DEV Community

Cover image for Who could your team not survive losing? Measuring the bus factor
Sascha Mirow-Stanitzok
Sascha Mirow-Stanitzok

Posted on Edited on Originally published at sascha-ms.de

Who could your team not survive losing? Measuring the bus factor

It is 11 pm, the payment integration is failing, and the only person who really understands it is on holiday. Most teams have at least one area like that. The uncomfortable question behind it is the bus factor: how many people could drop out tomorrow before something important stops moving? If the honest answer is "one", you have a knowledge silo.

As a team lead, I look at this from two sides: it is an operational risk, and almost always a people topic too. I wrote a full guide with a workshop format, a TypeScript script for Git analysis and a 30-day plan in Bus factor: how to break up knowledge silos in teams. Here are the main points.

The cost you don't see on the risk register

The obvious risk is someone being absent. The less obvious effects show up in normal weeks. The expert becomes a bottleneck, because every change in "their" area goes through them. They get called on holiday, receive the alerts at night and rarely get to pick up new topics. And an area only one person understands gets no real pushback in reviews, so shortcuts quietly accumulate.

The problem is common. A 2016 study by Avelino and colleagues found that 65 percent of 133 popular open source projects on GitHub had a bus factor of 2 or less. It also helps to remember there is no single bus factor for a team: your booking API may be well covered while the deployment pipeline depends on one person.

Two ways to make it visible

I recommend combining two perspectives:

  • A knowledge map workshop. Topics as rows, people as columns, everyone rates themselves from 0 ("no idea") to 3 ("I can explain it to others"). The bus factor per topic is simply the number of people at level 2 or 3. It also captures knowledge that never shows up in code, like on-call duty or contacts in business departments. It only works if everyone knows the table will not feed into performance reviews.
  • A Git history analysis. Research by Avelino, Passos, Hora and Valente describes how to compute a "degree of authorship" per person and file, then greedily remove the most dominant authors until fewer than half the files still have someone who knows them.

The core formula is short:

DOA = 3.293 + 1.098 ร— FA + 0.164 ร— DL โˆ’ 0.321 ร— ln(1 + AC)

FA = created the file (1 or 0)
DL = own changes, AC = changes by others
Enter fullscreen mode Exit fullscreen mode

Treat the result as a conversation starter, not a verdict. In Germany, an analysis of commits per person also touches co-determination rights of the works council, so keep it within the team and show results per directory, not as rankings.

Moving knowledge without devaluing anyone

Measuring is the easy part. In my experience, documentation alone rarely moves knowledge, because nobody reads the wiki until they have to. What works better is two people working on real tasks together:

  • Pairing with swapped roles: the person who does not know the topic holds the keyboard; the expert only explains.
  • Deliberate reviews and rotation: changes in a silo area get a reviewer who is still learning it, and the next ticket goes to someone else with the expert as contact.
  • The absence test: the expert is unavailable for a planned day and the team notes everything that gets stuck.

The human side matters most to me. "We need to raise your bus factor" can sound like "we are making you replaceable". I frame it as taking weight off someone's shoulders: whoever shares knowledge can go on holiday without a laptop and turns from bottleneck into multiplier. And I would never turn the bus factor into a KPI with a target, because then you get alibi commits instead of real learning.

What the full article covers

  • A step-by-step plan for running the knowledge map workshop in about an hour
  • A complete TypeScript script (around 120 lines) that estimates the bus factor from git log
  • Common pitfalls such as shallow clones, squash merges, bots and monorepos
  • More measures for spreading knowledge, from CODEOWNERS to runbooks and ADRs
  • A 30-day plan for team leads to get started next week

Read the full guide: Bus factor: how to break up knowledge silos in teams.

๐Ÿ“– This is the condensed version. The full article on sascha-ms.de goes deeper โ€“ with all code samples, diagrams and a short knowledge check.


๐Ÿ‘‰ Next step: Preparing a first one-on-one? My question cheat sheet puts the best questions on one printable page.

๐Ÿ“ฌ Stay updated: New articles on serverless, AI and leading tech teams via RSS, browser push or the weekly newsletter (in German).

Originally published on sascha-ms.de. More articles on serverless, AI in everyday development and leading tech teams: sascha-ms.de/en

Top comments (0)