Six Sigma
Six Sigma is a data-driven method for cutting defects out of a process, using Define-Measure-Analyze-Improve-Control to find the real cause of variation before you touch a fix.
Five stages march left to right, each one handing its output to the stage that follows it.
Reach for this when…
- The same defect keeps recurring and nobody agrees on why.
- You're about to fix a process based on a hunch, not data.
- Quality complaints are rising and 'trying harder' hasn't moved the number.
How to run it
- Define the problem, the process boundaries, and the customer requirement at stake.
- Measure the current process: collect real data on defect rates and variation.
- Analyze the data to find the actual root cause, not the obvious suspect.
- Improve: test and implement a fix targeted at that root cause.
- Control: build in checks so the process doesn't drift back.
A worked example
Situation. Daniel Ashworth ran Irwell Freight, a logistics company in Manchester, United Kingdom, where a growing share of shipments were arriving with damaged packaging and nobody could agree whether it was the drivers, the warehouse, or the boxes.
Applied. He ran a DMAIC pass: defined the defect precisely, measured damage rates by route and shift, and the data pointed to one loading dock with a specific handling problem, not the drivers at all.
Result. They retrained that one dock's team and added a control check at handover. Damage claims dropped by more than half within two months, and stayed down.
The catch
Six Sigma is built for processes with enough repeat volume to generate real data; run it on a one-off problem and you're just doing an elaborate guess. It's also heavier than most small teams need without dedicated training, and treating every variation as a defect to eliminate can strangle flexibility a business genuinely needs.
If you skip Measure and go straight from Define to Improve, you're fixing what you assumed, not what's happening.
Origin: Bill Smith; Motorola; popularised by General Electric