Six Sigma: why (and how) to use this methodology

In an operational excellence approach, we often start by picking the low-hanging fruits. Then, to continue the improvements, methodologies and tools must be used. Six Sigma is one of them, which has proven its worth. I give you the step-by-step methodology and explain in which case it works best.

Six Sigma Principle

Formalized by Bill Smith at Motorola in the late 80’s, the Six Sigma methodology combines several quality tools.

This methodology aims to improve processes in two ways:

  • reduce variability
  • increase the accuracy.

If I use it for archery, the methodology will help the archer send all his arrows in the same place (variability) and in the center of the target (accuracy).

Its name corresponds to the target standard deviation: +/-6σ. The process will give a result whose dispersion will be within an interval of six times the standard deviation of the target.

It uses various statistical tools to achieve and maintain improvements: FMECA, FIPOC, variance analysis, design of experiments and statistical process control, to name a few.

When Six Sigma methodology works wonders

The main advantage of this methodology is its rigour. Using mathematics and statistics, it is perfect in critical environments. I am thinking of the aeronautics, agri-food and pharmaceuticals, as well as the health sector.

If you work in an environment with people who pay great attention to details such as accountants or scientists, it will help you work with them.

Finally, if you produce standard goods or services, the Six Sigma methodology can help. By standard, I think of a spatial or temporal identical dimension. For example, the length of a piece, or its diameter, a call time lenght or a processing time.

Apply DMAIC step by step

DMA What? Pronounced “D-may-ick”, this acronym describes the steps of the Six Sigma methodology:

  • Define
  • Measure
  • Analyse
  • Improve
  • Control

Like any methodology, it follows the four stages of the Deming Wheel (Plan Do Check Act),focusing on the second part: Do. It helps to be rigorous in the implementation of the plan.


In Phase 1, the team will organize and identify the problem to be solved. To do this, it will use the classic tools of project management and problem identification: brainstorming, root cause analysis


Then comes the time of measurement, this is phase 2. This is the critical step for a successful project. It consists of gathering information, preferably numerical, about the process and what can impact it. The collection of information must be comprehensive and accurate. Otherwise, the next step, analyzing, will give bad results. We often talk about garbage in, garbage out, which means that the result cannot be better than the input data.

With today’s analysis capabilities, large volumes of data are no longer a problem. On the contrary, the accuracy of the analysis will be improved if you have historical data.


Phase 3 is the one that will help you satisfy regulatory authorities, extreme quality requirements and data aficionados. Without forgetting your biases and the difference between correlation and causation, you will analyze your data to make it talk. Two pieces of data can be correlated without having a causal link. Tyler Vigen has shown this with, for example, the number of movies Nicholas Cage appeared in and the number of pool drownings, or the divorce rate in Maine and margarine consumption.

Does room temperature have anything to do with part length or talk time? Take the time to look beyond correlation and look for causation in your data.


Based on your analysis, you can then improve (phase 4), using evidence. It is at this stage that you decide the solutions and put them in place.


Finally, the phase 5 is to monitor, preferably continuously, the results. You confirm your assumptions were correct. Above all, you rely on the data and analysis to predict when the process begins to deviate and risk to produce unacceptable products or services.

And now it’s time to move on to another Six Sigma project. Given the data collection and analysis, it takes between six and eighteen months to complete such a project. We are far from a quick win, but it is a win that will last over time.