Guide
Output Concentration and Bus Factor: Reading the Top 20% Share
Concentration asks how much output rides on a few people. It is a resilience signal, not a ranking. Here is how to read the top 20% share and Gini, and how to spread load safely.
Output concentration asks how much of a team's shipped work depends on a small number of people. It is a resilience question, not a performance one. A team where two people produce eighty percent of the output is fragile: a single departure, illness, or burnout can stall it. Measuring concentration lets you spot that risk before it becomes an incident.
This guide explains the two most useful lenses (the top 20% share and the Gini coefficient), how to read the healthy-elevated-concentrated bands, and, most importantly, how to redistribute load without turning the metric into a ranking of individuals.
Understand the top 20% share
The top 20% share is the fraction of total output produced by your most-productive fifth of engineers. In a healthy team it is meaningfully above 20% (output is never perfectly even) but not overwhelming. When the top fifth produces most of the work, the team is carrying concentration risk regardless of how talented those people are.
Read the Gini coefficient
The Gini coefficient summarizes the whole distribution in a single number from 0 (everyone contributes equally) to 1 (one person does everything). It complements the top-share view by capturing the shape of the middle, not just the tails. A rising Gini over time is an early warning that work is pooling.
Interpret the healthy / elevated / concentrated bands
Translate the raw numbers into bands so they are actionable. Healthy means output is reasonably spread and the team is resilient. Elevated means a subset is carrying more than is comfortable and you should watch it. Concentrated means real knowledge risk that warrants deliberate redistribution. Remember that small teams are naturally more concentrated, so read the band in context.
Connect concentration to bus factor
Bus factor is the number of people who would have to be lost before a project stalls. High concentration means a low bus factor. The two are the same risk viewed from different angles. Pair concentration with code-ownership data to find the specific areas where knowledge is dangerously thin.
Redistribute load without ranking people
The fix for concentration is structural, not personal. Pair on critical areas, rotate ownership, invest in documentation and onboarding, and give the people carrying the load room to hand it off. Crucially, never turn concentration into a leaderboard: the person producing eighty percent is usually a hero to support, not a problem to flag. The metric is about the system's resilience, not any individual's worth.