What does it mean that bottleneck measurement?

You cannot address a bottleneck you cannot measure — quantify the constraint.

Example: You suspect that context switching is your primary bottleneck. You feel scattered, you complain about too many meetings, and by Friday you can't point to anything substantial you shipped. But when you track your time for one week — actually log every transition between tasks — you discover that context switching accounts for 11% of lost time. The real bottleneck is decision latency: tasks sit in your queue for an average of 3.2 days before you even start them because you can't decide which one matters most. The bottleneck you felt and the bottleneck the data revealed were completely different problems requiring completely different interventions.

Try this: Choose the bottleneck you identified in L-0944 (or your strongest suspicion about what constrains your throughput). Design a measurement protocol: What specific metric will you track? What unit does it use? How will you collect it? For one full work week, measure that constraint daily. At the end of the week, calculate the average, the range (best day vs. worst day), and the trend (improving, stable, or degrading). Write one sentence: 'My [bottleneck] averages [X units] per [time period], ranging from [low] to [high].' You now have a baseline. Everything that follows in this phase depends on that number.