Sustain a Lean Improvement with a Process Control Plan
A welding-cell improvement can work during a Kaizen event and drift after a supervisor changes, a fixture moves, or a new operator joins the team. A process control plan makes the changed condition visible: what must stay true, who checks it, when to react, and how to confirm the correction. Use the plan to protect the specific gain and required safety and quality controls. Do not turn it into a generic audit score or assume that a lower setup time proves the process remains stable.

Choose the process condition that protects the improvement
Start with one process and the problem the team changed. If a welding-cell Kaizen reduced setup disruption, identify the fixture, approved work sequence, material location, training, and quality checks that made the new method work. If a 5S action moved a tool, specify its marked location and the reason the location matters. Avoid the word “sustain” without naming a condition that can be observed.
Separate the improved outcome from the method that supports it. A setup-time measure may show whether the change is holding. A point-of-use fixture check, first-piece record, and current work instruction may show whether people can repeat the approved method safely. Use both when they answer different questions. Keep customer specifications, the welding procedure, inspection requirements, equipment safeguards, and qualified-person requirements in the controlled system that owns them.
Write the baseline definition before setting a trigger. State the part family, shift, setup event, start and stop points, sample count, unit, and data owner. If the team did not collect enough stable observations for statistical limits, do not label a local reaction threshold a control limit. A trigger can still prompt an investigation, but its basis and limitations must be clear.
Build the post-Kaizen control plan
A control plan connects an important process condition to an observation and a response. Keep each row specific enough that an operator or supervisor can tell what to check and who can make the next decision. Define an owner for follow-up, not only for detecting a problem. If a reaction requires quality or engineering approval, name that role rather than asking an operator to make an unauthorized disposition.
| Process condition | Check and frequency | Owner | Trigger | Immediate reaction | Verify and close |
|---|---|---|---|---|---|
| Current approved fixture and instruction are available | At shift start and after a document revision | Cell lead | Wrong revision, missing fixture, or unclear method | Pause affected setup and contact the authorized process owner | Confirm correct controlled document and fixture before restart |
| Fixture remains in its marked, accessible location | At handoff and during the defined layered audit | Operator; supervisor verifies | Fixture is missing, obstructed, or stored outside its approved position | Find the cause and restore the approved location; do not use an unapproved substitute | Record the condition and check the next shift handoff |
| Required first-piece acceptance is complete | At each setup or changeover specified by the process | Operator and quality role specified by the control plan | Required result fails or record is missing | Contain affected work under site procedure and notify the quality owner | Resume only after documented authorized disposition |
| Setup follows the approved sequence | Record each defined setup; review trend weekly during the first month | Process owner | Local reaction threshold is exceeded or a step is skipped | Observe the next setup, confirm the data, and record the condition | Assign a cause and action; repeat observation after correction |
| Fixture and safety devices remain serviceable | Pre-use check and maintenance interval set by the approved process | Operator and maintenance | Damage, missing component, or failed safety check | Stop or secure work according to site procedure; call the responsible owner | Repair, inspect, and authorize return to service as required |
| Affected people can follow the current standard | After an approved revision, role change, or qualification change | Supervisor and training owner | Current instruction is unavailable or qualification is not confirmed | Assign qualified coverage and complete required training | Record training and verify the method at the point of use |
This is a planning example, not an approved welding control plan. The responsible site must define the actual welding procedure, inspection, safety, customer, and document-control requirements. A Lean worksheet cannot replace a process-specific safety or quality approval.
Use a run chart to notice drift
A run chart places observations in time order so a team can see a trend, shift, or unusual point. Record one consistent measure, such as setup minutes for a defined fixture change on the same product family. Annotate the date the approved method changed, when affected shifts were trained, and any relevant event such as a maintenance repair or revision update. Keep the sample count with each point if a point represents several setups.
Illustrative example — fictional setup observations, not plant data: the table below demonstrates an annotated time sequence. It does not establish statistical control. The example uses four setup observations before a fixture-staging change and four after. The process owner should select an appropriate chart and method before using real data to make a control decision.
| Sequence | Setup time (min) | Process annotation | What the team records |
|---|---|---|---|
| Baseline 1 | 52 | Approved prior method | Part family, shift, setup definition, sample count |
| Baseline 2 | 49 | Prior method | Same start and stop points |
| Baseline 3 | 51 | Prior method | Same product and measurement rule |
| Baseline 4 | 50 | Prior method | Baseline median is 50.5 min for this example only |
| Change point | — | Approved fixture-staging method begins; affected shifts receive required instruction | Change date, revision, owner, and approval |
| Follow-up 1 | 46 | New method | Continue observation; do not infer a lasting gain |
| Follow-up 2 | 44 | New method | Verify same setup and quality checks |
| Follow-up 3 | 61 | Unexpected rise; point is not a statistical signal by itself | Confirm measurement and observe the setup with the operator |
| Follow-up 4 | 48 | New method after investigation is scheduled | Record cause and any authorized correction |
The baseline median in this fictional set is (50 + 51) ÷ 2 = 50.5 minutes. The four follow-up values sort to 44, 46, 48, and 61 minutes, giving a median of (46 + 48) ÷ 2 = 47 minutes. That small example does not prove the process improved or stayed improved. The 61-minute point needs investigation; it is not a reason to invent a statistical limit or to blame the operator. Verify the measurement, method, product mix, equipment condition, and approved controls.
