Hey Phil, selection of non-functional requirements is an important decision as and when we plan for Non-functional requirements testing. These generally include the following:
-> Performance.
-> Load.
-> Stress.
-> Usability.
-> Portability.
-> Inter operations.
-> Usability.
-> How easily we can maintain.
Few of these requirements mentioned can be measured, rest totally depends upon the testers knowledge and experience.
Out of these characteristics mentioned which to consider depends upon the system under test.
So I would suggest to go for a risk analysis with your development team to chalk down the characteristics for which you need to have non functional requirements testing.
Till than cheers and wish you success in your efforts.
The main challenge is usually that the NF requirements are poorly specified, and it can then be difficult to get agreement to putting much effort into testing them. Often the NF requirements are unspecified and implicit. There are expectations of quality that have gone undocumented, but if you fail to meet them then all hell will break loose.
You need to go through the NF requirements working out how they can be tested. Often they will consist of bland, untestable statements. They need to be specific and ideally measurable. Challenge them if you can't see any way of testing them.
Use BCS set of test types, or the set provided by Sumeet, and check them against the requirements to see how many of the test types are both relevant and covered. There can be political problems if you start saying the NF requirements are rubbish and need to be revised. It can be more diplomatic to talk about the need to pin down the quality characteristics (effectively grouped by the NF test types) that the users require, and get some measurable targets that reflect the users' views in the light of the risk they're facing. It all comes back to risk. Defining the quality characteristics has to incorporate a risk analysis, as suggested by Sumeet.