Phase 5 · Control
How to build a control plan that holds the gains
The control plan keeps an improvement in place after the project ends. For each critical step it says what is measured, within which limits, how often, who checks it and what happens if it drifts.
In short
A control plan is a table that, for each critical step of an improved process, sets the metric to watch, its limits, how often it is checked, who owns it and the reaction plan if it goes out of limits. It is the main deliverable of the Control phase of DMAIC.
What it is
It is a short table, one row per critical step. The usual columns are the step, the metric, the limits, the measurement method, the frequency, the owner and the reaction plan.
It is what turns a project result into the normal way of working. It is handed over to the people who run the process, and it has to work without the project team.
When to use it
- In the Control phase, once the improvements are in place and their results confirmed.
- Before closing a project and handing the process back to operations.
- When a past improvement has drifted back and nobody noticed in time.
How to do it, step by step
- 1Choose the critical steps. Only the steps that bring the problem back if they slip. Usually the ones the improvements changed.
- 2Define one metric per step. Something that can be observed or counted, with the way to measure it written down.
- 3Set the limits. The range within which the step is under control. Outside it, someone acts.
- 4Fix the frequency and the owner. How often it is checked, and by which role. A role, not a name, so the plan survives a change of person.
- 5Write the reaction plan. What to do when a result is out of limits: the immediate action, who is told and how the cause is investigated.
- 6Hand it over and review it. Go through it with the process owner, and review it on set dates during the first months.
Worked example
Case: a plant that wants new operators to reach standard output sooner.
| Step | Metric | Limits | Frequency | Owner | Reaction |
|---|---|---|---|---|---|
| Assign a fixed buddy | New operators with the same buddy, on the same shift | All of them, for the first 8 weeks | Weekly | Shift leader | Reassign within 24 hours and record the reason |
| Follow the training path | Stage reached against the plan | No more than one week behind | Weekly | Training manager | Extra practice session and a review with the buddy |
| Protect practice time | Hours at the practice station | At least 4 hours a week | Weekly | Production manager | Free a practice station on the next shift |
| Keep work instructions current | Instructions at the station match the latest version | No out-of-date version in use | Monthly | Process engineer | Withdraw and update within one week |
| Release to production | Weeks to standard output | 9 weeks or fewer | At each release | Training manager | 5 Whys on the case if it exceeds 10 weeks |
Common mistakes
- Controlling every step instead of the few that bring the problem back.
- Limits with no reaction plan: everyone sees the result go out of range and nobody knows what to do.
- Naming a person as owner instead of a role.
- Writing it on the last day of the project, without the people who will use it.
Frequently asked questions
What is a reaction plan in a control plan?
It is the column that says what to do when a result goes out of limits: the immediate action to contain it, who is informed and how the cause is investigated. Without it, the plan only records problems.
What is the difference between a control plan and a control chart?
A control chart is a graph that shows whether a measure varies within its natural limits over time. A control plan is the wider table that says what to watch, how and who acts; a control chart can be the method for one of its rows.
How long should a control plan stay in place?
As long as the process runs. In the first months it is checked closely; once results are stable, the frequency can be reduced and the checks become part of daily management.