When the sample is small, use the chart as a conversation aid and collect more comparable points. A qualified person should decide whether a statistical process control chart is suitable, how to calculate limits, and whether the data meet its assumptions. Do not mix different setup types, shifts, or part families without labeling the change in the data.
Use layered audits to support the standard
An audit should verify the condition and help remove obstacles. It should not become a scorecard that rewards a perfect number while leaving a missing fixture or unclear instruction unresolved. Keep audit checks observable, short, and connected to the process risk. If a check fails, record what was seen, who owns the next action, and when the team will verify the correction.
| Review layer | Example timing | Responsible (R) | Accountable (A) | Consulted (C) | Informed (I) |
|---|---|---|---|---|---|
| Operator point-of-use check | At shift start and setup as specified | Assigned operator | Cell lead | Quality or maintenance when the check requires it | Process owner |
| Supervisor layered audit | Daily during the first week, then at the approved local interval | Supervisor | Production manager | Operator and process owner | Quality lead |
| Process-owner review | Weekly during the first 30 days, then based on risk and evidence | Process owner | Department manager | Quality, maintenance, and affected operators | Site leadership when escalation criteria are met |
| Quality-system review | At the interval and events required by the approved system | Quality representative | Quality manager | Engineering and production | Process owner |
RACI means responsible, accountable, consulted, and informed. A person can hold more than one role where the organization allows it, but the plan should still identify one accountable owner for the action. The example timing is not a universal requirement. The responsible managers should approve a cadence that fits the process, risk, customer requirements, and evidence.
Respond to repeated findings
Use a reaction that protects the process and turns the finding into a specific action. If a noncritical fixture-location issue appears in two consecutive local audits, an illustrative site rule might require the supervisor to assign an owner during the shift and review the cause within one working day. That one-day response is an example of a local rule, not a general welding or safety standard. More serious issues follow the site’s immediate stop, containment, reporting, and escalation procedures.
- Confirm that the observation and revision are correct.
- Protect people, product identification, and required quality checks.
- Observe the work and ask what condition made the approved method difficult to follow.
- Assign the action to a named process, maintenance, quality, training, or document owner.
- Set a due time and define the evidence that will show the issue is resolved.
- Repeat the check after the correction and update the controlled standard if approved.
Do not treat every deviation as a training problem. The fixture may be stored too far from point of use, the mark may be unclear, the standard may be outdated, or the planned method may not fit actual work. Ask the people doing the job to help test the cause and the fix. If the required process changes, use the site’s change-control and qualification path before making the new method standard.
Use staff changes as a review trigger
Turnover and role changes can expose a weak handoff. In a NIST MEP case, Woodside Industries reported that Lean practices had declined over time as experienced staff left and new people joined; the company later received additional Lean training for 15 employees. This source illustrates why training and method review may need to recur. It does not report a welding-cell control plan or prove that an audit schedule prevents drift.
When an operator, supervisor, or process owner changes, verify that the current instruction, qualification record, fixture location, reaction path, and escalation contact remain clear. Invite new and experienced staff to point out unclear steps. Update only the approved document and training record; do not rely on informal memory as the process control.
Keep this plan distinct from daily management
A control plan answers how a process owner detects drift in a particular changed condition and restores the approved method. A daily management routine is broader: it organizes regular team review, escalation, and follow-up across several measures or work areas. The companion daily management guide covers the operating rhythm. This article focuses on the welding-cell control, trigger, reaction, verification, and document update.
For a repeatable area-condition checklist and follow-up assignments, use the 5S audit guide. A 5S score alone does not replace a process-specific control plan; connect the audit finding to the actual process condition and its owner.
Start the control plan
- Name the process, product family, improvement, and measure the team intends to sustain.
- Record the approved baseline, method, revision, and evidence source.
- Define a practical trigger without claiming unsupported statistical limits.
- Assign one process owner and name the roles that must respond.
- Set an initial verification cadence and a review date after the first 30 days.
- Write the immediate reaction, escalation route, and closure evidence.
- Review the first drift or repeat finding with operators before revising the standard.
Record the first observation, its definition, and its owner. That gives the next supervisor a clear starting point and gives the team evidence to decide whether the improvement has held.
Sources
- Lean Enterprise Institute: Standardized Work — describes a shared work sequence and standard inventory as a basis for current process control and improvement.
- Lean Enterprise Institute: Kaizen — describes checking results, following up, and standardizing a demonstrated improvement.
- Lean Enterprise Institute: Leader Standard Work — describes regular management activities that support problem solving and accountability.
- NIST MEP: Lean Training and Implementation Pays Off in Big Results — Woodside Industries case describing practice erosion after attrition and new hires and later training; it is not evidence for the illustrative welding data.





