I’d add a fourth option: not over-trusting the methodology itself.

The world isn’t a perfect-information game, and many “problems” are defined under uncertainty.