For a complex product created in a development environment, you should create unit tests for individual parts of your product's key decision factor rules, using CER's support for rules testing.
You might consider creating end-to-end unit tests that test full scenarios involving the creation and activation of evidence, and the creation and activation of product delivery cases, to test that the overall key decision factor results are calculated as expected.
You might also perform manual testing of the online system to check that your overall key decision factor scenarios are handled as expected.
The Engine may encounter runtime problems when calculating key decision factors, due to calculation errors in CER attribute values.
If there is a runtime error in the calculation of a CER attribute value for a key decision factor, such as a reference not found (analogous to a NullPointerException in Java), or a division by zero, or any other calculation problem, then the Engine will throw an exception. The application logs will contain details of this exception including its stack trace. For CER calculation errors, the stack trace can include important information regarding the location within a CER rule set where the error occurred. To fix this, you will need to debug and retest your rules.