Six Sigma: In Information Technology (IT)

“Eighty-five percent of the reasons for failure to meet customer expectations are related to deficiencies in systems and process rather than the employee. The role of management is to change the process rather than badgering individuals to do better” – Dr. Deming Several process improvement methodologies like Six Sigma, Total Quality Management (TQM), Quality Circles, Taguchi, Statistical process control, etc. are being successfully implemented in the manufacturing industries sector. It was perceived that such improvement methodologies are ineffective in the IT services industry. GE, pioneers of Six Sigma implementation in a non-manufacturing set-up, has estimated benefits of the order of $10 billion during the first five years of implementation.
Some commonly made arguments against the effectiveness of Six Sigma in IT services sector were
• Software processes are difficult to measure.
• Software development is people intensive work that needs creativity.
• Software development is not a repeatable process.
• Six Sigma theories are based on assumption of normal probability distribution and Software processes cannot be included in this category.
Though these factors are true in some sense, the Six Sigma methodology can still be applied to IT processes.
The software processes are definitely difficult to measure but it’s not an impossible task. Industry leaders like IBM and institutions like Software Engineering Institute have designed and published many metrics for software processes for the benefit of the entire industry. Capability Maturity Models prescribe the quantitative management processes as one of the Key Process Areas at level 4. Lot of books and other material is available publicly to choose right metrics from. Six Sigma offers strong tools like Quality Function Deployment (QFD), CTQ flow-down and other templates to convert high-level VOC into measurable CTQs.
90% of the processes in a software services company are repeatable and can be improved by the process improvement DMAIC methodology. The DFSS methodology can be applied to the remaining 5-10 % of the processes, which involve creativity.
It is true that Six Sigma concepts evolved with normal distribution. But, Six Sigma tools can be easily adapted to handle processes having non-normal distribution
Having discussed the arguments supporting the applicability of Six Sigma to IT processes, let us make an attempt to understand the applicability of Six Sigma to the processes that are an integral part of IT services.

Core Delivery Processes
The software development life cycle (SDLC) consists of four phases – Analysis, Design, Coding and Testing. Along with these core phases, processes like defect prevention, project management, Software Quality Assurance (SQA), Reviews, etc. are an integral part of the Quality Management System of any IT service provider. The effectiveness of these core processes directly impact the CTQ parameters. There is a large scope for improvement in these processes in most IT companies. Six Sigma can be deployed to improve these processes.
One of the key factors in deploying Six Sigma is identifying the “Y” metrics (dependants). But for core processes this becomes simpler since historical data for key metrics such as review efficiency, review effectiveness, productivity, defect density, schedule variance and effort variance are already available. After prioritization, critical poor performing metrics can be taken as Six Sigma DMAIC projects.
Six Sigma DFSS methodology can be applied for software development projects. Six Sigma in SDLC helps in making the software manufacturing process more predictable and ensuring that all Customer CTQs are met. Some Sigma tools that can be applied in this methodology are –
• Quality Function Deployment (QFD) helps in converting the high-level customer requirements (VOC) into detailed program specifications. Use of QFD ensures that no requirements are missed and it also helps in prioritizing the software elements.
• Failure Mode Effect Analysis (FMEA) is a tool that provides effective risk management for the entire SDLC, and identifies the probable failure modes of software at design phase. This initiates corrective action on the design.
• Pugh matrix enables software developer / analyst to compare different concepts with reference to customer CTQs and create strong alternative concepts from weaker concepts Scorecard is a predictive tool used for:
• Predicting final quality (Y metrics) based on process (X) metrics
• Quantitative Risk Assessment Identification of High Defect Drivers Linkage from Customer CTQs at lower levels in a flow down
• Application of Design of Experiments in software testing is an emerging trend. Software testing based on orthogonal array, detects most possible defects at a fractional testing time.

Delivery Support Processes
The processes that are value enablers are equally important to consistently deliver best quality service to the customers. These processes consist of infrastructure and network services, Resource Management, HR processes, Finance and accounting, Training, Central Quality organization etc.
Efficiency and effectiveness of delivery support processes directly or indirectly contribute to the productivity of core delivery processes. Processes like infrastructure and network maintenance are extremely important for offshore development / BPO models.
Six Sigma DMAIC projects can be forked to improve any or all the processes mentioned above.
Some Y metrics for Delivery support processes are
• Resource turnaround time
• Cycle time for recruitment
• Defects in payroll processing
• On time invoicing
• Accuracy of invoicing
• Network response time
• Network utilization
• Training effectiveness
In effect, Six Sigma has a profound impact on the most critical resource in IT industry i.e. human resources.

Product Quality Attributes
It is of paramount importance to deliver a high quality software product. The application of Six Sigma to the above two areas – Core Delivery Processes and Delivery Support Processes, directly or indirectly contributes to product quality. Metrics like response time, resource usage (Memory / CPU), and resources availability are critical to the quality of a software application. Six Sigma methodologies can be molded to optimize performance in keeping with the required metrics.
The Six Sigma DFSS methodology enables us to predict product performance in the initial design stage so that adequate control measures are in place. Figure 3 depicts the impact of any error or missed requirement in design phase on the cost in the later phases of the software development lifecycle. It has been proved that time taken to fix a design or requirements defect during testing phase needs about 20 times of rework effort as compared to a defect fixed right at the induction phase. Here, deployment of Six Sigma can play a major role to reduce or control the development costs.

The DFSS methodology as applicable for software processes cannot be directly mapped to DFSS methodology as implemented in manufacturing processes. In manufacturing, a product once designed is produced for years together. Whereas, in case of software development, a software design is manufactured (coded, to be precise) only once. This makes the application of DFSS in software development tougher. In a typical manufacturing setup, the crux of DFSS lies in achieving manufacturability at Six Sigma quality levels. For a manufactured product, the design budget might be flexible but in the case of software solutions, the budget for design is very limited and all the CTQs must be met in the given budget. The DFSS rigor ensures that the software is designed, coded and approved with minimum rework.
The DMAIC methodology can be applied to improve the Product Quality Attributes of existing applications, too. Many a times, as the user base increases or if the application is deployed in a global environment, response time decreases. Round the clock availability of application has also become a critical issue in today’s global work culture and BPO scenario. DMAIC projects can be implemented to tackle such issues and find a cost effective fix. Improving reliability measures like MTBF (Mean time between failures) and MTTR (Mean Time to Repair) can be other focus areas of DMAIC improvement projects.

Customer’s Processes
Most IT companies provide “End to End” solutions to their clients and therefore enjoy a long-term relationship with their customers. This has benefited the service providers in acquiring significant domain knowledge. The consultants possess fairly good amount of tacit knowledge about the client’s core business processes in addition to IT skills. Six Sigma tools and techniques provide an excellent channel to develop a basis for solution based consulting.
Six Sigma methodologies can help core business processes as well as IT processes. Owing to the consultant’s exposure to customer’s processes through IT support, they are familiar with the best functioning processes, processes which are not operating efficiently and those processes which have reached entitlement. This enables prioritization to tackle the relevant processes and this prioritization of improvements makes implementation of Six Sigma easier.