- Mike Smith argues that software testing models often rely too heavily on test cases, which may not provide the best measures for control and risk management.
- An effective measurement framework separates objectives from initiatives and uses a complex model of relationships rather than a simple hierarchy. This provides better traceability and the ability to cope with change.
- Lessons can be learned across different domains of measurement and testing. An ideal testing model would incorporate concepts from performance management systems like the balanced scorecard to link testing to business outcomes.
- Many factors influence what level of measures and targets are suitable for a given situation, but the most important thing is that the model supports analysis and decision making to maintain control.