Chapter 047　What Does It Mean to Redefine the Organization?

What does it mean to redefine an organization? In most companies the question is quietly swapped for a different one: how should we redraw the chart? Boxes are merged, lines are moved, names are changed. A year later the front line works almost exactly as it worked before. What an organization looks like in the Age of AI has already been set out (→ Vol. II, Ch. 011). This chapter is about the act, not the state. On what criterion, in what order, and at what speed do we rebuild the organization we already have? Redefining an organization means answering those three questions.

1 The question — why it arises now

Reorganization is the most frequently executed act in management. Most companies merge, split, and rename units every year, or at least every few years. It is also the most visible evidence that management is doing something. Almost all of it is restructuring, not redefinition. The two are different acts. Management usage has called both of them “changing the organization,” and that single phrase has produced an enormous number of misses. The evidence is easy to find. After the change, did the agenda of the management meeting change? Did the path an approval travels change? Did what earns praise at year-end change? If none of the three moved, only the diagram moved. Until recently this did little harm. Changing the diagram stirs the air inside a company. It serves a purpose as a rotation of people. While the environment shifted slowly, a company could miss several times without taking a fatal wound. AI changed two things. First, the object that needs changing moved outside the structure. What used to determine an organization’s performance was the arrangement of who reports to whom. What determines it now is who can see what, and what gets rewarded. The cost of moving information has fallen by orders of magnitude. Rearrange the boxes while information and evaluation stay where they were, and performance does not move. Second, the organization acquired participants who hold no employment contract. AI agents are ceasing to be tools that wait for instructions. Given an objective, they work across process boundaries. Command lines, the location of responsibility, and the unit of evaluation all break down when the design assumes human beings only. When those two conditions overlap, a company can restructure repeatedly and still not move. The restructuring is worse than useless, because it buys the belief that something changed. So the question changes shape. It is not how to rearrange. It is what has to change for an organization to actually change — and in what order, and at what speed.

2 Conventional answers and their limits

Three answers circulate. Each is partly right. Each treats the organization as a single layer. The first answer: “Changing the organization means redrawing the chart” This is the most widely held assumption. A reorganization proposal reaches the board as one before-and-after diagram. The discussion turns on how many boxes disappear and how the reporting lines move. The work is not pointless. How the boxes are cut determines who talks to whom every day. People inside the same unit share information naturally; people in different units do not. Placement shapes behavior through the frequency of contact. But the effect is weak. Placement determines the range of possible behavior, not its direction. Direction is set by what gets rewarded. Change the boxes while the rewards stay put, and people carry on optimizing the old thing inside the new box. Within a few months the informal circuits come back. People ask the old counterpart directly and judge by the old standard. On paper a new organization is running. In substance the old one never left. This is what always happens when only the chart is redrawn. The second answer: “Change the structure first, and mindset will follow” The second answer puts structure first. Changing what people think is hard, so change the container, force the behavior, and let attitudes follow behavior. There is something to this. Behavior preceding attitude is observed on many front lines. Radical reorganizations have in fact triggered genuine shifts in mindset. But that happened only where evaluation changed at the same time as structure. The new unit was given a new definition of success. A structural change with the old evaluation intact forces nothing. Only the form is compelled; the real incentive stays where it was. Structure cannot beat evaluation. This is not a moral claim. It is a consequence of incentive design. The third answer: “Organizations change when mindset and culture change” The third answer takes the opposite position. Changing the container does not change people, so talk about values, run workshops, and build occasions for dialogue. Start there. The direction is right. Redefinition ends, finally, in a change to the criteria by which people judge. It cannot be achieved without speaking to people. Speaking is not enough. People respond to what is actually measured, not to what is proclaimed. Declare that risk-taking is encouraged, then ask only for explanations of missed targets at year-end, and the organization will not take risks. When the words and the measurements disagree, an organization always believes the measurements. The three conventional answers share one limit. All three treat the organization as a single layer. Each assumes that moving structure alone, or mindset alone, moves the whole. Organizations are not built that way.

