A common assumption about agentic commerce is that consumers will eventually hand purchasing decisions to AI and step back – that AI agents will act as fully autonomous personal shoppers that will need no oversight from their users. Tell an agent ‘buy me a nice jacket’ and a jacket in your size and style that’s within your budget will arrive the next day. New research suggests that is not what is happening.
Across four major Western markets, consumers are using AI assistants as instruments of comparison rather than substitutes for their own judgement and the trust signals that have governed retail for decades still decide which seller wins the transaction. This signals how agentic shopping will evolve in the future – a future that major payments companies, AI companies and merchants are already heavily bought into.
PSE Consulting recently surveyed 4,250 consumers in the UK, US, France and Germany who already use AI for online shopping. The headline finding is that 32% cite price as their primary decision driver when choosing between AI recommendations, more than double the 14% who simply accept the assistant’s top pick. US consumers are the most price-driven, with 37% making price their first filter. Adoption is also stronger than many had assumed: 43% report mostly positive experiences with AI shopping so far, while fewer than 3% report mostly negative ones.
That uptake matters in the context of McKinsey’s forecast that agentic commerce could orchestrate between $3 trillion and $5 trillion of global consumer transactions by 2030. The technology is no longer hypothetical. The question is how consumers behave once they are inside it.
AI is functioning as a price comparison engine
The first surprise in the data is how little consumers want to delegate. Only 14% follow the AI’s top recommendation without further filtering. The other 86% want to interrogate the shortlist, and price is the dominant lens for that interrogation. AI shopping is operating as a price comparison utility, not a delegated buyer.
The contrast with voice commerce is instructive. A decade of investment in voice-led shopping produced narrow adoption and persistent abandonment, with most smart speaker owners never making a purchase through their device. AI shopping has reached mainstream usage in months, with consumers actively returning to it. The reason is structural. Voice removed the visual price comparison that drives most online shopping decisions. AI assistants preserve it, often improving on it by surfacing options across retailers that a human searcher would never reach in the same time.
For merchants, the implication is that price competitiveness is a precondition for AI visibility, not an optional tactic on top of it. A product that does not appear in the AI’s shortlist on price grounds is invisible regardless of brand strength. But appearing in the shortlist is only the start of the contest.
Brand recognition and reviews close the sale
Once consumers see the AI’s shortlist, the signals that have governed digital retail for two decades reassert themselves. The PSE research finds 89% of respondents say recognising the seller’s brand is important or very important when acting on an AI recommendation. 92% say customer reviews matter. 68% would consider an unfamiliar seller, but only after checking reviews and ratings first.
These figures align with broader trust research. The 2025 Edelman Trust Barometer Special Report on Brand Trust positions trust as a third axis of brand competition alongside price and quality, with consumers narrowing their consideration set to brands that feel familiar and reliable. Harvard Business Review research reports that reviews remain one of the most persuasive inputs in purchase decisions, with their weight increasing for higher-stakes items. A peer-reviewed study published in Humanities and Social Sciences Communications found that consumers consistently prefer products with high star ratings and large review volumes from trusted platforms, even when cheaper alternatives are available.
The behavioural pattern is straightforward. AI compresses discovery and shortlisting. Trust signals then decide which item in the shortlist gets the click. A merchant brand that is unknown, has thin review coverage or visible negative feedback loses the transaction even when it wins on price.
Market differences sharpen the picture. UK consumers are the most seller brand-reliant of the four markets surveyed, with 15% citing brand recognition as their primary decision factor, almost double the 8% rate in the US, France and Germany. UK consumers also show the strongest attachment to existing loyalty ecosystems: 36% say losing loyalty points would make them less likely to continue using AI shopping tools, against just 14% in Germany. France, by contrast, emerged as the most AI-trusting market, with 53% reporting mostly positive experiences and the most evenly balanced spread of decision factors.
What merchants and platforms should do now
The implications for merchants are immediate. Structured product data and consistent pricing matter more than ever, because they are what AI agents read. Catalogues with patchy specifications, inconsistent taxonomies or unverified inventory will be filtered out before any human sees them. McKinsey’s work on European agentic commerce frames this as a move from search optimisation to agent optimisation.
Review density and quality become decisive in parallel. A seller with a thin or stale review profile cannot survive the consumer’s verification step, however attractive the AI’s recommendation. Active reputation management, rather than passive collection, becomes a baseline capability.
Loyalty integration cannot be an afterthought. The UK data shows that loyalty rewards are a meaningful brake on AI adoption when they sit outside the agent’s purchase flow. Merchants and payment providers who solve loyalty redemption inside agentic transactions stand to capture the most engaged segment of the market. Retail Dive’s analysis of related ICSC-McKinsey research underlines how quickly this revenue is moving from theoretical to budgeted, with 68% of consumers having already used at least one AI tool in their shopping in the past three months.
Platforms building agentic commerce infrastructure also need to expose brand and review signals prominently, rather than hiding them behind a single confident recommendation. The research is clear that consumers do not trust the recommendation alone. They want to verify it.
The settled assumption that changed
A year ago, the dominant assumption about agentic commerce was that consumers would gradually hand purchasing decisions to AI. The PSE data suggests something more interesting. Consumers are happy to hand over discovery and price comparison, which are tedious. They are not handing over the final decision, which carries financial and emotional weight. The signals that have always closed a sale, recognised brand and verified reviews, still close it.
For merchants, that is reassuring and demanding in equal measure. Brands that invested in reputation and review depth have a defensible position in the agentic era. Brands that did not are about to discover how exposed they are when AI removes the friction that previously hid their weaknesses.
