I mean honestly it makes sense if you have a force of 100k and lose 5k troops its not likely a big enough blow to have a large impact on weather or not you want to keep fighting, but if you only have 20k and lose 5k your nations military has been severely crippled and chances are will be a major factor in weather or not you want to keep fighting.
What you're describing is war exhaustion though, and that simply isn't modelled in CK2.
In my experience, of defending (on multiple occasions) against Catholic crusades, it felt wrong and played out horribly.
~30k vs ~30k victories, where I'd typically take 5k losses, for 10k kills, would earn me ~2WS.
The same battles where lost (with reversed K/D) would cost me ~10WS.
The innumerable stack wipes of badly managed AI troops, where my 10-30k stacks would wipe out 2-3k at a time, would barely register.
After several years of fighting, losing a single holding in the target would result in ~30% WS swings.
A complete nonsense as it's a trivial accomplishment for a 30k stack assaulting, and can just as easily be reversed a few days later.
When armies number in the hundreds to low thousands, the balance between battles and occupation WS works reasonably well.
However when the armies reach tens of thousands, and up, occupations & ticking warscore dominate.
The changes I'd suggest are:
1) A change to the way ticking warscore is applied.
It should accrue, or diminish, at a rate proportional to the % of the CB target held.
Hold nothing -> diminish slowly; 1 holding -> accrue very very slowly; hold entire target -> accrue rapidly.
Thus no more ridiculous warscore swings from capturing a solitary castle.
2) I don't think casualties should directly* factor into warscore, nor do I think the levy size of the attack or defender should matter.
Warscore from battles should be scaled according to the size of the CB target (number of holdings, or sum of the holdings' levies).
Thus if it's a crusade for a 1066 small, undeveloped Kingdom (e.g. Ireland), then battles involving thousands of defenders should give significant warscore towards resolving the dispute.
Whereas a crusade for a 1400 larger, more populous Kingdom such as Egypt, should require victory over much larger defending armies in order to earn decisive amounts of WS.
So something like: "defenders in battle" x "size of CB target scalar" x "+win or -loss" = WS
*(Casualties still indirectly impact WS, as they affect the participants' ability to effectively fight the war)