Six Sigma

How Six Sigma Works

Six Sigma was developed at Motorola by Bill Smith and Mikel Harry in 1986 as a systematic approach to reducing manufacturing defects. The name comes from statistics: sigma (the Greek letter) represents standard deviation, a measure of process variation. A process operating at six sigma has so little variation that it produces only 3.4 defects per million opportunities, which is effectively near-zero error for most purposes. General Electric’s high-profile adoption under Jack Welch in the 1990s spread Six Sigma beyond manufacturing into financial services, healthcare, retail, and eventually software operations.

Six Sigma is not a methodology for managing projects from initiation to delivery. It is a methodology for improving the performance of existing processes by identifying and eliminating the causes of defects and variation. The distinction matters: PRINCE2, Waterfall, and Scrum tell you how to manage work. Six Sigma tells you how to make the work produce better, more consistent outputs.

The primary implementation framework is DMAIC:

Define. Identify the problem, the process to be improved, the customers affected, and the business case for the improvement project. Produce a project charter and define the Voice of the Customer: what the customer actually requires, stated in measurable terms.

Measure. Collect data on the current process to establish a baseline. What is the defect rate today? What does the process actually do, as opposed to what the process map says it does? Measurement in Six Sigma is rigorous: the goal is to quantify the current state with statistical confidence, not to estimate it.

Analyze. Use statistical tools to identify the root causes of the defects and variation measured in the previous phase. Common tools include cause-and-effect diagrams (Ishikawa/fishbone), regression analysis, hypothesis testing, and process capability analysis. The output is a validated list of root causes, not a list of hypotheses.

Improve. Design and test solutions to the identified root causes. In Six Sigma, solutions are validated with small-scale pilots (Design of Experiments) before full deployment to confirm that the change actually produces the improvement predicted. This distinguishes Six Sigma from improvement initiatives that implement solutions based on intuition and measure success informally.

Control. Implement controls to ensure the improvement is sustained after the project team disbands. Control charts, standard operating procedures, and monitoring systems are deployed to detect if the process returns to its previous defect rate. Without this phase, improvements from the first four phases are commonly lost within months.

When to Use Six Sigma

Six Sigma produces its strongest results in high-volume, repeatable processes where defects are measurable, economically significant, and attributable to identifiable causes. Manufacturing quality control is the original and still strongest application. A production line producing one million units per day can justify the statistical rigor of DMAIC because even a half-percent defect reduction translates to thousands of units and significant cost savings.

Healthcare has adopted Six Sigma extensively for medication error reduction, surgical complication rates, patient wait time, and billing accuracy. Call center and financial services operations use it for error rate reduction in claims processing, loan underwriting, and transaction accuracy. Any high-volume process where defects are countable, the process is stable enough to measure, and the cost of defects is significant relative to the cost of the improvement project is a valid Six Sigma candidate.

Lean Six Sigma, which combines Lean’s waste elimination tools with Six Sigma’s statistical defect analysis, is the most common implementation in practice. Lean tools address the speed and flow of the process; Six Sigma tools address the accuracy and consistency. Most practitioners who work in manufacturing, healthcare, or service operations now encounter both as a combined system rather than as separate methodologies.

When Six Sigma Is Not the Right Approach

Six Sigma requires measurable, repeating processes. Creative work, research and development, software product design, and strategic consulting are poor fits because the processes are not standardized enough to establish a statistical baseline, and the outputs are not uniform enough to count defects in a meaningful way. You cannot calculate a sigma level for a branding campaign or a strategic recommendation.

Six Sigma is also expensive to implement correctly. Black Belt practitioners are scarce and command significant salaries. The statistical analysis requires tools and expertise that most organizations do not have in-house. And the DMAIC process is deliberate by design: the Measure and Analyze phases alone can take weeks to months for a rigorous project. Organizations looking for quick wins or facing fast-moving problems with unknown causes are often better served by simpler diagnostic frameworks before escalating to full Six Sigma rigor.

Finally, Six Sigma can produce perverse incentives when applied to the wrong metrics. Optimizing a call center for defect-free call handling can produce scripts so rigid that customer satisfaction drops. Optimizing a software development process for zero post-release bugs can produce release cycles so slow that competitors win on features. Six Sigma is a powerful tool that requires careful selection of what to optimize before it is applied.