What Is Six Sigma? A Plain-English Guide to the Method, the Numbers and the Criticisms
Six Sigma is a data-driven method for making a process produce fewer defects and less variation. The name comes from statistics, and the famous goal is 3.4 defects per million opportunities. This guide explains the idea without jargon, shows where the numbers come from, covers the history, belts and tools, and includes the criticisms you rarely see in training brochures.
The definition
ASQ, the professional body for quality, describes Six Sigma as “a method for disciplined quality improvement” that gives organizations tools to improve the capability of their business processes. It describes the approach as fact-based and data-driven, and as valuing defect prevention over defect detection.
In plain English:
- Measure how often a process fails.
- Find out why, using data instead of hunches.
- Fix the cause, and then make sure it stays fixed.
A defect is anything that fails to meet a customer requirement: a late delivery, a wrong invoice, a part outside its tolerance. An opportunity is a chance for a defect to occur.
Why variation matters
Averages hide problems. Imagine two pizza shops that both promise delivery in about 30 minutes. Here are five deliveries from each (made-up numbers):
| Shop | Delivery times (minutes) | Average | Spread (standard deviation) |
|---|---|---|---|
| Shop A | 28, 29, 30, 31, 32 | 30 | About 1.4 |
| Shop B | 15, 22, 30, 38, 45 | 30 | About 10.8 |
Both average 30 minutes. But Shop A is predictable, and Shop B leaves some customers waiting 45 minutes while others get their food in 15. Customers experience the spread, not the average.
Six Sigma attacks that spread. The less a process varies, the fewer results fall outside what the customer will accept.
What “sigma” means
Sigma (σ) is the symbol for standard deviation, a measure of spread. A “six sigma” process is one where six standard deviations fit between the process average and the nearest limit the customer will accept. When the spread is that small, defects become very rare.
Long-term, a six sigma process is assumed to produce at most 3.4 defects per million opportunities (DPMO). That figure includes an allowance for the process average drifting by 1.5 standard deviations over time. The table shows what each sigma level means under that convention. We calculated the values from the normal distribution, and they match the standard published table.
| Sigma level | Defects per million opportunities | Yield (good results) |
|---|---|---|
| 1 | 691,462 | about 30.9% |
| 2 | 308,538 | about 69.1% |
| 3 | 66,807 | about 93.3% |
| 4 | 6,210 | about 99.38% |
| 5 | 233 | about 99.977% |
| 6 | 3.4 | about 99.99966% |
Each step up cuts defects sharply. A 93 percent yield sounds good until you do the math: at 3 sigma, nearly 67,000 out of every million chances go wrong.
To calculate DPMO for your own process, see the formula in our DMAIC guide. Note that the 1.5-sigma shift is debated (see criticisms).
A short history
| Year | Event |
|---|---|
| 1986 | Engineer Bill Smith introduces Six Sigma at Motorola. |
| 1995 | Jack Welch, CEO of General Electric, makes Six Sigma central to GE’s business strategy. |
| Late 1990s | Wikipedia reports that about two-thirds of Fortune 500 organizations had begun Six Sigma initiatives. |
| Early 2000s | Six Sigma is combined with lean thinking under the name Lean Six Sigma. See Six Sigma vs. Lean Six Sigma. |
| 2011 | ISO publishes ISO 13053:2011, described as the first standard defining a Six Sigma process. |
Dates in this table come from the Wikipedia article on Six Sigma, except the lean merger, which is covered in our other guides. Savings claims made by companies are their own figures, so treat them with care.
Core ideas
| Idea | What it means |
|---|---|
| Customer first | Quality is defined by what the customer requires, not by what the company thinks is good enough. |
| Data over opinion | Decisions rest on measurements. If it has not been measured, it is a guess. |
| Process thinking | Results come from processes. Fix the process, not the person. See what a business process is. |
| Prevention over inspection | Stop defects from happening instead of finding them afterward. |
| Reduce variation | Consistent results are easier to predict, plan and trust. |
| Sustain the gain | An improvement is not finished until it has an owner and a control plan. |
The DMAIC method
A Six Sigma improvement project follows five steps, known as DMAIC:
- Define the problem, goal and customer requirements.
- Measure current performance.
- Analyze to find root causes.
- Improve by testing and putting in solutions.
- Control to hold the gain.
Each phase has its own tools and exit checks, covered in our DMAIC guide. For new products or services, a related method called DMADV is used. See DMADV and DMAIC vs. DMADV.
