A data-driven method for reducing defects and variation, run through five phases.
Six Sigma aims to make a process so consistent that defects become extremely rare. It treats variation as the enemy of quality and uses statistics to find and remove its causes.
Improvement projects follow a fixed sequence known as DMAIC, led by trained practitioners.
When a process produces too many errors or inconsistent results and the causes are not obvious, and when data is available to analyze.
An insurer finds that one in eight claims is paid incorrectly. Measuring by claim type shows that most errors come from two manual data-entry fields. Validating those fields automatically cuts the error rate by three-quarters, and a weekly control chart keeps watch.
Projects can be slow and heavy for simple problems, and the certification structure can become an end in itself. It is better at improving existing processes than at inventing new ones.
Engineer Bill Smith developed it at Motorola in 1986; General Electric made it famous in the 1990s under Jack Welch.
Profile: Jack Welch — The most imitated chief executive of his era: be number one or two, and differentiate people rigorously.
Facts on this page were checked against Wikipedia.