The founding myth of the small team is that it moves fast because it is small. This is half true. Small teams do move fast, and part of that speed is real. But part of it is an accounting trick — velocity borrowed from a future you have not yet met.
Understanding which part is which is the difference between a small team that scales into a great company and a small team that ships brilliantly for eighteen months and then quietly seizes up.
What small teams actually do faster
Some of the speed is genuine. Coordination cost scales super-linearly with headcount, and a team of four simply does not incur most of the overhead a team of forty does. There are no cross-functional syncs. There is no roadmap negotiation. Decisions get made in a hallway conversation and are in production before a larger org would have scheduled the meeting to discuss them.
This is real leverage. It shows up in the calendar and in the diff. A four-person team can, in a good week, ship what a forty-person team ships in a quarter, and the four-person team is often correct where the forty-person team is a compromise.
What small teams borrow against
The illusion begins where the writing stops. Small teams routinely skip the artifacts that make software legible to people who were not in the room: the architecture document that explains why the queue is where it is, the runbook that describes the last incident, the naming convention that keeps modules from colliding, the test that pins down the behavior everyone currently remembers.
These artifacts feel like overhead when the team is small enough to fit around one table. Everyone knows why the queue is where it is because everyone was there when it got put there. The overhead is invisible, so skipping it feels like a free win. It isn't. It is a loan against the version of the team that will exist in eighteen months.
The interest rate
The loan comes due in a specific way. Around headcount fifteen to twenty, the founding team's collective memory stops being a reliable substitute for documentation. New hires start making decisions based on guesses about why things are the way they are, because there is no written record. Some of those guesses are right. Some of them are subtly, expensively wrong.
The team starts to feel slower. Every project takes longer than it "should." Retrospectives blame process, or hiring, or focus. The real cause is usually simpler: the team is now paying the interest on eighteen months of undocumented decisions, and the interest rate is high.
What to write down anyway
You cannot write everything down. Trying to would slow the team down to the pace you were trying to escape. But there is a specific short list of things that repay the cost, even when the team is four people.
The first is the reason for every irreversible decision. Not the decision itself — that lives in the code — but the reason. Future you will remember the what. Future you will not remember the why, and the why is the only part that helps someone else evaluate whether the decision is still correct.
The second is the shape of your incidents. Every time something breaks in a surprising way, write down what surprised you. Not a full postmortem — just the surprise. Over time this becomes a map of where your system's mental model diverges from reality, which is the single most valuable artifact a growing team can inherit.
The third is the naming convention, whatever it is, written down and defended. Naming is where taste becomes contagious. If your naming is coherent, new hires absorb the coherence by osmosis. If it isn't, they invent their own, and you get a codebase that reads like it was written by strangers who share an office.
Speed that survives growth
The small teams that stay fast as they grow are the ones that treated their small-team phase as an opportunity to embed coherence, not just to skip overhead. They wrote down the things that would be expensive to reconstruct. They defended their naming. They resisted the temptation to treat "everyone knows" as a substitute for "it is written."
This is slower in the short term. It looks, on the calendar, like less velocity. But it is velocity that does not have to be repaid, and over a five-year horizon it is the only kind that compounds.
We think about this constantly at Fewzon. We are small, we are moving fast, and we are trying to be honest about which part of the speed is real. The parts that are borrowed, we are trying to pay back as we go.