Key takeaways
Short answer: A bottleneck is the physical resource with the least capacity versus demand — the slowest step that sets the pace of the whole line. A constraint is anything limiting the system, which may be a policy, a market or a supplier, not just a machine. Every bottleneck is a constraint, but not every constraint is a bottleneck — and improving a non-constraint adds cost with zero throughput. See also oee for manufacturing.
A bottleneck is the resource that cannot keep up — the station with the least capacity relative to what the line is asked to produce. It is physical, usually visible as a growing queue in front of it, and it dictates the maximum rate of everything downstream.
A constraint is broader: anything that limits system throughput. The Theory of Constraints points out that the real limiter is often not on the floor at all — it can be a scheduling policy, a market that will not absorb more output, or a single-source supplier who cannot deliver faster.
A plant believes its bottleneck is the slow CNC cell and is about to spend heavily on a second machine. But analysis shows the CNC is starved 30% of the time because the scheduling rule releases work in large weekly batches. The true constraint is the release policy, not the machine — change the policy and the existing CNC has plenty of capacity. Buying the second machine would have added cost and changed nothing, because it was never the limiter.
The Theory of Constraints says you improve the constraint and subordinate everything else to it. If your real constraint is a policy or a supplier, throwing capital at the floor changes nothing. Identifying the true limiter first is what separates expensive activity from actual throughput gain.
Look for the queue to find the bottleneck, but then ask whether a policy or external limit is really in control. Sometimes the floor is healthy and the constraint is order release, a quality hold, or a sole-source material. The cheapest improvements often come from elevating a non-physical constraint.
1. Assuming the constraint is always a machine. Policy and market constraints are common and cheaper to fix.
2. Improving non-constraints. Local efficiency that never reaches the customer.
3. Moving the bottleneck without noticing. Fix one and another appears; keep re-checking.
4. Optimising the bottleneck in isolation. The rest of the system must subordinate to it.
OEE on the bottleneck is the OEE that matters most — an hour lost there is an hour lost for the whole plant, while OEE on a non-constraint is largely free time. Focusing improvement and even maintenance priority on the constraint is how OEE work translates into real throughput.
Fabrico shows where losses actually cost throughput, helping you tell the bottleneck from the noise and protect the constraint. Book a demo to see constraint-focused OEE.
Overlapping — every bottleneck is a constraint, but not every constraint is a bottleneck.
On the bottleneck — its losses cost the whole plant; non-constraint losses are largely free time.
Yes — policies, market demand and single-source suppliers are common non-physical constraints.
You add cost and local efficiency with no throughput gain.
Yes — fix one and another usually appears, so re-identify it after each improvement.
Programați o întâlnire individuală cu experții noștri sau înscrieți-vă direct în planul nostru gratuit.
Nu este nevoie de card de credit!
Login
Blog
