AI Made Product Research Longer. The Ad Industry Is Selling Into the Gap.

RTB House published a survey on Monday saying that 42% of US shoppers think AI tools make them take longer to reach a final purchase decision, not shorter. Among Gen Z it’s 48%. The study, “Who’s Buying? Consumer Trust in the Age of Agentic AI,” ran through 1,840 respondents in the US, UK, France and Japan during June and July.

Jaysen Gillespie, VP of Product Marketing and Analytics at RTB House, framed it as the end of an era:

“For more than a decade, e-commerce growth was driven by buying attention on social feeds and optimising for instant checkouts. This research proves that era is coming to an end.”

He’s probably right about the direction. It’s also worth noticing who’s telling you. RTB House sells retargeting, and a longer research window is more days of inventory to buy against the same shopper.

That doesn’t make the finding wrong. Other people measured the same thing. Product.ai’s Trust in AI Commerce Report found 86% of US shoppers who used AI for product research checked the AI’s recommendation somewhere else before buying. IAB and Talk Shoppe put the same behavior at 89%. L.E.K. Consulting found 46% of AI users now begin their product research inside ChatGPT, Gemini, Perplexity or Claude, up from 25% in 2024.

Put those together and the mechanism is boring, which is how you know it’s real. The assistant is good at producing options and bad at closing. RTB House’s own numbers say 59% of US shoppers think AI surfaces brands they’d never heard of, and 63% have used it to build a product list. So the shopper arrives at the end of the chat with a longer shortlist than they started with, containing names they can’t evaluate, and then goes and checks. More candidates plus lower trust equals more homework.

The trust scores explain why the checking happens. In RTB House’s US sample, Google AI Overviews and ChatGPT each drew 43% trust for shopping advice, Claude 23%, Grok 21%. Friends and family drew 59%. AI has passed the social feed. It hasn’t passed a person the shopper knows.

Sky Canaves, Principal Analyst at EMARKETER, described what that does to the content side of the job:

“They’re looking for really in-depth product information that’s well-structured. They need a lot more content than what typically meets the eye on a product description page.”

Here’s where the vendor’s read and the operator’s read split. Gillespie’s advice:

“The longer research window gives marketers an opportunity to more deeply engage with timely and relevant offers across the entire shopper set. Our advice would be to take advantage of that extended window.”

More window, more ads. That is one honest answer, and for a company selling ads it’s the only one available. It’s not the cheapest answer for a 20-person team.

The other answer is that a longer consideration window is mostly a content problem wearing a media problem’s clothes. If 86% of shoppers leave the chat to verify, the thing that decides the sale is what they find when they check: the spec table, the comparison page, the third-party review, the pricing page that answers the question without a form. We’ve covered the conversion side of this, where AI-referred visitors convert better than search traffic precisely because they arrive having done the reading. The reading has to have something to land on.

For a B2B SaaS founder-marketer the translation is direct, because B2B never had an instant checkout to lose. A buyer who used to read three vendor sites now reads a chat summary of nine and then goes looking for evidence on the two that survived. Your product feed and your comparison content are now the sales call. The retargeting spend that follows them around the internet for the next three weeks is competing with their own research, which they trust more than either of you.

One caveat worth stating out loud. RTB House measured what people say about their own behavior, on a survey panel, not what they did. Self-reported “AI makes me slower” could just as easily mean “AI made me aware I was deciding.” The direction is corroborated by three other datasets; the size of it isn’t.

The era Gillespie says is ending was never about speed anyway. It was about buying the moment of intent cheaply enough. AI didn’t kill that trade, it moved the moment upstream into a conversation nobody in marketing can bid on, and left everyone a longer, more expensive stretch of road to cover afterward. Budget for the road, not the moment.