Why a 99% Yield Is Not Good Enough

I remember the not-too-distant past when a 99% yield rate would earn you bragging rights (myself included). Looking at what that really means by today’s metrics show that a 1% scrap rate converts to 10,000 defective parts per million (DPPM). As a customer, imagine a supplier striving to give you only 1% defective parts! World-class six sigma levels allow only 3.4 DPPM. If you are still hanging in there with me at this point, I would hope that you agree that 10,000 DPPM is totally unacceptable and are prepared to do something about it.

Note: It must be mentioned that the 3.4 DPPM attributed to six sigma levels was developed by Motorola, and based on the assumption that over time a process is likely to have a shift in the mean of up to +- 1.5 sigma. This potential shift is factored into the 3.4 DPPM. Statistical purists would argue that a six sigma level is actually .002 DPPM, but since the Motorola interpretation is universally accepted, I use 3.4 DPPM to represent a six sigma level.

When organizations like Motorola and General Electric began communicating six sigma expectations to their suppliers in the early 1980s, what began as a ripple quickly developed into a shockwave throughout the supply chain. To say that this concept was met with some resistance is a monumental understatement. Companies had absolutely no idea how they were going to affect a change of such magnitude that their process defect rate would drop from 10,000 to 3.4 DPPM. Through a slow and painful process, companies began to understand that the way to achieve these quantum paradigm changes was through lean best practices. The interesting paradox is that none of us would accept 99% in our personal lives, so why do we accept it in our businesses? The attached Figure shows what life would look like if we settled for having things right only 99% of the time in some areas we can all relate to. This kind of changes the perception that 99% is good enough, doesn’t it?

Contrast this with a six sigma level in which your local weatherperson’s forecast would be correct every single day for 795 years in a row! (talk about a pipe dream; but that could be another whole article)

What is Six Sigma?

Sigma (σ) is the eighteenth letter in the Greek alphabet, and is defined and used in two different ways: 1) As a mathematical measure of the amount of variation in a process. This is normally referred to as the standard deviation of a process; the lower the standard deviation, the better, and 2) To describe the quantity of defects a process will produce. This is normally referred to as the sigma level of a process and is a measure of process performance; the higher the sigma level, the better. Although statistics are usually associated with six sigma, that is only part of it; six sigma is the problem solving methodology called DMAIC (Define, Measure, Analyze, Improve, Control). DMAIC is a process that uses a collection of tools to identify, analyze, and eliminate sources of variation in a process. Six sigma can be an intimidating concept to grasp, particularly regarding the statistics and scary math part of the process. The key takeaway is that to achieve a six sigma level, process variation must be cut in half from a three sigma level.