The Anatomy of a defect opportunity
Dr. Mikel J. Harry
Setting the Stage
Since the practice of Six Sigma was first originated at Motorola (early 1980’s), the idea of a “defect opportunity” has been discussed and debated by many practitioners. As Six Sigma began to spread to other corporations (early 1990’s), the debate continued, but to such an extent; and filled with so much emotion, the controversy became known as the “opportunity wars.” During this period of intense examination, the key nuances and seemingly counter-intuitive principles that circumscribe a defect opportunity were ferreted out of the many papers, debates, presentations and hall-way discussions. The aim of this essay is to present those nuances and principles. In this context, we will examine the current anatomy of a defect opportunity.
For purposes of essay, we’ll use the industry term “product” to represent any type of deliverable – like a piece hardware, software program, or professional service, just to mention a few examples. We will also acknowledge that, at some point in the production cycle of a product, one or more defects can occur. To appropriately adjudicate those defects, it is often advantageous to report on their quantities in the form of a defect rate. A common and conventional metric that reports on the defect rate is called “Defects-per-Unit,” or simply DPU. Inherently, DPU is a sound and useful metric; and with proper attention, it has the ability to report on what it purports to measure – product quality, but only in a broad sense. At the same time, its primary limitation is that it cannot be used to compare two or more dissimilar products or report on the quality of in-process units.
One way to compare dissimilar products and the quality of in-process units is to employ a metric called “Defects-Per-Opportunity,” often given in abbreviated form as DPO. We recognize this metric in the form: DPO = D / O, where D is the total number of defects and O is the total number of corresponding defect opportunities. However, once the DPO metric is put into play, a recurring issue often arises. This issue can be nicely summarized in the form of a question: “What constitutes a defect opportunity and how should such opportunities be counted for purposes of quality reporting?” A reasonably accurate accounting of defects (D) is seldom an issue, but authenticating the total number of defect opportunities (O) can be problematic. In other words, the problematic nature of counting defect opportunities arises when there is an absence, shortfall or inconsistency of qualification criteria.
The Idea of a Defect
To understand the basic nature of a defect, let’s start with the dictionary. Turning to Webster’s, we find that a defect is described as: “An imperfection that impairs worth or utility.” It’s quite interesting to note that this particular definition links the idea of value (worth and utility) to the idea of imperfections. While this may initially seem like a small or rather insignificant point, the pragmatic implications are many and run very deep – at all levels of an enterprise. In other publications and white papers, this author has previously defined value as “the relative worth, utility or importance of something,” we can make a substitution of terms and refine the latter definition.
Thus, we may now say that a defect is: “Any nonconformance to standard that results in a loss of entitlement value to the customer or provider.” 2 Considering the latter definition, the term “entitlement” shall mean a “rightful level of expectation,” whereas the term “rightful” is established through the process of negotiation or what is conventionally accepted as “just.” The term “value” shall mean: “economic or utilitarian worth.” As one might expect, our discussion will proceed using these understandings as a source of reference. To better understand the nature of value, let’s consider the following conceptual relation: Value = (Quality x Quantity) / Cost. Given this highly generalized understanding, the numerator term is often called “Bang” and the denominator term is frequently referred to as “Buck.” Thus, value can be thought of a “Bang-per-Buck.” In this context, entitlement value can be loosely interpreted as the Expected Bang-per-Buck in the instance of Zero-Defects.
The Idea of an opportunity
Now that we have prescribed the nature of a defect, let’s turn our attention to the term “opportunity.” Broadly speaking, the Webster’s dictionary describes an opportunity as “a favorable juncture of circumstances.” Thus, a “defect opportunity” can be defined as: “a juncture of circumstances that makes possible the creation of a defect.” Given this definition, it can be argued that if all of the related circumstances were concurrently present, there would exist the possibility of creating a nonconforming attribute (i.e., defect). If the creation of such a defect is possible, then the potential for economic or utilitarian risk must also be possible. So, we now ask: “What circumstances must be concurrently present to create an opportunity for defect?” However, prior to fully answering this question, we must first acknowledge that each circumstance can be reshaped into a qualification criterion, but stated in the form of an action. If all of the qualifying actions were executed or otherwise fully satisfied, a defect opportunity would exist and could be subsequently authenticated. Conversely, if one or more such qualifying criteria were not executed or otherwise fully satisfied, a defect opportunity could not be made to exist and; as a consequence, could not be subsequently authenticated.
Qualifying a Defect Opportunity
At this point in our discussion, let’s start by understanding what it means to qualify something. Again calling upon the dictionary, we find that the word “qualify” is characterized by the definition: “to satisfy certain prescribed conditions or requirements.” In the context of our discussion, a defect opportunity would be only be qualified once all of the following actions have been verifiably executed in a quality manner:
- Select a product attribute based on its economic or utilitarian value. If a product attribute has not been identified and subsequently selected, there would be no opportunity for defect. If a selected product attribute has no economic or utilitarian value, there would be no logical or pragmatic reason for assessing and reporting its performance.
