The Motivation Inversion: When Personalisation Kills the Thing It's Trying to Serve

There’s a moment, and you’ve probably had it, where an algorithm serves you something so precisely targeted that it feels less like a recommendation and more like a diagnosis. Not creepy in the data-privacy sense. Creepy in the “you’ve reduced me to a pattern” sense. You were browsing, not confessing. And yet here is a system that has studied your clicks, mapped your preferences, and arrived at a conclusion about who you are that feels simultaneously accurate and completely wrong.

That feeling is the subject of this article.

According to Gartner’s 2026 Consumer Digital Behaviour Survey, 41% of consumers now describe personalised recommendations as “annoying.” In 2024, that number was 27%. That’s not a marginal shift in sentiment. That’s a tipping point. And like most tipping points, the interesting question isn’t what changed. It’s why it took so long.

The Comfortable Explanation

The standard reading of this stat is reassuringly simple. Privacy fatigue. Bad targeting. Too much data, not enough tact. People are tired of being tracked, tired of being profiled, tired of ads that follow them around the internet like a dog that’s learned the word “walk.”

All of that is true. But it’s the surface truth, the kind that lets marketers nod sympathetically while changing nothing. The deeper issue is motivational, and it sits underneath the privacy complaint where most people aren’t looking.

The promise of personalisation was always, whether anyone used the language or not, a psychological promise. It was supposed to serve three fundamental human needs. The need to feel competent, to discover things you didn’t know you’d love, to have your world expanded by someone who understood your taste. The need to feel connected, to be seen and known by a system that paid attention. And the need to feel autonomous, to have your time respected so you could choose what actually mattered.

That was the pitch. Here’s what happened instead.

The Narrowing

Personalisation systems learn. That’s the point. They observe your behaviour, build a model of your preferences, and optimise for prediction accuracy. The longer they run, the better they get at guessing what you’ll click on. And the better they get at guessing, the smaller the suggestion pool becomes.

What started as “here are ten things you might like” quietly becomes “here is the one thing the algorithm has decided you are.”

This is where the inversion happens. The three needs that personalisation was supposed to serve begin to degrade, one by one.

Competence goes first. If the system always knows what you want, there’s nothing to discover. The puzzle is solved before you’ve touched it. You stop feeling like someone with sophisticated, evolving taste and start feeling like a data point with a feedback loop. The recommendation engine was supposed to expand your world. Instead, it confirmed a version of you that existed six months ago and has been running on autopilot ever since.

Relatedness follows. Being known is a powerful thing. Being reduced is not. When a system says “we understand you,” the first hundred times it feels like recognition. The thousandth time it feels like surveillance. The line between “seen” and “classified” is thinner than most marketers think, and consumers have been crossing it in growing numbers.

Autonomy is the last to fall, and the hardest to recover. Narrowed choice doesn’t feel efficient to everyone. For some people, it feels like a cage.

The Canary

Not everyone experiences this at the same speed. Humans aren’t uniform in what they need from a system, and the ones who feel the narrowing first are the ones whose primary psychological need is freedom.

These are the people who walk into a shop and want to browse, not be led. Who open Netflix and scroll for twenty minutes, not because they can’t find something, but because the act of choosing is the point. Who, when presented with a perfectly optimised recommendation, feel not gratitude but claustrophobia.

For them, the algorithm’s precision isn’t a service. It’s a constraint. And their response isn’t to complain about it on social media. It’s to leave. Quietly, without explanation, for platforms and products and experiences that still feel open.

They are the canary in the personalisation coal mine, and their departure is a signal that the industry is misreading as churn.

Others tolerate it longer. The analytically minded initially appreciate the precision, the way a good data set rewards careful attention. But over time, the absence of challenge becomes its own kind of dissatisfaction. If the system always gets it right, what’s left to figure out? They don’t leave in protest. They disengage in boredom, which is worse, because it doesn’t show up in any metric until it’s too late.

The more practically oriented, the ones who value both freedom and reliability, take a different path entirely. They don’t disengage. They build.

The Workaround

When personalisation overreaches, a particular kind of person starts sabotaging it. Not out of malice. Out of need.

They open incognito windows. They create second accounts. They deliberately click on things they have no interest in, not because they want those things, but because they want to reset the algorithm. To pollute the signal. To feel, however briefly, like the system doesn’t know them anymore.

These aren’t edge cases. They’re an emerging behavioural pattern that the industry is misreading as privacy-consciousness when it’s actually something more fundamental. It’s autonomy-seeking. The same need that makes someone leave a party where they’ve been cornered by a bore, except the party is the entire internet and the bore is an algorithm that won’t stop finishing their sentences.

