Six Sigma: what it is, DMAIC and the belts, explained
Six sigma is 3.4 defects per million opportunities. That number is the whole method: measure first, then have an opinion. Never the other way round.
Six Sigma is a methodology for solving problems with data. Its aim is to cut a process's variation down to 3.4 defects per million opportunities.
Enter your email and you will get an alert when something new is published. Unsubscribe anytime.
Hence the name: six standard deviations (sigma) between your process mean and the tolerance limit.
It was born at Motorola in the eighties, an heir of the Total Quality Management of the fifties, and became famous when General Electric turned it into their way of working.
The six things worth knowing
1 · It's a method, not a philosophy. Lean is a way of looking at your factory. Six Sigma is a procedure with defined steps for attacking one concrete, measurable problem. Don't confuse them.
2 · The cycle is called DMAIC.
| Phase | What happens |
|---|---|
| Define | bound the problem and what a defect is |
| Measure | measure the current state with data, not impressions |
| Analyze | find the root cause of the variation |
| Improve | act on that cause |
| Control | make sure the improvement holds |
3 · It attacks variation, not the mean. A process with the right mean that swings wildly produces defects anyway. What Six Sigma is after is consistency.
4 · The belts are roles, not titles. Yellow belt knows the method. Green belt leads projects in their own area. Black belt runs cross-functional projects and trains others. They aren't medals for a CV.
5 · It demands real data. Without reliable measurement, DMAIC is a meeting of opinions with English names on it.
6 · It lives alongside lean. Lean removes what adds no value. Six Sigma reduces the variation of what does. Put them together and you get Lean Six Sigma.
The bridge to 2026
The phase that has always decided whether a Six Sigma project goes well is Measure. And it's also the one that's suffered most in practice: measuring costs time, it's done by sampling, and by the time you've got the data collected the situation has moved.
Well, that's exactly what has been turned upside down.
When the line records every piece, every stoppage and every intervention on its own, the measurement phase stops being a project of weeks and becomes the normal state of your process. And with data from a hundred per cent of production, the analysis phase finds causes a sample would never have seen.
It's the closest I've seen to DMAIC running at the speed of the process rather than the speed of the meetings.