The best advice/guidelines that I usually lay down for a modelling process especially a client facing data model is as follows :
- Model should be iterative
- Model should be dynamic
- It should be economical in data
- It should be economical in its views
- Concepts should be clear and recognized
- Model structure should be clear and recognized
- Models should be consistent to previously available models ie. - if they exist
- Try and make models as correct and complete as required
- The different concerns/viewpoints regarding the components should be treated orthogonally rather than in a scattered manner.