0:15

So what is this idea of Six Sigma and what does this mean?

And in terms of statistics, in terms of defects per million opportunities,

what does this mean and where did it come from?

So Six Sigma is commonly referred to as a standard.

Is a very high standard of quality that

is talking about 3.4 defects per million opportunities.

So this per million, not a percentage.

It's much more stringent than it being 3.4%.

And so when you think about 3.4 defects per million opportunities,

you are talking about a process that is 99.996% defect-free.

So we're talking about, again, a very high standard.

Now, although if you do put in 3.4 defects per million opportunities and

you know the statistics to go back and look at the number of defects,

let me warn you up front that you're not going to find it as being Six Sigma,

because there's some adjustments that we need to do for that.

And we'll talk about that in a little bit.

But for now,

let's just think of Six Sigma as being 3.4 defects per million opportunities.

And the point being, that it's talking about a very high quality standard,

a very low number of defects, right?

1:30

Okay, so before we start looking at some of the actual calculations for

Six Sigma, let's get some terminology straight.

So we said defects per a million opportunities, so you must be wondering,

what does that mean.

Well, a defect is a nonconformance.

So, traditionally we think about defects as being something that is

a defect in a part, in a physical part.

But it could be generalized to anything that a customer puts value on,

anything that a customer cares about, and

we can have a defect in that if the customer thinks it's a defect.

Defects per opportunity.

When we think about defects per opportunity,

you could take this clicker that I have in my hand, and the button that I have

to go forward may be referred to as a single opportunity for a defect.

So if there's a button that takes slides forward, and there's a button that takes

us slides back, there are two buttons, those are two opportunities for defects.

So what are we saying in a sense?

We're saying that a product can have multiple opportunities for defects.

And how do you define opportunities for defects?

Well that comes from, what does a customer care about in that product?

So I, care about these two aspects, these are what are going to be opportunities for

defects forming.

And then, what is this idea of defects per million opportunities?

So defects per million opportunities, is simply taking defects per opportunity, and

scaling it up to a million.

So instead of calling it a proportion, or instead of calling it a percentage,

we're going to say it's going to be out of a million, and

that's why we're saying defects per million opportunities.

So that's the idea that we're going to use when we calculate something like DPMO.

We're going to have the idea of opportunities, and

then scaling it up to defects per million opportunities.

3:23

So, the underlying idea for

Six Sigma comes from what we know as the standard normal distribution.

What you already may know from statistics or from taking other classes,

you would know that when you look at a standard normal distribution,

we know that is a certain area that is under the curve of the standard

normal distribution when you think about plus or minus three standard deviations.

And conventionally we talk about this, traditionally we talk about this from

the point of view of plus or minus three standard deviations.

Commonly, we say plus or minus three standard

deviations covers 99.7% of the area under the curve.

So 99% of the area will be covered within plus or minus three standard deviations.

So when we talk about Six Sigma, what we are essentially saying is that,

if you go six standard deviations from the mean, so

we're getting a little statistical here, we're getting a little bit technical here.

But what we're saying, is essentially that you have the mean in the center, and

you go all the way to the right, up to six standard deviations,

not just three that we commonly talk about.

But to six standard deviations, and

whatever area is left after that, that little small area that's left after

you've gone six standard deviations from the mean, represents the area of defects.

So that is what we are referring to when we are referring to a Six Sigma process.

So the defects are only to the extent of that small portion of the tail,

under the standard normal distribution,

that is beyond plus six standard deviations from the mean, right?

So first of all, you should be thinking, why is this stringent standard needed?

Why do we need a standard that is so

strict, that we're talking about 3.4 defects per million opportunities?

So, take a minute to think about that.

As to whether such a standard would be needed.

When would it be needed?

What are the implications?

And then we'll come back and see why Motorola started using this standard.

5:31

So first, if you are trying to answer the question of,

is such a stringent standard quality needed?

You would say, well it depends, like that's a standard idea,

that's a standard answer that we like to use in MBA school.

So it depends, well, it depends on what?

So, it depends on if you're talking about a process

where having such a high standard is important.

So, if you're talking about a nuclear power plant,

if you're talking about space exploration, if you're talking about airline flights.

There we are talking about a standard that

is needed to be at 3.4 defects per million opportunities.

Or even better, it probably needs to be even beyond Six Sigma.

And you need to achieve an even lower

level of defects than even 3.4 defects per million.

So you may say that, well, it depends,

it may be a process where you actually need such as stringent quality standard.

The other prospective that you can take is,

well, it depends on what's the cost for us to get to that standard.

So in some situations, in some contexts, you may say, well,

there may be a cost benefit analysis for us to get to such a high standard.

And another perspective that you might take, is that you may have a process

that is made up of many different steps, many different opportunities for error.

