AutoTune on the Yaw Axis. How to know if it needs to be improved

I successfully ran autotune on all three axes of my hexarotor and everything looks great except I am sceptical of the yaw axis. I am testing the new gains by deflecting the yaw stick, letting it snap back to center, and observing the overshoot on that axis. There appears to be no overshoot, but the drone slightly oscilates on the roll axis as it prevents overshoot and returns to level. What I am looking for is confirmation that this is just something I will have to live with or if there are any basic suggestions/links to approach resolving this observation.

Thank you.