3 Redefinition — redefining an organization means

realigning four layers

3.1 What separates restructuring from redefinition

Start by separating the two acts. Restructuring rearranges who does the work while treating the objective as given. What counts as an outcome does not change. Who delivers it does. Efficiency can rise. Duplication can fall. What the enterprise is for does not move. Redefinition changes what the organization exists to do, what it takes as its unit, who decides what, and what it calls good. The definition of an outcome is itself rewritten. The same people and the same assets then produce something different. Enterprise Redefinition takes the enterprise in five dimensions: Purpose, Business, Organization, Capital, and Leadership. The definitions belong to another chapter (→ Vol. V, Ch. 041). Redefining the organization is the third of the five. It is one dimension, not a standalone initiative. Organizational redefinition unaccompanied by redefinition of the business usually decays into restructuring. Three questions are enough to tell them apart. Did the criterion for cutting units change? Did the range of what can be decided change? Did the substance of what is evaluated change? Three noes mean restructuring. One yes means redefinition has started. One clarification, to be safe. Restructuring is not an inferior act. It is often necessary. What is dangerous is restructuring while believing you have redefined.

3.2 The four layers of an organization

To redefine an organization you have to take apart what an organization is made of. We treat it as four layers. The first layer is structure. Who belongs to the same unit as whom. How units are cut, and where they report. This is the only layer a chart can draw. The second layer is authority. Who can decide what. Up to what amount, across what scope, without whose approval. And who can overturn a decision once made. The third layer is the flow of information. Who can see what. Which numbers reach whom, at what granularity, at what speed. People cannot decide about what they cannot see. The fourth layer is evaluation. What is called good, and what is called into question. Promotion criteria, the basis for bonuses, the stories praised in meetings and the stories criticized. Not only the formal appraisal system: how executives spend their time belongs to this layer too. The four are ordered by depth. The higher the layer, the more visible, the easier to change, and the smaller the effect of changing it. The lower the layer, the more invisible, the harder to change, and the larger the effect. Reorganization debates concentrate on structure because structure is the only layer that fits on a sheet of paper.

3.3 Which layer, left alone, drags everything back

The answer is plain. Unless evaluation changes, an organization always returns to what it was. The argument is short. An organization is a set of behaviors, human and AI. Behavior flows toward what is rewarded. Structure is the vessel that holds the flow. Change the shape of the vessel and the water still runs downhill. Evaluation is the slope. So in an organization where only structure moved, events follow a fixed order. For the first few weeks people work to the new chart. Over the next few months they notice that the new chart makes results harder to produce. Being rational, they start using the old routes that do produce results. A year later the old order has recrystallized inside the new boxes. Take a concrete case. A company creates cross-business project units. The purpose is value creation across departmental lines. But participants are still evaluated on the budget attainment of the department they came from. What will they do? They will protect their home department’s margin ahead of the project’s success. Given the design, that is the correct behavior. The cause of failure is not the participants’ attitude. It is the design. Make the structure new while the evaluation stays old and this outcome is guaranteed. Reverse it — change the evaluation first — and people start moving even inside the old structure. They begin looking for their own routes to the new result. That does not license changing evaluation and leaving the rest. If evaluation changes but information does not arrive, people cannot confirm whether they are being evaluated at all. Without authority, they cannot perform the behavior being evaluated. With the old structure intact, they cannot reach the people they now need daily. In the end the four layers have to agree. So the accurate statement is this. Evaluation is the layer that drags everything back if it is not changed. The other three layers exist to be brought into alignment with it.

3.4 Why this layer structure bites harder in the Age of AI

Apply the Value Equation to the organization. Value = Purpose × Trust × Capability × Time The relationship is multiplicative. Because the relationship is multiplicative, value without purpose has no direction, without trust cannot spread through society, without capability cannot be realized, and without time cannot endure. If any single term is zero, the whole product is zero. Redefining the four layers bears most heavily on Trust. When evaluation cannot be accepted as fair, trust drains out of an organization quietly. In a company whose declarations and measurements disagree, everything management says is received at a discount. And the more AI enters, the harder evaluation becomes to design. Human workload stops functioning as a proxy for output. Items processed, hours worked, and headcount assigned all lose meaning as AI takes a larger share. An organization whose proxy has broken while its evaluation stays old falls into confusion, trying to measure what it can no longer measure.

