Service or Surveillance?
Shoppers say they want brands to know them and say they are taking steps to stop it, and the difference may lie in who holds the controls.

Consumers keep telling surveys that they want brands to know them and that they are taking steps to stop brands knowing them, and the difference between service and surveillance may depend less on the data than on who holds the controls.
Two messages arrive on the same evening: the first says that your usual coffee is on offer this week and asks whether you would like it added to Saturday’s delivery, and you tap yes without a second thought. The second opens with “We noticed you’ve been looking at…” and shows you the very pair of running shoes you browsed at lunchtime on another device, and you find yourself wondering, with a faint chill, who exactly “we” is and what else they have noticed. Both messages may rest on similar data and similar software, yet one feels like a shop assistant who remembers you and the other like a stranger who has been reading your post.
It is tempting to treat that difference as a technical problem, to be solved with better targeting, better timing or a better model. The STAR Framework starts somewhere else, with motivation, and in particular with the needs that Self-Determination Theory describes. My interpretation, which none of the research below tested, is that personalisation which cedes control to the customer tends to feel like service, while personalisation which removes control tends to feel like surveillance, and that the fix is therefore motivational before it is technical.
Wanting it and guarding against it
The surveys of 2026 keep finding both halves of the paradox, and Amperity’s 2026 State of Personalization in Retail reports, from a survey of 1,000 US consumers, that 83% want personalised shopping experiences (its press release describes the same figure as wanting retailers to remember them) and that 74% are more likely to purchase when they receive “truly personalized offers”, while 57% say their experiences still feel generic; beyond the sample size, its methodology sits behind a download form, so treat the figures as a vendor’s headline. Attentive’s 2026 survey of 1,050 shoppers in the US, UK and Australia, run online in February among people already signed up to at least one brand’s email or text marketing, found that 71% are taking action to protect their privacy, up from 64% the year before, by opting out of cookies, shopping in incognito mode or using separate email addresses. The same survey found that 69% of those privacy-conscious shoppers still want brands to learn from their shopping habits over time, much the same as the 66% of shoppers who are not privacy-conscious. An earlier Incogni survey of 1,006 American adults, conducted by Cint in autumn 2025, found that 95% were worried about their data ending up in a breach and that 78% were nonetheless willing to trade some of it for a shopping benefit.
All three are commercial surveys of self-reported attitudes and behaviour, and the companies behind them sell personalisation or privacy services, so the exact percentages deserve some caution. The pattern they share is the interesting part, and it is consistent: people want to be recognised and are wary of being watched, often in the same breath.
The industry’s answer is more
Meanwhile, the industry is scaling up, and in April Reuters reported that Tesco had partnered with Adobe to deepen its use of AI in analysing customer data, aiming to boost sales through more personalised marketing. Tesco’s announcement describes more than 24 million Clubcard households and plans to use Adobe AI and agentic AI capabilities “to better interpret and anticipate customers’ needs”, and says its teams will use customer intelligence “responsibly”. Becky Brock, Tesco Group Customer Digital Transformation Director, put the aim neatly: “we want customers to feel that the more they use their Clubcard, the more use it is to them.”
At the start-up end, citybiz reported on 1 October that the New York company OuterSignal had raised a $22 million Series A, announced the previous day and co-led by Long Journey Ventures and Abstract Ventures. OuterSignal connects to a brand’s commerce or CRM systems and combines them with publicly available information to build individual customer profiles, and its own announcement lists age, occupation, education, interests, family, hobbies and social presence among the details a brand can see. Both companies describe their aims in terms of relevance and service, and nothing here suggests either has handled data improperly; the point is simply that, in these two cases at least, investment is flowing towards knowing customers better at the moment customers report guarding themselves more.
Not quite a paradox
Researchers have studied the gap between stated concern and behaviour for years under the name “privacy paradox”. Spyros Kokolakis’s review of the research found the evidence conflicting, with results depending on method, context and definition, and recommended that future studies use evidence of actual behaviour rather than self-reports, which is a useful caution for the surveys above. Barth and de Jong’s systematic review of 32 papers, for its part, concluded that decisions to disclose tend to rest either on a weighing of risks and benefits or on a judgement that the risk is negligible. STAR’s Distortion Layer, its name for the cognitive bias pillar, is where I would look at that second route, since a risk judged negligible at the moment of clicking may look rather larger once the shoes reappear on another screen. My reading is that the paradox is partly a trade that people make on terms they do not fully see.
Two studies come closer to the question of control, and both measured behaviour or intentions under controlled conditions rather than asking people how they felt about privacy in general. Catherine Tucker analysed a randomised field experiment comparing personalised with generic ad text on a social networking site that, midway through the test, gave users more control over their personally identifiable information without changing how advertisers used the data, and after that change users were nearly twice as likely to click on personalised ads; the change in control was a site-wide policy, not one of the randomised conditions. Aguirre and colleagues ran three experiments showing that when firms collected information overtly, more personalised ads produced greater click-through intentions, whereas covert collection produced feelings of vulnerability that worked against them, an effect that trust-building cues could offset. Neither study tested STAR or measured psychological needs, but both fit the reading that how much say people have, and how openly they are asked, changes how the same personalisation lands.
