Happy Friday!

There's a very long-standing topic that folks have pretty strong opinions on, but rarely have I seen much consensus around, and that is how to organize teams and groups.

I thought I'd write a bit about the topic from what I've measured and observed over the years. While I'll say that right now, I lean toward cross-functional teams, that does not mean other methods cannot work.

I want to first talk about component/feature/matrix organizations as they all share a common trait of creating teams by function or skillset. You have development teams, operations teams, customer teams, quality teams, marketing teams, etc. Now, the theoretical benefits here are that specialization can lead to really highly efficient teams. This can happen because you can master the domain of skills needed for your team more deeply since it is the only thing in your environment.

Cross-functional teams, in contrast, have a mixture of skills and functions. A team may have developers, designers, infrastructure, quality, etc, all in one team. Their primary benefit is that they have few if any, dependencies and don't get stopped waiting on other groups.

I get that I'm only saying the good thing and not all the problems with either, but I'm focusing this article on efficiency very narrowly.

So which is better in terms of efficiency?

Well, let's define efficiency as how quickly something important moves through the organization. I'm defining it this way because no matter how quick an individual is, if the work never makes it out to the market it doesn't matter.

Now, let's pause quickly, and let me offer that you have dinner reservations tonight for just you. You can be either on time or late. That means in a very simplistic model you have a 50% probability to be either. Now lets say you're meeting someone. Each of you can be on time or late, and with that same simplistic model you have a 25% chance of both of you being on time. Add a 3rd party and now you're at 12.5%. A fourth drops it again to 6.25%.

I hope you can quickly see that in terms of how hard it is to get everyone to show up on time, more groups isn't easier.

Now, this is an overly simplistic model based on basic probability. What we would really need to see is actual performance from each group to adjust the model. But, here's the next riddle. How much more efficient does each individual group need to become to outperform the built-in probabilities of failure?

This is the magic question whose answer dictates success in feature/component/matrix groups. To make these structures work, each team either has to become so efficient that the gains outweigh the problem of the structure, or some very attentive people have to manage the space between them where things stop, slow down, get miscommunicated, or stuck waiting for the next group.

While these challenges are very solvable, most organizations don't invest nearly enough in the highly detailed work it requires. So when look back at the cross-functional team who seems inefficient from a staffing perspective and on an individual level may be slower, they tend to completely outperform other models because they start at a 50/50 chance of success.

I realize this framing is putting blinders on to a whole lot of other concerns and issues, but from what I've seen and measured underutilised individuals on a cross-functional team almost always outperforms at-capacity matrix teams.

What have you seen?

Sincerely,

Ryan