For the color balance node, now that blender handles the filmic color management I understood that one should rather use the Offset/Power/Slope correction rather than Lift/Gamma/Gain. It seems then much trickier to control and fine-tune the output. Is there any best practice and recommendation to start with ?
When it comes down to it, there is no specific method that works best. Lift/Gamma/Gain, Offset/Power/Slope, and Shadows/Midtones/Highlights are all just different mathematical models to do the same thing, just in different ways.
Read more about them here: https://www.provideocoalition.com/whats-in-a-name/
If you're comfortable with the Lift/Gamma/Gain model, I'd just stick with that because it won't really cause any noticeable difference in quality. But, if the curiosity is driving you crazy, and you really want to explore and master the other models of color correction, go for it! :)