Zip AI shopping research landed on July 21, 2026, when Zip Co Limited published survey findings from 1,755 US Zip shoppers on how artificial intelligence fits into the path to purchase. The headline finding is a trust gap. Shoppers happily delegate the legwork. They are far more hesitant to delegate the wallet.

What the Zip AI Shopping Survey Found

More than two-thirds of respondents, 68 percent, reported using an AI tool within the past three months, and 65 percent of those engage at least weekly. Among AI shoppers, 79 percent used it for research and learning while 66 percent used it to compare products, brands, or services. Some 77 percent say AI carries some or a lot of influence over purchase decisions.

Then the Zip AI shopping numbers turn. Nearly four in five, again 79 percent, say they want full control over final purchase decisions, and only 51 percent would be comfortable with AI autonomously completing a purchase on their behalf. Secondary findings fill in the picture: 48 percent say AI helped them decide faster, and 35 percent say it surfaced products or brands they would not have considered.

The Zip AI Shopping Number That Does Not Match the Headline

Read the bullet list carefully and a problem appears. That 68 percent figure covers using an AI tool, full stop. It is not shopping-specific. A separate finding in the same release states that one in four shoppers, 24 percent, used AI to help with a shopping or purchase decision in the past three months.

Those are very different claims, and the release headline picks the flattering one. Two-thirds of shoppers turning to AI before purchase decisions is the framing. The Zip AI shopping figure that measures shopping behavior is 24 percent. The 79 and 66 percent breakdowns then describe a subgroup labeled AI shoppers, which the release never defines against either base. Anyone citing this data should quote 24 percent, not 68 percent, when the subject is commerce rather than general AI adoption.

Who the Zip AI Shopping Survey Really Asked

The Zip AI shopping sample deserves equal scrutiny. These are 1,755 US Zip shoppers, surveyed online, drawn from a customer base of more than 4.6 million active US users. They are not a general population panel. They are people who already chose a buy-now-pay-later product, a group that skews younger and more digitally native than American shoppers overall.

That composition would push AI adoption figures up rather than down. So Zip AI shopping numbers describe how BNPL customers behave, which is genuinely useful to Zip and to merchants serving that segment. It is a weaker basis for the claim about American shoppers generally that the release opens with.

Zip AI Shopping Research Follows a Product, Not the Reverse

Here is the detail that reframes the whole release. Five weeks earlier, on June 16, 2026, Zip US announced it would support Stripe Shared Payment Tokens, the primitive that lets AI agents initiate transactions on a customer behalf without exposing payment credentials. Zip already built agentic checkout rails.

So this is not a company planting a flag while avoiding engineering commitment. The messaging is consistent across both announcements. Rory Herriman, US chief technology and operating officer, framed the Stripe work around a question of “who is really in control,” and Heck closed the survey saying shoppers want technology that “helps make smarter decisions, but still expect to remain firmly in control.” Zip AI shopping research is evidence gathered in support of a design choice already made, which is a more sophisticated move than thought leadership for its own sake.

The Honest Read on Zip AI Shopping Data

A buy-now-pay-later company publishing proprietary survey data about its own customers, with numbers that happen to validate its own product positioning, warrants the obvious discount. The 24 percent burial and the customer-base sampling are both real weaknesses.

None of that makes the finding worthless. The distance between broad comfort with AI assistance and narrow comfort with AI spending is a genuine design constraint, and it applies to everyone building agentic checkout. Stated preference is also cheap. People routinely say they want control and then take the convenient default, which is why 51 percent should be read as a snapshot of current sentiment rather than a forecast.

The draft-level reading that this is straightforwardly good news for BNPL deserves a caveat too. If checkout stays human, Zip keeps its moment. Yet Zip is plainly not betting on that, having already integrated with the rails that keep installment credit available when an agent does complete the purchase. Hedging both outcomes is the smarter position, and it is the one Zip took.

What to Watch on Zip AI Shopping Next

Watching the 51 percent move in a future survey cycle is reasonable, though survey comfort is a poor predictor of conduct. Stated willingness to let software spend money tends to lag actual behavior, in both directions.

The harder signal sits on the product side. Stripe indicated the Zip integration would go live soon as of June. Watch for the go-live and for any transaction volume flowing through agentic checkout with a BNPL option attached. That is behavior rather than sentiment, and it would test the 51 percent against what people do instead of what they tell a survey. If agentic volume climbs while the comfort number sits flat, the gap between stated caution and revealed preference becomes the actual story.

Sources: the research is published on the Zip newsroom, the Stripe integration was announced via BusinessWire, and trade coverage came from Chain Store Age.

Fintechbits covers consumer payments, agentic commerce, and buy-now-pay-later markets. Nothing here constitutes financial or investment advice. All analysis represents the editorial views of Fintechbits.