top of page

5.0 Control Phase

Control Phase Purpose of the Control Phase The Control Phase is the final phase of DMAIC, focused on sustaining the gains achieved in the Improve Phase. The aim is to prevent regression, stabilize improved processes, and ensure that performance remains within the desired limits over time. In this phase, improved processes are documented, monitored, and handed over for routine operation with clear controls that detect and respond to variation. Key objectives are: - Stabilize performance at the new improved level - Prevent backsliding to the previous state - Sustain gains through standardization and control systems - Transfer ownership to process operators and managers --- Linking Back to Define, Measure, Analyze, Improve Control does not start from scratch. It confirms and locks in what was achieved earlier. - From Define: Confirm the project goals and CTQs (Critical to Quality) that must remain in control. - From Measure: Reuse data definitions, operational definitions, and baseline metrics. - From Analyze: Focus on the key Xs (inputs) and Ys (outputs) that were found to drive performance. - From Improve: Monitor the implemented solutions and critical process changes. The Control Plan builds on these elements to maintain the solution. --- Control Plans Purpose of a Control Plan A Control Plan is a structured document that describes how to monitor and control the critical aspects of the improved process. It ensures consistency, visibility, and quick reaction to deviations. A Control Plan should: - Specify what to measure (CTQs, key Xs and Ys) - Define how and when measurements occur - Clarify who is responsible for monitoring and reaction - Describe what to do when performance goes out of control Key Elements of a Control Plan Typical sections include: - Process step – the specific activity being controlled - Characteristic – the variable or attribute monitored (e.g., cycle time, defect count) - Specification / target – nominal value, upper and lower limits, or defect criteria - Measurement method – how data is collected, with clear operational definitions - Sampling plan – frequency, sample size, and location in the process - Control method – charts, checks, mistake-proofing, or other controls used - Reaction plan – required actions when a signal or nonconformance occurs - Responsible owner – role accountable for monitoring and actions A good Control Plan is practical, visible, and tightly focused on the few critical items that truly drive process performance. --- Standardization and Documentation Standard Operating Procedures (SOPs) and Work Instructions Standardization locks in the improved method by defining the new “best known way” of working. The improved process must be clearly documented and made easy to follow. Key aspects: - Update existing documents – SOPs, work instructions, checklists, forms - Reflect actual practice – documents must match how work is performed - Include critical details – key steps, sequence, parameters, and tolerances - Clarify decision points – what to do in exceptional or borderline cases Effective standardization reduces variation caused by different people doing the same work in different ways. Visual Management Visual controls make the current state of the process and the standards obvious to anyone. Examples directly relevant to Control: - Process maps displayed at workstations to reinforce the standard flow - Visual work instructions with pictures, diagrams, or annotated screenshots - Status boards for key metrics, showing current values vs. targets - Color coding for tools, materials, or process states Visual management supports quick detection of deviation and aids adherence to the Control Plan. --- Process Capability in the Control Phase Confirming Capability After Improvements After improvements, the process capability should be reassessed using the same metrics and definitions used in Measure and Analyze. Goals of this reassessment: - Verify that the improved process meets specification limits - Quantify the new capability using indices such as Cp, Cpk, Pp, Ppk - Confirm that the distribution assumptions made earlier still hold - Ensure that the process is stable enough for capability indices to be meaningful If capability remains marginal, the Control Plan may require tighter monitoring or further improvement work before project closure. Short-Term vs Long-Term Capability In Control, distinguishing short-term from long-term performance helps anticipate real-world results. - Short-term measures (Cp, Cpk): - Limited sources of variation - Often measured under controlled conditions - Long-term measures (Pp, Ppk): - Include more sources of variation (shifts, drifts, environment) - Reflect what is likely to occur in day-to-day operation Control strategy should be based on realistic long-term capability, not only ideal short-term performance. --- Statistical Process Control (SPC) Purpose of Control Charts Control charts are central tools in the Control Phase. They: - Distinguish between common cause and special cause variation - Provide early warning of instability - Help ensure that process performance remains predictable over time A stable process with control charts in place is more likely to sustain the benefits achieved in Improve. Selecting Appropriate Control Charts Selection depends on: - The type of data (variable vs attribute) - The subgrouping strategy (individual vs rational subgroup) - The sample size Common charts: - Variable data: - X̄-R or X̄-S charts for subgrouped data - Individuals (I-MR) charts for single observations - Attribute data: - p-chart for proportion defective with varying sample size - np-chart for count defective with constant sample size - c-chart for count of defects with constant area of opportunity - u-chart for defects per unit with varying area of opportunity In Control, chart choice should match the way data will be routinely collected. Interpreting Control Charts Control chart interpretation focuses on: - Center line – the expected process average - Control limits – the expected natural variation range, not the specification limits - Signals of special cause variation, such as: - Points outside control limits - Runs of points on one side of the center line - Trends up or down across several points - Cycles or repeating patterns - Sudden shifts in level or spread When a signal appears, the Control Plan must specify how to investigate and respond to it. Reaction Plans for Out-of-Control Signals A reaction plan transforms chart signals into disciplined action. It should specify: - Immediate steps – stop, hold, or continue operations under defined conditions - Containment actions – quarantine suspect product, notify stakeholders - Diagnostic actions – identify potential causes, check recent changes, review logs - Correction steps – adjust settings, replace tools, retrain operators - Documentation – record the event, root cause, and resolution Consistent application of reaction plans prevents minor issues from growing into major performance problems. --- Control of Inputs (X) vs Outputs (Y) Managing Critical Inputs Improved outcomes (Ys) depend on controlling the critical inputs (Xs). Often, maintaining control of a few key Xs is easier than continuously inspecting all outputs. For each critical X: - Define acceptable ranges or settings - Create clear instructions for adjustment - Monitor adherence to setpoints - Include X-related checks in the Control Plan When Xs are kept in control, Y stability typically follows. Monitoring Outputs and Defects Monitoring outputs remains essential to: - Confirm customer requirements are met - Detect any unexpected shifts not captured by input controls - Provide evidence that the improved process remains capable For outputs, ensure: - Clear defect definitions - Consistent inspection criteria - Proper sampling methods aligned with process risk Both X and Y controls should work together to sustain performance. --- Mistake-Proofing (Poka-Yoke) in Control Purpose of Mistake-Proofing Mistake-proofing reduces the chance that errors will occur or escape detection. In Control, it: - Prevents reintroduction of old failure modes - Reduces reliance on memory, vigilance, or complex instructions - Strengthens the robustness of the improved process Mistake-proofing should target the high-risk steps identified in earlier phases. Types of Mistake-Proofing Relevant to Control Examples include: - Prevention devices – physical or logical constraints that make an error impossible (e.g., connectors that only fit one way) - Detection devices – automatic checks that identify errors early (e.g., sensors, validation rules) - Sequence enforcers – mechanisms that force a correct order of operations - Interlocks – conditions that must be met before a step can proceed In the Control Phase, document these devices in the Control Plan and ensure maintenance and testing procedures are in place to keep them effective. --- Control of Measurement Systems Ongoing Measurement System Monitoring The integrity of the Control Phase depends on reliable data. Key concerns are: - Repeatability and reproducibility over time - Calibration and maintenance of measurement devices - Consistent use of operational definitions for measurement If the measurement system drifts, control charts and capability indices can become misleading. Maintaining Measurement System Fitness Control of the measurement process itself can include: - Periodic checks or audits of measurement procedures - Scheduled calibration and verification activities - Refresh training for people performing measurements - Repeating MSA (Measurement System Analysis) when changes occur Measurement system control should be embedded in the Control Plan wherever key metrics are involved. --- Transition and Handover Embedding the Improved Process Into Operations Control is not complete until the improved way of working is fully embedded in routine operations. Critical steps: - Ensure updated SOPs, Control Plans, and visual controls are accessible - Confirm that data collection and control chart procedures are in place and understood - Validate that reaction plans are practical and actually followed - Integrate control metrics into regular performance reviews and meetings Embedding the process reduces the risk that improvements fade once the project team steps back. Training and Communication People who own and run the process day-to-day need clarity about: - What has changed in the process - Why the changes were made - How success will be measured - What their specific responsibilities are in monitoring and reaction Training should be focused, practical, and aligned with the Control Plan and standard work. --- Project Closure in the Control Phase Verification of Sustained Performance Before closing the project, confirm: - That performance improvements are statistically and practically significant - That control charts show stability at the new performance level - That capability meets or exceeds agreed requirements - That there are no unresolved high-risk failure modes Only after these checks is it reasonable to transition fully to routine management. Final Documentation Essential artifacts to complete and store include: - Final process maps and SOPs - Control Plans and reaction plans - Control charts and capability summaries (before and after) - Records of implemented mistake-proofing and standardization - Lessons learned for future improvement efforts Complete documentation supports long-term sustainability and helps future teams build on the work. --- Summary The Control Phase ensures that process improvements are sustained and that performance remains stable over time. Core activities include creating and implementing Control Plans, standardizing the improved process, confirming process capability, and establishing statistical process control with appropriate charts and reaction plans. Control of both critical inputs and outputs, supported by reliable measurement systems and mistake-proofing, forms the foundation of sustained gains. Effective transition, training, and documentation embed the improved process into daily operations and protect against regression.

