A Constraint is any factor that limits the throughput of a system, most often a bottleneck resource, but sometimes a policy, a market condition or a shortage of a specific skill. Managing constraints is the central discipline of the Theory of Constraints, and it applies far beyond the shop floor.
Physical constraints are tangible: a machine, a workstation, a supplier. Policy constraints are rules or decisions that limit throughput even when physical capacity exists, a batch-size rule, an approval process, a scheduling policy. Policy constraints are often the largest and the cheapest to remove, because removing them costs nothing but a decision.
Identify the constraint. Exploit it, get maximum output from it without spending more. Subordinate everything else to it, the rest of the system serves the constraint's pace. Elevate the constraint, invest in more capacity if the first three steps aren't enough. If the constraint moves, start again. The cycle never ends because as one constraint is broken, another emerges.
The system's throughput is set by its constraint. Improvements anywhere else don't change throughput; they only change local metrics. Focusing improvement effort at the constraint delivers system-level results; scattering it everywhere delivers busy-looking numbers and unchanged output.
Constraints are always there, remove one and another appears. The goal is not to eliminate constraints but to manage them consciously: know where the constraint is at any moment, protect it, exploit it, and move deliberately as it shifts.
