Sometimes when the opportunity presents itself (such as a weekend available to cache on a local machine) I like to save out the maximum density I can in a given timeframe simply because its easier to remove data than to not have it at all or build up density by interpolating between particles or the like.
In regards to simple workflows... I have a couple of easy-to-build compounds I keep handy (though its just as easy to build them as you go) one which assigns particles a random number between 1-100, another which tests for that value against a defined threshold for deletion or whatever. It makes it very quick to be able to build structures which act on a certain percentage of particles or to modulate other values with that number. By always using the same compound/logic I free myself from having to give it any particular attention or thought, I just drop the compounds in and know what I'm going to get. Simple but useful, and since it is (for better or worse) my own logic instead of one of the factory compounds I know what is happening under the hood and don't have to worry about unexpected results, pre-set contexts, or other caveats. Same goes for a number of other simple tools - the most useful being one which returns a uniformly random vector of a defined magnitude. The factory "randomize by cone" compounds irritate me. :)