Subordinate to the Constraint

Subordination is the third of the five focusing steps: once you have found the constraint and worked to exploit it, you make every other part of the system serve it. The non-constraints have spare capacity, so their job is not to run as fast or as busily as they can — it is to keep the constraint fully fed and never starved or disrupted. Subordination demotes local performance to a means, with the constraint’s output as the end.

CF stresses this because it is the antidote to the intuitive but mistaken drive to maximize each part on its own. Optimization away from the constraint is wasted: a faster non-constraint step does not raise throughput because material still queues behind the slowest link. Worse, an unsubordinated non-constraint actively harms the goal by overproducing — piling up inventory the constraint cannot process, generating cost and chaos without adding finished output. So subordination has two halves: protect the constraint from problems, and stop non-constraints from producing more than it can absorb.

This is the operational form of CF’s local-vs-global optimization argument. It explains why a balanced plant (every station near 100% utilization) fails: with statistical fluctuations and dependent events, full local utilization is not global efficiency. Subordination instead organizes the whole system, including deliberate excess capacity elsewhere, around one chosen weakest link.