Routine variation is a property of the system and must not be reacted to individually; only assignable variation warrants investigation.
Deming's demonstration was the red bead experiment: workers punished for variation they could not control. Reacting to common-cause variation as though it were assignable makes performance worse, not better.
In practice
- A sales manager reviews the worst performer every month and finds a different name each time, because the variation is the territory allocation.
- A hospital investigates each month's infection count in isolation and never sees that the number has been stable within its own limits for two years.