Roles and belts
Six Sigma borrows martial-arts language for its roles. The names vary between organizations.
| Role | What they do |
|---|---|
| Executive leadership | Sets the vision and clears barriers between departments. |
| Champion | An executive who sponsors a project, supplies resources and removes obstacles. |
| Master Black Belt | An experienced Black Belt who trains and mentors others. |
| Black Belt | Leads improvement projects, usually full time. |
| Green Belt | Works on project teams while keeping a regular job. |
| Yellow Belt | In some organizations, a basic level for team members. |
Roles come from ASQ and Wikipedia. For more, see Six Sigma belt rankings, the Green Belt course guide and Six Sigma certification.
Common tools
| Tool | What it is for |
|---|---|
| Process map | Draws the steps so everyone sees the same picture. |
| Pareto chart | Shows which few causes produce most of the defects. |
| Control chart | Shows whether a process is stable over time. |
| FMEA | Ranks possible failures by how serious and likely they are. |
| Design of experiments | Tests several factors at once to find what really matters. |
| Statistical process control | Monitors a process with statistics to catch problems early. |
Benefits
ASQ lists benefits that include fewer defects and errors, less process variation, and gains in profits, employee morale, customer satisfaction and product and service quality. It also notes reduced waste and resource use.
Practical examples are in our guides to benefits for engineers, benefits in finance and improving healthcare. These are organizational benefits claimed by the method’s supporters. Results depend on how well a project is chosen and run.
Criticisms
A fair guide should include them. According to Wikipedia’s summary of critics:
| Criticism | What critics say |
|---|---|
| Stifles creativity | A tight focus on reducing variation can hurt research and discovery. It is said to favor small, steady improvements over breakthroughs. |
| The 1.5-sigma shift | Statistician Donald Wheeler called it arbitrary. Whether the same shift applies to every process is doubtful. |
| Weak evidence | Researchers have said much of the literature lacks academic rigor, with case studies often supplied by consultants. |
| Does not create new products | It is best at fixing existing processes, not inventing new offerings. |
| Standard may be too lax for some products | Quality author Philip Crosby argued 3.4 defects per million is too many for items such as microchips. |
| Many projects fail | A 2010 newspaper report cited by Wikipedia said that more than 60 percent of projects fail. |
These are claims from critics, some from sources with their own interests, so weigh them as you would the supporters’ claims. A sensible reading: Six Sigma is a useful tool for reducing defects in repeatable processes, and a poor tool for creative work.
Where it is used
ASQ says Six Sigma applies wherever variation and waste exist and that every employee should be involved. It began in manufacturing, but the same logic fits any repeatable process:
- Manufacturing: part dimensions, scrap, rework.
- Healthcare: wait times, medication errors, discharge delays.
- Finance: invoice errors, loan processing time.
- Customer service: resolution time, repeat calls.
- Government: permit and application processing.
The industry examples are ours. Where a process has no clear repeatable steps, such as early-stage product design, other methods usually fit better.
How to get started
- Learn the vocabulary. Defect, opportunity, variation, process, customer requirement.
- Pick one small process you know. Count its defects for a few weeks.
- Draw it. Make a simple process map and note where problems appear.
- Try a Pareto chart. Find the few causes behind most problems.
- Take a course. Online providers such as Coursera and Udemy offer Six Sigma courses. We may earn a commission if you sign up, at no cost to you. We are not affiliated with any university.
- Apply it. A real project teaches more than any course.
If you also want to eliminate waste, see the Lean Six Sigma certificate guide.
Frequently asked questions
What is Six Sigma in simple terms?
A data-driven method for finding and removing the causes of defects and variation in a process, so results are consistent and meet customer requirements.
What does 3.4 defects per million mean?
It is the long-term defect level assumed for a six sigma process: no more than 3.4 defects per million opportunities. It includes an allowance for a 1.5-sigma shift in the process average.
Who invented Six Sigma?
Engineer Bill Smith introduced it at Motorola in 1986.
What does sigma mean?
Sigma is the symbol for standard deviation, a measure of how much results vary.
What are the Six Sigma belts?
Common roles are Champion, Master Black Belt, Black Belt and Green Belt. Some organizations also use Yellow and White Belts.
What is the difference between Six Sigma and lean?
Lean focuses on waste reduction. Six Sigma emphasizes variation reduction. Lean Six Sigma combines them.
Is Six Sigma still relevant?
It is still taught and used, though critics question some claims. Many organizations use it with lean methods and other approaches rather than alone.
Browse certificate, course and degree programs side by side.