A Support Engineer, trying to do the right thing, attempting to conduct support analyses, to carry out Through Life Cost predictions, in the earlier stages of a system’s life cycle will often be told that it is too soon for such activities.  The arguments for this seem reasonable enough, the design is not mature and hence there simply is no data.

Then Too Soon, suddenly becomes Too Late, with no apparent intervening interval, we now realise that the system design is now pretty much fixed, it’s now too late to influence the this design, and because the trials have not begun, we still have no data!