---
name: dmaic
description: Use when asked to run a Six Sigma DMAIC improvement project — Define, Measure, Analyze, Improve, Control — as a data-driven, project-scoped alternative to Kaizen's continuous, incremental cycle (see kaizen) or PDCA's general iterative loop (see plan-do-check-act).
---

# DMAIC

DMAIC is the core problem-solving methodology of the [[six-sigma]]
approach to continuous improvement — Define, Measure, Analyze, Improve,
Control — each stage building the data and consensus the next stage
needs, aimed at achieving a measurably higher level of quality and
efficiency in an existing process.

## The five stages

1. **Define** — clearly state the problem to be solved and the project's
   goals; assemble the team leading the effort, and create a project
   charter (see [[project-charter]]) outlining scope, objectives,
   timeline, and required resources. A DMAIC project without a clear
   Define stage tends to drift, since Measure/Analyze have nothing
   specific to be measured or analyzed *against*.
2. **Measure** — measure the current process and gather data on the
   problem: determine what to measure, then collect it from process maps,
   customer feedback, and statistical process control data. This stage
   establishes the factual baseline every later stage is judged against.
3. **Analyze** — analyze the collected data to find the problem's root
   cause, not just its symptoms — a detailed pass over the Measure
   stage's data specifically aimed at causal explanation, not just
   description.
4. **Improve** — develop and implement a plan addressing the identified
   root cause: brainstorm improvement ideas, then evaluate the candidate
   solutions to select the best one, rather than implementing the first
   plausible fix.
5. **Control** — ensure the improvement actually sticks: develop a plan to
   monitor the process and measure ongoing performance, confirming the
   change remains effective over time rather than reverting once
   attention moves elsewhere.

## Why the strict sequence matters

Each stage depends on the previous one's output: Analyze without a solid
Measure baseline is guesswork; Improve without a genuine root-cause
Analyze fixes a symptom, not the actual problem; skipping Control means an
improvement that looked successful in a pilot can silently regress once
the project team moves on. The discipline of *not* skipping ahead is
DMAIC's core value, more than any individual stage's specific technique.

## DMAIC vs. Kaizen vs. PDCA

All three are structured improvement cycles, but differ in scope and
cadence: DMAIC is a **project-scoped**, heavily data-driven methodology
typically run by a dedicated team for a defined, often significant
problem (see [[kaizen]] for the contrasting continuous,
everyone-participates, incremental-improvement model, and
[[plan-do-check-act]] for the more general, lighter-weight
iterative cycle DMAIC's own stages loosely map onto — Define/Measure to
Plan, Improve to Do, Control to Check/Act).

## Common pitfalls

- **Jumping to Improve before Analyze establishes a real root cause** —
  produces a fix for a guessed cause rather than the actual one, which
  often doesn't resolve the original problem.
- **No Control stage plan** — an improvement validated once but never
  monitored afterward is prone to silently reverting once attention
  moves elsewhere.
- **A vague Define stage** — without a clear project charter and scope,
  later stages lack a concrete target to measure and analyze against.
- **Treating DMAIC as the right tool for every improvement** — its
  project-scale, data-intensive nature suits a significant, well-defined
  problem; for smaller, everyday process tweaks, [[kaizen]]'s
  lighter continuous-improvement model is usually a better fit.

## Learn more

- [[six-sigma]] for the broader methodology DMAIC is the core cycle of.
- [[kaizen]] for the contrasting continuous, incremental improvement model.
- [[plan-do-check-act]] for the general iterative-improvement cycle DMAIC's stages loosely parallel.
- [[project-charter]] for the Define stage's charter artifact.