---
title: "Key person dependency in engineering teams | QMates"
source: https://qmates.tech/en/services/engineering-transformation/embedded-teams/key-person-dependency
description: "Parts of the system that only one person knows? Solo work, personal ownership and opaque code feed key person dependency. See how knowledge gets shared."
updated: 2026-11-21
---

Key person dependency

# Everything depends on two or three people

The people who know the critical parts get called in for every emergency and never find time to train anyone else, so they stay the only ones who know. Key person dependency feeds itself until a single absence stops a delivery. It is sustained by solo work, by parts of the system assigned to individuals, and by code only its author understands, and asking those people to write documentation just adds one more task to their load.

## Does this sound familiar?

-   Some parts of the system are known by one person only, and when they are away for a few days nobody touches them.
-   The key people are also the ones called in for every emergency, and they never find time to explain.
-   Reviews of the most delicate parts always wait for the same few people.
-   After someone left the company, certain changes stayed slow and risky for months.
-   People who hold the critical parts have little room for anything else, and the thought that they might leave weighs on decisions.

## What concentrates knowledge in a few heads?

The useful question is who else could change each critical part today, and why they do not. The answers almost always lead to a loop. Someone who works alone becomes the owner of a part, the owner has no time to teach, and so keeps working alone.

1.  ### Everyone works alone on their own part

    When each developer has their own area and review arrives once the work is finished, what the author knows never reaches the others. With review done while the code is being written and mob programming on the most delicate code, knowledge circulates as you deliver, with no separate effort to pass it on.

    Feeds into

    -   

    [

    Handled by

    Technical Coaching in XP and TDD](https://qmates.tech/en/services/engineering-transformation/technical-coaching-xp-tdd)

2.  ### Parts of the system are assigned to individuals

    When a part of the system has a single owner, everything about it goes through that person: decisions, reviews, emergencies. If each part belongs to a team, with several people who know it, responsibility stays and the bottleneck goes.

    Feeds into

    -   

    [

    Handled by

    Engineering Team Design](https://qmates.tech/en/services/engineering-transformation/engineering-team-design)

3.  ### Only the author understands the code

    Tangled modules with no tests mean that only those who know the history of the code dare to change it, and for everyone else each change is a risk they prefer not to take. With readable code covered by tests, knowledge moves from the person into the system.

    Feeds into

    -   

    [

    Handled by

    Legacy Code Refactoring](https://qmates.tech/en/services/engineering-transformation/legacy-code-refactoring)

4.  ### The people who know have no time to teach

    Key people are also the most in demand, and they spend their days solving problems only they can solve. Handing knowledge over gets postponed from week to week, until someone takes part of their workload by working at their side.

    Feeds into

    -   

    [

    Handled by

    Embedded Teams](https://qmates.tech/en/services/engineering-transformation/embedded-teams)

- [The service that tackles it](https://qmates.tech/en/services/engineering-transformation/embedded-teams)
- [Not sure which cause weighs most?](https://qmates.tech/en/services/engineering-transformation/software-delivery-assessment)

## What changes when knowledge is distributed

Each part of the system is known by several people, who rotate through the most delicate areas. An absence no longer stops a delivery, and reviews no longer wait for the same names.

When someone leaves the team, their parts pass to colleagues who have already worked on them, and delivery carries on. People who used to be indispensable get to choose what they work on.

- [SureVIVE: modernising the architecture without stopping delivery](https://qmates.tech/en/case-studies/surevive)
