In simple terms, process sigma is a measure of the variation in a process relative to customer requirements. We measure defects on a scale of defects per million opportunities (DPMO). Any instance of failing to meet customer requirements is a defect, so a high number for DPMO is undesirable. The originators of the Six Sigma methodology felt that in the manufacturing sector, the goal should be to have no more than a handful of defects for every million opportunities. Thus a process sigma of 6.0 was set as the goal because it corresponds to only 3.4 DPMO.
In terms of process performance, a level of only 3.4 defects per million opportunities is a defect rate of only 0.0003%, meaning 99.9997% of the time defects do not occur. In non-manufacturing operations, the goal is typically set not at six sigma but at four or five sigma. It would not be cost effective nor a good return on effort and investment to do what is necessary to achieve six sigma in service industries.
But since many processes operate at a sigma of only 3.5 or less, achieving 4.0 or 5.0 would be a substantial improvement. Even at 98% defect-free performance (sigma of 3.5), a company with a million sales has over 20,000 defects and probably 20,000 unhappy customers.
Note that the relationship between process sigma and DPMO is not linear. At a process sigma of only 2.0, the defect rate is over 300,000, meaning the defect rate is over 30%. Improving sigma from 2.0 to 3.0 results in a reduction of defects from over 300,000 DPMO to under 70,000 DPMO yielding a four- to five-fold reduction. An improvement from 4.0 sigma to 5.0 represents a decrease from over 6,000 defects to slightly over 200 defects, for a nearly 30-fold improvement.
The sigma scale provides a means of comparing performance of vastly different processes, as it can be applied to any process. Thus it can be used to determine which processes should receive priority for Six Sigma process improvement efforts, based on how well customer requirements are being met in each area.