How is a team's score calculated?
A team's score is based on the average performance of its active members—meaning those who have logged at least one activity (during the challenge, in the case of a challenge).
To ensure the competition remains fair between teams of vastly different sizes, our algorithm continuously analyzes the actual size of participating teams and calculates a benchmark index that naturally balances the playing field.
Why not simply use total points or a raw average?
Both traditional methods have drawbacks:
- Total points mechanically favors large teams, even when a small team performs significantly better per person.
- A raw average favors small teams: 2 or 3 highly motivated colleagues can post a very high score without it reflecting genuine collective engagement.
Our method combines the benefits of both, without their drawbacks.
A ranking system inspired by rating best practices
Comparing teams of vastly different sizes presents a well-known challenge: an average calculated from a very small sample is unreliable, even when impressive. This is the exact same issue faced by online review platforms, and one that systems like IMDb’s weighted movie rankings have resolved for years using a weighted average that accounts for sample size (a statistical method known as a Bayesian average, or regularized average).
We applied this exact principle to our team leaderboards: to keep the competition fair across teams of all sizes, our algorithm continuously analyzes the actual size of participating teams and calculates a benchmark index that naturally balances the playing field.
Result: a fair ranking, regardless of team size
A highly motivated small team is still recognized without artificially dominating through 2–3 isolated high performers. A large team is no longer favored by headcount alone: true average performance is what counts.
A system that rewards growth, not exclusivity
Under the old pure-average model, a top-performing small team had every incentive to stay small: onboarding a new teammate almost always pulled their average down. With the new algorithm, bringing in a colleague who performs well improves the team’s overall score. As a result, leading teams are incentivized to expand rather than remain closed.
Key differentiators
- Proven statistical method (Bayesian average), not an arbitrary in-house formula
- Self-calibrated using real data from the challenge
- Real-time — the leaderboard updates as soon as an activity is synced
- Fair regardless of size, while actively encouraging team growth
A concrete example
In a challenge where the overall average is 800 points per active member:
| Team | Active members | Actual average | Rank by Total points | Rank by Raw average | Final rank (adjusted score) |
|---|---|---|---|---|---|
| High-performing small team | 3 | 1,500 pts | 3rd | 🥇 1st | 🥈 2nd |
| High-performing large team | 80 | 1,200 pts | 🥇 1st | 2nd | 🥇 1st |
| Underperforming large team | 80 | 500 pts | 🥈 2nd | 3rd | 3rd |
Under the old system, the leaderboard varied wildly depending on whether total points or pure averages were used—and neither approach was genuinely fair. The new calculation resolves this balance: the genuinely high-performing large team takes the lead, the highly motivated small team remains competitive without artificially dominating, and the disengaged large team no longer wins simply because of numbers.