Skill Badges Raise Retry Rates 26% Over Leaderboards in 600 Puzzle Runs

What separates a player who tries a puzzle once from one who returns to it eleven times? Designers often reach for leaderboards first — visible rankings, public comparison, the promise of status. Yet a growing body of behavioral evidence suggests that the more private mechanism is the more durable one. In a 600-run puzzle study, participants who received skill badges retried 26% more often than those shown leaderboards, a gap that raises a pointed question: why does a small, personal signal of competence outperform the social drama of a ranking?

The Two Feedback Loops, Side by Side

Leaderboards and skill badges both deliver feedback, but they operate on different psychological machinery.

A leaderboard is a social comparison device. It tells you where you stand relative to others. Its motivational engine is status anxiety — the gap between your rank and the rank above you. That engine is powerful but brittle. It produces reliable effort among the small cohort who believe they can climb, and quiet disengagement among everyone else. Research on social comparison, going back to Leon Festinger's 1954 work, consistently finds that comparison with similar others motivates, while comparison with vastly superior others demotivates. A leaderboard makes both comparisons unavoidable.

A skill badge is a competence signal. It tells you something about your own mastery, independent of anyone else's performance. It answers a different question: not "am I beating them?" but "am I getting better at this?" The distinction matters because it changes what a failed attempt means. On a leaderboard, a failure is a rank drop — a loss. On a badge system, a failure is information about the distance remaining to the next marker.

The 26% retry gap in the puzzle study is best read as the difference between these two framings. The badge cohort wasn't rewarded more lavishly; it was given a progress narrative that survived failure.

Why the Ranking Frame Backfires

Kahneman and Tversky's loss aversion work established that losses loom larger than equivalent gains. A leaderboard converts every session into a potential loss: you can only fall. Even a strong performer experiences the ranking as something to defend rather than something to pursue. That's an exhausting posture, and it's one reason leaderboard engagement tends to spike early and decay fast.

There's also the matter of what a leaderboard makes salient. It directs attention to the scoreboard rather than the task. In puzzle terms, that means attention shifts from the mechanics of the problem to the arithmetic of the ranking — a shift that tends to hurt performance on anything requiring sustained reasoning.

Variable-Ratio Reinforcement and the Badge's Quiet Advantage

Behavioral psychology's most famous schedule — variable-ratio reinforcement, the intermittent reward pattern that produces high and persistent response rates — is often invoked to explain compulsion. But it's worth separating two things it can reinforce.

A leaderboard's rewards are intermittent in the worst way: whether you climb depends on who else shows up, how they perform, and where the thresholds fall. The reinforcement is variable but largely outside the player's control. That unpredictability sustains engagement through suspense, not through a sense of agency.

A badge system can be built on a schedule that is variable in difficulty but stable in meaning. You don't know exactly when the badge will trigger, but you know it tracks something you did. The reward is unpredictable in timing and predictable in relevance. That combination is far more sustainable across repeated sessions, because it reinforces the behavior you actually want — problem-solving — rather than the behavior of checking a ranking.

The Competence Loop

Deci and Ryan's self-determination theory identifies competence as one of three basic psychological needs, alongside autonomy and relatedness. Badges speak directly to it. Leaderboards speak to relatedness — but a thin, competitive slice of it, where the other person is mostly an obstacle.

This is the underappreciated finding in the puzzle data. The badge cohort wasn't more motivated in the abstract; it was motivated by a need that gets satisfied by the act of playing itself, not by the outcome of a comparison. That's a loop that can run indefinitely. A leaderboard loop requires an audience and a supply of rivals.

What the 26% Actually Measures

Retry rate is a deceptively simple metric. It captures the moment after failure — the hinge on which all practice-based learning turns. A player who retries is choosing to re-enter a state of not-yet-knowing. A player who leaves has decided the cost of that state exceeds its value.

The 26% gap suggests badges make that state cheaper to inhabit. Not because the puzzle got easier, but because the failure got reframed. A badge system implicitly says: the attempt counted, even though it failed. A leaderboard implicitly says: the attempt cost you a position.

This is consistent with Carol Dweck's work on growth mindsets, though the mechanism here is design-level rather than dispositional. You don't need to change a player's beliefs about intelligence; you can change what the interface tells them a failure means.

A Concrete Illustration

Consider a common puzzle format: a spatial reasoning task with escalating difficulty. In the leaderboard condition, a player who fails level seven sees a rank slip from 412th to 388th. The feedback is about their standing. In the badge condition, the same player sees a progress indicator: "Pattern recognition — 4 of 6 criteria met." The feedback is about their skill. The 26% retry gap is, in effect, the aggregate of hundreds of players choosing to try again because the second message left them with something to finish.

Designing for the Second Attempt

The practical implication isn't to abolish leaderboards. Rankings serve real social functions — they create conversation, they signal expertise, they give the top cohort a reason to stay. The implication is about sequencing and emphasis.

Put the competence signal first. Let the badge be the primary feedback, the leaderboard the secondary layer for those who opt into it. This preserves the social dimension without making status the default frame for every attempt.

Three design questions worth asking of any feedback system:

• What does this signal say about the player's own trajectory? If it can't answer that, it's not a competence signal.
• What does a failure look like through this lens? If failure is primarily a loss of position, the system is training avoidance.
• Does the loop require other people to function? Loops that run on the player's own progress scale better and decay slower.

The forward-looking version of this argument is that feedback design is becoming the main lever in practice-heavy learning environments — puzzle games, yes, but also training simulations, language apps, and any system where the goal is to get someone to try again after they've failed. The 26% figure is a reminder that the smallest design choice in that moment — a rank or a badge — may matter more than the difficulty curve itself. The next round of studies should test whether the effect compounds over weeks, and whether badges designed around specific sub-skills outperform generic achievement markers. That's the question worth running 600 more times.