Map
A picture is worth a 1000 words.
It couldn't be more true when it comes to D&D. Projecting your idea onto a 2-dimensional gridded sheet will have you questioning and planning out the space with all of the possibilities it may present.
Rivals
Not villains. Consider that throughout the campaign, the party may have a political, economical or social opponent. Not everything is about reducing each other to 0 HP.
Use Backstories
My story can be intricate and interconnected to the point of brain boil, but nothing gets the players on board as missions related to the backstory of the PCs (Player Characters). It isn't even a choice for me anymore: whatever narrative I want to craft, whichever TV show I got inspired by, it has to be served as an extension of the player-created origins and spiced with character-defining choices. Or they will call it "A filler session"... Which doesn't make any sense... Ungrateful hyenas.
Balance the Light & Darkness
In the end, balance slays the demon. Probably the most abstract, complex and philosophically charged of all the rules. But reminds me of what is the most important: trust between all participants of the session. If the main characters are cynical and care about no one, it may take burning the village to balance the story, but if characters are sensitive, have families and friends they care deeply about, then the DM should be careful even when making enemies capturing one of those special people as prisoners.
Quest Failing & Branching
The more important the quest is to the story, the more failable it should be. Due to the nature of the game, I think the following sessions can be less fun:
Knowing that you are in an "introductory" stage of the story.
Having the only 2 outcomes be "We save the world from the big evil boss or our characters all die".
Subtle choices that will influence the story.
It is just too easy to see DM as fully responsible for the outcome with his encounter balancing or subjective arbitration for what the right outcome of the choice should be. That is why I prepare the big juice confrontational sessions to be subject to various degrees of subjective failure. It would have been hard to do if the session were an introduction to a boss, for whom I have an entire castle built up. So the idea is the following: the mission can be failed, but not necessarily by defeating the players. The enemy can escape, plan not stopped in time, or investigation leading to a dead end. It is infinitely more fun when players know that the story will go in two very different directions based on the outcome of the session. This turns the 3 statements above into 3 more fun ones:
Players decide how they want to be involved.
It will be interesting to know how the world will change if we fail this.
The choice of lesser evil or the better reward is there to signify the price they pay to equalize the scale (Light & Darkness).
An example for this, probably, overly complex topic:
There is clearly an evil guy. Players decided to infiltrate his operations by working for him and feeding information to his opponents.
The evil gut wants to open some Eldritch gate (the usual stuff, you know).
Help close the gate or catch the bad guy who is running away with the scroll that can open more gates.
Revelations checklist
To be ready to improvise, I write a list of things that players may be able to learn about. As they weave their story, I use the list and see if some of that information can be revealed as a result of that action. It will also guide world-building. When it is done, just cross it out.
Hook & Quest
There is nothing wrong with the "Clear a mine for 100 gold". But when intricacies start to surface, consider asking two questions:
What characters can be asked to do?
What characters are motivated to do?
If those yield two different answers, even better. Now this quest can be failed by bad luck and intention. The safety of nearby villagers is a reason to take the job, but restoring mining operations may be detrimental to the bigger goal. Party may decide that an "accidental" collapse of the mine has "unfortunately" happened during the fight with Kobolds.
Subvert bias, not expectation
As you are the master of the world, twists aren't very impressive. Players will immediately jump to doubting if you gave them a fair chance to predict the surprise. What does feel good is when players get a reminder that this fantasy world obeys True rules.
Tavern half-orc cook doubles as a bouncer, as they prefer not to involve the city guards no matter how nasty the situation gets. Clientel would prefer an oily fist to the face rather then a sight of red cape with baron's sygil.
Tool vs. Character
It will take a while until this gets old: Have characters use Evil books, Evil crystals, Evil rituals or evil whatever to do something good. Infinity gauntlet, Necronomicon - objects that were used to do evil, soaked with "Evil vibes" are still just tools. When in desperation players go "What if we use it, but for good" is always a good reminder that it is always a choice that defines us, not the tool.
Deduce not reveal
Never spell out the thing, just give enough hints for them to piece it all together.
Closed-door
They are everywhere, and the door's traditional job is rationing the content, one room at a time. Consider every door on your map. Is there a possibility for added context to augment it into a symbol of mystery and anticipation? Are there any clues to what is behind the door? Are there apparent risks to proceeding to the next room? Maybe we can make it more optional to provide space for hesitation and risk evaluation.
As you enter the mansion, there is a large door directly in front. From the layout you can assume the main hall to be found beyond. Muffled sounds of conversation are reaching your ears. A lot of people are in there.
Miles and feet
It is great to leave things to the imagination. But if you describe something, make sure there is no ambiguity regarding the relative position of the things.
You enter the room. There is a table in the middle, candles on the walls and another door opposite to you. Some web is in the corner.
I don't know how big the room is, but it is not a problem. The description is solid, even without the size.
You enter a room, there is a table 10 feet from the wall.
Now I need to know the size and direction of the wall. Is this table to my left? To my right?
I never know how big and far things should be. Is 500 feet a lot? Should the caves be 1 mile from the village or 10? How wide is the mountain? It is safer to have players imagine what they consider to be the correct distance, rather than risk pulling illogical data out of thin air. But something magical happens when the description gives the player a specific location of an object. Otherwise, we tend to create a mental mark of uncertainty, which may prevent us from fully immersing ourselves in the situation. So, scale, speed and distance are an area of preparation and learning that is worth putting effort into, and this effort will pay off.
DMing is impossible without prioritizing, and some things will get more attention, and some will be skipped. And often, there is no right or wrong way. But making sure that the core is established without a shadow of ambiguity is always a good focus. DM may forget to mention if the city has 800 people or 10.000 and what people are wearing, but it will never be as important as knowing how far from the entrance the kid who is throwing stones at the window of the tavern.
