Six Sigma's rubbish. Isn't it?

Great Debates
21

So what exactly does the “Six Sigma” guy DO, in this process? Count the number of defects? Scream at the workers until quality improves? Does he (the SS guy) actually know more about the widget making process than the managers/workers on the line? In which case why not just fire your old managers and hire him to make the damn things? How the fuck is that even possible?

Or are you saying that I should pay this consultant millions so he can tell me “keep examining your process to eliminate defects” every morning? Like Xerxes’ slave who whispers “remember the Athenians” to him every night? :confused:

22

Shoot! where is the text book? [looks around, but it is late] Never mind, I’ll wing it AFAIK.

Yes, but not only that.

Not his or her job.

Now then, they are not Nazis! :slight_smile: , but this is the reason many institutions are teaching those techniques to many trades, while it is possible to get people that are very capable in quality control and would be good applying their knowledge on any field, it is recommended that the QA SS you get has previous experience in the related field.

It all depends on the contract. Clauses like the “I don’t do windows” (I do Linux :stuck_out_tongue: :wink: ) applies. I do remember there are 3 basic ways a Project manager will work in an organization [does some searching…] aha!: Functional Managers, Project Managers, Matrix Managers:

23

From what I was told , usually its a bunch of people that are already employed by the company that get sent on courses and six sigma is usually one of several programs that get implemented at the same time.

Every factory has a process flow for the products they make, which in my case starts with a steel truck that shows up and delivers steel coils to the factory.

These coils then go to the press room where stuff is stamped and then forwarded to the next link in the chain , which is either one of the lines or the sub assembly areas, which then forward that to the lines.

Every thing that comes from the receiving doors to the lines , has an internal price tag attached that increases for every process.

For every line that we are running , there is a shift count which includes parts that are scrapped for what ever reason, this count is taken for every shift and gets added to a data base. Over the course of a year enough is collected to be able to extrapolate numbers based on the scrap count, which is detailed enough to determine at what point in the process flow , was the part scrapped.

Appearance might include slug marks , or dents in the metal that are indicative of unsharp dies in the press at that time of the run.

Tight hinges failing an effort check , due to a bad spin means that the spinner needs to be throttled back or increased if the effort is too loose.

An average line has several human operator stations and the rest being automatic stations that do various things, and one of the things that the sigma people do is to view a line in totallity and either scrap the line and go with a better process or do a redesign in how the current line works.

If enough data is collected honestly , then it points a finger at where exactly the problems lie, and corrective action can be taken or budgeted for in the next purchase cycle.

Declan

24

Lots and lots of misconceptions about Six Sigma here. I’ve got some Six Sigma training, and have been involved in a number of Six Sigma projects.

Six Sigma is really just the same of a whole bag of tools useful in doing statistical analysis and/or following a set of best practices for design.

Six Sigma isn’t just statistics. What it’s really about is putting some rigor around the analysis of design and process control. It’s about using tools to help focus efforts in the right place and expose information.

There are two main branches of Six Sigma. One is more focused on statistical process control, called DMAIC. Ther other is DFSS or DMADV, and is a set of tools for aiding in the design of processes and products.

The names come from the following acronyms:

DMAIC:

Define - Define the project’s goals, measures of success.

Measure - Measure the process to establish benchmarks for current failure rates or success rates.

Analyze - Determine the cause of the defects.

Improve - Change the process to eliminate the measured causes of defects.

Control - put measurements and procedures in place to prevent the process from going in the weeds again.

Each one of these steps is basically a collection of various tools useful in helping achieve the objective.

As described, DMAIC sounds like something that would work fantastically for improving the performance of an assembly line where there are hard measures for tolerances, perfectly known rejection rates, plenty of samples for statistical analysis, etc. And that’s where Six Sigma has the biggest impact. There are plenty of great success stories of Six Sigma consultants going into factories and making large, measurable, and consistent improvements in quality. The tools work. Which you would expect since it’s just math along with some procedures and guidelines for how to apply it.

The other branch of Six Sigma is aimed at design:

Design
For
Six
Sigma

Or alternatively

Define - Define the goals of the project and what the deliverables will be.

Measure - Measure and determine customer needs, CTQs (Critical to quality - the things that, if they are defective, have the most impact on overall quality), and specifications.

Analyze - Analyze various options for meeting the customer’s goals

Design - Design the product or process using the information gathered above

Verify - verify the performance of the finished process or product against the identified CTQS, specifications, and customer needs.

