Nissafin's Paper

The goal of understanding quality costs is to analyze where you spend your time and money to get the most bang for the buck. It is well known that it is faster and cheaper to find and fix a bug during unit testing done by developers early in the development cycle. Should we then spend most of our time/budget on unit testing? No. There are many limitations to unit testing. Unit testing is not capable of finding many varieties of bugs, including graphical user interface (GUI) bugs, usability problems, end-to-end bugs, and configuration bugs. For most organizations, getting a better unit test effort will help you release a better product sooner. It is not a replacement for the test effort done by skilled software testers, but it may reduce the time that test effort takes. Understanding quality costs will hopefully help you shift some of your test effort to the most cost-effective places.

In the following, the total quality cost is shown in the upper bathtub-shaped curve. On the bottom axis is the quality of performance, ranging from totally defective to zero defects. On the left axis is the cost per good unit of product. You can see that with highly defective software, your prevention and appraisal costs are very low, but your failure costs are very high, yielding a high total quality cost. With zero defect software, likewise, your failure costs are very low, but your prevention and appraisal costs are very high. To optimize your total quality costs, you want to be between these extremes, at the bottom of the bathtub curve.

...