PROCESS

Here is How to Build

Design

Core

The direction

Once the core process is named, the first thing to design is not its detail but its direction. The direction is the core at its highest level, L1, a single clear statement of what the core is for and which way it goes: where it begins, where it ends, and the main path between. It is not a sketch of a lesser thing, and not a rough draft of the backbone. It is the core itself, seen at the top, before it is deepened. For a software subscription, the direction is one line: bring a subscription to an active state, deliver the software, and keep it licensed. That is the whole of L1, and everything below it will be a deepening of exactly that.

Setting the direction is deciding the shape of the whole before drawing any of its parts. It answers the questions that govern everything beneath: what does this core start from, and what does it deliver? Through what main stages does it pass? Where does it end? Answer those, plainly, in a line or two, and you have the direction. Leave them unanswered and begin drawing detailed steps, and you are detailing a shape you have not yet decided, filling in a picture whose outline is not drawn. The direction is that outline. It must exist, and be right, before the picture is filled.

This is why the direction is set first and on its own. It is small, a statement, not a diagram, and that smallness is its strength: it can be looked at whole, argued over, and settled, before any effort is spent below it. A core process with a clear, agreed direction has a top that holds, and everything deepened from it deepens something real. A core process whose detail was drawn before its direction was set has a bottom heavier than its top, detail resting on a shape no one agreed, and the first disagreement about direction brings the detail down with it.

The highest possible standard is to set the direction of the core first and alone, as a single clear statement of what it is for and which way it goes, before any detail is drawn, so that everything deepened below rests on an outline that has been looked at whole and settled.

Key Takeaway: Once the core is named, the first thing to design is its direction, not its detail. The direction is the core at L1: one clear statement of what it is for and which way it goes, where it begins, ends, and the main path between. It is the core itself seen at the top, not a lesser sketch. Setting it decides the shape of the whole before any part is drawn: start, delivery, main stages, end, answered in a line or two. Detail drawn before the direction is set rests on a shape no one agreed, and the first disagreement brings it down.

The direction is the core at L1, a single clear statement of what it is for and which way it goes, set before any detail.

MarvinPro · PROCESS · Here is How to Build · Design · Core · Discipline 3: The Direction (L1) · Section: The direction

MarvinPro | September 2026

marvinpro.com

Cheap at the top

Here is why the direction is settled before the detail, and it is a matter of cost. A direction is cheap to change while it is only a direction. It is a line or two; changing it is a conversation. But once the detail has been built beneath it, the backbone deepened, the steps drawn, the teams and systems mapped, changing the direction is no longer a conversation. It is the tearing-down and rebuilding of everything that was deepened from the old direction. The higher you change, the more falls with the change. At the top, almost nothing falls. At the bottom, everything does.

So the order is not a matter of tidiness but of arithmetic. Deepen a direction that turns out to be wrong, and the detail is built twice: once under the wrong direction, and again under the right one, after the wrong one is torn out. Settle the direction first, and the detail is built once, on ground that will hold. Every hour spent agreeing the direction at the top saves the far greater number of hours that a change of direction would cost once the detail is in place. This is the plainest economy in the design of a core: get the cheap thing right first, so the expensive thing is done only once.

And there is a second cost, quieter but just as real. Detail built under an unsettled direction is not only at risk of being rebuilt; it also gives a false sense that the core is progressing, when what is progressing is detail on a shape that may not survive. A backbone drawn under a direction no one has agreed looks like progress and is often rework in waiting. The discipline of settling the direction first protects against both: the literal cost of building twice, and the hidden cost of mistaking motion below for a decision above. Decide the direction, cheaply, at the top; then, and only then, spend the effort to deepen it.

The highest possible standard is to settle the direction while it is still cheap to change, before any detail is built beneath it, so that the detail is built once on ground that holds, and no effort is spent deepening a direction that has not been decided.

Key Takeaway: The direction is settled first because it is cheap to change while only a direction, a line or two, a conversation, but expensive once detail is built beneath it, when changing it means tearing down and rebuilding all that was deepened. The order is arithmetic, not tidiness: deepen a wrong direction and the detail is built twice; settle it first and it is built once. There is a second, quieter cost too: detail under an unsettled direction looks like progress but is often rework in waiting. Decide the direction cheaply at the top, then deepen.

A direction is cheap to change at the top and dear once detail is built beneath it; settle it first, so the detail is built once.

MarvinPro · PROCESS · Here is How to Build · Design · Core · Discipline 3: The Direction (L1) · Section: Cheap at the top

MarvinPro | September 2026

marvinpro.com

Agreed, and given time

A direction that is clear but not agreed is not yet a direction; it is one person's opinion of where the core should go. To be the direction, it must be agreed, by those who own the core business the core delivers, and by those who will build and live with the design beneath it. An outline one person holds and others reject will be fought at every level below, each detail a fresh argument about a direction that was never settled. So the work of L1 is not only to state the direction but to agree it, to bring those who matter to a shared line about what the core is for and which way it goes, before a single detail is deepened. The formal sanction of the whole designed core comes later, in its own discipline; what L1 needs is agreement on the direction itself, first.

And agreeing the direction, then deepening it, takes time, more time than those waiting for detail often expect. To set the direction, tune it, agree it, and then deepen it through its middle levels to the detailed backbone is real work, and for a core of any size it takes months, often several. This is not slowness. It is the method doing what it is for: closing, level by level, the gaps that rushing would leave open. A backbone deepened carefully from an agreed direction takes months and is built once; a backbone demanded in days is built on guesses and rebuilt for years.

