UK shoppers cap AI spending at GBP £150 per purchase

MACH Alliance has published research showing that UK shoppers cap autonomous AI spending at about £150 per transaction, underscoring continued caution about letting AI complete purchases without human approval.

The poll of more than 1,000 UK consumers found that 67% now use AI during the search and discovery stage of shopping. Adoption was markedly higher among younger groups, reaching 92% of millennials and 89% of Gen Z shoppers.

That willingness dropped once money was involved. A third of respondents said they were happy for AI to research a purchase but still wanted to retain control over the final product choice and spending decision.

Only 9% said they would be comfortable with AI completing a payment autonomously on their behalf. For more complex or high-consideration purchases, that figure fell to 4%.

On average, consumers said they would trust AI agents to spend £149.12 for them. The figures suggest a clear dividing line between using AI as a shopping assistant and handing it authority to complete a transaction.

Younger consumers appeared more open to that handover. Millennials said they would be comfortable with autonomous AI transactions averaging £235, while baby boomers put the limit at £67 per purchase.

Trust gap

The findings come as retailers test ways to move AI from product discovery to payment. Some merchants are exploring shopping journeys in which recommendations, checkout and payment are more tightly connected.

JD Sports is among the retailers moving towards more immediate AI-led payment experiences. Debenhams has also been testing AI-driven checkout within shopping environments run by Meta.

Even so, the data suggests broad public acceptance has yet to follow those experiments. Consumers appear more comfortable with AI narrowing choices or surfacing relevant products than with letting software make the final spending decision independently.

That distinction matters for retailers weighing how far to automate online commerce. Companies may see gains from AI-led search and recommendation tools, but the survey suggests fully autonomous purchasing remains constrained by confidence in accuracy, control and risk.

Jason Cottrell, president of MACH Alliance, linked that hesitation to the reliability of the systems handling shopping tasks from product data to fulfilment.

"Tackling the AI trust gap will depend on brands' ability to move agentic AI from influence to reliable execution across the entire buying journey," Cottrell said.

"As agents take on greater purchasing autonomy, the tech stacks underpinning them must enable them to act accurately and securely across everything from product information and real-time inventory to payments and fulfilment."

Retail response

The survey highlights a practical challenge for merchants and technology suppliers. If consumers are willing to accept AI guidance but not AI authority, retailers may need to design systems with clear approval points, especially for higher-value items.

That caution is likely to matter most in categories where purchases are infrequent, expensive or harder to reverse. The steep fall in acceptance for complex buying decisions suggests trust in autonomous checkout may develop unevenly across sectors rather than through a single shift in behaviour.

The gap between generations also points to a longer-term change in shopping habits. Younger consumers, already more accustomed to AI in digital services, may be more willing to let software take over parts of the transaction process, while older shoppers remain far more conservative.

That creates a mixed picture for retailers. Businesses targeting younger audiences may see stronger uptake for AI-led commerce tools, but a mainstream shift to autonomous payments would still require convincing a broader customer base that those systems can be trusted.

Cottrell said control over how AI systems behave would be central to that effort.

"Trust ultimately comes down to executional accuracy at every stage of the journey. Retailers need the right agentic capabilities and data, but they also need the flexibility to govern how those agents act. That's where open, composable architectures become critical, giving businesses the control and agility they need to adopt agentic commerce with confidence."