The depth of the testing
will depend both on the complexity of the change and the importance
of the area of functionality to the customer. From identifying what
the system should do, the criteria for user constraints should be
included. The test plan must include tests that reduce or eliminate
each identified risk. Each test applied should meet the 'S.M.A.R.T.'
criteria:
Specific: what
you are testing for must be clearly defined. The anticipated
result must be unambiguous.
Measurable: the
anticipated result must be capable of comparison with the
actual result to define pass or fail.
Achievable: do
not repeat the same test unless you are testing against new
criteria – large test matrices do not necessarily achieve
more than shorter, more specific, ones – they just take
longer.
Relevant: anticipated results must be those expected from the system. Do
not test for what is not there.
Timely: the test matrix should
naturally follow the process; anticipated results should be
ordered in the same way as the system would chronologically
produce them.