Why Leadership Decisions Destroy Business Value: Local Optimization and Hidden Anti-Value

Full text

1 Introduction

Many organisations show strong reported performance while their long-term ability to deliver value erodes. Revenue grows, dashboards turn green, each department hits its targets. At the same time coordination costs rise, customers grow less patient and risk builds up. Leaders sense the gap between measured success and organisational health, but their instruments do not show where it comes from.

This paper asks why leadership decisions that look correct within established metrics destroy business value. The answer proposed is that the cause usually lies not in poor execution or poor judgement, but in the value model a decision is evaluated against. Decisions are judged by visible outcomes inside the deciding unit. Their effects on value flows across time and across other value systems stay outside the frame.

The paper is part of a series on applying the Theory of Value Management (VMT) and the Æilus methodology in business. The vocabulary (value, anti-value, value flow, value system, value transformer, planned, realized and retrospective value) is defined in the first paper of the series (Bolshakov, ÆVRI, 2026) and is used here without re-derivation. The formalisation of the value currency Æ is given in "Value Management Theory and Employee-Centric Competitive Advantage: A System-Level Perspective" (Bolshakov, ÆVRI, 2026). OKRs and KPIs appear here only as the frame in which decisions are usually validated; they are treated in detail later in the series.

The paper treats leadership decisions as meta-converters, explains three mechanisms by which their anti-value stays hidden, and proposes a four-step decision model. It is conceptual and contains no measured data; the scenarios in Section 7 are hypothetical.

2 The problem: green indicators, losing company

2.1 A familiar pattern

Mismatched interests inside an organisation are ordinary. Each unit pursues its own efficiency and can lose sight of the efficiency of the company. High efficiency in one unit often appears as low results in another. A design team that wants to move fast needs complete upstream analysis; the analysis team, measured on throughput, ships what it can. Each is efficient on its own terms. The joint delivery suffers.

Management research knows this pattern. Goodhart (1975) observed that a regularity tends to collapse once it is used for control, and Campbell (1979) made a similar point about social indicators. Systems thinking (Meadows, 2008) and the theory of constraints (Goldratt, 1984) show that optimising a part does not optimise the whole. These works describe the symptom well. They do not provide one language in which the losses can be named, located and attributed to the decision that caused them.

2.2 Why more metrics do not solve it

The usual response is to add indicators, each again owned and evaluated inside a unit. A decision is still validated by those who made it, against indicators of the part of the world they control, over the horizon of their reporting cycle.

VMT reframes the question. It asks which value system a decision is evaluated in, which value flows it changes, and where the resulting value and anti-value are realized. In that frame "green indicators, losing company" is not a paradox. It is the expected outcome of evaluating decisions inside boundaries narrower than the flows they affect.

3 Leadership decisions as meta-converters

3.1 From transformer to meta-converter

In VMT a transformer is a role: an actor that consumes value and produces new value (A7, transformation). In a Value System Schema (VSS) this role is drawn as a converter connected to incoming and outgoing flows. An operational decision acts within one transformer. A leadership decision acts on the configuration of the system: it sets incentives, allocates resources between transformers, moves boundaries, chooses which indicators count and decides which flows are funded. This paper calls such a decision a meta-converter. It consumes value (attention, budget, trust) and produces changes in the structure of many flows at once.

3.2 Why meta-converters scale anti-value

Because a meta-converter acts on many flows, its anti-value is not produced in one place. A target that rewards a support team for short handling time changes the behaviour of every agent in every interaction. If the target creates anti-value for customers or for the engineers who receive escalations, that anti-value is produced at the scale of the flow, not of the decision. Growth multiplies the flows a decision touches. The decision is taken once and usually evaluated once; its effects are produced continuously. This is why anti-value from leadership decisions can grow faster than visible results in growing organisations.

3.3 Whose value is measured

Every statement here is made relative to a named value system, as A1 (contextuality) requires. Three systems are used throughout:

• The deciding unit: the department or leadership team that makes the decision and owns the indicators it is judged by.
• The company value system: all transformers of the organisation and the flows between them, with customers and suppliers as external actors.
• The customer value system: the customer as an actor who realizes value from the company's products, with its own criteria of improvement and degradation.

