Appropriate specifications at the IND stage

Biopharmaceutical product specifications are to be "scientifically sound and appropriate". However, how does a biopharmaceutical company determine what are acceptable specifications for its product at the early stages of clinical development? Are there really enough test data at the start of the IND trials to "establish" specifications? As an industry, are we doing too many tests at the IND stage (i.e. following the "must do every test on someone's published list" syndrome), and then finding that specifications need to be set for each test we run? Are we feeling pressurised (either by regulatory agencies or by our own corporate cultures) to set unrealistic and tight specifications at the IND stage, and then later regretting how to justify loosening them when we gain more experience with the release testing or stability of our products? A discussion on how to set practical product specifications for biopharmaceutical products at the IND stage is presented.