options before finalizing the purchase (26%)
Here’s where merchants are missing the mark. A quarter of consumers say they’d stop using an AI agent if a purchase was difficult to return or dispute, yet only 16% of merchants see returns and customer service as a barrier to their preparedness.
Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Big Disconnect: Expectations vs. Merchant Readiness
Here’s what actually happened. 72% of merchants think consumers will adopt agent-driven purchases faster than businesses can handle them. But consumer awareness is uneven—71% of 18-24-year-olds know about shopping agents, while only 19% of those 55+ do.
The trust gap is stark: 64% of younger shoppers are comfortable letting AI finalize a purchase, but 63% of older consumers are uncomfortable. A quarter of consumers say they’ll never delegate any purchase to an AI agent. This won’t be a quick or uniform shift.
US merchants are more bullish than their UK counterparts: 82% of American merchants see agentic commerce as the biggest disruption their industry has faced, versus 58% in the UK, where the fear of disputes and chargebacks is higher.
Merchants and Consumers Disagree on What to Delegate
Merchants expect delegation to start with complex, high-value categories: travel planning (35%), subscriptions and provider switching (33%), financial products (31%). But consumers want to hand over low-risk, everyday purchases first: groceries (41%), household items (31%), fashion (26%), beauty (25%). The logic is simple—you trust an AI to buy groceries before it alters your investment portfolio.
Liability Is the Real Dealbreaker
The report highlights a fundamental question: when an AI‑driven purchase goes wrong, who’s accountable? Merchants point to payment providers (35%), AI agent platforms (28%), then themselves (22%). Consumers blame the AI platform first (38%), then themselves (18%), merchants (10%), and payment providers (9%).
This confusion about accountability is why agentic commerce remains stuck in the experimental phase. Nobody wants to own the risk.
Consumer Demands Before Trusting an AI Agent
Before letting an agent buy on their behalf, consumers want concrete safeguards. Checkout.com’s data lists the essentials:
- A spending cap (30%)
- The ability to revoke the authorization instantly (29%)
- Simplified cancelation or returns (28%)
- Required validation above a certain amount (27%)
- Display of available options before finalizing the purchase (26%)
Here’s where merchants are missing the mark. A quarter of consumers say they’d stop using an AI agent if a purchase was difficult to return or dispute, yet only 16% of merchants see returns and customer service as a barrier to their preparedness.
Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Let’s look at the numbers. Despite the media noise, AI agents play a significant role in product selection and checkout in just 3% of online transactions in the US and UK. That’s from Checkout.com’s second edition of their agentic commerce report, based on 12,000 consumers and 400 payment executives.
I’ve seen this play out before. When a technology gets this much hype, the actual usage is always lower than the anticipation. OpenAI’s agentic capabilities, Mastercard’s agentic commerce infrastructure, Google’s Universal Commerce Protocol for Search AI Mode and Gemini, Visa’s Intelligent Commerce and Trusted Agent Protocol—all are in controlled experiments, not real-world scale.
Big Disconnect: Expectations vs. Merchant Readiness
Here’s what actually happened. 72% of merchants think consumers will adopt agent-driven purchases faster than businesses can handle them. But consumer awareness is uneven—71% of 18-24-year-olds know about shopping agents, while only 19% of those 55+ do.
The trust gap is stark: 64% of younger shoppers are comfortable letting AI finalize a purchase, but 63% of older consumers are uncomfortable. A quarter of consumers say they’ll never delegate any purchase to an AI agent. This won’t be a quick or uniform shift.
US merchants are more bullish than their UK counterparts: 82% of American merchants see agentic commerce as the biggest disruption their industry has faced, versus 58% in the UK, where the fear of disputes and chargebacks is higher.
Merchants and Consumers Disagree on What to Delegate
Merchants expect delegation to start with complex, high-value categories: travel planning (35%), subscriptions and provider switching (33%), financial products (31%). But consumers want to hand over low-risk, everyday purchases first: groceries (41%), household items (31%), fashion (26%), beauty (25%). The logic is simple—you trust an AI to buy groceries before it alters your investment portfolio.
Liability Is the Real Dealbreaker
The report highlights a fundamental question: when an AI‑driven purchase goes wrong, who’s accountable? Merchants point to payment providers (35%), AI agent platforms (28%), then themselves (22%). Consumers blame the AI platform first (38%), then themselves (18%), merchants (10%), and payment providers (9%).
This confusion about accountability is why agentic commerce remains stuck in the experimental phase. Nobody wants to own the risk.
