Last reviewed: October 2026
No Credit Card Required
This free Six Sigma DMAIC project checklist is for Green Belts, Black Belts and other project leads who run improvement projects, and for the champions who review them. One run is one DMAIC project, from the charter to handing the improved process back to its owner. Define writes the charter, the SIPOC and the customer requirements. Measure agrees how each measure is defined, checks the measurement system and measures the baseline and capability. Analyse maps the process and tests the possible causes with data until the root causes are verified. Improve selects solutions, runs a pilot and compares the result with the goal. Control rolls the change out, sets up the control plan and monitoring, updates the procedures and trains the people who run the process, and the project closes once the process owner accepts it and the benefits are confirmed. Each phase ends with a tollgate where the champion chooses Go, Recycle or Stop.
DMAIC is the five-phase method behind Six Sigma projects: Define, Measure, Analyse, Improve and Control. ISO 13053-1, the international standard on the DMAIC methodology, describes those five phases, how Six Sigma projects should be managed and the roles of the people who run them. What sets DMAIC apart is the order of work: nobody chooses a solution until the measurement system has been checked, the baseline measured and the root causes verified with data. That makes it slower than other improvement methods, and the right one when the cause is unknown and the problem is worth months of a trained lead’s time.
Use this checklist for one DMAIC project at a time. Smaller problems usually need a lighter tool: a continuous improvement cycle runs PDCA as a standing programme, and a kaizen event changes one process in a few days. A problem that needs containing and fixing for a customer fits a root cause analysis (8D), and a nonconformity from an audit or a complaint belongs in CAPA.
Use it when: a chronic problem has a cause nobody can prove, there is data to measure it, and the gain is worth several months of work.
Produces: verified root causes, a piloted solution, a control plan and confirmed benefits.
Decided at: five tollgates where the champion chooses Go, Recycle or Stop.
Use it when: a team improves its own process in small, quick cycles, one after another.
Produces: a steady stream of tested changes and an improvement register.
Rigour: data where it helps, without formal tollgates or a measurement system study.
Use it when: the problem sits in one process the team can see, and changes can be tried within days.
Produces: a changed process and new standard work, followed up 30 days later.
Not for: a cause nobody can prove. Analyse it first, then fix it in an event.
Six phases follow one DMAIC project: the five DMAIC phases, with Control split into running the controls and handing over. Each DMAIC phase ends with a tollgate where the champion chooses Go, Recycle or Stop. Recycle adds a rework task before the tollgate, and Stop adds a task after it to close the project properly; changing the decision hides them again. Nothing after the Improve tollgate can be completed until the champion decides. A continuous or discrete main measure adds the matching notes on the measurement system, capability and control chart tasks, and a failed measurement system study, a solution that needs change approval and claimed savings each add a task.
The project lead writes the charter and agrees it with the champion. Enter the planned date of each tollgate first: every phase’s due dates count back from its own tollgate, and move when it moves. Say whether the main measure is continuous or discrete, so the later tasks show the right notes.
Measure decides whether the rest of the project rests on facts. Define each measure, check that the measurement system can be trusted, then collect enough data to show how the process performs today. A measurement system that fails its study adds a task to fix it before the baseline is collected.
Analyse narrows many possible causes to the few the data supports. List the candidates widely, write each one as a hypothesis and test it, so the team does not spend money fixing a cause that was never proved.
Improve tests solutions on the verified causes, in a pilot that can be reversed. The Improve tollgate is the main gate: nothing after it can be completed until the champion records a decision, because full implementation changes the process for everyone and is the hardest step to undo. A solution that changes a safety control, a regulatory requirement or what the customer receives adds a change approval before the pilot.
Control makes the gain last after the team moves on. Roll the change out in full, write the controls into the documents people use, and watch the process long enough to show that the gain holds. Due dates count back from the Control tollgate.
The project ends when the process owner runs the controls and the benefits are confirmed, not when the solution goes live. Claimed financial savings add a check by finance. The champion’s final tollgate closes the project.
A tollgate is the review at the end of each DMAIC phase where the champion decides whether the project moves on. In this checklist the champion records one of three decisions: Go, Recycle (rework and resubmit) or Stop. The table shows what each tollgate asks and the evidence to bring. These are common practice, not the wording of a standard; agree your own criteria with whoever runs your improvement programme.
| Tollgate | The champion asks | Bring to the review | A common reason to recycle |
|---|---|---|---|
| 1. Define | Is the problem real, measured and worth solving, and can one team handle the scope? | The charter (problem, goal, scope, business case), the SIPOC, the CTQs and the team | A solution written into the problem statement, or a scope that covers half the business |
| 2. Measure | Can we trust the data, and how does the process perform today? | Operational definitions, the data collection plan, the measurement system result, the baseline, stability and capability | A measurement system nobody checked, or a baseline from too few data points |
| 3. Analyse | Which causes does the data support? | The process map, the cause and effect diagram, the test results and the verified root causes with their evidence | Causes chosen by opinion, with no test behind them |
| 4. Improve | Did the pilot meet the goal, and is it safe to roll out? | The solution selection, the risk assessment, the pilot results against the goal and the implementation plan | A pilot measured differently from the baseline, or no way to reverse the change |
| 5. Control | Will the gain last without the project team? | The control plan, the control chart, updated procedures, training records, the handover and the validated benefits | No owner for the control plan, or a gain measured over too short a period |
Why the Improve tollgate halts the checklist. Recycling at Define, Measure or Analyse costs analysis time. Full implementation changes the process for every shift and customer, spends the budget and is the hardest step to undo, so nothing after the Improve tollgate can be completed until the champion records a decision. The other four tollgates are decisions the champion records too, but they do not stop the checklist.