When companies first started implementing Six Sigma, they often went too far and tried to stick to every step and every statistical measurement in the process. We would get sheets on them with DMAIC in big letters across the top, and checkboxes with every process for every step, and we were expected to carry them out in detail. Needless to say, this led to some confusion, a lot of inconsistency of application, and not a little bemusement as people tried to figure out how to apply tools very literally to things that didn’t really lend themselves to statistical analysis.

But that’s not the faul of Six Sigma, it’s the fault of a bunch of managers who couldn’t think outside the box and realize that the tool had to change to fit the job. Today, we treat Six Sigma as a collection of useful tools, to be used or not as we see fit. Now, instead of just informally trying to make something ‘bulletproof’, we’ll do a formal FMEA (Failure Mode Effects Analysis) - stripped of the extra paperwork. If we’re evaluating a couple of competing designs, we’ll build a House of Quality, work out some Pareto charts showing the impact of various design decisions, etc. There’s no snake oil here or ‘fad of the month’ gimmicks - just a bunch of good mathematical and analytical tools which can be used appropriately or abused.

25

As for what Six Sigma guys do:

In many organizations there are Six Sigma ‘green belts’, and Six Sigma ‘black belts’. Green belts are engineers or other employees that have the basic training for applying Six Sigma to their own areas of responsibility. Six Sigma Black Belts are people with extra training that allows them to oversee large Six Sigma products, verify the methodology of Green Belt projects, train Green Belts, etc.

Here is a sample of a Six Sigma project to improve online banking.

The Six Sigma black belt defined the problems by collecting data from customers through surveys, found out what was critical to them, and where they found the most defects. From that he exposed the CTQs for the process, and measured current performance against them. Then he made recommendations for changing the process, and measured the results of the changes to verify that the new procedure was an improvement.

Here’s an example of improving wait times as a hospital with Six Sigma. This one shows a few more statistical tools used.

This is an excellent introduction to Six Sigma.

In software, we struggled to implement Six Sigma, and eventually gave up on some of it. However, we retained a lot of the tools and use them constantly. For instance, when defining a new project, we use a lot of the ‘Voice of the Customer’ tools - surveys, focus groups, usability tests of prototypes, etc. We use tools for gathering CTQs, and FMEAs when doing failure analysis. We use Pareto charts for evaluating the impact of various features and knowing which ones to cut.

None of these are unique to Six Sigma. But training in Six Sigma exposes you to them, and knowing which tools are available and using the right ones makes you a better designer or engineer.

26

A few months back, I explained Six Sigma for process improvement to Mom. “You take a snapshot of your current status, then you look at things you may want to change, then you choose which changes to apply, then you apply them, let things stabilize, rinse and repeat.”

She said “but isn’t that common sense?”

Me: stare.

She: “oh. Right, yes, the less common of all senses… :rolleyes:”

Six Sigma, ISO 9000, 5S… those things are just attempts to explain common sense. And like Voyager and most posters have already said, the problem is when they get applied badly, usually by people who think these tools are ends. No, they’re tools - to an end. If you lose sight of the end, you’re not an engineer, you’re a mole with a hardhat.

27

The “problem” with Six Sigma is not that the process is bad, it is that much of management runs on yelling a lot and telling people to “make it happen.” So CEO Alpha sees a report in Industry Weekly saying that Non-Competitor Zeta has saved umpteen billon dollars using Six Sigma in their manufacturing process. Alpha decides that even if his company only saves half of umpteen billion, it will still make him look like an all-star at his next Shake-Down-The-Board-For-More-Options meeting, so he orders the CIO and CFO to go forth and do Six Sigma. Unfortunately, Company Alpha is a service industry instead of a manufacturing outfit like Zeta, so when minions of the CIO and CFO from Alpha go talk to the people responsible at Zeta for implementing and running Six Sigma, most of what they are told makes no sense in terms of their actual jobs. However, they have a mandate from the CEO to make it happen, so, instead of going back to the CEO and insisting that he participate in the discussion so that he can recognize that it either is not appropriate to their industry or needs to be implemented following a different model, (for example, one based on a service industry company), they slap together an imitation plan based on manufacturing that they have “tailored” (management-speak for “create a façade”) to the service industry with no serious understanding of what they are doing. Then, when they go to implement it, they run into the problem that the employees who have to actually do the work resent the fact that they are still performing bullshit tasks for the façade of Total Quality Management that was implemented on the bones of the Quality Performance Review that preceded that and they execute the letter of the new rule in a half-assed fashion while ignoring the spirit, completely.

