PROCESS

Here is How to Build

Design

Sub

The first cut

When a process is divided, the first decision at every piece is the same: is this a sub-process, or a step? It is the most basic cut there is, and everything else about the division follows from it. Decide it rightly at each piece, and the process falls into the parts it is truly made of. Decide it wrongly, and the process is either broken into more pieces than it has, or lumped into fewer, and in both cases the division no longer matches the work.

The test was given in the last discipline: does the piece have an inside worth drawing? A step is a single act, done and passed on, with no inner flow. A sub-process is larger: a flow of several acts doing one coherent piece of work, with an inside of its own. So at each piece the designer asks whether there is a flow within it worth drawing out, and cuts accordingly: a single act stays a step; a flow of acts becomes a sub-process. This is the judgment that turns a process from an undivided whole into named parts, made one piece at a time.

And it is worth making well, because of what a sub-process lets you do that a step does not. A sub-process is a unit you can work on alone. When you need to change one part of a large process, you load just that sub-process, understand it, and work on it, without holding the whole process in your head, on your screen, or in the room. The rest stays as it is, trusted to do its part, while you work on the one part that needs you. A process cut into good sub-processes is a process you can work on a piece at a time; a process left undivided must be held whole every time it is touched. And there is more: a sub-process can be compartmentalised. It can be owned by one team, worked on without disturbing the rest, and shown only to those who need it: the people who should see how a payment is settled see that sub-process, without the whole core laid open to them. The cut into sub-processes is not only how a process becomes readable; it is how its parts become things that can be worked on, owned, and walled off on their own.

The highest possible standard is to make, at every piece of a process, the first cut of sub or step by whether the piece has a flow within it worth drawing out, knowing that a sub-process is what lets a part be worked on alone, owned apart, and shown only to those who need it, so the cut is made for what it makes possible and not by habit.

Key Takeaway: When a process is divided, the first decision at each piece is sub-process or step, the most basic cut, and the division follows from it. The test from the last discipline: does the piece have an inside worth drawing? A single act stays a step; a flow of several acts doing one coherent piece becomes a sub-process. It is worth deciding well because of what a sub lets you do: work on one part alone (loading just it, not the whole process), and compartmentalise it (own it, change it, or show it on its own, without laying the rest open). The cut is not only how a process becomes readable but how its parts become things you can work on, own, and wall off alone.

The first cut is sub or step, by whether the piece has a flow worth drawing; a sub is what lets a part be worked on alone, owned apart, and shown only to those who need it.

MarvinPro · PROCESS · Here is How to Build · Design · Sub · Discipline 2: Sub or Step · Section: The first cut

MarvinPro | October 2026

marvinpro.com

Worth, not just structure

The test is not whether a piece has any inside at all, but whether it has an inside worth drawing. Almost anything can be split into smaller acts if you look closely enough; writing a single value could be described as finding the field, checking it, entering the value, confirming it. But to draw that as a sub-process, with its own named flow, would be to add structure that does not earn its keep. The inside is there, technically, but it is not worth drawing: nobody needs to work on it alone, own it apart, or see its steps. It is a single act for every purpose that matters, and so it is a step.

So the judgment weighs the structure against its cost. Every sub-process you name has a price: it is one more thing to understand, to maintain, to keep consistent with the rest, one more named part in the picture. That price is worth paying when the piece has a real flow, when it can fail partway and the handling matters, when it is worth working on alone, owning apart, or walling off. It is not worth paying when the piece is a single act dressed up as a flow, when naming it adds a box to the diagram and nothing to the understanding. The question at each piece is not can this be divided but should it be: does the inside earn the cost of being drawn?

And the judgment runs the other way too. A piece that looks like a single act at a glance may hide a flow worth drawing, a flow nobody has looked at closely because it was always called one step. Settling a payment might be written as a single step (take the money), until you notice that taking the money is really a flow: take the payment details, request the charge, handle the decline, retry, confirm, each an act, with real handling between them. That flow was always there; calling it one step did not remove it, only hid it. Drawing it out as the sub-process it is makes the hidden structure visible, and the handling designable. So the judgment looks both ways: at the piece that pretends to have an inside it does not have, and at the piece that hides the inside it does.