4 Structure — sequence, and designing an organization

that includes AI

4.1 The design sequence is not the execution sequence

Here is the sequence most companies actually follow. Structure → Authority → Information → Evaluation Decide the structure, rewrite the approval thresholds to match, arrange the reports that are now needed, and finally ask HR to revise the appraisal system. Evaluation becomes a task for the following fiscal year, and is often forgotten there. Under this sequence the organization returns, for the reason given above. The design sequence runs the other way. Question → Evaluation → Information → Authority → Structure The starting point is the question this organization exists to answer. What is it to accomplish? Next, decide how you will confirm that the question has been answered. That is evaluation. Once evaluation is fixed, the information needed for confirmation is fixed. Once information is fixed, what the people who see it must be able to decide is fixed. Only once authority is fixed does the cutting of units follow. The chart is drawn last, as the conclusion of the design. It is not the starting point of a design; it is the record of one (→ Vol. II, Ch. 011). This is not the order of execution. Revising an appraisal system takes time. HR systems touch conditions of employment, so consultation procedures apply. In practice, design starts from evaluation while implementation often starts from structure. What matters is that when structure is implemented first, the design of evaluation is already complete. Moving structure toward an evaluation that has not been designed is the single largest failure in organizational redefinition.

4.2 When AI becomes a component, three things are rebuilt

When AI agents become components of the organization, three of the four layers are rebuilt from the ground: command, responsibility, and evaluation. Command. Command does not work on AI. More precisely, the form of a chain of orders carries no meaning. In a human organization, subordinates interpret a superior’s intent and fill the gaps according to circumstances. That filling does not happen with AI. What works is a written objective, a written constraint, and a written boundary of authority. Designing a unit that includes AI is therefore not a matter of drawing command lines. It is writing down what is to be maximized, what may not be crossed, and where control returns to a human being. Most organizations have run these statements tacitly. AI does not read the tacit. For the first time, an organization’s tacit knowledge is forced into writing. Responsibility. AI cannot take on responsibility. This is not a performance limit. First Principle 4 states it: AI Optimizes. Humans Define. AI optimizes; humans define value, purpose, and direction. Definition carries responsibility, and responsibility attaches only to people. So the design rule is single. Assign to every process AI handles the name of one human being who owns the result. Not joint responsibility. Not a committee. A process where one name cannot be written is not yet part of the organization. Put AI into a blank in responsibility and nobody looks at it until an accident happens. Evaluation. This is the hardest of the three. Whose result is a result produced by AI? Three errors are common. The first is to book it nowhere. AI’s cost vanishes into an expense line and the output belongs to no one. Then the person who used AI to produce a result is not rewarded for it. Rational employees stop using AI. The second is to evaluate AI utilization itself. The means becomes the end, and AI is forced into processes that never needed it. Early in a deployment this is sometimes unavoidable. It must not run past a year. The third is to make people and AI compete on the same yardstick. On processing speed, humans always lose. Designing that contest pushes the human role back toward doing the work. There is one right direction. Move the object of human evaluation from volume of work to the value produced by the territory a person holds, and to the design of that territory. What was connected to what, where a human judged, and what was handed to AI. The quality of that design is what evaluation will be about. And the concept of headcount breaks. Number of people and processing capacity stop being proportional. Measuring the size of an organization in people carries no information in the Age of AI. What should be measured is the size of the question a unit can take on.

4.3 Redefinition is an act of design by the executive

Place the Leadership Formula. Leadership = Purpose × Question Design × Capital Allocation × System Architecture × Trust The fourth term, System Architecture, is organizational redefinition itself. This equation is multiplicative as well. If that term is zero, no size of the other four makes management work. Note what the term is not. It is not an HR term. Two of the four layers, authority and evaluation, can be changed by no one but the executive team. What HR can design is the vessel of the system, not the criterion of what counts as good. Setting the criterion is the executive’s work.

