Theory of Constraints
Why it matters
A founder has limited hours and money, and cannot improve every part of the business at once. The theory turns a long list of possible improvements into one problem to solve now. It also explains why effort so often feels wasted: speeding up a step that is not the constraint only builds a bigger pile of work in front of the one that is.
How to apply it
- Find where work waits longest. A queue in front of a step usually points at the constraint.
- Check the guess against the output: if that step doubled in speed, would the whole process produce more?
- Keep the constraint busy and fed with good-quality work, since an hour lost there is an hour lost for the whole business.
- Resist adding people or tools elsewhere until the constraint moves.
What it is
Eliyahu Goldratt set out the theory in his 1984 novel The Goal. Its central idea is simple: any process has one step that limits how much the whole thing can produce. That step is the constraint. Output of the entire chain is capped by it, however fast the other steps run.
Goldratt described five focusing steps. First identify the constraint. Then exploit it, which means getting the most out of that step as it stands, before spending money. Then subordinate everything else to it, adjusting the other steps so the constraint is never starved or blocked. Then add capacity to the constraint itself if it still limits output. Then go back to the start, because once one constraint is cleared, another appears.
Common mistakes
- Improving the wrong step. Optimising a step that is not the constraint adds no output. Prove the constraint before spending.
- Naming a symptom as the constraint. "We are too busy" is not one. Find the specific step where work waits.
- Adding capacity first. Exploit the constraint before buying more. Often it loses time to waiting, rework or poor inputs.
- Starving or blocking it. If the constraint waits for work or cannot hand work on, the whole process loses time. Keep it fed.
- Declaring victory and stopping. Once one constraint is fixed, another becomes the limit. Go back to step one.
- Treating people as the constraint. It is usually a step or a policy, such as an approval rule. Fix the system before blaming individuals.
- Applying it to everything at once. The method is for one limit at a time. See constraint and bottleneck resource for the individual ideas.
