Six Sigma

Six Sigma is a method for improving processes by reducing variation. It treats every defect, delay or error as the product of a process that varies more than it should, and it uses data and statistics to find the causes of that variation and remove them. The name refers to a statistical target: a process operating at six sigma produces no more than 3.4 defects per million opportunities.

Where Six Sigma comes from

Six Sigma was developed at Motorola in the mid-1980s, when engineer Bill Smith formalised a way of measuring defect rates that could be compared across very different processes. Motorola credited the programme with large quality gains and cost savings, and in 1995 General Electric adopted it company-wide under Jack Welch, tying it to management bonuses and making it a standard part of corporate improvement. From there it spread to manufacturing, services, healthcare and finance.

The method has evolved since. Its statistical core is unchanged, but the way it is taught and deployed has been reshaped by its combination with Lean, which is the form most organisations use today.

What “six sigma” means

Sigma (σ) is the statistical symbol for standard deviation, a measure of how much a process varies around its average. A process whose output fits comfortably inside its specification limits has a high sigma level; one that regularly produces results outside them has a low one.

The sigma scale converts defect rates into a single comparable number:

• 3 sigma — about 66,800 defects per million opportunities (93.3% yield). Typical of an unmanaged process.
• 4 sigma — about 6,200 defects per million (99.4%). Common in reasonably well-run operations.
• 5 sigma — about 230 defects per million (99.98%).
• 6 sigma — 3.4 defects per million (99.9997%). The target the method is named after.

The point of the scale is not that every process must reach six sigma. It is that defect rates become measurable, comparable and improvable, whether the process assembles circuit boards or handles insurance claims.

How a Six Sigma project works: DMAIC

Six Sigma projects follow a five-phase cycle called DMAIC: Define, Measure, Analyse, Improve, Control. Each phase has a purpose and a set of tools, and a project does not move to the next phase until the current one is complete.

• Define the problem, the customer, the process and the goal. A SIPOC and the voice of the customer set the boundaries.
• Measure the current performance with reliable data. This is where measurement system analysis and process capability come in.
• Analyse the data to find the root causes of variation, using tools from the Pareto chart and Ishikawa diagram through to hypothesis testing and regression.
• Improve the process by acting on the confirmed causes, testing solutions before rolling them out, sometimes with design of experiments.
• Control the improved process so it does not drift back, with control charts, standards and a control plan.

A second cycle, DMADV (Define, Measure, Analyse, Design, Verify), is used when the goal is to design a new process or product rather than improve an existing one. It is the basis of Design for Six Sigma.

The statistical toolkit

What distinguishes Six Sigma from other improvement methods is its insistence on statistical evidence. Decisions rest on data, not on the loudest opinion in the room. The main tools are:

• Descriptive statistics to characterise a process: mean, standard deviation, distribution shape.
• Process capability indices (Cp, Cpk) to compare what a process does against what it is required to do.
• Hypothesis tests (t-tests, ANOVA, chi-square) to establish whether an observed difference is real or noise.
• Regression and correlation to quantify how inputs drive outputs.
• Design of experiments to test several factors at once and find the settings that matter.
• Statistical process control to distinguish normal variation from a genuine change that needs a response.

The depth of statistics used depends on the problem and the practitioner. A Yellow Belt project may need only a Pareto chart and a run chart; a Black Belt project may need a designed experiment.

Belt levels

Six Sigma borrowed its skill levels from martial arts. Each belt corresponds to a level of statistical competence and project responsibility, and each level includes the ones below it.

• Yellow Belt — understands the method and its basic tools; contributes to projects or leads simple ones.
• Green Belt — leads improvement projects using DMAIC and standard statistical analysis, usually alongside a day job.
• Black Belt — leads complex, cross-functional projects using advanced statistics; often full-time; coaches Green Belts.
• Master Black Belt — trains and mentors Black Belts, owns the deployment programme and advises leadership.

Belt titles are not regulated, and their meaning varies by provider. That is why certification standards exist: the LSSx.0 standard defines what a holder of each belt is expected to know and to be able to do.

Six Sigma, Lean and Lean Six Sigma

Six Sigma and Lean address different problems. Six Sigma reduces variation: it makes a process consistent. Lean reduces waste: it makes a process fast and free of activities that add no value. A process can be consistent but slow, or fast but unpredictable, and most real processes are some of both.

That is why the two were combined. Lean Six Sigma uses Lean’s tools for flow and waste alongside Six Sigma’s tools for variation and root cause, inside the same DMAIC structure. In practice almost all training and certification today is in Lean Six Sigma rather than Six Sigma alone.

When Six Sigma is the right tool, and when it is not

Six Sigma works best on problems where the process exists, runs repeatedly, produces measurable output, and the cause of poor performance is not obvious. A defect rate that nobody can explain, a cycle time that varies wildly between identical orders, a quality problem that has survived several fixes: these are Six Sigma problems.

It is the wrong tool for problems whose cause is already known (just fix it), for one-off situations with no data, and for processes that need to be redesigned rather than tuned. It is also often over-applied: a three-week DMAIC project on a problem that a one-hour 5 Whys would have solved is a waste of the method.

Getting certified

Six Sigma competence is demonstrated through certification: training followed by an examination, and at Green Belt and above, a completed project. Lean Six Sigma International delivers certification programmes from Yellow Belt to Master Black Belt under the LSSx.0 standard, in Belgium, France and online. Each level builds on the one before, and the full pathway runs from a two-day introduction to a thirteen-day Master Black Belt.