As a rule, a PRA is first performed at the master test plan level, i.e. early on in the total testing process. To ensure the PRA's usefulness, the scope of the testing process and the business requirements have to be clear and more detailed requirements, designs or acceptance criteria have to be well on their way. Insofar as necessary, supplements and further detailing are provided at the level of the separate test levels. If there is no master test plan, a PRA is performed at the level of the separate test level. The test manager must ensure that the PRA covers only the scope of the test level.

The result is incorporated in the (master) test plan and constitutes the basis for the subsequent decisions in strategy as to non, light or thorough testing of a characteristic (e.g. a quality characteristic) or object part (component) of the product to be tested.

You should realise that a PRA is a snapshot. In the course of the subsequent process, it will be found that some risks are overestimated and some have been underestimated or neglected altogether. The test manager must adjust the test strategy and associated estimated effort, as well as the planning.

1. Determining participants
2. Determining the PRA approach
3. Preparing session/interviews
4. Collecting and analysing product risks
5. Completeness check.

The description of activities below can make it appear as if the organisation and performance of a PRA are not very complicated tasks. In practice, however, many test managers struggle with the tasks. This chapter attempts to describe the process as correctly and transparently as possible, but we should emphasise that much depends on the specific test management skills of the test manager. For instance communication skills, diplomacy and far-reaching knowledge of the possible risks and characteristics (to be tested) and being able to make a (rough) estimate of the costs, time and practical feasibility of testing.

Iterations

When working with iterations or increments, the question is: what is the scope of the PRA. Is this the system to be realised, or is it an intermediary product at the end of the iteration or increment? In such a case, the preferred option is to perform an initial PRA covering the defi nitive system, followed by a small-scale additional (mini) PRA focusing on the iteration/increment.

Example - A system will consist of 2 functional subsystems. The PRA shows that, in addition to functionality, performance and security must also be tested. Subsystem 1 is realised in iteration 1; subsystem 2 is added in iteration 2. Performance and security can be tested starting from iteration 3. The mini PRA and related test strategy for iteration 1 determine that subsystem 1 is tested thoroughly from the functional perspective. The mini PRA and the test strategy for iteration 2 determine that subsystem 2 must be tested with medium thoroughness, and subsystem 1 only with light thoroughness for regression. It is decided that for iteration 3, functional integration and regression tests plus performance and security tests will be executed.