---
name: six-sigma
description: Use when asked about the Six Sigma quality-management methodology as a whole — belt-level roles, sigma levels/defects-per-million, and choosing between DMAIC (improving an existing process, see dmaic) or DMADV/DFSS (designing a new one) — as distinct from running the DMAIC cycle itself, which this skill defers to for the step-by-step improvement procedure.
---

# Six Sigma

Six Sigma is a data-driven quality-management methodology aimed at
reducing process variation and defects to a statistically rigorous
standard — the name refers to a process performing within six standard
deviations ("sigma") of the mean between the process center and the
nearest specification limit, equivalent to about 3.4 defects per million
opportunities.

## The two core methodologies

- **DMAIC** (Define, Measure, Analyze, Improve, Control) — for improving
  an *existing* process; see [[dmaic]] for the full five-stage
  procedure. This is the methodology most Six Sigma projects use, since
  most improvement work targets a process that already exists and is
  underperforming.
- **DMADV / DFSS** (Design for Six Sigma: Define, Measure, Analyze,
  Design, Verify) — for designing a *new* process or product to Six
  Sigma quality from the outset, used when no existing process can be
  incrementally improved to the required performance level, or when the
  product/process doesn't exist yet.

## Belt levels

- **Yellow Belt** — basic Six Sigma awareness; participates in projects
  as a team member.
- **Green Belt** — leads smaller improvement projects part-time,
  alongside their regular job.
- **Black Belt** — leads major Six Sigma projects full-time, mentors
  Green Belts, and typically has deep statistical-analysis training.
- **Master Black Belt** — trains and coaches Black Belts, sets Six Sigma
  strategy across the organization, and consults on the most complex
  statistical problems.

The belt structure mirrors martial-arts ranking deliberately: it signals
that competence is progressive and trained, not a one-time certificate.

## Sigma levels and defect rates

| Sigma level | Defects per million opportunities |
|---|---|
| 2σ | ~308,000 |
| 3σ | ~66,800 |
| 4σ | ~6,210 |
| 5σ | ~233 |
| 6σ | ~3.4 |

Most organizations that haven't invested in process improvement operate
around 3–4σ; moving even one sigma level higher typically requires
substantially more process discipline and measurement rigor than the
previous level, which is why Six Sigma treats it as a genuine
methodology rather than a target to simply declare.

## Six Sigma vs. Lean vs. Vanguard Method

Six Sigma focuses on reducing *variation* through statistical analysis;
Lean focuses on eliminating *waste* and non-value-adding steps — the two
are frequently combined as "Lean Six Sigma." Both originated in, and are
most directly suited to, manufacturing and other high-volume,
repeatable-process contexts. [[vanguard-method]] specifically
argues that applying either without first studying a service
organization's demand and flow can miss the larger source of
"defects" in service work: system-design failure demand, not process
variation.

## Common pitfalls

- **Choosing DMAIC when the process doesn't exist yet, or DMADV when it
  does** — the two methodologies solve different problems; using the
  wrong one either forces an "improve" project to design from nothing or
  a "design" project to redundantly re-derive a process that already
  exists.
- **Chasing a sigma-level target as a number rather than as evidence of
  reduced variation** — see [[vanguard-method]]'s broader warning
  about target-driven management distorting what people actually attend
  to.
- **Applying Six Sigma's statistical, manufacturing-derived toolkit to
  highly variable, judgment-heavy service work without first checking
  whether the fit is right** — see [[vanguard-method]] for the
  detailed argument against this.
- **Treating belt certification as equivalent to project success** — the
  belt structure signals training, not a guarantee that a given project
  will be executed well.

## Learn more

- [[dmaic]] for the full Define-Measure-Analyze-Improve-Control procedure.
- [[critical-to-quality]] for translating customer requirements into the measurable characteristics a Six Sigma project targets.
- [[vanguard-method]] for a systems-thinking critique of applying Six Sigma uncritically to service organizations.
- [[kaizen]] for the contrasting continuous, incremental (rather than project-scoped, statistical) improvement model.