And even if you get the line workers of a manufacturing company to nod their heads and agree to try the new methodology, you still have the problem that the new methodolgy calls for the line to be shut down to address the problem if bad parts go up by .003% in an hour, but the plant manager’s review has never been changed by the upper brass when the upper brass ordered the new methodology, and he is still going to lose his annual bonus if the line is shut down for repair, losing 300 parts per hour, even if it results in a significant improvement in salable merchandise, so he is going to order the line to continue running until the bad part rate hits 30% in an hour.

Well, in the .000001% of American companies that actually operate on the assumpotion of doing it right, the SS guy will work with the manufacturing engineers and trhe manufacturing operators to discover why a particular part is failing. Is the machinery not sufficiently maintained? Are the line operators sufficiently trained to perform their tasks correctly? Can the design of the manufacturing process be improved in a way that will result in fewer bad parts?
If you are familiar with manufacturing, you are aware that there is always a tension between the engineers and the people on the line, with each claiming that the other does not pay enough attention to “how it should be done.” (There are good and bad engineers and good and bad operators and a good (not bad) SS specialist will find ways to make sure that the process is driven by the “good” ones, not the lazy jerks.)

28

Since nobody’s done it yet I’ll post a Dilbert link, although it’ll only be there for 20 more days…

http://www.dilbert.com/comics/dilbert/archive/dilbert-20070119.html

29

It’s a cult of bullshit, paperwork, and wasted time/money. That’s my impression from seeing it grossly misapplied to many areas. Yet another “silver bullet” and fad that fails to deliver on the promises of its high-priced advocates.

30

I’m a software developer and a Six Sigma Greenbelt. Stand back! My calculator is a registered weapon. My company adopted Six Sigma methodology in a serious way eight years ago. We process payment transactions for banks, tens of millions of them a day. Originally, we applied a Six Sigma statistical analysis to the success rate and quality of our transactions. After establishing a baseline, the ongoing success rate and quality numbers became part of our Service Level agreements with our bank customers. If the numbers dip, we pay penalties. The customers have access to the raw numbers; it’s all pretty transparent.

After a couple of years of measuring existing processes, the company began to apply Six Sigma methodology to software maintenance and new development projects. This means that when we design a new process or a change to an existing process, we try to identify points at which we can measure success or failure in the process, some measures by which we can quantify quality. It appears to work in my environment. I can easily envision it being rubbish if misapplied.

31

I can’t speak to Six Sigma (not to be confused with GI Joe Sigma Six) specifically but I have spent about 10 years in management consulting. I would imagine that it is like any other methodology, process or performance metric a company tries to implement. Whenever you change the way a company does something, you will always have simple-minded people who argue the changes won’t work because “this is the way we’ve always done it”. Well the way you’ve always done it sucks. That’s why your competetors are outperforming you.

On the other hand, you want to avoid situations where you spend so much time performance metric-ing and project managing that all your people have time for is filling out procedural forms. And it has to make sense. It’s relatively easy to measure manufacturing processes but how do you measure services like finance or IT?

32

Six Sigma? That is so 1995! I thought America capitulated in the Quality Wars of the 90s.

33

I am an engineer who is a Six Sigma blackbelt (not certified) at my company. Sam Stone gave a very good overview of what Six Sigma methodology and projects look like. No need to add to that.

This is a pdf of the ASQ Certification Requirements – basically an overview of the statistical methods and quality procedures that a project blackbelt might be expected to use to solve a problem.

You’ll notice it’s quite a grab-bag of stuff: from how to write a team charter, to use of Taguchi methods. Individually, they’re reliable and proven methods – I don’t expect anyone will suggest that t-tests are “rubbish”. It’s in the implimentation that Six Sigma has problems, not in its methods.

34

I’ve worked mainly on the technology side in development, so I can’t speak from direct experience. But… I can see where Six Sigma could easily be misapplied to non-manufacturing areas. One thing it can force you to do though, is to write down and analyze your processes (how you get stuff done) and look for areas where “defects” are most likely to occur. But if there is an overemphasis on the quantitative side of the equation then it’s easy to get bogged down in the details or to just make stuff up so that it looks good on paper but has little relation to the reality of the work being done.

35

Ford, Chrysler, and GM used to think so a few decades ago. They’re still catching up.

36

To be fair, I have never heard even the wildest of Six Sigma advocates claim that it can be used to improve a product design. The process of designing yes, but not the design itself. Six Sigma does involve collecting the voice of the customer (at least the way we do it) which should help, but it is not like Six Sigma invented this. Six sigma should have improved the electrical performance of the kettle, but I’m not so sure about the design of the handle.

37

I’d think call centers would be good places to apply it. I suspect you can model the incoming call rate quite nicely, and that most calls would be pretty standard. I could imagine that a “defect” would be the need to kick the call up a level.