This is why the designer needs their manager's trust. The pressure, always, is to see detail quickly, the steps, the diagrams, the instructions, and a manager who does not understand the method will press for the bottom before the top is set, and mistake the months the method needs for delay. The designer's manager must trust the designer enough to grant the time the levels take, and to accept that the early months produce direction and structure, not yet the detailed how. The manager who grants that time gets a core built once, on a direction that holds. The manager who will not forces the detail rushed beneath a direction not yet agreed, and buys, in the end, the very rework the method exists to prevent. Trust at the top buys soundness below.

The highest possible standard is to agree the direction with those who own the business and those who will live with the design before deepening it, and to be given, by a manager who trusts the method, the months that setting and deepening the levels truly take, so the core is built once on an agreed direction rather than rushed and rebuilt.

Key Takeaway: A direction clear but not agreed is one person's opinion; to be the direction it must be agreed, by those who own the core business and those who will build and live with the design, before any detail is deepened, or it is fought at every level below. And the method takes time: setting, agreeing, and deepening the direction through the levels to the detailed backbone takes months, often several, not slowness but the method closing gaps rushing would leave. So the designer needs their manager's trust to be granted that time; the manager who grants it gets a core built once, the one who will not buys the rework the method exists to prevent.

Agree the direction before deepening, and be given the months the levels take; trust at the top buys a core built once.

MarvinPro · PROCESS · Here is How to Build · Design · Core · Discipline 3: The Direction (L1) · Section: Agreed, and given time

MarvinPro | September 2026

marvinpro.com

A real example

Return to the software company, and set the direction of its core before any detail. The core business is the subscription; the core process delivers it. So the direction, at L1, is a single line: bring a subscription to an active state, deliver the software to the customer, and keep it licensed for as long as it is paid for. That is the whole of L1, no steps yet, no teams, no systems, just the shape of the core, stated plainly enough to be looked at whole and agreed.

And it was tuned while it was still cheap. The first draft of the direction said only bring a subscription to an active state and deliver the software, three verbs, subscribe, activate, deliver, and it looked complete. But in agreeing it, someone who owned the subscription business pointed out that a subscription is not a sale but a standing promise: the software must keep working only while it is paid for, and stop when it is not. So a fourth beat was added, keep it licensed, verify and hold back, and the direction became subscribe, activate, deliver, verify. Changing it cost nothing, because it was still only a line; no detail had been built beneath the three-verb version to tear down. Had the backbone already been drawn for subscribe-activate-deliver, adding the verify beat would have meant reworking all of it. Caught at L1, it was a single added clause.

Then, agreed, it was given time to deepen. Those who owned the business and those who would build and run the core agreed the four-beat direction, and only then did the designer begin deepening it, through the middle levels toward the detailed backbone, work that took months, not days. The manager who oversaw it understood that the early months would produce direction and structure, not yet the click-by-click detail, and granted the time. What came out at the end was a backbone built once, on a direction that held, because the cheap thing had been made right, and agreed, and given its time, before the expensive thing was built at all.

The direction was one line, tuned while it was cheap, agreed by those who owned and would run it, and only then given the months to deepen, so the backbone was built once.

MarvinPro · PROCESS · Here is How to Build · Design · Core · Discipline 3: The Direction (L1) · A real example

MarvinPro | September 2026

marvinpro.com

The Direction (L1)

Once the core is named, the first thing to design is its direction, not its detail. The direction is the core at L1: a single clear statement of what it is for and which way it goes, where it begins, where it ends, and the main path between. It is the core itself seen at the top, not a lesser sketch. Setting it decides the shape of the whole before any part is drawn, and detail drawn before the direction is set rests on a shape no one agreed, brought down by the first disagreement about where the core should go.

The direction is settled first because it is cheap. A line or two is cheap to change while it is only a line; a change is a conversation. But once the detail is built beneath it, changing the direction means tearing down and rebuilding all that was deepened from the old one. So the order is arithmetic: deepen a wrong direction and the detail is built twice; settle it first and it is built once, on ground that holds. And detail built under an unsettled direction gives a false sense of progress, motion below mistaken for a decision above. Decide the cheap thing right, so the expensive thing is done only once.

But a direction is not settled until it is agreed. Clear but not agreed, it is one person's opinion, and it will be fought at every level below. So the work of L1 is to agree the direction, with those who own the core business and those who will build and live with the design, before any detail is deepened. And that work takes time: to set, tune, agree, and deepen the direction through the levels to the detailed backbone is months of real work, often several, not slowness but the method closing the gaps that rushing would leave. So the designer needs their manager's trust, to be granted the time the levels take, and not pressed for the detailed how before the direction is set. The manager who grants that time gets a core built once; the one who will not buys the rework the method exists to prevent.

So the discipline is this: set the direction of the core first, one clear line of what it is for and which way it goes. Tune it while it is cheap, before any detail rests on it. Agree it with those who own the business and those who will live with it. And be given, by a manager who trusts the method, the months that setting and deepening the levels truly take. Do this, and the direction holds, the backbone below is built once, and the whole core rests on an outline that was made right, agreed, and given its time before the first detailed step was ever drawn.

Set the direction first, tune it while it is cheap, agree it, and give it time; then the backbone is built once on ground that holds.

MarvinPro · PROCESS · Here is How to Build · Design · Core · Discipline 3: The Direction (L1) · Chapter Outcome

MarvinPro | September 2026

marvinpro.com

Think Simple.