The Dangers of the Requirements Coverage Metric
When testing a system, one question that always arises is, “How much of the system have we tested?” Coverage is defined as the ratio of “what has been tested” to “what there is to test.” One of the basic coverage metrics is requirements coverage-measuring the percentage of the requirements that have been tested. Unfortunately, the requirements coverage metric comes with some serious difficulties: Requirements are difficult to count; they are ideas, not physical things, and come in different formats, sizes, and quality levels. In addition, making a complete count of “what there is to test” is impossible in today’s hyper-complex systems. The imprecision of this metric makes it unreliable or even undefined and unusable. What is a test manager to do? Join Lee Copeland to discover how to reframe the idea of coverage from a quantitative to a qualitative measure, recognizing that coverage does not measure our testing of a product, but our testing of a model that represents the product, with many details omitted.