FREE IMPROVEMENT WORKSPACE
DMAIC
Build the investigation from problem definition to sustained checks.
How this works
What you’ll get · Keep the question, evidence, trial and control plan connected. Each phase needs a review; a completed box is not approval.
Use the tools and save work for free. Data exports and reports require Plus or Teams.Compare plans →
How to run a DMAIC project that holds up
DMAIC improves an existing process in five phases, each closed by a tollgate review with the sponsor. The studio keeps one measure from the charter to the control chart, so every phase answers the same question: is the problem smaller, and will it stay smaller?
Define
Write the problem as facts — what, where, since when, how much — without a cause or a solution. Set a goal on one measure with a date, say what is in and out of scope, and turn the voice of the customer into one CTQ requirement. A SIPOC of 4–7 steps keeps the team on the same process. Example: 4 % of cartons need a label correction; the goal is 2 % by the end of the quarter.
Measure
Write an operational definition so two people count the same thing, then check it: in an agreement study two people rate the same items, and 90 % agreement or more means the data can be trusted. Record the baseline period by period. For a share the studio gives DPMO and process sigma — 4 % defective is 40,000 DPMO, about 3.25 sigma — and a control chart shows whether the baseline is stable.
Analyze
List possible causes and test each one. A cause is supported only with evidence; a data check makes it stronger. Compare two groups on the same measure — with and without a changeover, cavity 4 against the rest — and read the p-value: below 0.05, a difference that large is unlikely to be chance. Then confirm the mechanism at the process.
Improve
Choose solutions aimed at supported causes and rate them on the countermeasure ladder: training and reminders are the weakest rung, making the problem impossible is among the strongest. Pilot the change, mark when it started and compare before and after. The studio tests whether the improvement could be chance before you claim it.
Control
Hold the gain with a control plan: what is checked, how, the limit, how often, who, and what happens when it drifts. A control chart after the change shows the new level and its limits. Hand the plan to the process owner, record the benefits and how they were worked out, and close the project with the lessons learned.
DMAIC or A3?
Use an A3 for a problem one team can solve in weeks by direct observation. Use DMAIC when the problem sits in a process with data, the cause is not obvious, and the result has to be proven and held. An A3 can escalate to DMAIC: the studio carries the improvement case across.
The tollgate check
Before each tollgate the studio checks the storyline: nine required links — charter, measure, baseline, supported cause, solution, plan, result, control plan and decision — and nine advised ones, such as a trustworthy measurement, a stable baseline, a significant data check and a significant improvement. Each missing link takes you to the field that completes it.
Check the method with a worked example
Leave evidence for each stage decision, from the scoped problem to the response plan. This workspace is not a statistical capability suite or certification.