The fastest product decision I ever watched was made in a stairwell, by two people, in under a minute.
It was a pricing change, and not a small one — it touched every customer on the platform. In a larger company it would have meant a deck, a meeting to schedule the meeting, a finance sign-off, a legal review, three weeks of calendar friction before anyone typed a character. Instead, the head of product caught the engineer who owned billing on the way to the parking garage, said the thing, got a nod, and by the time they hit the street the work was scoped. It shipped that night. It was right.
I have spent a long time since then trying to explain why that company moved the way it did, and why so many others — staffed with smarter people, better funded, holding more of the market — moved like they were wading through wet cement. The honest answer embarrassed me, because it had nothing to do with talent. It had to do with plumbing.
Pressure finds the crack
Water is the most patient force on earth and the most relentless. Put a dam in front of it and it does not argue. It does not schedule a review. It presses, evenly and without ego, against every surface at once, and it routes through whatever gap you left it. A river is just water that found a channel narrow enough to move through.
A team is the same. Every team carries pressure — a backlog of things people know need doing, decisions waiting to be made, the held breath of work that everyone agrees on but no one has been allowed to start. The question is never whether the pressure exists. It always exists. The question is how wide the channel is, and how many obstacles sit between the pressure and the open field on the other side.
A small team is a narrow channel. The pressure has almost nowhere to disperse, so it concentrates, and concentrated pressure moves. Two people in a stairwell are a river. A twelve-person steering committee is a reservoir — vast, calm, and going nowhere in particular.
The variable nobody measures
When founders ask why a competitor is shipping faster, they almost always reach for the same two explanations: the other team is more talented, or the other team is working harder. Both are flattering to believe, because both are things you can fix by hiring or by suffering. Both are usually wrong.
The variable that actually governs a team's speed is the number of people who have to agree before something can happen. Call it the agreement count. It is not how good your people are. It is how many of them stand between an idea and its execution.
This is why adding people so often makes a team slower, which feels like a paradox until you do the arithmetic. Fred Brooks named the mechanism fifty years ago: communication overhead grows faster than headcount, because every new person has to coordinate with everyone already there. Five people have ten lines between them. Ten people have forty-five. You did not double the team. You quadrupled the number of conversations that have to go right before anything ships.
But Brooks was describing a cost. I think the deeper point is about routing. Each person you add to the agreement count is not just another conversation — it is another surface for the pressure to spread across, another place the water can pool instead of cut. The dam does not get taller. It gets wider. And a wide, shallow dam holds back exactly as much water as a tall narrow one, while feeling, to everyone standing on top of it, far more reasonable.
What looks like caution
Here is the part that took me years to see, because it wears such a convincing disguise. The widening almost never announces itself as bureaucracy. It announces itself as care.
Someone gets surprised by a decision they wanted to weigh in on, so next time they get looped in. Reasonable. A launch goes sideways, so a review gate goes up to catch the next one. Also reasonable. A senior hire, quite sensibly, wants visibility into what touches their domain. Each addition is defensible on its own — and each is one more person who has to agree, one more inch of width on the dam. No single decision built the reservoir. The reservoir is what you get when twenty reasonable decisions accumulate and nobody is measuring the thing they add up to.
I have done this myself, more than once, and called it diligence. It felt like maturity at the time. It felt like the responsible thing a real company does. What it actually did was take a team that moved like water and teach it to move like a committee, one well-meaning sign-off at a time.
The narrow channel is a design choice
If speed comes from a narrow channel, then the job is not to find faster people. It is to keep the channel narrow on purpose, against the constant, decent, well-argued pressure to widen it.
Concretely, that means a few things I now treat as load-bearing. Give a decision an owner, singular, not a forum. A forum diffuses pressure; an owner concentrates it. Default to the smallest set of people who genuinely must agree, and be ruthless about the word must — wanting to know is not the same as needing to approve, and most "stakeholders" want the former while being granted the latter. When you do add a gate, put an expiration date on it, because gates installed in a panic outlive the panic by years. And separate, always, the people who need to know from the people who need to consent. Confusing the two is how a five-person decision quietly becomes a fifteen-person one.
None of this is a license for recklessness, and I want to be careful here, because the easy misreading is that small teams move fast by skipping the thinking. They don't. The stairwell decision was made in under a minute because the two people in it had earned the context to make it — they knew the customers, the system, the numbers, the blast radius. The channel was narrow, but the water in it was deep. Speed without depth is just a leak.
The shape of what's coming
There is a structural reason this matters more now than it did when Brooks was writing, and it has to do with where the cost of building has gone. The expensive part of making something used to be the making. You needed a lot of hands, and the agreement count was almost an afterthought, because coordinating ten people was cheap relative to the work those ten people did.
That ratio has been quietly inverting for a decade. The tooling keeps getting better at the making, which means the binding constraint on a team is shifting, hard, toward the deciding. When execution gets cheap, coordination becomes the whole game. The companies that win the next decade will not be the ones with the most people. They will be the ones who kept the agreement count low while the headcount, inevitably, grew — a harder problem than hiring, and one almost nobody is treating as a discipline yet.
I suspect, though I can't prove it, that we will come to see the org chart — the pyramid we inherited from the railroads and the army — as a machine optimized for a world where labor was the bottleneck. It is a beautiful machine for moving lots of hands in the same direction, and a terrible one for moving a decision quickly, because every layer is another surface for the water to spread across. We built the dam wide on purpose, back when we needed the reservoir. We may not need it much longer.
What to carry out of here
So if your team feels slow, resist the first two explanations. You probably do not need better people, and you almost certainly do not need them working longer hours. Go count instead. For the last real decision your team made — not the trivial one, the one that mattered — count the people who had to agree before it could move. Then ask, of each of them, honestly: did this person need to consent, or did they just need to know?
The number you get back is your speed. Not your talent, not your funding, not your hours. The number of people standing between the pressure and the open field.
Keep the channel narrow. Keep the water deep. And the next time someone reasonable asks to be added to the approval list, remember that the dam never got tall overnight. It got wide, one good intention at a time.