Prepare the mix
Keep dialogue and music on clearly identified tracks. Check that the dialogue range actually contains audible speech and that mute or solo states are appropriate for the analysis.
Set a reasonable base music level before applying ducking. If the unducked music is excessively loud, the amount of required attenuation can make the result sound unstable.
Review the ducking behavior
Automatic ducking lowers music around dialogue according to the selected setup. Listen to the approach to each line, the speech itself, and the recovery between phrases. Short pauses should not always cause a distracting rise in music.
Inspect quieter words and breaths, which can reveal where the chosen settings need refinement. Keep the intended musical phrase and the intelligibility of speech in balance.
Revisit after editing speech
Moving or deleting dialogue changes the timing relationship with the music. Recheck the ducking after a substantial speech edit rather than assuming an earlier result still fits.
Play the full sequence with every intended track active, then measure the finished mix before export.