Groceries Yes, Holidays No
The same shopper may hand over the washing powder and keep the holiday for themselves, and what is at stake in a purchase may sort customers as usefully as the year they were born

Regulatory focus theory, from Tory Higgins, describes a prevention focus in which people concentrate on safety and on avoiding mistakes, and I would expect that focus to strengthen as the cost of a mistake rises. If that is right, one person could be perfectly happy to let a machine reorder laundry detergent and still insist on choosing their own holiday, because a wrong box of detergent is a small annoyance while a wrong holiday costs money, time and leave that cannot easily be repeated. Daniel Kahneman’s account of dual process thinking describes a fast, automatic System 1 and a slower, effortful System 2, and I would expect much routine buying to sit with System 1, so handing those choices over may feel like very little change, whereas a costly choice may call for slower attention. Both ideas are my starting hypotheses and not findings, and a new survey from Adyen is a useful place to look for hints of them, because it asked about specific categories of purchase as well as AI in general.
What shoppers say they would hand over
Adyen’s research, which Censuswide ran among 11,000 nationally representative consumers aged 16 and over around the world and which Adyen announced in a release dated 3 November 2026, asked which purchases people would let an AI assistant make automatically, without seeking approval first. Across all ages, standard weekly groceries (22%) and household essentials such as laundry detergent (22%) came top, followed by regular takeaway orders (20%) and routine pet supplies (16%), while only 13% were comfortable letting an assistant book flights, hotels or holidays and 7.5% would allow automatic purchases of luxury goods. The headline Adyen chose is about age, because it says 72% of twenty-somethings are comfortable letting an assistant buy for them, against 21% of shoppers over 70.
The age gap is large, but the same release shows young shoppers sorting purchases by stakes as well. Among 20 to 29 year olds, 30% would automate groceries, 30% household essentials and 28% takeaway orders, while 19% would let an assistant book their travel and 14% would let it buy designer apparel and homeware. The 72% and the 30% are different measures and the release does not say how they relate, although Adyen’s French release, which reports French respondents, describes its equivalent figure as the share who would let an assistant buy at least one of the listed product types automatically (74% of French 20 to 29 year olds, with 33% choosing groceries). I read the global 72% the same way, but that is my inference and the global release does not state it.
How far to trust the numbers
The survey was published by a payments company, and the release includes a quotation from its Global Head of Agentic Commerce saying that seamless payment infrastructure is key to capturing agentic commerce, so I would treat it as a large commercial survey and not as academic research. It records what people say they would allow, which is different from what they do, and the release gives no fieldwork dates and no country breakdown for the global figures. It also contains an oddity, because the summary says 21% of over-70s are comfortable while the body says nearly 72% of over-70s refuse. Both could be true if ‘refuse’ means would not let AI buy any of the listed items, which is how Adyen’s French release uses the word, but they leave about 7% unexplained and the release does not say, so I would hold the age figures loosely until Adyen publishes full tables.
The release also includes no cross-tabulation of age by category, so the data cannot rank age against stakes, and “stakes matter more than age” should not be treated as settled. National figures can also sit below the global ones: Adyen’s French release, which reports French respondents only, gives 19% for weekly groceries and 12% for flights, hotels or stays among all adults, and among French 20 to 29 year olds 33% would automate groceries against 18% for travel.
Two ways of cutting the same customers
My own view is that age and stakes are not rivals, and that the numbers are compatible with both. Age seems to tell you how likely someone is to say yes to automation at all, while stakes seem to tell you where they say it, and a plan built on age alone would treat a willing twenty-something as willing across the whole basket, even though in the global data only 19% of twenty-somethings would extend the same permission to travel, against 30% for groceries, and the release does not say how far those two groups overlap.