Continuous or discrete data. The main measure decides several tools. For a continuous measure such as time, weight or length, check the gauge with a repeatability and reproducibility study, express capability as Cp and Cpk, and monitor with an individuals or X-bar chart. For a discrete measure such as pass or fail, check the inspectors with an attribute agreement study, express performance as a defect rate or yield, and monitor with a p chart or a chart of defect counts. NIST’s e-Handbook of Statistical Methods defines Cp and Cpk for a stable, normally distributed process and suggests about 50 independent values as a large enough sample, so check stability before you quote capability.
The standards. ISO 13053-1:2011 covers the DMAIC methodology and ISO 13053-2:2011 the tools and techniques for each phase. iso.org lists both as first editions, last reviewed and confirmed in 2022. They are paid standards; this checklist follows common practice and does not reproduce either one.
The champion is chosen on the first task, and each tollgate is assigned to them alone. Go moves the project on, Recycle adds a rework task before the tollgate, and Stop adds a task to close the project properly, with the reason recorded. Nothing after the Improve tollgate can be completed until the champion decides: the approval step that protects full implementation.
Enter the planned date of each tollgate on the first task, and each phase’s tasks fall due in the weeks before its tollgate. When a tollgate moves, change its date and its tasks move with it. The Reports Grid shows every project in progress and which tasks are overdue, so whoever runs the improvement programme sees where each project stands.
Say whether the main measure is continuous or discrete, and the measurement system, capability and control chart tasks show the matching notes. A failed measurement system study adds a task to fix it, a solution that touches safety, regulation or the customer adds a change approval, and claimed savings add a finance check, all set up with conditions in the template designer.
CheckFlow does not run your statistics. Do the gauge study, the capability analysis and the hypothesis tests in your own statistics software or spreadsheet, and attach the output to the task that asked for it. CheckFlow runs the project around the analysis: who does each step, by when, with what evidence, and what the champion decided at each tollgate. CheckFlow for manufacturing shows how plants run their other quality checks the same way.
Rate the risks of a new solution with the FMEA Checklist, write the new procedures with the Process Standardization Checklist, and send a customer problem that needs containing to the Root Cause Analysis (8D) Checklist or the CAPA Checklist. For quicker improvement, use the Kaizen Event Checklist or the Continuous Improvement Cycle Process Checklist, and read the guide to continuous improvement for where DMAIC sits among the other models.
Describe the process in a sentence and the AI checklist generator drafts the phases and tasks for you.
DMAIC stands for Define, Measure, Analyse, Improve and Control, the five phases of a Six Sigma improvement project. Define sets the problem, goal and scope. Measure checks the measurement system and measures the baseline. Analyse finds the root causes and verifies them with data. Improve tests solutions in a pilot and rolls out the one that works. Control puts the controls in place so the gain lasts, then hands the process back to its owner.
A tollgate is the review at the end of each phase where the project’s champion checks the phase’s work and decides whether the project moves on. In this checklist the decision is Go, Recycle or Stop. Recycle sends the team back to rework something specific and resubmit; Stop ends the project with the reasons recorded. A good tollgate checks evidence, not slides: the charter, the measurement system result, the test results, the pilot data.
Both are improvement cycles. PDCA (Plan, Do, Check, Act) is short and repeated: a team plans a small change, tries it, checks the result and adopts or drops it, then starts again. DMAIC is one longer project with tollgates, a checked measurement system and root causes verified with data before any solution is chosen. Use the continuous improvement cycle checklist for PDCA and this checklist for a DMAIC project.
Use a kaizen event when the problem sits in one process, the team can see the waste and changes can be tried within days. Use DMAIC when the cause is unknown, the problem varies in ways people argue about, and the data needs analysing before anyone can say what to fix. A DMAIC project often finds a cause that a later kaizen event then fixes in the area.
The business case; a problem statement in measured terms with no solution in it; a goal statement with the measure, baseline, target and date; the scope, with the first and last process steps and what is out of scope; the team and their time; the champion and the process owner; and the planned tollgate dates. Keep it to one page, and update it when the Measure phase replaces the estimates with measured figures.
Because the baseline, the capability and every test that follows are only as good as the measurements. A gauge repeatability and reproducibility study for continuous data, or an attribute agreement study for pass or fail judgements, shows how much of the variation you see comes from the measurement itself. If the measurement system is not acceptable, fix the gauge, the method or the training and repeat the study before you collect the baseline.
Both compare the spread of a stable process with its specification limits. Cp uses only the width: the distance between the limits divided by six standard deviations. Cpk also allows for where the process is centred, using the distance from the mean to the nearer limit divided by three standard deviations, so it is never higher than Cp. NIST’s e-Handbook of Statistical Methods gives both definitions and notes that they assume normally distributed data.
Yes. ISO 13053-1:2011 describes the DMAIC methodology, including how Six Sigma projects should be managed and the roles and training of the people involved, and ISO 13053-2:2011 describes the tools and techniques for each phase. iso.org lists both as first editions, last reviewed and confirmed in 2022. This checklist follows common practice and does not reproduce either standard.
14-day free trial, no card required. The Business plan is $10 per user per month after the trial. Full details at checkflow.io/pricing.
14-day free trial, no credit cardFull refund in your first 30 daysUS hosting, EU on request