The workaround is the signal. When your most engaged users are actively sabotaging your recommendation engine to feel free, the engine is the problem. And the fact that they’re willing to invest effort in circumventing it tells you exactly how much the narrowing costs them.

The Filter Failure

There’s a framework that explains why personalisation systems get this so wrong, and it has to do with how they communicate.

Any message, any recommendation, any piece of content, resonates differently depending on what the recipient values. Some people respond to evidence and logic. Some respond to novelty and possibility. Some respond to connection and belonging. Some respond to reliability and precedent.

Effective communication addresses all four. Personalisation systems almost never do.

They’re built to nail the first one. Here’s the data, here’s the evidence, here’s why this recommendation makes logical sense based on your history. That works for the analytically minded, at least for a while.

They sometimes attempt the fourth. Here’s what people like you tend to buy. Here’s what’s proven and popular. Here’s the safe choice. That works for the stability-oriented, at least for a while.

But they almost never address the second or third. They rarely say: here’s something completely outside your pattern that might surprise you. They almost never say: here’s what people you admire are exploring right now. The first would require the system to intentionally introduce uncertainty. The second would require it to understand social dynamics, not just individual behaviour.

The result is an imbalance. Two filters working, two filters silent. And the two that are silent happen to be the ones that serve autonomy and connection, the very needs that are degrading.

This is why the 41% number keeps climbing. It’s not that personalisation is getting worse. It’s that the humans it’s serving are getting more aware of what’s missing.

The Reversal

The answer isn’t to abandon personalisation. That would be like abandoning navigation because the first maps were wrong. The answer is to reverse the direction of travel.

Most personalisation systems are built on a single implicit assumption: that the best recommendation is the one most similar to what you’ve already chosen. That similarity equals relevance. That prediction accuracy is the ultimate metric.

But similarity isn’t relevance. It’s confirmation. And prediction accuracy, pursued to its logical extreme, produces a world where you never encounter anything that challenges, surprises, or changes you.

The alternative is personalisation that optimises for expansion rather than confirmation. Same data, same algorithms, completely different direction.

Instead of “based on your pattern, here’s more of the same,” move toward “based on your pattern, here’s what you haven’t tried yet.” Use the model not as a mirror but as a launch pad. Treat the recommendation engine not as a feedback loop but as a telescope, something that helps you see further, not just more clearly into what you already know.

Practically, this means introducing controlled randomness. Not chaos, but calibrated serendipity. A percentage of recommendations that are deliberately outside the user’s pattern, framed not as mistakes but as explorations. The algorithm saying, in effect: “I know what you like. Now here’s something that might change what you like.”

It means making the algorithm visible. Let users see the logic. Let them adjust the dials. Let them override the system. Autonomy doesn’t require the absence of curation. It requires the option to refuse curation. The difference is everything.

And it means optimising for breadth, not just depth. Instead of “you watched five cooking videos, here’s a sixth,” try “you watched five cooking videos, here’s something from a completely different domain that shares a structural similarity you wouldn’t have found on your own.” The system knows your pattern. The question is whether it uses that knowledge to close doors or open them.

The Real Tension

This article isn’t really about personalisation. It’s about a tension that runs through every system that tries to serve a human being.

Systems optimise for prediction. Humans need surprise.

Systems optimise for efficiency. Humans need challenge.

Systems optimise for relevance. Humans need space.

The Gartner stat isn’t a personalisation problem. It’s an optimisation problem. And as artificial intelligence becomes the primary interface between brands and consumers, between platforms and people, this tension will only intensify. Every improvement in algorithmic accuracy narrows the gap between what the system predicts and what the human experiences, and in doing so, it narrows the space where autonomy lives.

The brands that navigate this well won’t be the ones with the best algorithms. They’ll be the ones that understand something the algorithms can’t: that you cannot optimise for freedom, because the moment you do, it stops being freedom.

The 41% aren’t rejecting personalisation. They’re rejecting the version of themselves that personalisation has decided they are. The question for every brand, every platform, every system that claims to know its users is simple: are you expanding their world, or shrinking it?

If you can’t answer that, the number will keep climbing.

David Chadderton spent his twenties and thirties teaching people how to make life-or-death decisions at forty thousand feet. He now applies the same principles to consumer psychology, which, depending on the brief, can feel equally high-stakes. He’s the creator of the STAR Framework and the author of The STAR Framework: Rewriting the Rules of Consumer Engagement (NYC Big Book Award 2025), The STAR Operating System: Decode Mindset, Understand Motivation, Transform Human Behaviour, and Dear Algorithm, It’s Not Me, It’s You. By day, a Chief Marketing Officer. By night, a behavioural science obsessive who writes The Unoptimised Human because he can’t stop thinking about why people do what they do.