Do consumers want to approve AI shopping purchases before they happen?

RTB House, an ad-tech retargeting company, reports from its own research that "the willingness to delegate purchasing power to an AI shopping assistant varies by generation, safeguards, and price thresholds," and that among the safeguards tested, a pre-transaction approval step is the one consumers want most. RTB House says "human oversight before a transaction is placed is the main driver," calling it "the leading preference for 35% of consumers," and that this preference is "particularly important to baby boomers," 44% of whom "would prefer to sign off before an agentic AI makes a purchase decision." Baby boomers also showed a second, related safeguard preference more strongly than other respondents: 35% of them want purchases restricted to trusted retailers, versus 29% of all respondents.

RTB House frames this as a generational split in comfort with transactional autonomy rather than a uniform rejection of agentic buying: the piece notes elsewhere that "some shoppers are even comfortable granting transactional autonomy" and that a "subset of super-trusters" would let AI buy without typical protections. The sign-off and trusted-retailer findings sit at the cautious end of that same spectrum — the segment, skewed toward baby boomers, that wants a human checkpoint and a restricted vendor set before ceding control of a purchase. RTB House's stated reason for publishing the breakdown is commercial: it argues brands "need to be responsive to the new paradigm" and should design for these safeguard preferences to "build trust," rather than assuming uniform comfort with autonomous purchasing across age groups.

This generational sign-off preference lines up with the mechanism What makes people distrust AI agents they delegate to? identifies at the task level: in that controlled study, an irreversible, externally visible action (sending an email) produced the sharpest trust drop and highest demand for approval, regardless of output quality. A completed purchase is exactly this kind of action — irreversible once placed and visible in its consequences — so RTB House's survey finding that 35% of consumers (44% of boomers) want to approve the purchase moment specifically, rather than earlier research and comparison steps, is a market-level echo of that lab-level pattern, with age substituting for task type as the variable that predicts where oversight is demanded. It also extends the trust-as-gate framing in Why does AI trust keep falling even though it matters most? from general AI acceptance to a specific commerce safeguard: trust here isn't just correlated with use, it is being traded for a concrete control mechanism (sign-off) at a measured rate.

RTB House's excerpt gives no sample size, country list, or survey methodology, and as a company that sells retargeting technology to retailers, it has a direct commercial interest in framing brand responsiveness to these preferences as necessary — the recommendations that follow (seek "advertising partners who have the technology to make sense of" a more complex data landscape) point toward its own line of business. The 35%/44%/29%/35% figures describe stated preferences on a survey, not observed behavior when an actual AI agent is about to spend a respondent's money, and the excerpt does not say whether "sign off" means a one-tap confirmation, a detailed review, or something else. Within those limits, the finding supports treating pre-purchase approval as a strong default for agentic commerce UX aimed at older users specifically, rather than a single autonomy setting applied uniformly across a customer base — a design implication other notes in this library reach from lab and workplace studies (What makes people distrust AI agents they delegate to?) that this survey now extends to a market segment.