Habit-Forming Design Without Manipulation

Something that took me a while to say out loud: I have designed engagement loops that worked a little too well. Screens tuned to pull people back, streaks and nudges and just-one-more-tap moments. At the time it felt like good product work, because the numbers went up. Retention improved, sessions climbed, and everyone in the review nodded.

The uncomfortable part came later, when I started asking a different question. Not "did this bring people back," but "did people come back because the product helped them, or because I engineered a compulsion they did not choose?" Those are not the same thing, and for a stretch of my career I treated them like they were.

The problem is not habits. It is whose interest the habit serves.

Here is the reframe that changed how I work. A habit is not inherently manipulative. Brushing your teeth is a habit. Checking your bank balance before a big purchase is a habit. Opening a language app every morning is a habit, and it is a good one. The goal of habit-forming design is not the villain here.

The line is about whose interest the loop serves. A habit that helps the user reach a goal they actually have is a service. A habit that serves the company's metrics at the user's expense is a trap. The mechanics can look identical from the outside. A daily streak can build a genuinely useful routine, or it can weaponize your fear of losing progress to keep you logging in past the point of value. Same pattern, opposite ethics.

This is why I lean on real behavioral principles rather than growth-hacking tricks. I wrote the full argument in my piece on why behavioral design is the most valuable UX approach, and the short version is that understanding why people act lets you design with them instead of against them. The same understanding that lets you build a compulsion is what lets you refuse to.

Dark patterns are not an edge case anymore

If you think this is a fringe concern, look at the numbers. In July 2024, the FTC, ICPEN, and GPEN reviewed 642 websites and apps offering subscriptions and found that nearly 76% used at least one possible dark pattern, and nearly 67% used more than one. That is not a few bad actors. That is the default.

And the cost of that default is climbing. In September 2025, Amazon settled an FTC case over its Prime subscription for $2.5 billion. The regulators alleged that signing up took two clicks while canceling ran a four-page, six-click, fifteen-option maze full of retention offers and guilt trips, and that the design produced more than 35 million nonconsensual enrollments over seven years. Somebody designed that cancellation flow. Somebody sat in a room and made canceling deliberately harder than joining, and shipped it.

I think about that a lot. Because the designers who built that flow were almost certainly competent, and almost certainly told themselves a story about why it was fine.

The test I actually use

So how do you build something people love coming back to without becoming the cautionary tale? I use three practical checks.

The first is the one Nir Eyal, who literally wrote the book on habit-forming products, calls the regret test. Would your users regret taking the action if they fully understood what you understand about the design? If the honest answer is yes, you are not designing a habit, you are designing a manipulation. A streak that helps someone learn Spanish passes. A streak engineered mainly so they dread quitting fails.

The second check is symmetry. Whatever you make easy to start, make equally easy to stop. Easy to subscribe should mean easy to cancel. Easy to enable notifications should mean easy to silence them. Amazon's $2.5 billion problem was, at its core, a symmetry problem. The moment your on-ramp and your off-ramp look nothing alike, you have your answer.

The third is what I call the daylight test, and it is the simplest. Would you be comfortable explaining this design choice, out loud, to the actual user affected by it? "We put the cancel button here because it is where you would look for it" survives daylight. "We buried it because retention" does not. If a pattern only works because the user does not notice it, that is not clever design. That is the definition of the problem.

Retention you do not have to hide

When I worked on the Monety fintech micro-loans product, engagement mattered enormously, because a lending product only works if people actually manage their repayments. But the goal was never to trap anyone in the app. It was to make the genuinely useful moments, a payment reminder before a due date, a clear view of what you owe, easy to return to. The habit and the user's interest pointed the same direction. That is the whole trick. When your retention strategy and your user's actual goals are aligned, you do not need dark patterns, because the honest version already works.

This is the same principle I keep coming back to across everything I build, whether it is AI features or engagement loops. I made the case in thoughtful AI product design that solves real problems instead of adding gimmicks, and it applies just as cleanly here. And if you want the mechanics underneath all of this, the specific cognitive levers at play, I broke down the psychology principles behind great UX recently. Those levers are neutral. What you do with them is not.

The best habit-forming products are not the ones that trap you. They are the ones you would choose again if you saw exactly how they were built. Design for that, and you never have to hope your users do not read the news.

Would you be comfortable showing your last engagement decision to the person it was designed to affect? That is the only test that really matters.