When I was learning guitar growing up my teacher used to tell me the hardest part about song writing and playing was choosing where to go when faced with different creative options. His point was a lot of people struggle not because they can't come up with anything but because they have too many ideas vying for control. They just have trouble discerning the best path and going forward. Guitar playing is a lot different than game mastering but I think the issue of choice is similar.
When I run into trouble in a campaign or a session it is often a product of failing to make a concrete decision about a key event (did Lady Plum Blossom go to Mystic Sword Temple to wipe out the remaining disciples and if so how). I may even think I've made a choice but fail to make decisions about the details. This is is the worst case scenario because it creeps up on you and you have to spit out an answer when it comes up.
As a GM you are constantly making choices. A lot of this is before the game starts, during prep. But I think most of your choices actually arise in play as events unfold. Obviously you want to make the best decisions possible, and agonizing a little over choice can be a good thing, but I think sometimes you just need to pick something and move ahead.
The biggest issue here is what I call the half-choice, where you leave the outcome of an event fuzzy in your mind, so it has a shape but no real definition. For example, if the players just stole the Wind Saber of Sunan from Lady White Blade, it would be very easy to decide that she reacts by sending people after them, but not follow up with any clear details (who does she send? Where? When?). You can wing this, just throwing a bunch of Mystic Sword Sect assassins at the party every so often so they feel the threat is real, but what happens when they take one of their attackers prisoner and demand answers? Granted the assassin may have a limited view of the big picture, but you should at least know what that view is. He at least knows when he was sent, who told him to go after the PCs and why. If you've only made a half choice on the matter then you have to make up those details as the players question the guy, and while that can work it can also lead to all sorts of contradictions.
So what I find helpful is to simply make decisions about this kind of thing at the appropriate time, hopefully before I a situation arises where the PCs start asking questions. I'll just jot a note to myself during the game regarding Lady White Plans intentions, directives and actions. It is really helpful to have a campaign calendar here. I always keep one for my games, and while I don't always remember to make the best use of it, when I do it is one of the most helpful things for managing on-going events. This allows me to mark the day that something happens, pencil in a potential date for a certain event to take place (for instance if Lady White Blade sends someone to intercept the party in Chen, I can look at the map and estimate their travel time,then mark their arrival time on the calendar).
Again, you can also improvise. Sometimes you have to improvise. You should improvise. But even during improvisation, you have to make choices and the point of making decisions for me is 1) consistency and 2) making player character decisions matter. If you make choices and stick with them, these root your NPCs and their actions in time and space just as the player characters themselves are rooted. Time becomes important to play. If the players decide to spend a day at the Silk Tavern, that decision could mean the bad guy gets away or that they arrive after the their Sifu has been dropped into a pit of venomous insects (not before or during).
Making decisions, sticking to those decisions, is also a self imposed limitation on the GM that makes the world feel real. The GM can always cheat for a more dramatic result. While I am not saying the GM should avoid drama (I love when drama arises in play) or that he should always abstain from coincidences that keep things entertaining, I do think striving for believability is crucial to player-buy-in. And I feel like one important aspect of this suspension of disbelief is seeing that the outcomes the GM probably desires are not always occurring (sometimes NPCs make choices that work in the favor of the PCs and don't fuel challenge or drama).
I want to be clear that I am not saying the aim should pure simulation of a world, where the players are reduced to dealing with everyday-life frustrations and annoyances rather than going on fabulous adventures. I am just saying it doesn't have to feel like Raiders of the Lost Ark all the time. That is a movie, meant to entertain viewers and not interactive like an RPG. I think in an RPG you need to make sure that player character choices matter and that the randomness of the dice matter. Not fudging and avoiding railroads is one part of this. A more subtle thing is the GMs role in managing the background details of NPC choices, reactions and events. The more concrete your choices the more solid the setting will feel.