My DVR went out last night, and I called the Dish Network support line. It automatically went through a bunch of fixes that, while not relevant to me, were actually reasonable. Then it told me how to reboot the receiver, which did work. I suspect someone there went through a lot of effort understanding the main things that fixed the problems people called in with. Not getting a person worked for me (there was a way of skipping the step) and saved them money. I don’t know if they used a quality improvement method for this, but they could have.

38

Once upon a time, the members of a prestigious consulting firm in Boston were invited to attend a conference on Quality Analysis to be held in Philadelphia. They decided to travel by train. When they arrived at the train station on the day of travel, they discovered that a number of academic experts in several fields of statistics from the various universities in the area had been invited to the same conference and were waiting to board the train. The consultants noticed something peculiar about the statisticians.

“We notice,” said the consultants, “that while we consultants each purchased a ticket for the train, you statisticians have only one ticket among you. How do you expect to get from Boston to Philly with only one ticket? Surely you will all be thrown off the train!”

The statisticians smiled and said, “We have our methods!”

The train began to board, and both groups entered the same car. Shortly after leaving the station, they heard a conductor traveling down the adjacent car asking for tickets to be punched. As soon as they heard him, the statisticians jumped out of their seats, ran to the end of the car, piled into one of the tiny bathrooms at the end of the car, and locked the door.

When the conductor entered the car, he saw that the bathroom was occupied, so he knocked on the door and said, “Ticket, please!” The door opened a tiny crack, a ticket was pushed through, the conductor punched it, and the ticket was pushed back through the door. After the conductor had punched everyone else’s tickets and left, the statisticians returned to their seats, whereupon the consultants congratulated them on their cleverness.

After the conference, both groups again met up at the train station. This time, the consultants had only one ticket among them. They were surprised to see, however, that the statisticians had NO ticket!

“We notice,” they said, “that while we have purchased one ticket to use among us, you have NO ticket! How do you expect to get from Philly to Boston with no ticket? Surely this time you will all be thrown off the train!”

The statisticians smiled and said, “We have our methods!”

After boarding the train, they heard a conductor traveling down the adjacent car asking for tickets to be punched. Promptly, the consultants jumped up out of their seats and piled into one of the tiny bathrooms at the end of the car, and at the same time the statisticians jumped out of their seats and began to pile into the other tiny bathroom at the end of the car. As soon as the consultants had locked the door to their tiny bathroom, one of the statisticians opened their door, reached across the aisle to the consultants’ bathroom, knocked on the door, and said, “Ticket, please!” The consultants opened the door a tiny crack and pushed the ticket through. The statistician grabbed the ticket, went back into the other bathroom, and locked the door. When the conductor arrived, he punched the ticket for the statisticians and threw all the consultants off the train.

MORAL: Never attempt to apply statistical methods unless you understand the underlying principles.

39

Pretty much. I’ve been through training and seen it both successfully and unsuccessfully used.

When it is used in its intended fashion, as a toolset of data analysis techniques, and not dogmatically thrown at a business system as a mission-statement driven cureall, then it is highly useful.

Unfortunately, consultants (and many blackbelts I’ve worked with) push Six Sigma with dogmatic infomercial furor, and have left groups and companies in a dijointed mess trying to retool everything (factory, HR, facilities, finance, sales) to fit a vision that is really tailored for manufacturing.

40

One of my problems with non-manufacturing Six Sigma projects has been the pigheaded resistance from non-operations groups to using any Six Sigma techniques, because “that’s for operations”.

There is no magical divide between manufacturing and non-manufacturing applications. If you have a process that you can gather a reasonable amount of data on, you can apply Six Sigma methods to find the causes of variation in that data set.

This is generally the sort of data that is routinely collected on manufacturing processes, because those processes usually have some sort of quality control program already in place. HR, Finance, Sales, etc. generally have no QC system in place, and accordingly don’t think it’s even possible for them to collect data so that they could use Six Sigma. So why should they even try?

The one transactional Six Sigma project I’ve been involved in was trying to cut down the number of customer quality complaints. None of the data we needed to understand what was going on were collected in the same place. One person could tell me how many customer complaints we had, another person could tell me from which customers those complaints came, etc. Ironically, our CTS folks complained from start to finish of that project, but once we’d Pareto’d out the customers’ products that resulted in the majority of customer complaints, CTS was more than happy to bask in the glory of reducing their complaint frequency.

The odd thing is, that transactional Six Sigma projects like that (usually involving data transactions, rather than physical manufacturing) are generally easier on the statistics required. The real difficulty is in getting the transaction measured so that you have data which can usefully tell you what is happening. That’s not difficult to do; it’s just tedious.