The building was never designed to remember

Think of a company's decision-making as a building. Not a metaphor about ambition or growth, but a literal architectural question. How was it designed. Where does the load sit. What happens when one wall shifts.

Most companies were not designed at all. They were built the way houses are built in a hurry: room by room, problem by problem, one extension at a time. The first room was fine. The founder made every decision, remembered every one, and the house held. The second room worked too. Then a floor was added, then a wing, and somewhere around three hundred people, the building is large and busy and nobody can say with confidence which walls are load-bearing and which ones could come down without consequences.

The leadership team sits at the top of this building. Every consequential decision passes through them. And the building was never designed to help them be consistent. It was designed, if it was designed at all, to be fast.

That is the architecture. The question is what it costs.

How the building was designed (the past)

In the early days, consistency was free. It came from the founder's memory. Every decision was made by one person, or by a handful of people who sat close enough to overhear each other. The pattern of prior decisions was not written anywhere because it did not need to be. It lived in one head, and one head was enough.

This worked beautifully, and people remembered it working, and that memory became the building's first load-bearing wall: the belief that decisions should be fast, informal, and held in the heads of the people who made them. Anything else was bureaucracy.

As the company grew, new rooms were added. A VP was hired. Then another. Then department heads, regional managers, people whose job it was to say yes to things the founder no longer had time to see. Each of them inherited the architecture: decide fast, remember what you decided, don't make a process out of it.

But now the decisions were spread across a dozen heads instead of one. And a dozen heads cannot synchronise their memories. The VP who approved eighteen percent for a plant supervisor in July does not share a memory with the VP who approves twelve percent for a nearly identical case in October. They might be the same VP. It does not matter. Without a record, each decision is made against a blank, and the blank is filled by whatever the person happens to be carrying that day.

Nobody chose this architecture. It simply formed, the way buildings do when each room is added to solve today's problem without asking what it does to the structure. The company grew into a building where the most consequential decisions rest on the least reliable material: the memory and mood of whoever happens to be in the room.

And the building held. For a while. The way buildings do before the leaks start.

Where the money leaks (the present)

The leaks do not announce themselves. There is no single dramatic failure. There are a hundred small ones, running continuously, none large enough to investigate, all of them flowing from the same crack in the foundation.

The consistency leak. A VP approves eighteen percent in July and twelve in October for the same situation. He does not know he has contradicted himself because he cannot see his own prior decision. Three months later, two supervisors compare notes. Now the company has a grievance to manage, a trust problem in the plant, and a precedent it cannot explain. The cost is not the increments. The cost is that every future increment request in that plant will be calibrated against perceived fairness rather than policy, and perceived fairness is shaped by the contradiction the VP does not know he made.

The mood leak. The same leader, the same question, a different day. Monday morning, rested, after a good quarter review, the answer is yes, go ahead, invest in it. Thursday afternoon, behind on numbers, after a difficult call with a distributor, the answer is let's wait. Both answers feel right in the moment. Neither is wrong in isolation. Together they teach everyone below that the answer depends on the weather upstairs. So people start timing their requests. They learn which days to approach and which to avoid. They read mood instead of policy, because mood is the actual variable. That is not politics. That is rational behaviour inside a building that runs on one person's emotional weather. The company is paying full salaries for people to spend a portion of their intelligence managing upward rather than working outward.

The closed-room leak. This one is the slowest and the most expensive. The VP has an operations head who disagrees on the July increment. She recommends twelve. She explains the precedent risk clearly. The VP hears her, weighs it, and overrules her. That is his right.

But her input vanishes. It was offered in a room and it left with the room. When October comes, nobody is there to make the counterargument. The VP does not even remember there was one.

Over time, the operations head notices. Not that she is being overruled, that is tolerable, but that her input has no weight beyond the meeting it was offered in. If the decision goes wrong later, there is no record she warned against it. If it goes right, there is no record she shaped it. Her perspective is structurally weightless. It exists for the duration of the conversation and then it is gone.

