Don't pay people to do what they already love.

There’s a move every builder reaches for when engagement dips: add points. A streak, a badge, a leaderboard, a little coin that lands with a satisfying sound. It feels like free motivation — you’re not changing the product, just paying people a bit of attention-currency to keep coming back. Most of the time it works. But there’s a specific case where it quietly backfires, and it’s the case builders least expect: the behaviors your users already loved.

In the early 1970s, psychologists ran a study that should be on the wall of every growth team. They found preschoolers who drew with markers purely because they liked to. Then they split them up. One group was promised a fancy “Good Player” certificate for drawing. One got the same certificate as a surprise, after the fact. One got nothing. A couple of weeks later, when the markers were quietly left out during free play, the children who had been promised the reward drew noticeably less than they used to — less than the kids who were never rewarded at all. The reward hadn’t added motivation. It had replaced it, and when it went away it took some of the original joy with it.

That’s the overjustification effect. The brain is always trying to explain its own behavior, and it prefers the simplest story. If you’re drawing for no reason, the story is “I like drawing.” Introduce a gold star, and a new, louder story becomes available: “I’m drawing to get the star.” Once that story is installed, remove the star and the logic is merciless — no star, no reason. An interest that used to run on its own now needs a coin in the slot.

The deeper frame here is Self-Determination Theory: people have intrinsic motivation when an activity feeds their sense of autonomy (I chose this), competence (I’m getting good at this), and relatedness (this connects me to others). A reward that feels controlling — expected, contingent, “do this and you’ll get that” — chips at autonomy. It reframes a freely chosen act as a transaction someone else is running. A large meta-analysis of decades of these studies landed on the uncomfortable version of the finding: tangible rewards that people expect for doing a task tend to undermine intrinsic motivation for it. Not always, not catastrophically, but reliably enough that “just add points” is not the free lunch it looks like.

So the build decision isn’t “gamification good” or “gamification bad.” It’s a triage question you ask before you add the mechanic: is this behavior already carrying its own motivation? If users journal in your app because journaling helps them think, or post because they genuinely have something to say, be careful — you’re holding a lit match near the thing that already works. Rewards are best spent on behaviors with no intrinsic pull yet: the tedious setup step, the invite you need them to send, the habit that hasn’t formed. There, a nudge can bootstrap something into existence. On the things people love, protect the autonomy instead of buying it.

And if you do reward, the details decide the damage. Unexpected rewards — the surprise certificate — barely hurt, because they can’t reframe a decision that was already made. Rewards that signal competence rather than control (“you’ve gotten faster at this”) feed the very thing intrinsic motivation runs on. The failure mode is the expected, dangling, if-then carrot on a behavior that never needed one.

The uncomfortable takeaway for builders: the strongest engagement you have is the kind you didn’t engineer. When you see users doing something for their own reasons, the instinct to “capture” it with a reward is often the instinct to convert an asset into a liability. Sometimes the most sophisticated growth move is to notice a fire already burning and simply not throw money on it.

The science, to look up: the overjustification effect and the undermining effect; Edward Deci’s 1971 experiments; Lepper, Greene & Nisbett’s 1973 marker study; the 1999 Deci, Koestner & Ryan meta-analysis; and Self-Determination Theory (Deci & Ryan) for the autonomy/competence/relatedness frame.