5 What it looks like in practice — speed, and where to

begin

5.1 An organization cannot be changed all at once

If all four layers could be changed simultaneously across the whole company, this would be a short chapter. It cannot be done. There are three reasons. First, the constraint of learning. Understanding what a new evaluation means takes people time. Distributing a policy document does not transmit it. People believe a criterion only after they have seen one cycle actually run under it. The change in evaluation therefore takes at least one full cycle to reach behavior. Second, the constraint of trust. A change in evaluation always destroys someone’s standing. The people who scored highest under the old criterion score lower under the new one. Push forward without explaining that pain, and what is lost is not confidence in the system. It is confidence in management. Third, the constraint of operations. Companies have to change while running. You cannot stop delivering to customers in order to rebuild the organization. Dismantle all four layers at once and there is a gap between the disappearance of the old order and the arrival of the new. During that gap, judgment stops. So the practical question of organizational redefinition becomes: where do we begin, and how fast do we widen?

5.2 Change one flow deeply rather than the whole company

shallowly There is a principle for choosing the starting point. Take the one flow of value that is most congested. Not the company. One flow. For that flow, align all four layers completely. Set the question, rebuild the evaluation, open the information, move the authority, and cut the unit last. The scope may be narrow. The depth may not be compromised. Why does this shape work? Because redefinition needs evidence. The fact that results actually appeared under the new evaluation is what persuades the next scope. Change the whole company shallowly and no evidence is produced. Everyone pretends to change a little and nobody can show a result. A word on speed. In what we observe across many companies, structural change can be implemented within a quarter. Delegation of authority takes about six months before it operates in practice. Rebuilding the flow of information usually requires system work, so allow around a year. Changing evaluation takes half a year to design and more than a year from implementation to settlement. That asymmetry is what creates the sequencing problem. The layer with the greatest effect is the layer that takes the longest. Unless you start from the design of evaluation, you will not arrive in time. Companies that start from structure return to the old order before evaluation catches up. Once the first flow moves, widen to the second. For many companies one to two flows every six months is a realistic pace. It varies with size and industry. What matters is not the count but that all four layers are present in each one.

5.3 Five shapes of failure

When organizational redefinition fails, the shapes are remarkably alike. Shape 1: redrawing the chart and nothing else. The most common. Box names change and reporting lines are redrawn. The approval path, the visible numbers, and the substance of evaluation do not. Six months later people are working the old routes. This is not redefinition, and it is not even restructuring. It is renaming. Shape 2: creating cross-cutting units while leaving evaluation vertical. As described above. Participants protect their home department, and management concludes that “collaboration is not in our culture.” What is missing is not culture. It is design. Shape 3: pushing authority down without pushing information down. The company declares that the front line will decide. But the costs and the full customer picture needed for the judgment remain visible only at headquarters. The front line cannot decide, so it asks upward anyway. Delegation survives in form, and decisions get slower than before. Shape 4: changing the language of evaluation without building the instrument. The company announces that risk-taking will be rewarded. At year-end the only figures actually compiled are the old ones. Evaluation becomes a double ledger, and people believe the ledger that is compiled. Within a year nobody repeats the announced words. Shape 5: changing everything at once. The trap for ambitious executives. Structure, authority, information, and evaluation all switch on the same April 1. If any one of them slips, the whole thing stops. Something always slips. After a few months of confusion, most such programs are reverted. All five share one thing: one of the four layers is missing. An organization returns to its old state through the missing layer.

5.4 What an adaptive organization is

The Enterprise Redefinition Maturity Model (ERMM) puts the Organization dimension as a single question. “Can structures adapt rapidly to technological change?” That question summarizes everything above. What determines the speed of adaptation is not the number of reporting lines. It is whether the four layers are written down and consistent with one another. An organization running on tacit arrangements has to begin every attempt at change by decoding its own present state. Level 4, the Continuous Redefinition Enterprise, lists adaptive structures among its characteristics. Adaptive does not mean soft. It means that the procedure for rearrangement has been designed in advance. It is the state in which management knows what changing one thing will affect. The gap from Level 3, the Transformation Enterprise, shows up exactly here. A Level 3 enterprise can also change its organization substantially. But it treats each change as a one-off project. So the next change starts again from nothing. Three notes travel with the model and must not be dropped. Progression is not linear: organizations frequently display characteristics from multiple levels simultaneously, and an enterprise may hold Level 4 AI capability while remaining Level 2 in leadership. The model evaluates organizational coherence rather than isolated excellence. Maturity is read across all five dimensions in balance, because exceptional technological capability with weak leadership redesign cannot produce higher maturity. And Level 5 is not a target to be reached as fast as possible; different industries require different levels of organizational adaptability.