A decision may produce realized value in the first system and anti-value in the second or third. T2 (flow inversion) states that one flow can carry opposite interpretations in different systems. T15 states that changing the system boundary changes the conclusions without any change in reality. Much of what follows applies these two results.

4 How anti-value hides

By A5, anti-value is an independent category, not an absence of value. A decision can produce real value and real anti-value at once. The problem is that anti-value is usually less visible. Three mechanisms of concealment can be distinguished.

4.1 Lagged anti-value

Planned and realized value do not coincide (A4), and zero divergence is an unstable special case (T3). A decision is typically evaluated on its planned value and its first realized effects; its anti-value is often realized later. T4 states that flow resistance accumulates historically, and T6 that accumulated value distortion reduces the capacity to realize future value. A decision that skips a review step, defers maintenance or thins a support function can deliver its planned saving now while adding resistance that is realized as anti-value over many later periods. By then the decision has been validated and rewarded. Retrospective value is the category that captures this: a decision judged positive at the time can be re-evaluated later as negative. Without that re-evaluation, lagged anti-value is never attributed to its source.

4.2 Displaced anti-value

By A10, value systems interact only through external flows, and by T10 a system's state depends on inflows and outflows to other systems. A unit can improve its own balance by pushing anti-value into a flow that leaves its boundary: shifting work to internal customers, lowering the quality of hand-overs, or promising what delivery cannot provide. The deciding unit realizes value; the receiving system realizes anti-value. This is the content of T5: local optimisation does not guarantee global improvement. The Æilus principle that answers it is to reduce flow resistance, not shift it.

Displacement can cross the company boundary. A decision can be positive for the company and negative for the customer. When the customer's participation condition (A8) is no longer met, T7 applies: the actor leaves or changes its form of participation. The anti-value returns as lost revenue, with a lag and without a visible link to its cause.

4.3 Anti-value invisible to the deciding unit

The deciding unit cannot see anti-value realized by actors it does not observe, in flows it does not model. T9 states that if structure does not explain stability, there are latent flows or the boundaries are wrong. For leadership the corollary matters: if a decision appears to have no cost, the cost is probably realized outside the model. T14 adds that value without interpretation degrades into lost value; anti-value that nobody interprets is likewise unrecorded and unmanaged. T17 states that interpretational distortion, where influential sources make claims without confirmation by realization, accumulates anti-value. A leadership team that validates its decisions through its own reports is such a source.

4.4 Summary

5 Efficiency as a profile

Local optimisation is usually justified by efficiency, a word applied to people, teams, businesses and states without saying efficiency of what and at whose expense. Æilus separates the levels of transformer, value flow and value system (and value element, for which "efficiency" is the wrong question). Confusing the levels is one way leadership decisions go wrong.

5.1 The transformer: the only level with a formula

Efficiency = (produced value - produced anti-value) / (consumed value - consumed anti-value)

The formula applies to people, teams, services and departments, inside a specific value system or without reference to one. A transformer need not be efficient in all systems at once: a department can be highly efficient in its own reporting system and destructive in the company system. Æilus keeps these figures separate rather than averaging them. The formula also shows why local figures mislead. A unit that leaves out anti-value realized elsewhere reports a higher efficiency than it has. The ratio is only as honest as the boundary used to count anti-value.

5.2 The flow: convergence, losses, dynamics

A value flow delivers value over time in a required volume and form. Its efficiency has three components: convergence of planned and realized value, losses (delays, distortions, transaction costs, anti-value) and dynamics (stable or growing value). A large flow can be inefficient if it is unpredictable. Shared steps need special care. Authorisation in a digital product, or an API gateway, produces no value of its own but takes part in every flow; removing it stops all of them. Its efficiency can be judged only by the aggregate value of the flows it serves.

5.3 The value system: balance, growth, structural fitness

A value system has no single efficiency coefficient. It is under no obligation to be efficient, stable or useful; it can degrade, act as a parasite or exist only temporarily. What it has is a viability profile:

• Balance: can the system feed its transformers without permanent deficit, growing anti-value or exhaustion of internal resources?
• Growth: is there positive dynamics in the volume of flows, the capacity of transformers or total system value?
• Structural fitness: is the system resilient to failures, free of critical bottlenecks and able to adapt?

