Reward Outcomes, Not Output: Recognition Systems That Quietly Filter Community Noise

Most community leaderboards measure what is easy to count, so members optimize for countability. In one brand community I audited, the top ten contributors by post volume were also the ten most muted by other members. The single most-cited practitioner in that space ranked 74th — invisible, because the scoreboard only saw frequency, not effect.

The fix is to separate activity from contribution and give them different status. Activity — streaks, event attendance, upvotes given — stays visible but low-weight. Contribution is verified: an accepted solution, a member confirming they were helped, a resource reused by peers. Impact is peer-cited: someone references your answer in a later thread or links it in their own work. Build the leaderboard on the second and third tiers only, require the asker to mark the solve, and apply a quarterly decay so last year's glory does not carry this year's authority. Volume stops paying.

Weights are a design choice, not a neutral default, and they will be gamed the moment they are legible. So publish the rubric, rotate one metric per quarter, and watch who rises. If the newly recognized names are the ones people already quietly ask for advice, the system is working.

Which signal in your community is hardest to fake — and are you actually measuring it, or just counting posts?