Design for six sigma: software product quality

• Authors: Ajit Ashok Shenvi
• Venue: ISEC '08 Proceedings of the 1st India software engineering conference
• Year: 2008

"Product quality" is the biggest influencer that determines the bottom-line in a commodity business like Consumer Electronics (CE). Typically CE products are made and sold in mass quantity, wherein the margin per device is already low. A Cost of Non Quality (CONQ) due to field problems and usability issues, adding to that, erodes it further directly impacting the profits thus reducing the competitive advantage. This also has a negative influence on the "brand image" and customer loyalty. To overcome this challenge, CE embraced the "DFSS" (Design for Six Sigma) methodology to boost product quality and hence reduce CONQ On one hand, software development is strongly governed by "software engineering principles" applied through a SEI-CMM model. This has a strong emphasis on establishing sound process framework that would ensure a good software quality. On the other hand, the core principle of DFSS is to minimize variation. However, software is mainly digital in nature (it works or it does not, yes/no) and is expected to be 100% predictable with "no variation" by itself -i.e. the same software would have exactly the same behavior under same environmental conditions/inputs. In addition, there is nothing like "samples" with respect to software development process. It is the same piece of code that evolves right from development phase to maturity phase. With all this, the very concept of "statistics" and correspondingly the various fundamental DFSS metrics like the Z-score, etc start to become fuzzy in case of software. Moreover "software does not wear out with time, it only becomes obsolete" This raises some fundamental questions - Can DFSS be really applied in software development? If yes, does it really help? How does it fit in with the existing process framework based on CMM/CMMi? Etc. etc&& This paper is an attempt to answer these questions by sharing the experiences of using the DFSS methodology in a case study of real-life software development project of DVD-Hard disk recorder product. The paper covers in detail starting from application of concepts like Voice of customer, to Critical To Quality parameters (CTQs), how the various DFSS phases were mapped into existing software development phases, how change management aspects were handled, training needs taken care of, and results obtained. The paper concludes with a summary of learning's and some points to ponder This paper is written as a case study with the intention of sharing the best practice of DFSS deployment in a software development for product quality improvement.