Highly depends on the target and what you want to capture. In general, I would split the target into it’s components: individual narrowband channels and continuum/broadband.
Take M31 for example:
Just going for the main disk? An equal split of RGB is enough
Want want to capture the halo and ifn? Now you should capture some L too. The halo and ifn aren’t too faint, so it’s probably enough if you do a 2:1:1:1 LRGB split.
Want to capture the extragalactic Ha? Add at least a couple hours of Ha data.
Want to capture the foreground Ha? Make that couple dozen hours instead.
Ambitious enough to go for the Oiii arc on top of that? You’ll likely need tens of hours of Oiii data at least.
The precise data amounts per filter are highly variable. To estimate how much time you need per filter you either look at other images of that target, or look at NSNS and compare the brightness of your target to ones you’ve done previously.
Two good rule of thumbs I personally follow are don’t exceed like 4:1:1:1 LRGB split (after that point your color data will not support the lumiance enough imo), and keep the Narrowband/continuum ratio below like 10:1 so continuum subtraction doesn’t introduce significant noise.