What Is Six Sigma?

Six Sigma is a data-driven method for improving quality by reducing variation and defects in a process. A process operating at Six Sigma quality produces no more than 3.4 defects per million opportunities. It was developed at Motorola in 1986 and uses a defined project structure, statistical tools, and trained practitioners to find the causes of variation and remove them. It is a separate methodology from Lean, and the two are frequently combined as Lean Six Sigma.

What "six sigma" actually means

Sigma (σ) is the statistical symbol for standard deviation, a measure of how much a process varies around its average. A "six sigma" process is one where six standard deviations fit between the process mean and the nearest specification limit. The further the spec limits sit from the natural variation of the process, the less likely any single output falls outside them.

Strictly, a centered process with six standard deviations to the nearest limit would produce about 2 defects per billion, not 3.4 per million. The 3.4 figure comes from a Motorola assumption that any process mean drifts by up to 1.5 sigma over time, so a process that measures six sigma in the short term performs like 4.5 sigma in the long term. The 1.5 sigma shift is debated, but 3.4 DPMO is the number the method is built around.

Most manufacturing processes run between three and four sigma. The gap between 93% yield and 99.99966% yield is the business case for the method.

An opportunity is any chance for a defect to occur: each critical dimension on a part, each field on a form, each solder joint on a board. Counting opportunities rather than units lets processes of different complexity be compared on the same scale.

A short history

Six Sigma was developed at Motorola in 1986, with engineer Bill Smith generally credited as its originator. Motorola won the first Malcolm Baldrige National Quality Award in 1988 partly on the strength of the program. General Electric adopted Six Sigma company-wide in 1995 under Jack Welch, tied it to management bonuses, and reported billions in savings, which is what turned it from a Motorola practice into a global standard. Through the 2000s it spread to healthcare, finance, and services, and merged with Lean into Lean Six Sigma.

The DMAIC methodology

DMAIC (pronounced "duh-MAY-ick") is the five-phase structure for improving an existing process. Every Six Sigma improvement project follows it.

1. Define. Write the problem statement, the project scope, the customer requirements (critical-to-quality characteristics), and the business case. Output: a project charter.
2. Measure. Map the current process, decide what to measure, confirm the measurement system is reliable (measurement system analysis), and collect baseline data. Output: a current sigma level and a validated data set.
3. Analyze. Find the root causes of variation. This phase uses root cause analysis tools such as fishbone diagrams and the 5 Whys, Pareto charts to prioritize, and statistical tests (hypothesis testing, regression, ANOVA) to confirm which inputs actually drive the output. Output: verified root causes.
4. Improve. Design, pilot, and implement solutions that address the verified causes. Design of experiments is often used here to find the input settings that minimize variation. Output: an improved process with measured results.
5. Control. Lock in the gains: control charts to monitor the process, updated standard work, a response plan for when the process drifts, and handoff to the process owner. Output: a control plan.

DMADV for new processes

When the goal is a new product or process rather than fixing an existing one, Six Sigma uses DMADV (Define, Measure, Analyze, Design, Verify), also called Design for Six Sigma (DFSS). The first three phases are similar; Design replaces Improve with building and optimizing the new design, and Verify replaces Control with pilot runs and handoff to production.

DMAIC in practice

A bottling line is rejecting 2% of units for under-fill. Define: reduce under-fill rejects on Line 3 from 2% to under 0.2% in 90 days. Measure: gauge R&R confirms the fill-weight scale is reliable; two weeks of data show rejects cluster on the first hour of each shift. Analyze: regression shows fill weight tracks product temperature, which is low after the line sits idle; the fill valve timing was set for warm product. Improve: add a temperature-compensated fill timing table; pilot on one shift; rejects drop to 0.1%. Control: add fill weight to the hourly control chart, update the setup standard, train all shifts.

Six Sigma belts

Six Sigma borrows martial arts belt levels to describe practitioner training and roles.

