Efficiency is usually treated as an unquestioned good: machines should not sit idle, people should not wait, and every minute should be used. In an interdependent system, however, keeping every local resource busy often moves waste instead of removing it.
Activity is not throughput#
When one step finishes work that the next step cannot accept, the apparent output becomes inventory, queues, and coordination overhead. The utilization report improves while the result customers need arrives no sooner.
Activity asks whether a resource is moving. Throughput asks whether the whole system is getting closer to its goal. The two measures agree only when work can flow through the entire system.
High utilization magnifies variation#
A resource operating near full capacity has little room for ordinary variation. One rework loop, urgent request, or brief outage creates a queue that propagates downstream. A little spare capacity can therefore be a source of resilience rather than waste.
The more uncertain the work, the more valuable that room becomes. A perfectly loaded plan may look disciplined while producing unreliable delivery.
Metrics shape local behavior#
When teams are judged only by individual output, machine uptime, or departmental cost, each local group rationally protects its own number. It produces early, refuses interruptions, and pushes problems across the boundary. Nobody intends to damage the whole, yet the combined result can be worse.
A better question is not “who is still idle?” but “what constrains the flow of valuable output?” Shared measures such as lead time and system throughput help local decisions serve the common goal.
Efficiency needs an object#
Before optimizing efficiency, name what should be accomplished more effectively. Keeping every minute occupied is not the same as delivering valuable work faster and more reliably.
The goal is not universal busyness. It is sustained attention on the constraint that matters most.