It really should be a fairly straightforward process (note this does not imply simple). I suggest the problem is too much thinking around the requirement and a want of diligence in sticking to time and budget constraints, both of which should be defined early on (especially the time aspect).
Also, perhaps, confusion over whether the objective is to have the best possible example of whatever-it-was-you-wanted or a replacement that is somewhat better than what it replaces and on time.
‘The perfect is the enemy of good enough’ should be tattooed on every Sandhurt graduates forehead.
The British Army beat the Germans with good enough.