Purchasing reports success. Due to larger purchase volumes, unit prices have decreased, and the cost key figure is the best in years. Production reports success. Machine utilization is at a record high; no expensive machine stands idle. Logistics reports success. Trucks are full, and freight costs per unit have fallen. Three departments, three green key figures, three satisfied managements. Only the overall result is red: inventory levels are high, capital is tied up, delivery times have lengthened, and an important client has left. Everyone fulfilled their task. The company as a whole lost.
A system composed of nothing but peak performances is far from an optimal system. Where each department maximizes its own key figure, the overall system suffers the damage.
An industrial company that I advised on this matter was precisely in this situation. Every function achieved its goals, yet the overall margin declined for three consecutive years. The analysis revealed the correlation that the departmental reports obscured: favorable purchasing prices forced quantities that strained the warehouse. High machine utilization generated semi-finished goods that no one had ordered. Full trucks resulted from shipments being held back until full capacity, leading to some clients being supplied late. Every single saving was real. In total, they cost more than they yielded.
This pattern has a name, and those who understand it manage differently. Three levers lead out of the trap.
Eliyahu Goldratt, with the Theory of Constraints, formulated a simple but profound idea: the throughput of a system is not determined by the sum of its parts, but by its narrowest point, the bottleneck. Any improvement at a point that is not the bottleneck does not increase the throughput of the whole. At best, it fizzles out. At worst, it creates inventory, waiting times, and interface costs that burden the overall result. Goldratt’s core statement for executives is: Tell me how you measure me, and I will tell you how I behave. Those who measure departments by departmental key figures achieve departmental optimization, even if it harms the whole.
Here lies the clear distinction that is constantly confused in practice. A local optimum is the best achievable state of an individual department, measured by its own key figure. An overall optimum is the best achievable state of the system, measured by the common overarching goal. These rarely coincide. Maximizing each individual department regularly leads away from what improves the company as a whole.
| Characteristic | Departmental Optimum | Overall Optimum |
|---|---|---|
| Metric | Utilization and costs of the department | Throughput of the overall system |
| Incentive | everyone maximizes their own | everyone aligns with the bottleneck |
| typical consequence | inventory, waiting times, interface costs | smooth flow to the client |
| Impact on the whole | the sum remains suboptimal | coordinated result |
A look at practice shows that suboptimization rarely arises from negligence. On the contrary, it is the result of diligence, of dedicated departments doing exactly what they are measured and rewarded for. This is precisely what makes it so difficult to resolve. No one acts incorrectly, yet the whole loses.
Before improving any department, clarify two questions: What is the goal of the overall system, and where is the point that limits this goal? Only these two answers give direction to any optimization. An increase at the bottleneck benefits the entire company. An increase at any other point brings nothing to the whole and, as soon as it creates inventory or effort, even adds an extra burden.
This leads to an unusual consequence: departments that are not the bottleneck should not maximize themselves, but rather support the bottleneck. A machine that is not the bottleneck does not have to be fully utilized. A purchasing department that supplies the bottleneck with material serves the whole more than one that achieves the lowest unit price and sacrifices delivery reliability for it. This subordination contradicts every reflex for full utilization, and precisely for this reason, it requires a conscious decision from leadership.
The bottleneck is often not a machine, but a rule. An approval threshold, a reporting cycle, or an outdated metric often limit throughput more than any physical capacity, and precisely such bottlenecks remain undiscovered because they do not belong to any single department. Those who only look for the technical bottleneck overlook the more expensive one in their own rules.
Suboptimization is usually not a behavioral problem, but a measurement problem. As long as a department is measured by its own utilization or cost ratio, it will improve this metric, even at the expense of the whole. The most effective correction, therefore, does not start with an appeal, but with the key figures. Supplement each departmental key figure with a metric that reflects the contribution to the overall result, and in case of conflict, prioritize the overall metric.
This requires measuring what matters, not what benefits the individual department. A plant manager whom I accompanied for several months replaced the central control variable of his production. Instead of machine utilization, he henceforth measured on-time delivery to the client and the lead time from order to delivery. Within half a year, inventories significantly decreased, and the utilization of individual machines consciously declined. Although the plant appeared less utilized on paper, it operated profitably for the first time in years in terms of overall results.
The problem lies deep within traditional cost accounting, which even rewards local overproduction. If production manufactures goods that no one has ordered, unit costs decrease on paper because fixed costs are spread over a larger quantity. Full cost accounting also records this overproduction as inventory assets, i.e., an asset. The balance sheet shows an asset, while dead capital is tied up in the warehouse. Therefore, anyone who wants to eliminate suboptimization must also question this logic of accounting, not just the control key figures.
As long as each department is optimized in isolation, the overall optimum arises by chance, if at all. The solution is not to drive individual departments harder, but to align them with a common goal and manage their transitions. It is there, at the interfaces between departments, that the costs of suboptimization arise, and it is there that they are addressed.
This is primarily a leadership task, not a structural question. It is about building bridges between silos, instead of pitting each department against the next in a race of individual key figures. Where departments align their collaboration across their own boundaries with a shared result, suboptimization loses its breeding ground. The efficiency gained in this way no longer secretly makes the system more expensive, but rather frees up resources that finance future growth.
First: Take the most important efficiency key figure of a single department. Does its increase demonstrably improve the overall result, or only the balance sheet of that department? If you hesitate, you already know the answer.
Second: Where is the bottleneck in your value creation system? And do your other departments support this bottleneck, or do they compete with it for resources and attention?
Third: Take the three most frequently reported departmental key figures in your area of responsibility and check for each whether it measures departmental activity or contribution to the whole. Replace any that only reflect the department before agreeing on the next goals.
A system composed of nothing but partial optima is not an optimal system, but an expensive one. The leadership task is not to drive every department to peak performance, but to define the common goal and the bottleneck and to put every departmental key figure at their service. Efficiency that only shines locally is unfortunately not a strength but a shift of inefficiency to another level.
The most expensive saving is therefore the one that is appreciated in one department and implicitly reappears as higher expenditure in the next. It does not appear in any departmental report because it arises between departments.
Efficiency is not effectiveness – Before optimizing a department, the question is whether its activity serves the whole at all.
Operational Excellence – Operational strength arises from coordinated interaction, not from the sum of isolated peak performances.
All Insights can be found in the overview.
Proven tools and models for self-application are available under Solutions.
If you want to take these thoughts further for your company, a no-obligation initial conversation is worthwhile.