Practical Case: Control Phase A mid-sized hospital reduced outpatient lab turnaround time from 6 to under 2 hours using DMAIC. After Improve, leadership wanted to ensure the gains held during staff rotation changes and seasonal peaks. Context and Problem The Improve phase had standardized: - a new specimen triage sequence, - a batching rule for high-volume tests, - and a priority lane for STAT samples. Within three weeks, average turnaround began creeping above 2 hours on evening shifts. Supervisors reported “everyone does it slightly differently again.” How Control Phase Was Applied The Black Belt and lab manager focused Control activities on making the new way the default: - Standard work documents: One-page job aids with the exact triage sequence and batching rules were posted at each accessioning station and embedded into the LIS workflow screens. - Visual controls: Simple color-coded racks: - red for STAT, - yellow for routine queued for next batch, - blue for overflow. Any mis-sorted rack was immediately visible. - Process control chart: A weekly X‑bar chart for turnaround time by shift was created from LIS data and automatically emailed to the manager and charge techs every Monday. Upper and lower control limits were set based on the new stable performance. - Response plan: A short “if-then” playbook: - If two consecutive points on the control chart were above the upper control limit on a given shift, the charge tech would: - audit 10 consecutive samples for correct triage and batching, - give 5-minute refresher coaching if compliance was below 95%, - record actions in a simple log reviewed at the daily huddle. - Handoff and ownership: The Black Belt transferred monitoring to the lab manager, who added reviewing the control chart and playbook log as a fixed agenda item in the weekly staff meeting. Result Over the next three months, all shifts stayed within control limits. Occasional spikes during flu season were traced to noncompliance with batching rules; the predefined response plan corrected behavior within days. The hospital removed the project from active monitoring after six stable months and retained only the automated chart and weekly review, with no loss of performance. End section

