When to split teams

Before I discuss when to split teams, let's clarify why we split them. There are two inherent limits to how people can collaborate efficiently. Let's examine these limits one by one.

The first limitation is how many direct reports a leader can have and still do a good job. Every leader has a slightly different threshold for direct reports, but everyone has an upper limit. There is no hard rule for this, so there is no math formula. For example, I experienced that each direct report requires 10 to 15 percent of my time every week. Six to eight direct reports leave me with no time to take care of "everything else". Based on that, I believe it's hard to imagine that anyone could run a team with more than 12 people and still do a good job.

The other limit concerns the size of a single team before the communication overhead becomes too much of a burden. The limit is much stricter here and a direct consequence of communication complexity growth. I call this "The links formula". The formula gives a strong indication: five to six people is a reasonable upper limit for how big a single team can be before things get too complicated.

These limits tell us two things:

• A team should have no more than 6 people to function well.
• One engineering manager can't run more than two or three teams.

Does that mean you should split as soon as you hit these limits? Unfortunately, it's not that easy. If you hit the limit, it's already too late to change ("better late than never" still applies). Furthermore, because we're late, we start rushing into taking a decision and then end up with a split that isn't good enough. The first few months after a rushed split look like this: people have no clue who is doing what, product managers don't know anymore what backlog they are supposed to fill, engineering managers worry that nothing is happening in such a chaotic situation, so they constantly check on the progress and end up annoying everyone. In short, everything was nicer before we split. To make things even more complicated, you can't split as soon as you have 7 people in one team. You would end up with teams that are too small and too hard to operate. For example: a team of three people almost never functions properly. It lacks resiliency to holidays, people getting sick, and so on.

This situation seems to have no solution:

• You can't split up too late because it leads to chaos.
• You can't split up too early because tiny teams don't function well.

It's tricky because we frame structural changes as a point-in-time change. There is a "before we split" and an "after we split" phase. The key is to frame the problem as evolution of the structure. Evolutions are slow but constant processes: we constantly work on splitting teams.

Share your ideas with the team, give everyone the time to understand how you want to split and why you need to do it. Keep your split proposals in mind when hiring, share with your potential hires how you intend to split teams, ask for feedback; outsiders' feedback can be extremely valuable. This continuous exercise prepares you and the team for what I call "the hard split". It's a plain name for the actual day in which you start using new backlogs and roadmaps, change the sitting plan (if relevant), and formally introduce the new teams. The goal is to shorten as much as possible the transition period that follows the hard split.

Some confusion is inevitable because structural changes are complicated. Don't expect the smoothest process, prepare as much as possible for it. Here's what to look out for:

• Who joins what team?
• What's the mission of the new teams?
• What are the names of the new teams?

The first point is a matter of communication and feedback. As a leader, you surely can come up with a good new structure, but asking the people involved is a necessary step. Good leadership isn't convincing people to go where you want them to go. It's understanding where people want to go and helping them get there. Sometimes asking for feedback will surprise you. The reasons why some people may want to move to another team are gold for leaders. They provide great insights into your teams and how to better organise them.

The last two points are about how to split teams. The topic is worthy of its own paragraph.