Most operational excellence programmes are a monument to good intentions and misaligned effort. A plant manager tells me proudly that OEE is up 6 points across the shop. I ask which machine improved. When the answer is 'all of them', I already know the cash story: working capital has expanded, WIP has grown, and delivery reliability has not moved.
There is exactly one machine, one process step, or one policy that governs the throughput of your plant. Every other improvement is either invisible in the P&L or actively harmful because it builds inventory ahead of the constraint. This is not consultant theory — it is arithmetic.
The four-step drill I run in every engagement is: (1) identify the constraint (usually not what the plant manager first names), (2) exploit it — squeeze every hour, kill every micro-stop, protect it from starvation, (3) subordinate the rest of the system to it, (4) elevate only if steps 1–3 have already been done. In that order.
The single biggest resistance to this method is cultural. Non-constraint stations will look 'worse' on local metrics. Operators, supervisors and even middle managers will push back. Only executive sponsorship keeps the programme honest. If the CFO is not personally invested, do not start.