Act as a Lean Six Sigma Black Belt helping me scope and run an improvement project. Build a practical DMAIC plan from the information below. Do not assume the solution is known or that automation is the answer. Project context: [PROCESS AND PROBLEM] Baseline evidence: [BASELINE DATA] Target and business case: [TARGET AND BUSINESS IMPACT] Team and constraints: [TEAM AND CONSTRAINTS] Timing: [TIMELINE] Produce a working project charter and action plan in these sections: 1. Problem statement, goal statement, scope boundaries, customers, CTQs, and a clear operational definition for each primary metric. Separate symptoms from the suspected root cause. 2. A DMAIC table. For each phase, list deliverables, 3-6 concrete activities, accountable owner, consulted roles, evidence required to exit the phase, and a realistic date or duration. 3. A measurement plan specifying numerator, denominator, unit, data source, sampling period, data owner, baseline, target, and likely data-quality problems. Include process time, defect/rework, and customer-impact measures where relevant. 4. Analyze phase hypotheses to test, including stratification by product, shift, channel, supplier, or handoff where relevant. Recommend appropriate basic tools such as SIPOC, process map, Pareto, run chart, fishbone, 5 Whys, control chart, or capability analysis, and state what decision each tool supports. 5. Improve-phase experiment plan: change, prediction, pilot population, success criteria, guardrail metrics, owner, and rollback rule. Do not present a countermeasure as proven before a test. 6. Control plan with standard work, monitoring cadence, control limits or escalation triggers, response owner, and a 30/60/90-day sustainment check. Flag missing baseline data, invalid comparisons, vague targets, and any claim the supplied evidence cannot support. Before finalizing, check that every activity has one accountable owner and that the target is measurable without changing its definition mid-project. Ask up to 3 clarifying questions only if a required input is missing.
Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [PROCESS AND PROBLEM] | Describe the process, where it starts and ends, who is affected, and the observed problem. | Inbound claims intake from receipt to assignment; 18% need rework because key fields are missing. |
| [BASELINE DATA] | Provide current volumes, timing, defect or error rates, complaints, and how the data was collected. | 1,200 claims/month; median assignment time 2.8 days; rework sampled from April-June 2026. |
| [TARGET AND BUSINESS IMPACT] | State the desired measurable outcome and the cost, revenue, service, or risk impact. | Reduce rework to 8% and median assignment time to 1.5 days; delays create overtime and complaint risk. |
| [TEAM AND CONSTRAINTS] | List available roles, decision makers, systems, budget limits, and non-negotiable constraints. | Operations manager, intake lead, two analysts, IT liaison; no new software before Q1; union work rules apply. |
| [TIMELINE] | Give the deadline, key dates, and how much staff time the project can use. | Twelve weeks, with pilot decision in week 8 and results reviewed by September 30. |
How to use
- Paste the prompt and replace every bracketed field with your project facts.
- Keep the measurement-plan definitions; they prevent a team from claiming improvement with inconsistent data.
- Check that the proposed analysis tests causes before it recommends fixes.
- Follow up with: “Turn the DMAIC table into a weekly meeting agenda and RACI for the next 12 weeks.”
Variations
Project charter
Use this when the problem is real but the project has not been approved yet.
Create a one-page Lean Six Sigma project charter for [PROCESS], where [PROBLEM] is causing [BUSINESS IMPACT]. Use this baseline: [BASELINE]. Define in-scope and out-of-scope work, customers and CTQs, primary and guardrail metrics with operational definitions, a SMART goal, sponsor, project lead, team roles, assumptions, risks, and a 90-day milestone plan. Distinguish facts from hypotheses. Flag any goal that lacks a denominator, time period, or data source. Ask up to 3 questions only if a required field is missing.
Root cause analysis
Use this after you have mapped the process and collected early evidence.
Help me plan a root-cause analysis for [DEFECT OR DELAY] in [PROCESS]. Available data: [DATA]. Current process steps and handoffs: [PROCESS MAP NOTES]. Suspected causes: [INITIAL HYPOTHESES]. Produce a cause-and-effect structure organized by people, process, technology, materials, measurement, and environment; then rank which hypotheses deserve testing first. For each, name the data slice, comparison, chart or test, decision rule, and confounder to watch for. Do not call a cause confirmed without evidence. Flag data gaps and ask up to 3 questions if needed.
Control plan
Use this after a pilot has shown a change can improve the process.
Create a Lean Six Sigma control plan for [IMPROVED PROCESS] after this pilot result: [PILOT RESULTS]. The critical measures are [CTQS AND METRICS], and the process owners are [OWNERS]. Produce a table with each metric, operational definition, data source, sample frequency, target or control limit, dashboard owner, escalation trigger, immediate containment action, root-cause response, and review cadence. Include standard-work changes, training evidence, and 30/60/90-day audits. Flag any result that is too short or too small to support a sustainment claim. Ask up to 3 questions if necessary.
Tips
- Write the problem statement in terms of a measurable gap, not a preferred fix; “missing fields cause rework” is more useful than “we need a new form.”
- Define defects and denominators before comparing periods, teams, or suppliers, especially when volumes vary.
- Use a pilot with guardrail measures so a faster process is not accepted if accuracy, safety, or customer outcomes deteriorate.
- Assign one accountable owner per action; a department or committee cannot close an action item.
FAQ
Can AI run the statistical analysis for my Six Sigma project?
It can help choose analyses and interpret results you provide, but validate calculations, sampling, and statistical assumptions with your data analyst or Black Belt.
Should I use DMAIC for every process issue?
Use DMAIC for an existing process with a measurable performance gap. A new process design may fit DMADV or a simpler discovery and pilot approach better.
What data should I collect first?
Collect a consistently defined outcome, volume or opportunity count, timestamps, and meaningful segments such as channel, shift, product, or handoff.