Jona brought up some interesting ideas. I would fully support making battle server work like that and have the siege server work a bit differently where the other team has a better defensive position, but lack something. In most games they have worse equipment and so on, but in cRPG that wouldn't work so I'm not sure how exactly balance it out.
There's a problem with the balance though. Stacking could be a huuuge issue if there's no autobalance. This would drive the server populations down in 30-60 min rounds, so there has to be a way to balance mid-round. Dunno if it's possible to balance players that are alive to the other team.
The XP system in the battle type server would be easy to set up as Jona pointed out. Combined with a decent auto-balancer it could be great.
On Siege type game, it would have to be the flag caps that matter. Imo this would require the map makers to be extremely careful not to make some flags too hard to cap and some flags too easy. It could be set up with a constant XP for defenders of around x2 and then bonuses for team having a good KD. These could go up all the way to x3 or x4. On the attacker side it would be all about the flags, with base XP being around x1. Take one flag and whole team gets XP and the people currently on the flag get a small, maybe 2-3k bonus. On a map with total of 7 flags for example in a 30 minute round the flags could be worth 30*x4.5/7=19x per each flag capped. That would give out great XP when capping all flags quickly, but really low when capping no flags at all.
Let's say Team Grey Stack would take all 7 flags in 20 minutes. They get the base XP of 1k for 20 minutes = 20k plus 7 flags=133k=7.65k/min plus bonuses for the flagcappers. Defenders would get 2k/min plus bonuses
Let's say Team randomers would fail to take but one flag in 30 minutes. They get base XP of 30k plus 1 flag=19k which totals to 1.6k/min plus bonuses for those pro players with high KDs that suffered in the losing team
Pros: Gives incentive to be quick about the flag cap and not leech with a constant multiplier
Cons: Stacks could be horrendously effective