combat:
For minion mobs, do keep in mind that the trade off for the GM is that they’re trading action economy for simplicity. I probably wouldn’t put more than 3-5 minions in a single mob.
For planning combat, the conflict rank of your NPCs in the scene should total to be equal to or around the total school rank of your PCs. You can start to increase the total conflict rank by 1 per each loot above basic loot that each PC gains over the course of the campaign. For PC Shugenja, I almost feel like their school rank should be treated as double for the purposes of calculating balance vs NPC conflict rank.
I also like the idea of throwing out the gridded map entirely and using “Fate-Style” combat. Fate is another TTRPG, but I think it’s combat works well when we think about range bands for 5e L5R. What you do is, lay out a few index cards, these represent different sections or “zones” of the battlefield/arena/building that the conflict takes place in. Write down on each index card at least 2 ”aspects”, these are terrain qualities, traps, or something useful the PCs can use. Then the PCs might add a third quality if they spend Opportunities to do so. Then use tokens or write down the names of different mobs, adversaries, and PCs to track what sections they’re in. It takes 1 rangeband to get within range 0-1 of a target that shares the same zone as you, 2 rangebands if the target is using some kind of Aspect in that zone to give themselves height or obstacles. It takes 2 rangebands to move to another zone.
This is more flavor than mechanical, but I also like the idea of the Player describing their action rather than saying what action or technique they want to use. Since players should be rolling for their actions before the start of their turn, and they have access to all the information surrounding the NPCs, they should know exactly how to describe the outcome.