How to Choose

Two ways:

1. Evaluating tradeoffs, ideally with some kind of data. Every configuration has tradeoffs, and there (probably) is no such thing as a single perfect design. A default-matrix organization will be deeply in-sync for major global initiatives, but will have a hard time being customer-focused or inventing the next amazing thing. Autonomous units will be more customer-centered, but will be “messy” in comparison to a matrix. IMO the best data to inform different organization designs is time allocation, mostly because it’s our only non-renewable resource.

2. Trying different alternatives and seeing what works (which helps with option 1). This is always happening in organizations, but almost never with the kind of formality and intention that we’d like.

Why is this important? Because it defines the product we make. Tesla’s case, above, is a perfect illustration of Conway’s Law.

Any organization that designs a system (defined broadly) will produce a design whose structure is a copy of the organization's communication structure. — Melvin E. Conway

Said differently: organization configures the product; organization design is product design.