Six Sigma

You may have struggled to get through my rather lengthy (I didn’t name the blog Hot Air Buffoon for nothing), and technical explanation of the sigma rating system yesterday but I felt it was a relevant bit of background for today’s, less maths-heavy, post.

In the mid 80s Motorola developed a new management strategy designed to elevate all of their production processes to a 6 sigma rating and, through this, six sigma became a term used to describe this management strategy. Since it’s inception the Six Sigma management strategy has been adopted by many companies across the world, including those outside of manufacturing. Companies that have successfully implemented Six Sigma include Amazon.com, Boeing, DELL, Ford, Pepsi, and essentially the entire US armed forces. And those are just the companies that I imagine most people have heard of; there are many more companies with insanely huge outputs that also use Six Sigma. This, however, isn’t to say the system isn’t without its faults.

As I mentioned above, Six Sigma is a management strategy with the aim of achieving a rating of 6 sigma in all of a manufacturer’s processes. This target is achieved through the use of two methods: DMAIC (pronounced duh-may-ick) is designed for improving current processes, and DMADV (pronounced duh-mad-vee) also sometimes referred to as Design For Six Sigma or DFSS (not to be confused with the furniture retailer) which is used to create new processes. Here is a fairly quick breakdown of DMAIC and DMADV, using the old toast example

DMADV stands for Define, Measure, Analyse, Design, and Verify; which, funnily enough, are also the five steps involved in the method. Rememeber: this is the method used to create new processes.

The first step is to specifically Define all aspects of the process such as the customer's requirements, and the project boundaries. This is a little like when my picky little princess asks me to make her some toast before work in the morning and tells me she wants it well toasted (not burnt) with just a very thin layer of butter (because she doesn’t want to get fat), and for it to be warm, with the butter all lovely and melted (because she’s a fussy little git). And she needs it quick because her train is at eight fifteen. From this I have a well defined specification and timeframe in which to produce the product.

The second step is to accurately Measure the customer’s needs and specifications. This would be where I ask her highness exactly how hot she’d like her toast to be, how much butter she would like, and how quickly she needs it.

The third step is to Analyse the different options for the production stage. Should I try and eyeball the butter for the sake of speed or should I use something to measure it precisely. Should I put the bread into the toaster on a medium setting, or blast it with the blowtorch from the shed.

The fourth step is to Design the process to meet the customer’s needs. So it’s a blast with the blowtorch to get the toast to the required forty five degrees Celsius followed by three grams of butter, which has been weighed out on the scales, thinly spread over the top.

The fifth and final step is to Verify the process. Hmm, it only took me thirty seconds to do this toast but it tastes a little too paraffiny, and footy. The bathroom scale may be sensitive but isn’t hugely accurate, or beneficial to the overall flavour. The blowtorch could need rethinking too. I think we need to move on to DMAIC

DMAIC stands for Define, Measure, Analyse, Improve, and Control; and is used to improve current processes.

The first step is to specifically Define all aspects of the process such as the customers requirements, and the project boundaries. This would be: hot toast at fourty five degrees Celsius, with three grams of melted butter, in three minutes.

The second step is to accurately Measure all the key aspects of the process and collect any relevant data. Given time, the process I’d use to produce toast for her highness could be monitored and I would make detailed notes on things such as the performance of the blowtorch in getting the toast to the correct toastyness, the accuracy of the scale, time taken, and so on.

The third step is to Analyse the data collected and determine the primary causes of any defects as well as identify areas for improvement. I might be burning the toast a bit and the amounts of butter might be a little inconsistent.

The fourth step is to Improve the process by finding ways to fix or prevent problems. Perhaps I could buy some kitchen scales to measure the exact amount of butter, or just use the toaster to stop burning the toast. It might take longer but, as long as it toasts in less than three minutes, we’re still ok.

The fifth and final step is to Control the improvements and make sure that the new process is maintained. In other words, keep using the toaster and don’t do stupid things like weighing food on the bathroom scales. After a bit of faffing around, my little bunnykins might even stop hitting me for trying to feed her blowtorched toast.

So that’s a quick rundown of the basic Six Sigma method, but I did say it has its faults. Some people claim that Six Sigma has negative effects such as stifling creativity and limiting the levels of innovation within a company. Another feature of Six Sigma is that it creates a hierarchy of roles based on martial arts rankings, with green belts and black belts being examples of terms used to describe people at different levels of Six Sigma proficiency within a company. The use of black belts as consultants travelling to different companies has aided the growth of training and certification, leading some people to claim that many consulting firms will oversell Six Sigma as an area of expertise for them when they only really have a basic understanding of the process.

Perhaps the biggest question about Six Sigma comes from its primary aim of achieving a process quality rank of 6 sigma. It may be ok for some processes to produce 3.4 defects per million opportunities, it might not make sense for others. Only 3.4 out of a million laptops failing might be great but would you want to be on a life support machine with the same chance of defect. It is also accepted that the quality of a process will deteriorate over time and so a sigma shift of 1.5 over time is considered acceptable. This means that a process with a long term sigma rating of 4.5 could still count as a 6 sigma process. That life support machine isn’t looking so life supporting now is it.

Well, I have once again managed to skip though a fairly deep subject with reckless abandon, and now I must compose some kind of dry and witty remark with which to sign off.

Bye!