Need support on a metric that will quantify 'Quality of Deliverables-First time Right' in software industry
Sudhir PorayathSenior Manager - LS compliance| Cognizant Technology SolutionsChennai, Tamilnadu, India
Looking for metrics that will project Quality of deliverables or First Time right in a typical software development project or testing project Saving Changes...
Rarely will you get "first time right" when dealing with software unless we are talking about a very simple enhancement or change to an existing application.
Quality is measured based on adherence to functional and non-functional requirements but if those requirements are not well understood by stakeholders or are evolving, it becomes a moving target.
Defect counts are useless as they might represent the tip of the iceberg. Rather, you need some way to quantify the overall cost of quality and incorporate the perception of stakeholders.
I would look at it a bit differently than my colleagues.
Test versions of software typically go through multiple iterations prior to release into an operational environment. Quality expectations differ significantly between developmental and released versions so I would recommend establishing a maturity milestone after which point issues count towards your "first time" quality. For example, that could be when you deliver a version to your customer or freeze the version to start certification with a regulatory agency.
After you have passed the maturity milestone such as the formal release of a product version, you might consider categorizing problems into multiple types. Some issues might be critical such as if they affect safe operation of a product and require an immediate fix. Others might be unsatisfactory to the customer but not urgent. Yet others might be invisible to the customer such as internal processes to collect performance data.
With different category types, you can not only count the number of issues after your "first time" but it also gives you a sense of their importance. Critical errors would raise questions about how they passed regression testing. Metrics of nuisance-type errors provide information of whether your testing plan is an acceptable balance of cost and time to release, vs. passing some errors on to your customer. Saving Changes...
"Nearly every great advance in science arises from a crisis in the old theory, through an endeavor to find a way out of the difficulties created. We must examine old ideas, old theories, although they belong to the past, for this is the only way to understand the importance of the new ones and the extent of their validity."
Community Champion