Every system also has a survival threshold, a minimum level of flows below which its configuration cannot reproduce itself. Below it the system does not become inefficient; it begins to die.

The profile view matters for leadership. A decision that raises growth while lowering balance or structural fitness is neither efficient nor inefficient as such. T13 states that there is no universal criterion of optimality. The decision is a trade-off, and the question is whether it was made explicitly by someone responsible for the whole profile, or implicitly through a local indicator.

6 A value-based decision model

The model applies to any leadership decision before it is taken and again after it has had time to act. It adds to existing performance frameworks a layer that names whose value is at stake and who confirms it.

6.1 Step 1: name the value system

State the value system in which the decision is evaluated: actors, flows, criteria of improvement and degradation, interpretation rules and time horizon. Then state which other systems the decision touches; the minimum set is usually the deciding unit, the company and the customer. A decision evaluated in only one of them is incompletely evaluated by construction (A1, A2, T15).

6.2 Step 2: map the affected flows

Identify every flow whose volume, form, timing or predictability the decision changes. In a VSS the decision is placed as a meta-converter and the flows it reshapes are marked. Attention goes to flows that cross the unit's boundary, to shared steps, and to flows whose consumers are not represented in the decision forum (T9).

6.3 Step 3: estimate planned value and planned anti-value

For each affected flow, state the planned value and planned anti-value, the system where each is expected to be realized, and when. Estimates may use Æ where value and anti-value are commensurable. Where they are only partially commensurable or incommensurable (A13), the trade-off is stated in words rather than forced into one figure. Anti-value is estimated as a separate quantity (A5), not netted silently against value.

6.4 Step 4: define who validates realized value, and when

For each flow, name the actor who will confirm realized value and anti-value, and the time at which they will do so. The principle is direct: the output value of a process must be assessed by the one who consumes it, checked by someone else, and reassessed regularly because contexts change. This step breaks the loop in which the deciding unit validates itself (T17), schedules a retrospective re-evaluation that can catch lagged anti-value (T4, T6), and gives receiving systems a voice before their participation condition fails (A8, T7). T16 states that scaling through interpretation without confirmation by realization is structurally unstable; Step 4 is the minimum structure that prevents it.

7 Illustrative scenarios

The following scenarios are hypothetical. They are constructed to show how the mechanisms and the model apply, and they do not describe any real organisation or report any observation or result.

7.1 Hypothetical scenario A: support optimised for handling time

Leadership in a hypothetical software company sets a target to reduce the average handling time of support requests. In the support unit's value system the decision works: more requests closed per agent, lower cost per request. Step 1 adds the customer system and the engineering system that receives escalations. Step 2 shows two flows crossing the support boundary: resolved problems to customers and escalations to engineering. Step 3 records that requests closed without resolution may return as repeat contacts, and incomplete escalations may raise diagnostic load on engineers: displaced and partly lagged anti-value (T5, T10, T4). Step 4 assigns validation to the customer, through confirmation that the problem was solved, and to engineering, through a review of escalation quality after a set period. The decision may still be taken, but as an explicit trade-off with a scheduled check.

7.2 Hypothetical scenario B: product teams rewarded for output

A hypothetical product organisation rewards teams for features delivered per quarter. In the teams' system, realized value is high. Step 2 maps the reshaped flows: features to users, maintenance load to the platform team, complexity into the product. Step 3 notes that a feature's value for users is realized only when it is used (A6), while its maintenance cost continues regardless. Unused features become elements whose potential degrades into lost value (T14) while their cost remains. The anti-value is lagged (T6) and largely invisible to the teams, because users who ignore a feature send no signal into the teams' indicators (T9). Step 4 moves validation to users and the platform team, with a retrospective review of delivered features.

7.3 Hypothetical scenario C: cutting a shared step

A hypothetical leadership team proposes to cut the budget of an internal authorisation service used by all customer-facing products and accept lower reliability. Step 1 shows the service cannot be evaluated in its own system, since its value is realized only in the flows it serves. Step 2 finds that every product flow passes through it. Step 3 estimates that a small loss of reliability produces anti-value in every flow at once, and that sudden unpredictability costs customer trust. In profile terms the proposal improves balance in one cost centre and lowers structural fitness of the whole system (T8: sustainability is lost when a transformer role can no longer be performed). Step 4 assigns validation to the owners of the dependent flows. The model does not decide; it shows that this is a system-level trade-off to be owned at that level.

