Teams rarely struggle because they lack effort. They struggle because effort is pointed at the wrong place. The order desk is buried, so you hire another coordinator, and three weeks later the desk is buried again. The queue moved; the constraint did not.
The loudest problem is rarely the binding one
A bottleneck announces itself through symptoms downstream of where it actually lives. Late shipments look like a warehouse problem. Trace them back and the delay was a quote that sat unapproved for two days. The warehouse was never slow; it was starved.
Before changing anything, find the step where work waits the longest, not the step where people complain the loudest. Those are usually different places.
A short test
For one week, timestamp a single unit of work at every handoff:
- When did it arrive at each step?
- When did someone actually start on it?
- When did it leave?
The gap between "arrived" and "started" is your real signal. Long gaps mark the constraint. Short gaps with long processing times mean the step is genuinely at capacity, which is a different fix entirely.
What holds up after you find it
Once the binding constraint is clear, the durable fixes are almost always one of three:
- Remove the wait (approvals that batch, queues with no owner).
- Protect the constraint so it never sits idle.
- Only then add capacity, if it is still the limiting step.
Adding capacity first is the expensive habit. It feels like progress and usually just relocates the queue.