The highest possible standard is to decide sub or step by whether the inside is worth drawing, weighing the real structure of the piece against the cost of naming it, so a trivial inside is left a step and a hidden flow is drawn out as the sub-process it truly is, and no structure is added that does not earn its keep nor left buried that should be seen.

Key Takeaway: The test is not whether a piece has any inside, but whether it has one worth drawing. Almost anything can be split finely, but drawing a trivial inside as a sub-process adds structure that does not earn its keep: no one needs to work on it, own it, or see it alone, so it stays a step. Every sub-process named has a price (one more thing to understand and maintain), worth paying when the piece has a real flow, can fail partway, or is worth working on, owning, or walling off alone. The judgment runs both ways: a single act dressed as a flow stays a step; a flow hidden inside a supposed step (the handling buried) is drawn out as the sub-process it is.

Decide sub or step by whether the inside is worth drawing, not whether it exists: leave a trivial inside a step, and draw out a hidden flow as the sub-process it truly is.

MarvinPro · PROCESS · Here is How to Build · Design · Sub · Discipline 2: Sub or Step · Section: Worth, not just structure

MarvinPro | October 2026

marvinpro.com

Too fine and too coarse

There are two ways to get the cut wrong, and they are opposites. The first is cutting too fine: making sub-processes where steps would do, naming an inside that is trivial, breaking the process into more named parts than the work has. This is over-dividing, and its result is clutter. A process cut too fine is harder to read for all its apparent organisation: the eye is lost among a crowd of tiny named parts, each a box to open, none worth the opening. Structure that does not earn its keep does not merely fail to help; it actively obscures, burying the real shape of the process under a litter of divisions that mean nothing.

The second is cutting too coarse: leaving as a step something that is really a flow, lumping several acts into one supposed act, letting a sub-process hide inside a step. This is under-dividing, and its result is hidden structure. A process cut too coarse looks simpler than it is, and that false simplicity is a trap: the lump that is called one step is really a flow with its own branches and failures, and because it was never drawn, it was never designed. The handling that should have been built is missing, not because anyone decided against it, but because the structure that would have shown the need for it was never cut. Under-dividing does not remove complexity; it conceals it, until it surfaces in the running as the thing nobody designed.

So the discipline is to cut at the right grain: fine enough that every real flow is drawn as the sub-process it is, coarse enough that no single act is dressed up as one. And the signals help find it. A piece is a sub-process when it has a flow of several acts, when it can fail partway and the handling matters, when it is reused in more than one place, when it is owned or performed by a distinct team within this process, or when it is worth working on, owning, or walling off on its own. A piece is a step when it is a single act, done and passed on, that cannot meaningfully fail partway and that no one needs to open, own, or see alone. These are signals, not a formula; the judgment is still the designer's. But cutting by them, and watching for both failures, keeps the grain right: neither a clutter of trivial parts nor a lump with its structure buried.

The highest possible standard is to cut a process at the right grain, neither so fine that it clutters with trivial sub-processes that do not earn their keep, nor so coarse that real flows stay hidden inside supposed steps, using the signals of flow, failure, reuse, ownership, and the need to work on a part alone to tell a sub-process from a step.

Key Takeaway: There are two opposite ways to get the cut wrong. Too fine (over-dividing): sub-processes where steps would do, a trivial inside named, the process cluttered with tiny parts that obscure its real shape. Too coarse (under-dividing): a real flow left as a supposed step, its structure hidden, so the handling that should have been built is missing, not decided against but never shown, surfacing later as the thing nobody designed. So cut at the right grain, guided by signals: a sub-process has a flow, can fail partway, is reused, is owned apart, or is worth working on, owning, or walling off alone; a step is a single act, done and passed on, that none need open, own, or see alone.

Cut at the right grain: fine enough that every real flow is a sub-process, coarse enough that no single act is dressed as one, guided by the signals, not a formula.