Consumer Demands Before Trusting an AI Agent
Before letting an agent buy on their behalf, consumers want concrete safeguards. Checkout.com’s data lists the essentials:
- A spending cap (30%)
- The ability to revoke the authorization instantly (29%)
- Simplified cancelation or returns (28%)
- Required validation above a certain amount (27%)
- Display of available options before finalizing the purchase (26%)
Here’s where merchants are missing the mark. A quarter of consumers say they’d stop using an AI agent if a purchase was difficult to return or dispute, yet only 16% of merchants see returns and customer service as a barrier to their preparedness.
Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Reading time: 5 min
Key Takeaways
- AI agents drive only 3% of online transactions in the US and UK, despite massive industry buzz and heavy investment from tech giants.
- Consumers demand control: Spending caps, instant revocation, and purchase validation are non-negotiable before they’ll delegate buying decisions.
- Merchants are unprepared: 72% believe consumers will adopt agentic commerce faster than they can handle it; clean data and clear liability are the biggest hurdles.
AI Agents Are Still Marginal in Real Transactions
Let’s look at the numbers. Despite the media noise, AI agents play a significant role in product selection and checkout in just 3% of online transactions in the US and UK. That’s from Checkout.com’s second edition of their agentic commerce report, based on 12,000 consumers and 400 payment executives.
I’ve seen this play out before. When a technology gets this much hype, the actual usage is always lower than the anticipation. OpenAI’s agentic capabilities, Mastercard’s agentic commerce infrastructure, Google’s Universal Commerce Protocol for Search AI Mode and Gemini, Visa’s Intelligent Commerce and Trusted Agent Protocol—all are in controlled experiments, not real-world scale.
Big Disconnect: Expectations vs. Merchant Readiness
Here’s what actually happened. 72% of merchants think consumers will adopt agent-driven purchases faster than businesses can handle them. But consumer awareness is uneven—71% of 18-24-year-olds know about shopping agents, while only 19% of those 55+ do.
The trust gap is stark: 64% of younger shoppers are comfortable letting AI finalize a purchase, but 63% of older consumers are uncomfortable. A quarter of consumers say they’ll never delegate any purchase to an AI agent. This won’t be a quick or uniform shift.
US merchants are more bullish than their UK counterparts: 82% of American merchants see agentic commerce as the biggest disruption their industry has faced, versus 58% in the UK, where the fear of disputes and chargebacks is higher.
Merchants and Consumers Disagree on What to Delegate
Merchants expect delegation to start with complex, high-value categories: travel planning (35%), subscriptions and provider switching (33%), financial products (31%). But consumers want to hand over low-risk, everyday purchases first: groceries (41%), household items (31%), fashion (26%), beauty (25%). The logic is simple—you trust an AI to buy groceries before it alters your investment portfolio.
Liability Is the Real Dealbreaker
The report highlights a fundamental question: when an AI‑driven purchase goes wrong, who’s accountable? Merchants point to payment providers (35%), AI agent platforms (28%), then themselves (22%). Consumers blame the AI platform first (38%), then themselves (18%), merchants (10%), and payment providers (9%).
This confusion about accountability is why agentic commerce remains stuck in the experimental phase. Nobody wants to own the risk.
Consumer Demands Before Trusting an AI Agent
Before letting an agent buy on their behalf, consumers want concrete safeguards. Checkout.com’s data lists the essentials:
- A spending cap (30%)
- The ability to revoke the authorization instantly (29%)
- Simplified cancelation or returns (28%)
- Required validation above a certain amount (27%)
- Display of available options before finalizing the purchase (26%)
Here’s where merchants are missing the mark. A quarter of consumers say they’d stop using an AI agent if a purchase was difficult to return or dispute, yet only 16% of merchants see returns and customer service as a barrier to their preparedness.
Brand Loyalty: The French Exception
The study also looks at brand loyalty. Globally, 57% of consumers would let an AI switch to a better offer. But the French? Only 43%—close to the UK (46%) but far behind the UAE (71%), China (68%), and Brazil (65%). If an AI runs your errands in France, it better stick to the usual brands.
Three Priorities for Merchants
So what’s the play? The report advises merchants not to bet on a single protocol—the market is too fragmented. Focus on three foundations:
- Make your catalogue AI-readable: structured product feeds, rich metadata, clear pricing. Protocols like OpenAI’s ACP and Google’s UCP need machine-readable data. The cleanest product sheets will get more recommendations.
- Adapt payment infrastructure: tokenization and secure storage of credentials, but don’t lock into one protocol or provider during this unstable phase.
- Set clear agent policies: explicit permissions, spending controls, category restrictions, dispute management, and consent frameworks.
The data is clear. Agentic commerce is coming, but it’s not here yet. The winners will be the ones who prepare without chasing the hype.
Building websites since before Google existed. I’ve run SEO, growth, and content for startups across California — and I’ve watched every ‘revolutionary’ tactic eventually expire. What doesn’t expire: understanding systems, compounding effort, and thinking slower than everyone else.