A team is not a number

Around the middle of 2022 I watched a team I was close to grow from seven people to about a hundred and forty in roughly eight months.

I was supposed to be excited about this. Funding had come in earlier that year. We had real product-market traction. Every hire we made was someone we had been wanting to bring on for a while, in some cases for years. From the outside the team looked like it was going well, and on paper it was.

What I noticed instead was that we were quietly getting slower.

The Tuesday in question

I remember a specific Tuesday morning.

There was a small product decision that needed to be made. Nothing dramatic. It was the kind of thing that, six months earlier, would have happened in the corner of our old office. Someone would have walked over to two other people. The three of them would have argued about it for ten minutes, decided, and gone back to their desks. By lunchtime, the affected piece of the product would already have been updated and the new version would be in front of whoever needed to see it.

On that Tuesday, the same kind of decision needed two meetings, four people on Discord, a written proposal, two rounds of comments on the proposal, and a senior person reluctantly making the call after a third meeting because nobody else felt comfortable owning it.

The decision itself was identical to the one we would have made six months earlier in the hallway. The slow part wasn't the thinking. The slow part was finding out who was supposed to be in the room.

Nobody on the team was doing anything wrong. We had simply added enough people that coordination cost something it had not cost before, and we had not built any new habits to absorb that cost. Each individual person felt like they were doing the right thing. The system as a whole had quietly stopped being able to make small decisions quickly.

What I had actually been measuring

I sat with this for a long time. The honest thing I kept coming back to was that I had been measuring growth in the wrong unit.

Seven people had been able to do certain kinds of work because they were seven people. We could fit the whole product in one conversation. We could change our minds on a Monday and have the change shipped by Friday. We had specific instincts about who would push back on what, and we adjusted in real time without ever talking about it.

A hundred and forty was not a bigger version of that. It was a different organism.

The instincts that had served seven of us were not portable. They had relied on everybody being two desks away from everybody else, and that condition was no longer true. Some of the instincts didn't even survive Discord, because Discord carried a tone of voice the original conversations had never needed.

Treating a hundred and forty as a slightly larger version of seven is where most of the friction came from. I had been calling it growth. It was more accurately a transition. We had quietly become a different kind of team and were still trying to use the old team's habits.

Shrinking back

A year and a half later, partly by design and partly by circumstance, the same group had moved back down to about forty.

Different work, mostly. Different rhythm. The reasons were a mixture. Some product bets we'd made hadn't landed the way we'd hoped. The macro environment had shifted. We had had to decide where to spend the smaller energy budget we had left.

Some of it was painful at the time. The people we let go were people I had recruited personally and stood up for in promotion conversations. There was no clean way to do it. There still isn't. The people who stayed were carrying something we did not talk about often enough.

But forty turned out to be the size where we could do the kind of work we actually needed to do.

Decisions started happening in hallways again, sometimes literally. Mistakes were getting caught earlier because you could feel them in the room rather than discover them in a quarterly review. The product I'm proudest of having shipped during this period was shipped at this size, not at a hundred and forty. The product would not have been possible at seven, and would not have shipped on time at a hundred and forty. It needed roughly the size we were at.

That was not a coincidence. It was the shape of the team finally matching the shape of the work.

The thing I want to remember

I don't have a clean lesson from this. I don't really want one. The version I keep coming back to is just: the team is not a number.

Whatever it is now, it is a different organism than it was two quarters ago. Pretending otherwise costs a lot. It costs decisions, morale, and the specific kind of trust that small teams have and large teams have to work harder to recreate. It costs the things that don't show up in any dashboard, which means by the time it shows up, you've usually already paid.

The honest measure of team health for me, after all that, isn't headcount. It is how easy it still is to make a small decision in a hallway. When the hallway gets quieter, something has changed. When the hallway gets noisier in a useful way, something is working.

I don't think I'll measure team health by headcount again.

That's mostly what I wanted to remember.