Control: Lock In the Gain and Hand It Over
What Control must produce
Control keeps the improved process from drifting back. Concepts: a control plan stating what is monitored, by whom, with what tool, at what frequency, and the reaction plan when a signal appears; updated standard work and documentation; training; SPC chart selection - variable charts for continuous data (Xbar-R for subgroups of roughly 2–9, Xbar-S for larger subgroups, I-MR for individual measurements) and attribute charts (p proportion defective, np number defective, c defects per unit with constant opportunity, u defects per unit with varying opportunity); control limits computed from the process as the plotted statistic ± 3 standard errors, never from specification limits; out-of-control rules - a point beyond a limit, a run of seven or eight on one side of the centreline, a trend, two of three points beyond 2 sigma; ongoing capability monitoring; and project closure with handoff to the process owner, lessons learned and financial validation.
Traps questions exploit
- Drawing specification limits on a control chart, or treating control limits as customer requirements.
- Using a p chart when the count is defects per unit rather than defective units; using c when exposure varies (use u).
- Choosing Xbar-R when each subgroup is a single invoice or a single batch (use I-MR).
- Declaring victory because the process is in control when it is still incapable - stability and capability are separate questions.
- Omitting the reaction plan; a chart nobody acts on is decoration.
Worked example
Auditing 5 invoices a day and counting errors per invoice → c chart (constant opportunity). Tracking individual customer cycle times → I-MR. Tracking the fraction of orders late when daily volume varies → p chart with limits that change with n.
Rule of thumb: control limits come from the voice of the process; specification limits come from the voice of the customer. Never mix them.