Some years ago, I was involved in a major process reengineering project for a services business. We kept worrying about what we do when something out of the norm happened. This meant a huge amount of rework trying to cater for automating every eventuality until we decided it made more a lot more sense to ignore the exceptions (which were around only 2% of volume). When exceptions happened, we’d flag them and find a manual “work around”
The result was a much more efficient business by concentrating on what normally happens instead of constantly trying to cater to what very occasionally happened.