Ask an operations team to raise output and it will usually find plenty to improve. A faster machine here, an extra person there, a better system for one of the steps. Much of that effort changes nothing, because the output of a process is set by its slowest step, and most improvements are made to steps that are not the slowest.
Find the constraint
In any sequence of steps, one has the least capacity relative to the demand placed on it. That step — the bottleneck, or constraint — determines the throughput of the whole process. Work queues in front of it, and the steps after it wait for its output.
The constraint is often visible if you look for it: it is where work accumulates, where overtime is concentrated, and where people say they are always behind. It is not always where the most expensive equipment or the most senior people are.
Improving anywhere else
Once the constraint is known, the logic of improvement becomes clear, and it is counter-intuitive:
- An hour lost at the constraint is an hour lost for the whole process. It cannot be recovered elsewhere.
- An hour saved at a step that is not the constraint saves almost nothing. The work simply waits longer in front of the constraint.
- Keeping every step busy is not the goal. Running non-constraint steps at full capacity builds inventory and queues in front of the constraint without raising output.
The practical steps follow from this. Make sure the constraint never waits for work or stands idle through breakdowns, changeovers or missing inputs. Move work that does not need the constraint to other resources. Check quality before work reaches the constraint, so that its capacity is not spent on items that will be rejected. Only then invest in adding capacity to it — and when you do, expect the constraint to move to another step.
Utilisation and queues
A related misunderstanding is the belief that high utilisation is always efficient. As any resource approaches full utilisation, the time work spends waiting in front of it rises sharply, because there is no slack to absorb variation in arrivals and processing times. A team running at near full utilisation will have long and unpredictable lead times even when it is working flat out.
Operations that must respond quickly — customer service, maintenance, emergency work — deliberately hold spare capacity for this reason. Spare capacity at a non-constraint step is not waste; it is what keeps work flowing to the constraint.
Measure the right things
Measures that reward local efficiency — output per hour at each step, utilisation of each machine — encourage exactly the behaviour that builds queues without raising throughput. More useful measures describe the system: throughput of the whole process, lead time from order to delivery, and the performance of the constraint itself.
The principle is simple enough to state in a sentence and difficult enough to apply that it is widely ignored. The capacity of an operation is the capacity of its constraint. Improvement starts by finding it.