Six Sigma: what it is, the principles, DMAIC and the belts

What is Six Sigma?

Six Sigma is both a measure and a method. As a measure, it describes how often a process produces defects. As a method, it is a structured way of running improvement projects with data.

The core idea is that customers feel variation, not averages. A delivery that is on time on average but sometimes a week late still lets customers down. Six Sigma aims to make the process consistent, then keep it there.

Six Sigma projects are run by trained practitioners, with a sponsor from management and a clear financial or customer goal. They often use statistical tools such as hypothesis tests, regression and designed experiments, alongside simple ones such as the fishbone diagram.

What does 6 sigma mean in numbers?

Sigma (σ) is the Greek letter used for standard deviation, a measure of spread. A process at six sigma has its specification limits six standard deviations from the mean, so very little of its output falls outside them.

By convention, Six Sigma tables allow for the process mean to drift by 1.5 sigma over time. With that shift, six sigma equals about 3.4 defects per million opportunities (DPMO). Four sigma is about 6,210 DPMO and three sigma about 66,807.

How you count opportunities changes the answer. Count 10 fields per invoice instead of 5 and the same 37 defects give 3,700 DPMO, which looks better without anything changing. Agree the opportunity list once and keep it fixed while you compare results.

You can turn your own defect counts into a sigma level with the Six Sigma calculator, and check how a measured process fits its limits with the process capability calculator.

Where Six Sigma came from

Six Sigma was developed at Motorola in the mid-1980s. Engineer Bill Smith is usually credited with the approach, and Motorola registered Six Sigma as a trademark.

General Electric adopted it across the company in the mid-1990s under chief executive Jack Welch, which made it widely known. It then spread to other manufacturers, banks, hospitals and service companies.

Its statistical roots are older. Walter Shewhart's control charts from the 1920s and W. Edwards Deming's teaching on variation are the base that Six Sigma builds on. See statistical process control.

Six Sigma principles

Most descriptions of Six Sigma come back to a handful of principles.

1. 01

   Focus on the customer

   Define quality by what is critical to the customer, often written as CTQs (critical to quality requirements). The DMAIC page shows how a complaint becomes a CTQ.

2. 02

   Use data

   Measure the process and verify causes with data before acting, rather than relying on opinion.

3. 03

   Reduce variation

   Work on the spread of results, not only the average, because customers feel the bad days.

4. 04

   Improve the process

   Treat defects as products of the process, and fix the process rather than blaming people.

5. 05

   Sustain the gain

   Build in controls, such as control charts and standard work, so results do not slide back.

DMAIC: the Six Sigma cycle

1. Define

   Sets the problem, scope, goal and customer requirements in a project charter.

2. Measure

   Checks the measurement system and builds a baseline of how the process performs today.

3. Analyze

   Finds and verifies the root causes of defects and variation.

4. Improve

   Tests and implements solutions that remove those causes.

5. Control

   Keeps the gain with a control plan, monitoring and updated standards.

The DMAIC page covers each phase, its tools and the tollgate reviews in more detail. New processes use a variant called DMADV: define, measure, analyze, design and verify.

A DMAIC project: an illustrative case

Here is how one problem might move through all five phases. The numbers are made up to show the flow, not taken from a real company.

1. D

   Define

   Customers complain about late invoices. The charter states the problem as "23% of invoices go out more than 2 days after shipment" and sets a goal of under 5% within four months. The sponsor is the finance manager.

2. M

   Measure

   The team checks that shipment and invoice dates in the system match the paperwork, then pulls three months of data as a baseline. Late invoices run at 23%, with a wide spread from day to day.

3. A

   Analyze

   A Pareto chart shows that 70% of late invoices wait for a missing purchase order number. Testing the data shows that most of them come from customers who order by phone, where no one asks for the number.

4. I

   Improve

   The team adds the purchase order number as a required field on the phone order form and runs a two-week pilot with one sales desk. Late invoices on that desk fall to 4%.

5. C

   Control

   The required field goes live for all desks. A weekly control chart of the late invoice rate goes to the finance manager, with a rule to investigate any week above 7%. The project closes when the rate has held for two months.

Six Sigma belts

1. White belt

   A short introduction to Six Sigma terms and roles, for anyone working near a project. Training is usually short and has no project. See the white belt guide.

2. Yellow belt

   The basics of DMAIC and simple tools, so a person can take part in projects as a team member, often collecting data for a green belt. See the yellow belt guide.

3. Green belt

   Enough statistics and project skills to lead smaller projects, often alongside a regular job. Many providers expect one completed project, and some don't. See the green belt guide.

4. Black belt

   Advanced statistics and change leadership, to lead larger projects full time and coach green belts. Certification usually asks for project experience as well as an exam. See the black belt guide.

5. Master black belt

   An expert who trains and mentors black belts and helps choose projects for the organization.

Above the belts sits the champion, or sponsor: a senior manager who picks projects that matter to the business and removes the barriers a team cannot. The Six Sigma belts guide compares all five levels and the champion role.

Six Sigma certification

There is no single official body for Six Sigma. Certifications come from professional associations, universities, consultancies and employers, and they differ in exam rigor, project requirements and cost.

Before choosing, check whether the program needs a completed project, whether the exam is proctored, and whether employers in your field recognize it. The Lean Six Sigma certification guide compares providers.

What is Lean Six Sigma?

Lean Six Sigma combines the two approaches. Lean removes waste and speeds up flow, and Six Sigma reduces variation and defects. Together they address both how fast and how reliably a process delivers.

In practice, Lean Six Sigma projects follow DMAIC and use lean tools such as value stream mapping, 5S and kanban alongside statistical ones. The lean vs Six Sigma guide explains where each fits.

Is this a Six Sigma problem?

A Six Sigma project takes months of a trained person's time, so check the problem first. It is a good candidate when several of these are true:

• It keeps coming back. The same defect or complaint returns after each quick fix.
• Results vary widely. Performance differs between shifts, sites, teams or systems.
• It can be measured. You can count the defects or time the process, or could start doing so.
• The cause is unclear. People disagree about why it happens.
• It crosses departments. No single team owns the whole process.
• It shows up in the numbers. Rework, complaints or lost orders are visible to customers or in costs.

If the cause and the fix are already obvious, skip the project and make the change. If the problem is work waiting between steps rather than variation, start with lean, as the next section explains.

When Six Sigma is the wrong tool

Six Sigma works on variation in processes that repeat often and produce data. It fits poorly in three common cases.

Common Six Sigma myths

Six Sigma has a reputation for being heavy and statistical. Some of that is earned, but four common beliefs get in the way of using it well.

Six Sigma tools and software

Common Six Sigma tools include SIPOC diagrams, process maps, Pareto charts, fishbone diagrams, measurement system analysis, capability studies, hypothesis tests, designed experiments and control charts.

Statistical packages and project tracking tools support this work. The Six Sigma software and SPC software buying guides list options, and the Lean Six Sigma tools hub covers the rest of the stack.

Frequently asked questions