Comparing people to peers in their own bracket beats one global leaderboard

What it is

Global leaderboards' two major side effects (demotivating the tail, inducing gaming) share a structural root: a single global sequence puts incomparable participants on one comparison axis and shows only absolute position. The improvements restructure the comparison: grouping lets participants of similar ability, starting point, or activity compete with each other (attainable upward comparison), and relative progress makes the participant's reference point their own past rather than others' present. Both keep the incentive structure of ranking while removing the structural triggers of "never catch up" and "any means for rank."

Why it happens

Grouping works by restoring upward comparison's attainability condition. Social comparison's motivational effect depends on reachability: comparing to adjacent-level peers makes gaps analysable (what did they do differently) and paths visible (possibly overtaken next week), keeping expectancy and instrumentality positive—that is the design logic of segmented leaderboards (percentile bands, within-cohort ranks). Relative progress works by moving the evaluation anchor from others back to oneself: breaking a personal best, comparing to last week's self, rising completion rates—this feedback offers guaranteed instrumentality (one's own effort necessarily changes the number) independent of others' performance. Their combination (progress boards within same-level cohorts) balances competition's social charge with comparison's fairness. Empirical support comes from health-behaviour and education research: feedback framed as progress against one's own past sustains long-term adherence better than comparison with others, and small same-level competition groups raise participation across the whole distribution rather than only at the top.

Where it stops holding

Grouping and relative progress are not superior in every context—competitive contexts' essence is global comparability (esports, sport), where participants come precisely for absolute rank, and segmenting would break the competitive point. Grouping brings new gaming surfaces: once grouping criteria are understood, "choosing one's own cohort" becomes strategic (deliberately underperforming into a weaker group), so grouping needs dynamic adjustment (periodic promotion by performance) and entry verification. Relative progress's motivational ceiling is also lower—self-comparison lacks social drama (no thrill of defeating an opponent) and appeals less to strongly competitive participants; a hybrid (personal progress as default view plus optional cohort boards) covers a wider motivation spectrum. Cohort granularity is a balancing art: too large and it degenerates into a global board (demotivation returns), too small and competition feels thin (few people per cohort); granularity must tune to participation scale and distribution.

Applying it

• Group dynamically by ability or activity level (e.g., percentile bands) with within-group ranks plus between-group promotion and relegation, keeping every participant in a reachable competitive band.
• Make the personal-progress view the default (comparison with one's own past) and cohort boards an optional view that strong competitors enter by choice, so comparison is never imposed on everyone.
• Verification: compare per-percentile participation retention between a global board and the grouping-plus-progress hybrid. Success requires both improved retention in the bottom 50% and maintained participation at the top—averages alone will miss the distributional gain.

Related

• Same group: W11.03.1 Bottom-half participants are discouraged, not motivated · W11.03.2 Rankings induce metric gaming
• Nearby: W9.01 Matchmaking and balance · W11.03 Leaderboards' side effects · U1.05 Self-comparison presentation of personal data
• Search terms: cohort leaderboard · relative progress · percentile ranking · leaderboard alternatives