From AI-assisted discovery to purchase: Closing the consumer trust gap

The future of retail may depend not on whether consumers use artificial intelligence, but on how far they are willing to let it act on their behalf.

The data suggests shoppers who use AI build larger baskets and make higher-value purchases. A Bayes Business School study of more than 12,000 supermarket trips in Germany found that shoppers using AI-assisted smart carts spent up to one-third more than those who did not.

Adobe’s transaction data reflects the same pattern. During the 2025 holiday season, AI-referred traffic converted 31% more frequently than other sources, and Adobe’s July 2026 reporting showed those visits generated 53% more revenue per visit and converted at a 60% higher rate than visits not directed by AI.

While consumers are increasingly willing to use AI to research products and guide purchase decisions, they remain significantly less comfortable allowing AI to complete transactions on their behalf.

The opportunity for retailers lies in understanding where consumers embrace AI, where they hesitate and how to bridge that gap to unlock growth. The retailers that engage customers through AI will be rewarded.

Where the hesitation shows up

Trust depends on what consumers ask AI to do. Adobe found that 95% of surveyed consumers who turn to AI consider its responses at least as trustworthy as traditional search engine results, a figure that rose to 97% for regular users. Consumer hesitation grows as AI moves from suggesting products to selecting and purchasing them. As consumers move through the buyer journey, they ask: How is my data being used? Whose interests shaped this recommendation? And how much control am I giving up?

Riskified’s quarterly tracker found that while AI has become integral to product discovery, consumers’ comfort with agents making purchases fell from roughly 70% in late 2025 to just 45% by early 2026. Checkout.com reported that 24% of surveyed consumers indicated they will never delegate purchases to AI. In Alchemer’s 2026 retail report, nearly half of surveyed shoppers had used AI to research a purchase, yet they trusted its recommendations less than online reviews or word of mouth.

Why are consumers wary of how these tools use data, generate recommendations and exercise autonomy? The data points to a few key areas prevalent across age groups.

• Privacy: Consumers wonder how the data behind personalized recommendations is used and who has access to it.
• Recommendation quality: Three-quarters of consumers surveyed said brand-sponsored recommendations would reduce their trust in both the AI agent and the paying brand, according to Chain Store Age. Those findings challenge the “retail media inside the agent” model many platforms are relying on for monetization.
• Control: Braze found that only 10% of consumers surveyed are willing to let agents operate fully independently.

While consumers readily use AI to explore their options, usage drops sharply as they move toward completing the transaction.

Winning trust

The gap between AI-assisted research and AI-driven purchasing represents one of retail’s biggest opportunities. Success will belong not to the companies that automate checkout fastest, but to those that earn enough trust for shoppers to let an agent act on their behalf.

Retailers that build transparency, control and accountability into customer experiences will gain a lasting competitive advantage. Organizations should consider doing the following:

• Audit product data before buying technology: AI agents recommend only products they can reliably evaluate. They rely on clean, protocol-compliant data; poor data can prevent products from being considered, regardless of the technology’s sophistication.
• Support the major AI protocols: ACP (Agentic Commerce Protocol), UCP (Universal Commerce Protocol) and MCP (Model Context Protocol) help connect merchants to the leading AI shopping and assistant ecosystems, making products available to AI agents and increasing products’ potential visibility. While protocol support enables agent access, agent recommendation depends on accurate product information, strong content and a compelling customer experience.
• Change what you measure: Sessions and bounce rate say little here. Track whether products appear on priority queries, where they rank and their share of AI-generated recommendations relative to named competitors.
• Focus on providing AI-assisted research before earning AI checkout: Roughly a quarter of shoppers use AI to decide what to buy, while almost none currently plan to purchase through AI agents. Invest in the stage where consumers already use AI.
• Make AI visible and personalization a choice: Privacy is the leading AI concern in every age group, and unlabeled AI reads as a hidden agenda. Personalization a customer has opted into feels helpful, not intrusive.
• Ground recommendations in sources shoppers already trust: Verified reviews and real purchase data provide credibility that helps AI recommendations earn consumer confidence.
• Rethink attribution for the agentic era: As AI agents increasingly complete purchases on behalf of customers, businesses need new ways to identify the channels, experiences and recommendations that created demand in the first place.