She adjusts. Not as a protest, not dramatically, but by degrees. The objections get softer. The counterarguments arrive later, or framed more carefully, or not at all. She starts reading the room before she speaks, calibrating whether today is a day the VP will hear a dissenting view. She is not becoming less competent. She is becoming more efficient within a building that has taught her that strong input and weak input produce the same result: both dissolve after the meeting.

The company is now paying for a senior operations head and receiving a fraction of her judgment, because the architecture discards everything that is not the final answer. The perspectives that would keep the leader consistent, the objections that would catch the mood-driven calls, the alternatives that would make the decision richer, all of it flows out through the same crack. Offered, heard, dissolved, gone.

The compounding leak. Each of these leaks feeds the others. Inconsistency teaches people that decisions are temporary, so they hedge. The mood dependency teaches them to manage upward, so they spend energy on timing instead of substance. The closed room teaches them not to push hard, so the leader hears less. The leader, receiving less pushback and unable to see their own pattern, becomes more inconsistent. The cycle tightens. The building gets more expensive to operate every quarter, and the expense does not appear on any line item because it is absorbed into the salaries of people who would have been paid anyway.

A company five years into this pattern is not carrying five years of the same cost. It is carrying an accelerating cost, because every unrecorded decision, every dissolved perspective, every mood-driven contradiction makes the next one more likely and more expensive.

What the building actually needs (the future)

The instinct, when you see the leaks, is to fix the leader. Send them on a course. Give them a coach. Talk about consistency as a leadership competency. Run a 360 review.

None of that touches the architecture. The leader is not the problem. The leader is an unassisted human being making hundreds of decisions a year inside a building that gives them no memory, no baseline, and no access to the perspectives that were offered last time. Fixing the leader is like repainting a wall to stop a leak in the foundation. It looks better for a quarter.

The investment is architectural, and it is smaller than it sounds.

A visible record of consequential decisions, including the reasoning and including the dissent. Not every choice. Not a form for everything. Just the ones that pass the test: could someone, in eighteen months, reasonably ask who approved this and why. When the VP opens the October case and sees that in July, for the same situation, he approved eighteen, and that his operations head recommended twelve for these specific reasons, and that he overruled for these specific reasons, then October takes thirty seconds and the two supervisors have nothing to compare except two answers that make sense together.

A preserved record of the perspectives that shaped the decision, not just the outcome. The operations head's twelve-percent recommendation lives next to the VP's eighteen-percent approval, with both sets of reasoning. Not so the VP can be checked. So the VP can see what he weighed last time, and so the operations head knows that her strongest argument will outlast the meeting. She brings her best thinking every time because the architecture tells her it matters.

A visible pattern, so the leader can see their own track record. Not as a judgment. As a tool. The VP who can see his last five decisions in this category, with their reasoning and their context, does not need a coach to be consistent. He has something better than a coach: his own prior judgment, preserved and retrievable, which is the only thing that can make the four hundred and first decision look like it came from the same person as the four hundredth.

The investment is not in technology. Not in process. Not in governance. It is in making the building remember what happened inside it, so that the people making decisions are not working from a blank sheet every time.

The return

Here is why this is an investment and not a cost.

Every rupee spent on a decision that contradicts a previous one is a rupee wasted. Every hour a senior person spends managing upward instead of working outward is an hour lost. Every quarter where the operations head brings a softer argument because the architecture told her the strong one would dissolve is a quarter of diminished judgment. Every grievance that arises from an inconsistency that nobody can explain is a trust withdrawal from a bank that takes years to build and months to empty.

These are not hypothetical. They are running right now, in most companies past about a hundred people, and they compound quarterly.

A company that invests in its decision architecture, that gives its leaders a memory and its advisers a voice that survives the meeting, does not just stop the leaks. It gets something it has never had: the ability to compound its own judgment over time. Decision number four hundred and one is better than decision number four, not because the leader got smarter, but because four hundred prior decisions are visible, with their reasoning and their context and their dissent, and the four hundred and first stands on all of them instead of standing alone.

That is the difference between a company that gets older and a company that gets better. The building is the same. The architecture is what changes.

Part of The Other Books, an ongoing series on the decisions companies forget.