- Assign a scale of measure to the selected product attribute. If a product attribute cannot be measured, it would not be possible to assign a performance specification. Without a specification, an opportunity for defect cannot exist. Hence, a defect would not possible.
- Affix a performance specification to the defined scale of measure. If a product attribute has no performance specification, a defect would not be possible.
- Produce the selected product attribute. If a product attribute is not manifested into real time and space, the attribute cannot be qualified. As a result, a defect would not be possible.
- Measure the attribute’s performance. If a product attribute is not measured, a defect opportunity would not exist. As a consequence, a defect would not be possible.
- Compare the attribute’s performance to the related specification. If the performance of a product attribute is not compared or otherwise contrasted to its performance specification, it 3 would not be possible to assess the quality of the given attribute. Owing to this, a defect would not be possible. For any given case, a prospective defect opportunity can only be duly qualified when all of the aforementioned actions are executed or otherwise satisfied in a quality way. Therefore, only qualified defect opportunities should be included in a Quality Information and Reporting System (QIRS).
The Defects-Per-Opportunity Metric
Now, let’s take a deeper look at the Defects-Per-Opportunity (DPO) metric. Considering the denominator term of this performance metric (O), there are only two possibilities; either 0 = 0 or 0 = 1. Given the case where 0 = 0, the prospective opportunity cannot be qualified. In this event, the corresponding numerator value would be indeterminate. Therefore, the DPO would be indeterminate. On the other hand, if the denominator term was given as 0 = 1, the prospective opportunity would necessarily have been qualified. Based on this logic, the numerator term could assume the values 0 or 1. If the numerator was given as 1, the attribute should be treated as a nonconformance (defect). On the flip side, if the numerator was given as 0, the attribute would be treated as yield.
The Complexity Dilemma
The reader should not confuse the following three ideas: a) attribute count; b) defect opportunities and c) complexity. Recall that a product attribute is a performance characteristic, like a part’s width. In some cases, practitioners have mistakenly associated the total number of design features with the total number of qualified defect opportunities. Generally speaking, the total feature count is significantly greater than the qualified defect opportunity count. Often times, the complexity of a product is given by the total number of nodes and connections. Again, for most cases, this number will be significantly different from the feature and opportunity count. For example, consider the instance of a single product unit. In this case, we’ll say that the total defect count as determined to be D = 50. Furthermore, we’ll say the total number of design features was given as 600; the total number of qualified defect opportunities was reported as 200; and the complexity was estimated to be 1,400. Thus, there would be three ways to compute DPO:
1) DPO = 50 / 600 = .0833 | DPMO = .0833 x 10^6 = 83,333 | 2.88 Sigma.
2) DPO = 50 / 200 = .2500 | DPMO = .2500 x 10^6 = 250,000 | 2.17 Sigma.
3) DPO = 50 / 1,400 = .0357 | DPMO = .0357 x 10^6 = 35,714 | 3.30 Sigma.
As you can see from this example, the reported differences are quite profound. So, it should now be apparent why great care should be exercised when defining the anatomy of a defect opportunity.
One of the conventional practices in the field of Six Sigma is to focus our efforts on those things that prove critical to quality, such as high-leverage defect opportunities. Such an opportunity is often called a “Critical to Quality” characteristic, or CTQ in short. As might be expected, not all defect opportunities are critical to product quality. Generally speaking, we employ the Pareto Principle (or 80/20 rule) to classify what is and is not critical. In the instance of defect opportunities, use of this principle would reveal that only 20% will prove to be amid the “Vital Few.” The remaining 80% would be among the “Trivial Many.”
In some operational or business scenarios, it might be advantageous to only include the Vital Few opportunities (for purposes of quality reporting, analysis, and management). However, if this practice is implemented in one segment of an organization, it should be made applicable to all such segments. The same can be said across products. Establishing rational definitions, guidelines, and rules related to the counting of defect opportunities goes a long way toward ensuring that the composite quality system (and its centralized or distributed databases) remains unbiased in terms of data integrity and computational consistency.
Otherwise, the stealthy and deceptive practice of “Denominator Management” could easily become an underground means to inappropriately and unrightfully “increase the quality numbers,” so to speak. If the definitions and practices associated with the denominator term of the DPO metric are authenticated and standardized for all organizational segments (across all product types), the only way to report a higher level of quality improvement is by way of “Numerator Improvements.”
As a final comment, the reader should recognize that this essay only scratches the surface of the subject at hand. Again, the intent is to summarize the anatomy of a defect opportunity, not discuss every detail and constraint. Consequently, this essay should be viewed as a high-level primer, not a comprehensive manifesto on the subject.