The DMAIC Methodology

DMAIC stands for Define, Measure, Analyze, Improve, Control. It is the standard problem-solving framework in Six Sigma, designed for situations where a process exists but is not performing to the level you need.

The five phases are sequential. Each one builds on the output of the previous one. Skipping a phase, which is common when people are impatient for results, typically means the solution addresses symptoms instead of causes.

The Define phase answers three questions: What is the problem? Who does it affect? What does success look like?

You write a problem statement, identify the customer (internal or external), and set a measurable project goal. "Reduce scrap" is not a project goal. "Reduce scrap on Line 3 injection molding from 4.2% to below 2.5% within 90 days" is.

The output is a project charter that the team and management agree on before any data collection begins. Without this agreement, scope creep will eat the project alive.

The Measure phase collects baseline data on the current process. You need to know how bad the problem actually is before you can claim improvement.

This phase forces you to define your measurement system: what exactly are you measuring, how, and how often? A surprising number of "quality problems" turn out to be measurement problems. If two inspectors measure the same part and get different results, your measurement system needs work before you touch the process.

In the Analyze phase you look at the data to find root causes. Tools like Pareto charts, fishbone diagrams, scatter plots, and hypothesis testing help you separate real causes from assumptions.

The discipline here is resisting the urge to jump to solutions. Most teams arrive at the Analyze phase already "knowing" what the fix should be. The data often tells a different story. A machine operator might blame raw material quality while the data shows that 70% of defects occur during the first 15 minutes after a changeover.

The Improve phase tests and implements solutions that address the root causes you identified. Pilot runs, designed experiments, and before/after comparisons provide evidence that the change works.

The Control phase locks in the gains. You update standard operating procedures, set up monitoring (control charts, dashboards), and define response plans for when the process drifts. This phase is where most improvements die. Without control mechanisms, processes revert to their old behavior within weeks.

That reversion is not a people problem. It is a systems problem. If the control plan lives in a spreadsheet nobody checks, it is not a control plan.