And the process is going to be a combination of all those opportunities for

error, and

how is that going to play into the total output quality of that whole process.

So, what are we going to get from that process?

And that's another perspective that you want to take when you're thinking about,

is 3.4 defects per million opportunities too stringent a standard?

So let's take this third perspective and

study it from a numerical perspective, right?

So you have a 3 sigma process.

Now we're moving from Six Sigma to a 3 sigma process, and

a 3 sigma process, without getting into the details of it,

3 sigma represents 66,803 defects per million opportunities, right?

Six Sigma represents 3.4,

3 sigma represents 66,803 defects per million opportunities.

So, if we have a process that has 20 tasks in it, right?

Or you can think of a physical product that is made up of 20 different

sub-assemblies, 20 different parts that are going into that particular product.

And each of these is being manufactured, or

each of the steps in a process are being run at a rate of 3 sigma.

So each one is

giving a 66,803 defects per million opportunities kind of output, right?

So if there are 20 steps, what will happen to the output rate

from the whole process that has 20 steps, or if there are 20 parts?

What will happen to the product quality for the product that is made from putting

those 20 parts, those 20 sub-assemblies together?

So, if you take a look at the calculation,

it's basically saying you take 0.933197, you multiply it by 0.33197.

When you're talking 2 steps, you multiply it again, when you´re talking 3 steps, and

you keep doing that until you get to 20 steps.

So it´s raising it to the power of 20, which gives you a defect-free rate,

or a likelihood of finding a defect-free product at the end of 20 steps, or

20 parts, of only 25.09%.

8:57

So, this should give you some appreciation of when there are many different parts,

if there's a complex product like a car, or a cell phone even.

And you're talking about many different parts going into that particular product.

Each part being at a 3 sigma level performance would mean that you're

going to get a product that's going to be 25% defect free,

75% chance of having a defect.

Which no customer would be willing to accept,

no company would be willing to accept.

So, from that perspective, a stringent standard such as Six Sigma is required

when you have highly complex products, when you have highly complex services,

highly complex physical goods that are being produced.

All right, so that's why Six Sigma is something that Motorola went after.

9:42

Now, what does this mean from things that we know from our daily experiences?

So in terms of comparing a Six Sigma performance with a 99% good performance,

so 1% level of errors.

In 300,000 letters delivered, 3,000 missed deliveries versus 1 missed delivery

when you're talking a Six Sigma performance.

In 500,000 computer restarts, 5,000 crashes when you're

talking about 99% good, versus 2 crashes when you're talking about

Six Sigma good, or 3.4 defects per million opportunities good.

In 500 years of month-end closings, you are talking about 60 months

being not balanced versus 0.02 months not being balanced,

when you're talking about Six Sigma good.

10:30

Week of TV broadcasting, 99% good, 1% defect,

you are talking about 1.68 hours of dead air and

when you're taking about 3.4 defects per million opportunities.

When you are taking about 3.4 being the bad part of it and rest being good,

you're saying only 2 seconds of dead air in a week of TV broadcasting.

So here you can also see that the idea of measuring a process based on it's DPMO,

it's sigma level, is something that can be translated into any kind of process.

And that's one attraction of this metric of Six Sigma,

that you can use it for different kind of processes and

have a conversation about at what level of sigma that process is.

And these could be many different processes of many different contexts, and

this gives you a common metric.

Comparing, on the other hand, between a four sigma versus a Six Sigma level of

performance, if you're talking about cell phone service, a four sigma is

4.46 hours of no service, a Six Sigma is 8.8 seconds of no service.

Missed putts in 100 rounds of golf.

We're talking about 100 putts that are being missed in the 1,800 putts that

you're going to make, based on 100 rounds of golf, and 18 holes.

Versus, in Six Sigma, you're talking about a minuscule number of putts,

0.006 putts being missed in the 1,800 opportunities that

you're getting from 800 rounds of golf.

So what you can also see from this comparison,

is that going from four to Six Sigma is a quantum leap, it's a huge jump.

And this is exactly what Motorola was going after.

They were going after a ten-fold improvement in quality.

A ten-fold reduction in defects when they were trying to improve their

quality from what it used to be, before they started implementing Six Sigma.

So going from a 3 to 4 sigma level, to be more specific, is a 10-fold improvement,

4 to 5 is a 30-fold improvement, 5 to 6 is a 70-fold improvement.

And that's the kind of quantum improvement that you can get when you're going from

a small sigma level to a high sigma level.

Now why Six?

Why not seven, why not 4.5?

And that's something that has been popularized by Motorola.

And, you may have heard this earlier, Six Sigma is actually

a registered trademark of Motorola, so they stuck with Six Sigma.

But there's no reason for you to have that as being a magic target.

It could be higher or lower based on where your process is,

where your company is trying to achieve quality performance at.