The Numbers Behind the Shift
RTB House surveyed more than 1,800 consumers across the United States, United Kingdom, Japan, and France and found that AI tools have moved past influencers, traditional media, and social platforms as a trusted source for shopping decisions. Among U.S. respondents, 59% still trust friends and family most, but 44% now say they trust AI tools, a figure that edges out every other category the survey tracked. Sixty-eight percent of respondents said they had used at least one AI platform for shopping within the past three months, and 62% had specifically used AI to compare prices or brands.
The discovery numbers are just as notable. Fifty-nine percent of U.S. consumers credit AI with introducing them to brands they had not previously known, and 42% said AI tools extend how long they spend deciding on a purchase because the tools surface more options than a typical search or social feed would. That second finding cuts against the assumption that AI shortens the path to purchase. For a meaningful share of shoppers, AI is making the funnel longer and wider before it gets narrower, which has direct implications for how retailers measure the return on AI-driven traffic rather than treating it like a faster version of paid search.
Trust Comes With Conditions
The willingness to let AI actually spend money is real but bounded. Forty-two percent of American millennials said they would allow an AI agent to complete a purchase up to $250 without further approval, as long as a seven-day return policy was in place to backstop the decision. That dollar ceiling and that return-window condition both matter: shoppers are extending trust incrementally, tied to specific guardrails, rather than handing agents a blank check.
Across the full sample, 35% of respondents said they wanted a human review step before an AI agent finalized any transaction, and that share rose to 44% among baby boomers specifically. The generational gap is consistent with what retailers have seen in other AI adoption data: younger shoppers move faster to autonomous tools, older shoppers want a confirmation click before money changes hands. Retailers building agentic checkout flows need both paths available in the same product, not a single default that assumes universal comfort with full autonomy.
Where the Traffic Is Actually Going
Google AI Overviews and ChatGPT each reached 43% of U.S. shoppers surveyed as the AI tools they had used for shopping, putting the two roughly even as the dominant entry points. Claude followed at 23% and Grok at 21%, both meaningful shares for tools that were not built as shopping products first. That distribution tells retailers something concrete about where discovery budget needs to move: general-purpose AI assistants, not shopping-specific apps, are where most consumers are actually starting their research.
Amazon's own data reinforces the scale of the shift inside a single retailer. The company reports that its Alexa for Shopping feature now serves 350 million users, with a fivefold increase in shopping-related interactions over the prior comparison period. That scale puts Alexa for Shopping well past pilot status. It is evidence that one of the largest retailers in the world has already moved conversational AI shopping from an experiment to a mainstream surface, and competitors without an equivalent product are ceding ground on the exact channel this survey says shoppers increasingly trust for research and comparison.
The End of Social Feed Discovery, According to RTB House
Jaysen Gillespie, vice president at RTB House, framed the finding bluntly, saying the research shows the era of social feed driven e-commerce discovery is coming to an end as consumers shift research and comparison behavior toward AI tools. That is a strong claim from a company with a commercial interest in AI-driven retargeting, and it deserves some skepticism. But the underlying numbers, AI tools now edging out influencers and matching or beating media outlets on trust, support at least a directional version of the argument.
Retail media networks and brand marketing teams built the last five years of strategy around influencer partnerships and social commerce integrations. This survey suggests that budget allocation needs to shift toward structured product data and content that AI systems can parse and recommend accurately, since AI Overviews and chat assistants pull from indexed product information rather than from a curated feed. Brands with messy or thin product data will show up less often in AI-generated recommendations regardless of how strong their influencer relationships remain.
What a Trillion-Dollar Forecast Means for Retail Priorities
McKinsey and the International Council of Shopping Centers project that U.S. agentic commerce, transactions initiated or completed by AI agents rather than direct human browsing, will reach $1 trillion in volume by 2030. That figure is a projection, and projections this size carry real uncertainty, yet it lines up with the adoption curve this survey documents: two thirds of shoppers already used AI shopping tools in the past quarter, and a meaningful share are willing to let those tools transact within defined limits.
For retail technology leaders, the practical response is not a single big bet but a set of near-term investments: clean, structured product data that AI systems can retrieve accurately, checkout flows that support both autonomous and human-confirmed purchase paths, and clear dollar-threshold and return-policy signals that give AI agents the confidence to complete a sale instead of stalling out mid-funnel. The survey's own numbers show shoppers are already comfortable with AI up to a defined limit. Retailers that make those limits explicit and generous, backed by real return policies, are the ones positioned to capture transactions instead of just impressions.
