It's the correct approach, though, Morrus.
If you make one change, and something goes wrong, you know what caused it (broadly speaking).
If you make two changes, and something goes wrong, the amount of work involved in tracking down the source of the problem just doubled. That's a best-case scenario.
And with little things like COMPLETE DATABASE REWRITES and COMPLETE HARDWARE REPLACEMENTS, you know things are going to go wrong. It's never going to be a question of "hoping it all works" -- the question has to be "making it as simple as possible to figure out why whatever went wrong went wrong".
Any time you change your software, you introduce issues. Guaranteed. The trick is to manage those issues and make it as simple as possible to solve them.