A two-stage strategy for anti-spam is actually not a good idea if you have a top-notch email security appliance at the border. The amount of leftover spam is very low and you’re likely to have a similar proportion of additional false positives compared to any additional caught spam. Two stages are only really needed if your border solution stinks.

A related point about end-user rules. I’d suggest that most people stick with blocking senders and domains they don’t like. Anything beyond that is probably inviting more false positives than not.