Running a DMAIC Improvement Cycle

Five DMAIC phases in order — and the decision points where a smaller problem should drop to the 4-stage PDCA loop instead.

DMAIC is the Six Sigma project cycle: Define, Measure, Analyze, Improve, Control. Its finish line is statistical — a process running at no more than 3.4 defects per million opportunities.

Define and Measure turn a complaint into a number: what the problem is, who it hurts, and how many defects per million opportunities the process produces today. If the cause is already obvious at this point, switch to PDCA — a full Analyze phase wastes time on a problem you already understand.

Analyze finds the root causes behind the numbers, and Improve removes them and verifies the gain. This pair of phases justifies DMAIC's cost: when causes hide inside variation, only data separates them.

Control locks the fix — new standards, checks and named owners — so the defect rate stays down after the project team leaves. A project without a Control phase is a temporary improvement.

PDCA — Plan, Do, Check, Act — has 4 stages and suits fast, well-understood fixes: plan a change, try it small, check the result, act to adopt or adjust. Run DMAIC for heavy, data-driven problems; run PDCA loops inside Control to keep tuning.

Six Sigma arrived at Motorola in 1986 and went company-wide at General Electric in 1995–1996, and the method has stayed in use ever since. For the management system that surrounds these projects, see ISO 9001 Certification Step by Step.

• Define the problem and the customer it hurts.
• Measure the current defects per million opportunities.
• Analyze the data for root causes — or switch to PDCA if the cause is already clear.
• Improve the process and verify the gain.
• Control the new process so the gain survives the team.

Further reading

• Six Excellence Frameworks Compared
• ISO 9001 Certification Step by Step