MarvinPro · PROCESS · Here is How to Build · Design · Sub · Discipline 2: Sub or Step · Section: Too fine and too coarse

MarvinPro | October 2026

marvinpro.com

A real example

Return to the software core's activate beat, and make the cut piece by piece. Settling the payment is a sub-process. Look inside it and there is a real flow: take the payment details, request the charge, handle a decline, retry or confirm, each an act, with handling that matters between them: a payment can fail partway, and what happens then must be designed. It is worth working on alone: the person fixing the retry logic should be able to load just this, not the whole subscription core. So settling the payment earns its drawing: it is a sub-process, and cutting it out as one makes its flow visible and its failures designable.

Recording the entitlement is a step. Once the account is open, the entitlement is written, a value set, a flag raised, that lets the software run. Could it be split? Technically, into find the record, write the value, confirm the write. But that inside is not worth drawing: it cannot meaningfully fail partway in a way that needs its own handling, no one needs to work on it alone, and naming it as a sub-process would add a box to the diagram and nothing to the understanding. It is a single act for every purpose that matters. So it stays a step.

Now see the two failures against the same beat. To cut too fine would be to make recording the entitlement a sub-process, and then its three imagined parts sub-processes again, until activate is a thicket of tiny named flows, each a box to open, the real shape of activation lost in the clutter. To cut too coarse would be to call settling the payment a single step (take the payment), hiding its flow: and then, when a payment declines in production, there is no drawn structure for the retry, no designed handling for the failure, because the step that was really a flow was never cut open. The right grain sits between: settle the payment is a sub-process, record the entitlement is a step, and activation divides into the parts it is truly made of, each cut for what it is and what it lets you do.

Settle the payment is a sub-process, a real flow that can fail and is worth working on alone; record the entitlement is a step, a single act not worth drawing; the right grain is the one that matches the work.

MarvinPro · PROCESS · Here is How to Build · Design · Sub · Discipline 2: Sub or Step · A real example

MarvinPro | October 2026

marvinpro.com

Sub or Step

When a process is divided, the first decision at every piece is sub-process or step, the most basic cut, from which the rest of the division follows. The test is whether the piece has an inside worth drawing: a single act, done and passed on, stays a step; a flow of several acts doing one coherent piece becomes a sub-process. And the cut is worth making well because of what a sub-process lets you do: work on one part alone (loading just it, not the whole process), and compartmentalise it (own it, change it, or show it only to those who need it, without laying the rest open), which a step does not.

The test is worth, not mere structure: not whether a piece has any inside, but whether it has one worth drawing. Almost anything can be split finely, but drawing a trivial inside adds structure that does not earn its keep, one more part to understand and maintain, and nothing gained. So weigh the inside against the cost of naming it. And weigh it both ways: a single act dressed as a flow stays a step, while a flow hidden inside a supposed step, its handling buried, is drawn out as the sub-process it truly is.

Getting it wrong has two opposite shapes. Too fine, over-dividing, clutters the process with trivial sub-processes that obscure its real shape. Too coarse, under-dividing, leaves a real flow hidden inside a supposed step, so the handling that should have been built is missing, never decided against but never shown, surfacing later as the thing nobody designed. So cut at the right grain, and let the signals guide you: a piece is a sub-process when it has a flow, can fail partway, is reused, is owned apart, or is worth working on, owning, or walling off alone; a step when it is a single act, done and passed on, that none need open, own, or see alone.

So the discipline is this: at every piece of a process, make the first cut of sub or step by whether the inside is worth drawing, weighing the real structure against the cost of naming it, and watching both failures, the clutter of too fine and the hidden flow of too coarse. Cut by what a sub-process makes possible (a part worked on alone, owned apart, walled off), not by habit. Do this, and the process divides at the right grain into the parts it is truly made of, each a sub-process or a step for good reason, which is the ground the rest of the dividing, how to cut each part cleanly, is built on.

The first cut is sub or step, judged by whether the inside is worth drawing; cut at the right grain for what a sub makes possible, and the process divides into the parts it is truly made of.

MarvinPro · PROCESS · Here is How to Build · Design · Sub · Discipline 2: Sub or Step · Chapter Outcome

MarvinPro | October 2026

marvinpro.com

Think Simple.