Then playing again and again from start is metagaming as well...
And I mean not save scumming (like reverting bad consequences), but making learn-ahead time point and returning to it as many times as you need to learn variants of doing something (not outcomes!).