This might be a good time to start stealing concepts from neighboring disciplines. There are lots of templates and standards around capturing reliability analysis results and potential issues. From ishikawa ("fishbone") diagrams showing intefactions and potential operations for failure, to FMECA ( Failure Modes, Errors and Criticality Analysis), to even Sigma analysis (as in SixSigma), there are lots of great starter kits to work from. It's just kind of strange that IT all too often does not take the time to do such things.

Simply taking the time to think through the issues is great, but capturing the results in an easily disseminated format will help a LOT. it's harder to "crucify" someone for a bad outcome if the potential for bad outcome was documented and lsigned off before the work started (or went into play).