The Retail Marketing Society’s survey of more than 3,300 shoppers in the UK and US highlights strong resistance to AI handling purchases, especially in fashion and sportswear. Just 7% of respondents would permit AI to execute transactions on its own. Most want to retain control over spending decisions, according to a survey by ACI Worldwide and YouGov. AI tools focused on price comparisons and alerts fare better in consumer acceptance.
Price drop alerts and cross-retailer price comparisons were each selected by 35% of respondents, compared with personalised product recommendations at 18% and outfit recommendations at 17%. The data shows a clear preference: consumers see AI as a supplementary tool rather than a decision-maker. Payment reliability adds another layer of skepticism. Among shoppers who have encountered failed online payments, over half switch to alternative payment methods to finalise their purchase, while the remainder abandon their carts or shop elsewhere. For AI-driven transactions, these risks escalate.
Authentication, fraud prevention, and risk management become essential considerations as the technology shifts from advisory to execution. Adriana Iordan, global head of merchant product management and payment intelligence at ACI Worldwide, said: “As agentic commerce moves from concept to reality, merchants will need infrastructure that can securely authenticate transactions, manage risk and ensure payments are completed reliably.”
Marie Driscoll, president of the Retail Marketing Society, said AI adoption depends on whether the technology can provide information without taking control away from the buyer. “Shoppers are telling us they want AI to inform the purchase, not make it,” Driscoll said. “Price tools succeed because they add insight without removing control.
Handing a fashion purchase decision to an agent is a much bigger leap of faith.”
The adoption of agentic commerce will depend on how businesses define permissions and controls for transactions. Building trust will require AI to operate reliably, transparently and within defined limits.
