“Organizations which design systems are constrained to produce designs which are copies of the communication structures of these organizations.”
— Conway’s Law
In other words: the way the company is structured – hierarchy, teams, divisions, etc – determines how the product will be divided and worked on too. It’s the organizational, positive equivalent of the old Latin saying “divide et impera“. The way your company is organized is the way everything the company produces will be organized.
This is probably one of the most important laws that connects business organizations and design, and should me something that every manager, every CxO, every HR person, and in general everyone that organizes teams and people should be very aware of.
This law is so important because it has a lot of important facets:
- It tells you to start with people first.
- It tells you that you can’t change the architecture of a software if you don’t change at the same time how the people working on it are organized.
- It explains why certain systems developed in a certain way: just look at the people.
- It tells you to hire consultancies that work in a way you’d like to work, not companies that you don’t want to resemble.
- It hints you on how to structure new projects, starting by thinking how groups are organized.
- It tells you to not buy a software that implies a different practice than the one your team is comfortable to work with.
The beauty of this is that can work well on things as structured as engineering and as loose as organizing a party.
An important consequence of understanding this law is how obviously strict, formal, traditional hierarchies can’t work on new product, new solutions, innovation and changing markets. Organizational hierarchies should be able to change as fast as either the need of internal innovation (internal pressure) or the market transformation (external pressure).
- A. MacCormack, J. Rusnak, C. Baldwin (2011) Exploring the Duality between Product and Organizational Architectures: A Test of the “Mirroring” Hypothesis.
Thanks to Joan Rho for discussion and sources.