Practice question: Control Phase A manufacturing team has completed an improvement project that reduced defect rate from 4.5% to 1.2%. The sponsor requires ongoing visibility and rapid detection of performance shifts. Which Control tool is most appropriate to ensure early detection of process deterioration? A. SIPOC diagram B. X̄–R control chart C. Pareto chart D. Fishbone diagram Answer: B Reason: An X̄–R control chart monitors process mean and variability over time, enabling early detection of special causes and deterioration after improvements are implemented. SIPOC and fishbone are design/analysis tools, not ongoing control tools; Pareto shows aggregated defect distribution, not real-time stability. --- A process shows a long-term DPMO of 6,210 for a critical CTQ. Assuming a standard normal distribution and a 1.5 sigma shift, what is the approximate long-term sigma level of the process, commonly used for process capability reporting in Control? A. 3.0 σ B. 3.5 σ C. 4.0 σ D. 4.5 σ Answer: C Reason: A DPMO of ~6,210 corresponds to a short-term Z of about 4.5, and after applying the 1.5 sigma shift for long-term performance, the reported long-term sigma level is approximately 4.0 σ. The other values do not match the DPMO–Z relationship commonly used in Six Sigma conversion tables. --- A Black Belt is designing a control plan for a transactional process with low daily volume and continuous, approximately normal cycle time data. What is the most appropriate choice to monitor stability of this CTQ in the Control Phase? A. p-chart B. u-chart C. I–MR chart D. c-chart Answer: C Reason: I–MR charts are suitable for individual, continuous, approximately normal data with low subgroup sizes (often n=1), which fits low-volume cycle time monitoring. p-charts and u-charts are for attribute data; c-charts are for count of defects, not continuous time data. --- In the Control Phase, a team must ensure that frontline operators consistently follow the new standardized work for a critical setup activity. Which approach best addresses long-term sustainment of the changed method? A. Issue a one-time email with the new work instructions B. Integrate the new method into standard work, training, and layered process audits C. Conduct a single kaizen event focused on the setup activity D. Rely on the project charter to enforce compliance Answer: B Reason: Embedding the new method into standard work documents, formal training, and layered process audits creates structural reinforcement and feedback loops, which are key to long-term sustainment in Control. A one-time communication, a one-off kaizen, or relying on the charter alone does not provide robust, ongoing adherence mechanisms. --- A process has been improved and is now monitored with an X̄–S control chart. After several months, the chart shows points within control limits but a sustained run of 10 points in a row above the center line. What is the most appropriate Black Belt action in the Control Phase? A. Ignore the pattern because all points are within control limits B. Adjust the specification limits to match the new pattern C. Investigate for a potential process shift and consider recalculating control limits if stable at the new level D. Immediately revert to the pre-improvement process Answer: C Reason: A run of 10 points on one side of the center line is a common non-random pattern rule indicating a process mean shift; the Black Belt should investigate causes and, if the new level is stable and desirable, update control limits to reflect the new process performance. Ignoring the pattern, changing specs arbitrarily, or reverting the process are not appropriate evidence-based control actions.

bottom of page