Two hundred engineers. It is the first thing you tell people about the company, the way you would give its revenue or its age.
When something needs building, the first question is how many engineers it will take. The size of the team is the plan, and it is set before anyone has decided what to build.
A bigger team feels like a stronger one, so the reqs stay open. A good year means you double the team and say so at the all-hands. Growing the company and growing the headcount have been the same sentence for so long that you have stopped hearing two things in it.
No one ever shipped an org chart. Headcount was only ever the price of building, and building has stopped being something you pay people for.
For fifty years the number was honest. Software was built by hand, so more hands built more, and headcount was another word for how much you could make.
Coordination was the cost. Every hand you added had to be aimed, and aiming them bought a layer of managers and a standing meeting to keep the work from colliding with itself. You paid it gladly, because under a certain size the hands still built faster than the overhead could slow them.
The org chart was a way to cut the work into pieces small enough to hand out. It rested on one fact so plain that no one thought to say it. A person is how software gets built.
A person is no longer how software gets built. The machine builds it now, in an afternoon, for the cost of the power it draws, and the hands you spent fifty years hiring have nothing left to do.
What is left is the coordination. Fred Brooks named the trap in 1975: add people to a late software project and you make it later. Every person you add multiplies the talk, not the work. Two people keep one line straight between them; fifty keep more than a thousand. You paid that cost for decades because the hands under it were building. Now the building is the machine’s, and the coordination is all a new hire adds.
The proof came early, in the companies that never grew the overhead. Facebook paid nineteen billion dollars for WhatsApp when it had fifty-five employees and four hundred and fifty million users. Instagram had thirteen people when it sold for a billion. You called them freaks and kept hiring. They were not freaks. They were small enough that almost no coordination stood between the work and the people who wanted it. That was rare then. It will be ordinary now.
You have seen this before, and that is the problem. Assembly gave way to C, C to Python, Python to the framework that wrote the boilerplate for you. Each layer killed a kind of handwork, and each time the industry hired more engineers, not fewer, because cheaper building only meant more building to do. So the same comfort is there for the taking now. Just another abstraction. Move up a level and hire as before.
Those layers left you building, only from a higher level. This one takes the building itself. There is nothing above it to climb to, because the thing above building was never more building. It was the decision of what to build, and that is judgment. The climb that always carried your engineers up to the next kind of work ends there, and there is room on it for very few.
It is also the one thing hiring cannot add. Every voice in the room pulls the decision toward the middle. Taste is a minority position, usually one person who is right before the others can see why, and a room of forty is built to talk that person down. The more you hire, the safer and duller the call, until a company that once decided things sharply decides them by committee.
Growth does not add judgment. It outvotes it.
So stop hiring engineers. Not the ones you have, and not with a severance memo. Stop reaching for a person every time you need something built, because building is the one thing a person no longer adds.
Hire for what is left. The taste to know what to build, the judgment to kill the rest. Then hire almost none of it, because it does not scale by the head either. One person with taste and a machine will outbuild a floor of forty.
The right size of a team is the smallest group that can still hold the whole thing in their heads. The first hire past that is not more capacity. It is the first brick of the structure you will build to manage the people you did not need. You will hire them anyway, because getting bigger is the only way you know to get stronger. That is the flinch. Not that the small team fails, but that it works, and you do not feel big while it does.
The org chart is a fossil, the shape of a bottleneck that has moved out of building for good. Stop feeding it. The staffing firms will keep selling engineers by the dozen, the same trade as selling hours: a price on work that is no longer scarce. Let the companies that still count heads call it strength. Keep yours small enough to think in, hire the judgment, and point the machine at the rest.
We build. They bill.