This is a question that gets asked very frequently especially at the end of projects that have not gone well.
While there are many people who will advocate one methodology or another for testing and tell you their way is the only way, every method must adapt to the project at hand, the risks and the desired outcomes. There is no one way to test but there are ways to improve the testing.
The first need is time at the end of the project to evaluate what went right and wrong before people forget.
Some common concerns:
- Too much documentation: It is acceptable to reduce documentation if the same testers will be available for the next project but not so effective if the next project will be staffed with new people unfamiliar with the project.
- Too little documentation: This can be solved but it takes time or can be fed through AI to generate the complete documentation. Always review anything produced by AI.
- Too much repetition: This is a process error that can be solved by RCA and then implemented but is not likely to have much effect for at least a project or two.
- Unused testcases: This is a classification issue that can be solved with AI at this point.
- Too many escaped defects: Again RCA with the aid of AI can address this but it is too late to implement the solution for the existing project.
