Something I am a bit surprised that nobody has mentioned yet is that you probably don't want to put indicator terminals inside of case statements or event structures, or at least not ones that aren't guaranteed to execute (i.e. if the only way to stop a VI is to go through a "Stop" frame in the structure).
The reason for this is that the VI will behave differently when one of the terminals is wired to the connector pane as an output. If you run the VI with an unconnected terminal and the case with the terminal in it doesn't execute, the value remains the same as it was before the VI ran. However, if you run that same VI as with the connector pane hooked up and the terminal's case doesn't execute, the default value from the terminal is the output instead.
This is mostly a problem if you're using a VI that you sometimes run as a sub VI and sometimes as a top VI (i.e. for debugging) but can also cause issues if you're using local variables or property node to read back the values of the indicators on the screen (though as was said before, this should be rigorously avoided).
Plus, if you bury your output terminals in frames where you can't see them for SubVIs with multiple outputs, it's possible to forget about that one particular output and fail to check to see if you have to adjust its value when altering the VI to change some of the other output values.
So for that reason I always put my indicator output terminals at the right side of the screen outside of all frames, but I will quite often bury input controls inside the code whereever they need to be used (especially if there's only one relevant frame of a case or event structure that uses that input).
