Subscribe to DSC Newsletter

Six Sigma is a quantitative approach to problem solving - to solve certain types of problems. At the root of Six Sigma is an improvement methodology that can be described by the acronym DMAIC: define, measure, analyze, improve, and control [1]. Those interested in reading up on Six Sigma might consider the book for dummies, which I found fairly succinct. Those wondering what I mean by "certain types of problems" should consider how to apply the approach to their own business circumstances. I don't mean to the general business but specific problems troubling the business. For me, it is quite easy to identify some conceptual design issues, which I will be discussing in this blog.

Long before the math is reached, there is an issue of problem recognition, which occurs at "D" - design. The presumption is that people who enabled, contributed, or caused the problem have the ability to recognize it in a manner that leads to its resolution. If this is true, then I agree the quantitative analysis might provide some useful insights, assuming the problem can be quantified and handled in a quantitative manner. If the assumption is false - and it probably is - then the ensuing calculations would be pointless. I don't believe statisticians necessarily appreciate my observation: just because there is a problem to solve doesn't mean it's the same problem that needs a solution. Consequently, the solution gained might not solve the underlying problem.

Consider an obvious question: "Why aren't people buying our products?" Now, I don't know exactly how a team of analysts would pose this question in a manner that can be addressed quantitatively.  However, I want to emphasize that corporate overlords are probably interested in tangible results. As such, there is a limit to the extent to which performance failure can be obscured by the haze of analysis. "We have increased clicks to these buttons by at least 500 percent!" Yet if revenues still aren't being generated by the website, it would seem that the analysts forged a narrative that in reality is faulty: that revenues were being impaired by lack of web traffic to specific links. The question is, why aren't people "buying" the products.

The methodology in Six Sigma is primarily a matter of definition rather than discovery. Six Sigma is a fairly scientific approach: a person discovers by testing whether assertions are correct. I realize that for those without alternatives, such an approach might seem like the only reasonable way. In the case of discovery, the narrative is formed by the client. I notice that more effort is being made for businesses to interact with the client's narrative rather than the one imposed by researchers. However, there are methodological hurdles in terms of dealing with the unstructured nature of client data. I have addressed this issue using a method of codified narrative called BERLIN: Behavioural Event Reconstruction Linguistic Interface for Narratives. Those interested in reading more should search Google. Interestingly, I seem to have gained ownership over the topic.

Sometimes it is not practical to ask clients directly for their perspective. So I just want to mention that for certain businesses, it is necessarily to gain knowledge by inference: scientists might try adjusting the lighting, background music, smell in the isles, attire of sales people, product arrangements, painting of walls, hand soap in washrooms, toilet paper in washrooms, placement of shopping carts, and so forth. This isn't actually an exercise in discovery but rather behavioural modification. Now, a scientific approach to multifarious events seems problematic in that the data is fundamentally non-quantitative. If scientists tried to pose the problem as quantitative anyways, it would be a never-ending process of defining and postulating. I am suggesting that Six Sigma represents a poor tool for inferring relationships extending from routine events.

On the other hand, I created a tool that is specifically designed to deal with multifarious events: the Crosswave Differential Algorithm. (Again check Google for more details.)  In this case, no quantities are needed. I am merely pointing out that alternatives exist not just to Six Sigma but the entire rationale behind its methodology. In closing, I find it necessary to caution people about operating a business from a string of assertions potentially far disassociated from reality. I know that the status quo might seem perfectly normal. The entire discourse surrounding decentralization brings to light the problems of central control. However, the fact that an operation might be decentralized doesn't necessarily mean that it is operating in a decentralized manner. Just to confuse the issue, I would even say that a centralized operation does not necessarily have to operate in a centralized manner. Decentralization is premised on the need for increased interaction with the client narrative. An organization hardly comes any closer to that narrative by substituting it - imposing its own.

[1] Craig Gygi, Bruce Williams, and Neil DeCarlo. Six Sigma for Dummies, 2nd ed. (Hoboken, New Jersey: John Wiley & Sons, Inc., 2012), 44.

Views: 1145

Tags: analysis, arts, charlatans, decentralization, devolution, fads, gimmicks, martial, methodology, methods, More…popular, quantitative, research, scientific, shamanism, sigma, six


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Dragos Bandur on February 13, 2017 at 10:07am

Interesting approach, but probably my education in systems theory prevents me from seeing right away its benefits.

As a six sigma practitioner I cannot help to observe that - in DMAIC as well as in DFSS - "D" stands for "Define" which precludes "Design" and also that Six Sigma has a systemic approach (inputs, process, outputs and all that).

Built on similar principles as its predecessors (Voice of the Customer, QFD etc.), Six Sigma was attempting the deployment of "engineering statistics" within operations (production, services). The emphasis was on "deployment" as enabling everyone within the operations to participate in measurable continual improvement and simplification of processes without necessarily having had advanced statistical background.

Finding and implementing the right solution however required (cross)validation which is unfeasible outside the quantitative (numerical) realm.

At its time, six sigma seemed to be a robust liberation from the "produce-then-inspect" reactive quality and the bureaucracy of TQM, and - similar to other improvement management systems - been having sustainability issues: as the big cost incurring problems were vanishing, the elbow in the imaginary chart of "operations cost-reduction v. cost of six sigma deployment", showed that the approach reached its justification limit.

In my opinion, a computer- assisted quantitative approach (already in place in some corporations) eliminates the abovementioned elbow, and gives predictable reliability.

After all, it is the quality which attracts the Customers and the reliability which keeps Them. 

Comment by Don Philip Faithful on February 6, 2017 at 12:29pm

James, I consider that precisely the type of situation where Six Sigma might be suitable: repetitive manufacturing. It can address anomalies and other issues pertaining to conformance. I'm in quality control myself, and I don't consider Six Sigma particularly worthwhile for the work that I do; this is because attribution is so clear and strong in my case. One would tend correct problems directly rather than by formulating and testing hypotheses. But I can see how a large operation might need an intermediate process.

Jim, I will investigate further when I have the opportunity.  But I certainly appreciate the additional details you posted.  I hope readers follow up.

Comment by Jim Massey on February 6, 2017 at 12:25pm
I have more than 10 years experience using lean 6 sigma methodologies and tools to improve business process performance. I believe your dummies book may be very outdated or incomplete since modern approaches leverage qualitative and quantitative tools. 6S is a process centric methodology, so it's only suitable for businesses who document the processes they follow to produce their service. Ask the folks at iSixSigma for advice; they're helpful, intelligent and may be interested in the tool you've developed.
Comment by James Theobald on February 6, 2017 at 11:04am

I am not trained in Six Sigma, but most references to Six Sigma that I find within my limited real world experience are closely related to highly controlled production or manufacturing facilities where product quality is in question.  The processes measured are highly repetitive, long term with-multi year to decade long histories, and with 12-24 hour runs.  Most of the processes are automated and mechanical.


  • Add Videos
  • View All

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service