The leaderboard scores each developer by balancing Quality (AI review scores) and Impact (code complexity), so it rewards engineers who ship robust code, not just those who ship the most lines. Find it under Code Review → Leaderboard once your team has a few reviewed PRs.
Why the leaderboard
Team-relative scoring — both metrics are normalized against your own team’s average, so scores stay meaningful in your contextBalanced assessment — quality and impact contribute equally (50/50), so high-volume output can’t mask low-quality workFair comparison — large refactors count proportionally more than trivial changesQuality incentives — the 1–10 Quality Score turns code review into a positive feedback loopDiagnostic, not just a ranking — it surfaces workload imbalances and quality trends that PR counts alone don’t show
Quality score Each PR receives a Quality Score on a 1–10 scale. The score starts at 10 and is reduced by the number and severity of findings — critical issues carry a higher penalty than low-severity ones. The Team Quality Score shown on the dashboard is the average across all developers.
Contribution Score = (Normalized Quality Score + Normalized Impact Score) / 2
Where:
Normalized Quality Score = Author’s Average Quality Score / Team Quality ScoreNormalized Impact Score = Author’s Total Impact / Team Average Impact
Contribution Score Label Interpretation ≥ 1.5 Excellent Significantly above team average 1.0 – 1.5 Good Above team average 0.8 – 1.0 Average At or near team average < 0.8 Needs Improvement Below team average
A score of exactly 1.0 means the developer sits exactly at the team average.
Impact calculation Each merge request’s impact score measures the complexity of the change:
MR Impact = (files_changed × 6.0) + (lines_added × 0.14) + (lines_deleted × 0.28)
Metric Weight Rationale Files Changed 6.0 Cross-file changes indicate higher complexity Lines Added 0.14 New code requires understanding and integration Lines Deleted 0.28 Deletions often require more careful analysis (2× addition weight) Minimum Impact 1.0 Floor value to prevent division issues
The impact formula is derived from the Oobeya GitWiser Coding Impact Score methodology.
Example calculation Team data: Author Avg Quality Score Total Impact Alice 8.5 450 Bob 7.2 280 Carol 9.0 120
Team averages:
Team Quality Score = (8.5 + 7.2 + 9.0) / 3 = 8.23
Team Avg Impact = (450 + 280 + 120) / 3 = 283.33
Alice’s Contribution Score:
Normalized Quality = 8.5 / 8.23 = 1.03
Normalized Impact = 450 / 283.33 = 1.59
Contribution Score = (1.03 + 1.59) / 2 = 1.31
Alice scores 1.31 — 31% above the team average.
Reading the leaderboard Use the score combinations as a diagnostic for engineering health, not just a ranking:
Pattern Signal Action High Quality + High Impact Top performer Ideal mentor, lead reviewer, or owner of complex refactors High Impact + Low Quality Possible burnout — shipping fast, cutting corners Check workload and deadlines High Quality + Low Impact May be stuck on a hard problem or under-utilized Check task allocation Uneven Impact distribution High “Bus Factor” — knowledge concentrated in one person Redistribute knowledge and workload
Review the leaderboard weekly to track quality trends and workload balance across your team.
Skill matrix See per-developer proficiency across Security, Performance, Correctness, and Patterns domains
Code review analytics Track code review trends, developer performance, and team health over time
