The most expensive way to build something large is to build all of it before any of it works. Design the entire thing, configure every module, integrate every system, and flip the switch on the whole apparatus at once. It is the most natural plan in the world to draw on a whiteboard, and it concentrates every risk you have into a single day eighteen months out.
There is a steadier way, and it is not the same as lowering your ambition. You ship something small that genuinely works, end to end, and then you grow it — by choice, in a sequence you control, each step earning the next. Big things that last were almost always small things that worked, extended on purpose.
Small means a working slice, not a half-built whole
The word "small" does a lot of work here, so it is worth being exact. Small does not mean half of everything, built halfway. It means one slice that runs all the way through — a single process that posts, reconciles, and closes for real, even if it is the only one live. A narrow thing that works end to end teaches you something a broad thing that half-works never can: whether the foundation actually holds when real data and real users hit it.
That feedback is the entire point. When the first slice goes live, you learn what your design got wrong while the cost of being wrong is still small — before you have poured the same foundation under a dozen other processes. Each increment is also a place you could stop and still have something of value, which is a very different risk posture from a program that delivers nothing until the end and then delivers it wrong.
- End to end, not half-deep. One process that runs the full distance beats five that each stop in the middle.
- Live and used, not demoed. Real data, real users, a real close. A demo proves nothing about whether it holds.
- A chosen next step, not a leftover. You decide what grows next because it is the right next thing — not because it is what happens to be left.
The line between this and drift
This sits right next to a warning from earlier in the series — that drift is not progress — and the two can look alike if you squint. Both involve a project that keeps growing. The difference is the whole game. Drift is expansion nobody chose, arriving through the side door. Growing on purpose is expansion you sequenced deliberately, where each step is a decision with a reason behind it. Both grow. One grows on purpose.
So start with the smallest thing that reconciles. Prove it in production, with real numbers and real people. Then choose — deliberately, out loud — the next piece to add, and the one after that. The vision can be as large as you like. It is the path to it that should be built one working slice at a time.