In all three scenarios the original decision is reasonable within its frame, and nobody needs to be incompetent. Value is destroyed by the evaluation frame: one system, one indicator, one point in time, validated by the decider.

8 Implications for leadership practice and governance

For leaders. The question "is this decision efficient?" is replaced by more precise ones: in which value system, through which flows, above or below the survival threshold, and at whose expense? Because leadership decisions are meta-converters, the cost of a weak value model grows with the organisation. Leaders of growing organisations gain most from naming the value system early.

For governance. Three structural requirements follow. Someone must be responsible for the value system as a whole: in Æilus, the Value System Owner (VSO), whose scope includes system-level anti-value, persistent outflows and protection against collapse through local optimisation. Each transformer needs an owner of its realized value and of the anti-value it produces: the Value Transformer Owner (VTO), who is obliged to increase realized value and reduce anti-value, not to keep inflating planned value. And the organisation needs regular events where planned, realized and retrospective interpretations of key flows are compared, so that interpretational distortion is detected. These roles and processes are the subject of a later paper in the series; without them the decision model in Section 6 has no owner.

For indicators. The analysis does not argue for removing KPIs or OKRs, but for placing each indicator inside a value frame, linked to the value system it measures and to the flows it can distort.

On ethics. A value system can be efficient internally and still produce anti-value for other systems. Æilus records this structurally and separates it from moral judgement, which requires a different level of analysis and is outside the scope of this paper.

9 Limitations and research agenda

This paper is conceptual. Its claims are derived from VMT axioms and theorems and the Æilus methodology and illustrated with hypothetical scenarios. It reports no empirical data and does not show how often the concealment mechanisms occur, how large hidden anti-value typically is, or whether the model improves outcomes. Estimating planned anti-value requires knowledge the deciding unit may lack. Consumer validation depends on consumers' willingness and ability to report. Aggregation in Æ across partially commensurable domains rests on a policy that can itself be distorted. And the model adds cost; for small, reversible decisions the cost may exceed the benefit.

Open research directions include:

• retrospective re-evaluation of recorded leadership decisions, comparing planned with realized and retrospective value across several value systems;
• prospective application of the four steps in participating organisations, recording whether anti-value identified in Step 3 is confirmed in Step 4;
• the typical lag between a decision and the realization of its anti-value, by type of decision and flow;
• VSS-based tracing of which boundaries anti-value most often crosses, and which structures favour displacement.

10 Conclusion

Leadership decisions destroy business value not mainly because leaders are incompetent, but because the value models used to evaluate decisions are inadequate. Decisions are meta-converters: they reshape many flows at once, and their anti-value is produced at the scale of those flows. That anti-value is lagged, displaced or invisible to the deciding unit, mechanisms grounded in A4, A5, A10, T4, T5, T6, T9, T15 and T17. Efficiency is a profile rather than a single ratio. The value-based decision model follows: name the value system, map the affected flows, estimate planned value and anti-value separately, and define who validates realized value and when. It does not decide for leaders. It makes the cost of a decision visible in the system where it will be paid.

References

• Bolshakov, V. Value Management Theory and Employee-Centric Competitive Advantage: A System-Level Perspective. ÆVRI, 2026.
• Bolshakov, V. Value, Anti-Value and Value Flows: A Working Vocabulary for Practitioners. ÆVRI, 2026.
• Bolshakov, V. Assessing the Value of Processes: Efficiency, Risks, and Anti-Value. aeilus.tech, 2025.
• Bolshakov, V. Rethinking Efficiency: Value Systems, Flows, Converters, and Elements. aeilus.tech, 2026.
• Campbell, D. T. Assessing the Impact of Planned Social Change. 1979.
• Goldratt, E. M. The Goal. 1984.
• Goodhart, C. Problems of Monetary Management: The U.K. Experience. 1975.
• Meadows, D. Thinking in Systems. 2008.