In Dear Algorithm I wrote about one person behaving differently at different moments: “Tuesday Dave and Saturday Dave are the same person. They just have different needs at different moments.” A shopper who automates the weekly shop and then spends a Saturday afternoon choosing a holiday is the same person in two situations, and a profile built only from the first will misread the second. Adyen itself frames the finding in a similar way, saying that consumers are “increasingly open to delegating routine, low-risk purchases, while retaining manual control over higher-value transactions”, and its Global Head of Agentic Commerce, Karan Katyal, adds that “trust must be built from the ground up”. Neither statement tells us why people draw the line where they do, so the prevention-focus and System 1 explanations above remain hypotheses that a researcher could test and that this survey does not.
What to do with it
The STAR Framework describes four lenses, which are different ways of looking at a decision, and they may offer a way to think about why stakes change what people want. The Realist lens foregrounds security, stability and reliability, along with simplicity and ease, so a routine purchase handed to an assistant may look like a reliable convenience while a costly one may call for reassurance first. The Thinker lens foregrounds competence, understanding and considered analysis, so a big purchase may bring the comparison of options back in. The Adventurer lens foregrounds autonomy, exploration and freedom to act, so choosing a holiday may be part of the pleasure and not a chore to delegate, and the Socialiser lens foregrounds connection, inclusion and relationships, so a purchase that involves other people, such as a trip taken together, may matter in a way that a box of washing powder does not. These lens readings are my own interpretation and I have not tested them against the Adyen data, so I offer them as prompts for questions to ask and not as predictions, and they are not labels for shoppers.
For a team planning AI shopping features for the busy season, I would start with four steps. First, segment by category and stakes as well as by age, using your own purchase history to see where individual customers already hand over routine buying, where they keep control, and where the answer differs from what their age group would lead you to expect. Second, offer full automation where the choice is routine and an approval step where a mistake would be costly, because that matches the line that Adyen’s respondents drew between groceries and holidays, with 22% comfortable letting an assistant buy groceries and 13% letting it book flights, hotels or holidays. Third, ask customers directly which purchases they would never delegate, since the survey only tells us about the list that Adyen chose. Fourth, test these figures in your own market before they shape a plan, because they come from one survey of 11,000 people worldwide with no country breakdown, and Adyen’s French release reports lower figures for French respondents. None of this requires abandoning age as a variable, only adding the question of what the customer is being asked to hand over, which is easy to put into a brief and costs little to test.
If you are curious about your own decision-making style, the STAR Cognitive Profile Assessment (33 questions) is at starassess.com.
David Chadderton is a behavioural strategist, Chief Marketing Officer and creator of the STAR Framework. He writes The Unoptimised Human.
References
- Adyen (2026) “Adyen Research Finds 72% of Young Shoppers Trust AI to Buy on Their Behalf, but Big-Ticket Purchases Lag”, 3 November. https://www.adyen.com/press-and-media/adyen-research-young-shoppers-trust-ai-to-shop-on-behalf
- Adyen (2026) “Commerce agentique : les trois quarts des jeunes acheteurs français sont prêts à déléguer leurs achats du quotidien à l’IA”, 12 November. https://www.adyen.com/fr_FR/presse-et-medias/adyen-etude-les-jeunes-consommateurs-font-confiance-a-l-ia-pour-faire-leurs-achats
- Chadderton, D. (2026) Dear Algorithm, It’s Not Me, It’s You: Putting the Personal Back Into Personalisation. 2nd edn. 16 May 2026.
- Higgins, E. T. (1997) “Beyond pleasure and pain”, American Psychologist, 52(12), pp. 1280 to 1300. https://doi.org/10.1037/0003-066X.52.12.1280
- Kahneman, D. (2011) Thinking, Fast and Slow. New York: Farrar, Straus and Giroux.
Eight minutes. Thirty-three questions.
A precise picture of how you operate: motivations, decision style, and blind spots. Free to start.
- S
- T
- A
- R
The canonical source.
Essays, when there is something worth sending. Choose the subjects you want. I will ask you to confirm the address before I write.