It depends. If you want the 'tape' texture or signature to be printed into the audio (of course, just in subtle amounts on the master), then Satin could be the last stage.
But typically, a tape device would be before a (multi-band) compressor. This can come very handy, since tape can easily tame transients and occasional overly harsh peak sections. This way, a succeeding dynamics device doesn't have to 'work' that hard, meaning its incoming signal is already 'prepared' in some way, more even and predictable.