Autonomy, and the urge to push back
Self-Determination Theory, as Richard Ryan and Edward Deci set it out, describes three basic psychological needs: autonomy, competence and relatedness. STAR adds a fourth category, security, as its own extension, so it should not be read as part of SDT. A separate theory, Jack Brehm’s theory of psychological reactance, describes the motivation to restore a freedom that feels threatened. My reading, again untested, is that personalisation which anticipates every move can narrow the space in which a person feels they are choosing, which touches autonomy, and that the urge to opt out, browse privately or give a separate email address looks a little like reactance in action.
Three of STAR’s four Mindsets work here as lenses on the design question, with no customer assigned to any of them. The Adventurer lens foregrounds autonomy, exploration and freedom to act, and it reacts against being too well known, asking whether the customer still feels free to surprise the brand. The Thinker lens foregrounds competence, understanding and considered analysis, and it tends to welcome relevance when the logic is legible, asking whether the customer could explain why this message arrived. The Realist lens foregrounds security, stability and reliability, along with simplicity and ease, and it asks whether the data feels safe and the controls easy to use. A coffee reminder passes all three questions comfortably, while the shoes from another device fail at least two of them, and in my reading more data would not have rescued the second message, since what it lacked was any sign that the customer had a say in it. None of these lenses was tested by the research above; they are ways of reading it.
A toolkit for personalisation that feels like service
- Show your working. A short “because you bought…” line turns an inference into an explanation the customer can check and correct.
- Ask before you infer. Where you can, let people tell you their preferences directly; in Attentive’s survey, the most common thing that made marketing feel personalised, chosen by 48%, was using preferences shoppers had shared.
- Hand over the dial. Make controls visible and easy to adjust, since Tucker’s field experiment suggests a greater sense of control can make personalised ads more likely to be clicked rather than less.
- Collect in the open. Tell people what you gather and why at the moment you gather it, because covert collection is where Aguirre and colleagues found vulnerability.
- Measure the flinch as well as the click. Track opt-outs, unsubscribes and complaints alongside conversion, because the cost of surveillance-flavoured personalisation may show up there first.
The test question is one any team can ask of a message before it goes out: could the customer explain why they got this, and switch it off if they wanted to?
The coffee reminder and the running shoes may come from the same data and the same software, and the difference between them is that one left the customer holding the controls while the other made it plain that someone else had them. Knowing a customer well was never the problem, since good shopkeepers have always done it; the trouble starts when the customer can no longer see what is known, or say no.
If you would like to see which Mindsets tend to lead for you, the STAR Cognitive Profile Assessment at starassess.com takes 33 questions, and the assessment and its full report are both free, which may be a useful starting point for noticing when being known feels like service to you and when it starts to feel like being watched.
David Chadderton is a behavioural strategist, Chief Marketing Officer and STAR Dynamics. He writes The Unoptimised Human.
References
Aguirre, E., Mahr, D., Grewal, D., de Ruyter, K., & Wetzels, M. (2015). Unraveling the personalization paradox: The effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 91(1), 34–49.
Amperity. (2026, 27 January). 2026 State of Personalization in Retail. Amperity.
Barth, S., & de Jong, M. D. T. (2017). The privacy paradox: Investigating discrepancies between expressed privacy concerns and actual online behavior. A systematic literature review. Telematics and Informatics, 34(7), 1038–1058.
Brehm, J. W. (1966). A Theory of Psychological Reactance. Academic Press.
citybiz. (2026, 1 October). OuterSignal raises $22M Series A to expand AI-powered personalization marketing. citybiz.
Haynes, A. (2026, 30 September). OuterSignal raises $22M so every customer is Somebody. OuterSignal.
Incogni. (2025, 10 November). How much do Americans value their personal data? An analysis of the online shopping habits of US consumers. Incogni.
Kokolakis, S. (2017). Privacy attitudes and privacy behaviour: A review of current research on the privacy paradox phenomenon. Computers & Security, 64, 122–134.
Reuters. (2026, 13 April). Tesco partners with Adobe to ramp up AI-driven personalised marketing. Reuters.
Rollins, A. (2026, 11 September). 2026 personalization trends: What 1,000+ shoppers expect from brands. Attentive.
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68–78.
Tesco PLC. (2026, 13 April). Adobe & Tesco enter strategic AI partnership to personalise experiences and reward loyalty for Tesco’s customers. Tesco PLC.
Tucker, C. E. (2014). Social networks, personalized advertising, and privacy controls. Journal of Marketing Research, 51(5), 546–562.
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.