• White Belt. Basic awareness; can support a project team.
• Yellow Belt. Understands the fundamentals and participates in projects as a team member.
• Green Belt. Trained in DMAIC and the core statistical tools; leads smaller projects, usually part-time alongside a regular job.
• Black Belt. Full-time project leader with advanced statistical training; leads complex, cross-functional projects and coaches Green Belts.
• Master Black Belt. Trains and mentors Black Belts, owns the program methodology, and works with leadership on project selection.
• Champion or Sponsor. An executive who selects projects, removes obstacles, and is accountable for results.

There is no single governing body. ASQ and IASSC offer the most widely recognized certifications; many large companies run their own internal belt programs.

Core Six Sigma tools

• Statistical process control (SPC) and control charts. Plot process output over time with calculated limits to distinguish normal variation from a real change.
• Process capability (Cp, Cpk). Compare the spread of the process to the width of the specification. Cpk of 2.0 corresponds to six sigma.
• Measurement system analysis (gauge R&R). Confirm that the measurement itself is not the source of the variation.
• Hypothesis testing, regression, ANOVA. Confirm statistically which inputs affect the output.
• Design of experiments (DOE). Vary several inputs at once in a structured way to find the settings that minimize variation.
• FMEA. Rank potential failure modes by severity, occurrence, and detection to prioritize prevention.
• Fishbone diagrams, 5 Whys, Pareto charts. The same root cause analysis tools used throughout Lean.

Six Sigma vs. Lean

Lean and Six Sigma attack different problems. Lean removes waste to improve flow and speed; its tools are visual and are applied daily on the floor by the people doing the work. Six Sigma reduces variation to improve quality; its tools are statistical and are applied in defined projects by trained practitioners. Lean asks "why does this take so long?" Six Sigma asks "why is this inconsistent?"

Lean Six Sigma combines them: Lean to strip non-value-adding steps out of a process, Six Sigma to make the steps that remain predictable. In practice most programs start with Lean basics such as 5S and kaizen, because a disorganized process generates variation that no amount of statistics will fix, then apply Six Sigma projects to the chronic quality problems that remain.

Criticisms and limits

Six Sigma has real limits. Projects are slow and resource-heavy, which makes the method poorly suited to small problems or small companies. Rigid application can crowd out quick, informal improvement. The 1.5 sigma shift is a convention, not a law. And some high-profile adopters later scaled back their programs after concluding the structure was discouraging innovation. The method works best for chronic, high-cost variation problems where the cause is not obvious, and worst as a mandatory framework for every improvement.

Frequently asked questions

Is Six Sigma part of Lean?

No. They are separate methodologies with different origins (Toyota for Lean, Motorola for Six Sigma) and different goals. They are often combined as Lean Six Sigma.

What does 3.4 defects per million mean?

A process at Six Sigma quality has a 99.99966% yield: 3.4 defects for every million opportunities for a defect to occur. The figure assumes a 1.5 sigma long-term drift in the process mean.

Do I need a belt certification to use Six Sigma?

No. The tools can be learned and applied without certification. Certification matters when an employer or customer requires it, or for leading formal projects.

What is the difference between DMAIC and DMADV?

DMAIC improves an existing process. DMADV (Design for Six Sigma) designs a new process or product to meet Six Sigma quality from the start.

Which industries use Six Sigma?

Manufacturing, especially automotive, electronics, aerospace, and pharmaceuticals, plus healthcare, financial services, and logistics.

Is Six Sigma still relevant?

Yes, though it is now usually practiced as Lean Six Sigma rather than on its own. The statistical tools are standard in quality engineering regardless of what the program is called.

Related tools and resources

Six Sigma sits alongside the other quality methods covered here: Total Quality Management, Zero Defects, quality control, and quality assurance. For the full body of knowledge, see ASQ's Six Sigma resources. For the visual workplace that Lean Six Sigma starts from, 5S Today stocks Lean manufacturing tools and 5S supplies.