6 Questions for the executive

The argument, in one line. Redefining an organization is not redrawing the chart. It is realigning four layers — structure, authority, the flow of information, and evaluation — toward a new question. At the center of that work sits evaluation. An organization that changes structure without changing evaluation always returns. This is not a matter of will. It is a consequence of design. First Principle 6 states it. Enterprise Exists to Redefine Itself. Continuous self-redefinition is its essence. The organization is the vessel through which that redefinition is carried out, and the vessel is itself an object of redefinition. Three questions to close. Each can be answered at your next executive meeting. Question 1 — In your most recent reorganization, did the substance of what is evaluated change? What will the head of the new unit be asked about a year from now? Is it different from what the head of the old unit was asked? If it is not, the change was restructuring. That is not wrong in itself. Believing it was redefinition is what is dangerous. Question 2 — For every process AI now handles, is there the name of a human being who owns the result? List the processes and write in the names. Blank cells will remain. Those cells are the blanks in your organization. A blank is not found until an accident happens. And when the accident happens, nobody can take it on. Question 3 — How many redefinitions with all four layers in place are running right now? Counting company-wide initiatives spread thinly tells you nothing. Count the flows in which question, evaluation, information, authority, and structure are all present. Zero means the organization has not begun. One means there is something to widen from. None of the three questions asks about the shape of the organization. All three ask what this organization rewards. Structure is visible. So management wants to touch it. What actually moves an organization sits in the layers you cannot see. Who can see what. Who can decide what. And what is called good. AI can help with this design. It can make the current flow of information visible, test whether evaluation metrics are consistent, and estimate the reach of a proposed change. It cannot decide what counts as good. That is a choice of value, and a choice of value carries responsibility. Redefining an organization means taking that responsibility and deciding again what will be rewarded.

In brief

Redefining an organization is not redrawing the chart; it is realigning four layers.
The four layers are structure, authority, the flow of information, and evaluation. The lower the layer, the less visible it is and the larger the effect of changing it.
Unless evaluation changes, an organization always returns. This is a consequence of design, not of will.
The more AI takes over processes, the further the old assumption collapses that workload is a proxy for output.

Key concepts

Enterprise Redefinition / Value Equation / Question Design / Future Value / Enterprise Redefinition Maturity Model (ERMM)

The chain of ideas

Purpose → redesign of evaluation → alignment of authority, information, and structure → Trust → Future Value

Related first principles

Principle 6 — Enterprise Exists to Redefine Itself. Principle 4 — AI Optimizes. Humans Define. Principle 9 — Leadership Means Designing the Future.

Related chapters

Vol. II, Ch. 011 “What Is an Organization in the Age of AI?” — draws the whole picture of the organization this chapter acts on
Vol. V, Ch. 048 “What Does It Mean to Redefine Talent?” — reopens the question of who the organization is made of
Vol. V, Ch. 041 “What Is Enterprise Redefinition?” — why Organization sits third among the five dimensions
Vol. VI, Ch. 054 “What Does It Mean to Redefine Corporate Culture?” — how a change in evaluation settles into culture

Papers and companion volumes

Kadowaki, N. (2026a). Future Value Theory: A Management Framework for Enterprise, Capital, and Society in the Age of AI. VURA Working Paper Series. SSRN: https://ssrn.com/abstract=7120980 / Zenodo: https://doi.org/10.5281/zenodo. 21255662
Kadowaki, N. (2026b). Enterprise Redefinition: Toward an Enterprise Evolution Theory for the Age of AI. VURA Working Paper Series. (Published on Zenodo; under review at SSRN)
100 Questions on Management in the Age of AI, #057 “How Does the Shape of an Organization Change in the Age of AI?” / #044 “Why Do Employees Wait to Be Told?”

Chapter 047 What Does It Mean to Redefine the Organization?