A bottleneck is the one part of a system that limits everything else — a system can never move faster than its slowest constraint, so improving anything other than the bottleneck rarely improves the whole.
The idea
In any process made of sequential or interdependent steps, one step has the least capacity and sets the pace for all the others. Speeding up a non-bottleneck step just creates idle capacity or a pile-up in front of the bottleneck; it does not raise total output. This is the core insight behind the theory of constraints: find the constraint, exploit it fully, then subordinate everything else to it.
When to use it
- Output or throughput has plateaued despite effort spread across many parts of a process.
- Diagnosing why a team, pipeline, or system underperforms its apparent capacity.
- Deciding where to invest limited time or budget for the biggest overall gain.
How to apply it
- Map the process end to end and measure the capacity or speed of each step.
- Identify the step with the least capacity relative to demand — that is the bottleneck.
- Protect and fully utilize that step first (remove waste, avoid starving it of input).
- Only after the bottleneck is relieved, look for a new one — it will have moved elsewhere.
Watch out for
Optimizing a non-bottleneck step feels productive but does not change total output, and can even hide the real constraint. Bottlenecks move once relieved, so this is a repeating diagnosis, not a one-time fix.
Related models
- Critical Path Analysis — identifies the sequence of dependent steps that determines total project duration.
- Root Cause Analysis — a complementary way to find what is really limiting a system, not just its symptoms.
Sources
Eliyahu M. Goldratt’s Theory of Constraints, introduced in his book “The Goal” (1984), is the foundational treatment of bottlenecks as the single limiting factor in a system’s throughput.