In the last year, i’ve been playing D&D both as the Dungeon Master and as a Player, and here are some of the things i’ve observed which I think creates a successful session or dungeon. It’s based around the notion that the three pillars of D&D, are:

Roleplay
Combat
Exploration

Rather than have separate scenes for each of these, I tend to think it’s possible and desirable to have all three as options in all Dungeons. Not being familiar with terms, i’m going to use the word ‘Dungeon’ to refer to any distinct part of a campaign that happens at a single location and the word ‘Scene’ to refer to a thing that happens in a single room with a collection of characters and objects.

Give the Players Lots of Problems

…rather than just a single mission to accomplish, the players can be given multiple problems to solve at once, sometimes these should be competing with each other for the players attention, actively working against each other and stacking up, and all of which can be played using any of the pillars. The following are examples of the kinds of things that should be happening not one at a time, but have several of these layered on at the same time:

The Players must rescue a Dark Elf Prisoner from a Castle by negotiation, combat or stealth, their choice.
The Custodian of the Castle is obsessed with adding to his collection of Powerful Magical Objects he has ordered local innkeepers to confiscate any Magical Objects from the players.
One or all of the Player are being pursued by an assassin / wanted by the local police for some reason related to a previous encounter.
When the Players arrive, a prison breakout is already underway, the Prisoner has set himself up as a mad cult leader making complex demands.
A secret entrance is a puzzle involving levers and passwords and stuff, but the players should never be left alone to solve it: they’re being sniped at all the while by either literal snipers or just a stupid goblin in a cage who is constantly insulting them and demanding release in order to give them clues.
Minor characters in bad situations, at very inconvenient moments.

Give the Players a Toy Box

Once the players have a mission to accomplish, the environment should be rich with stuff the players can use. Multiple levels, things to hide behind, destroy or climb over. Bridges and Balconies are always good. It is not the DM’s job to figure out how this stuff gets used, or even to worry about it. The players will always find things to do when under pressure.

If the players see a very expensive chandelier in a room near a balcony, above the entrance, that is clearly stolen from Elven culture, at a point where Orcs are imminent, there are at least three or four ways to use that.
A magic grappling hook/glove thing with a high chance of entertaining failure. This is a good tool, as it allows the players the opportunity to work with the environment and alter the topology of combat.
Barrels full of Very Expensive Flammable Booze.
You can’t beat an angry and belligerent Goat: fight, befriend or enrage for tactical opportunities.
Objects which have vaguely suggested interactions with other objects seem helpful, but don’t be prescriptive: all of the above seem to me to be combinable in interesting and alarming ways, alongside whatever other nifty items the players have on them.

Interesting Antagonists

Partly, the above points about complexity and toy boxes is to enable you as a DM to give the players more interesting villains for low level players and the equipment to deal with them outside of regular combat and weapons:

A badly wounded Liche who is bargaining with the players for its life and the players own while it is also under seige by a Half-Orc Crime Syndicate turned vigilante force gives the players a chance to encounter big villains without being completely outclassed in combat, as they would be against any one of these characters under normal circumstances.

There was an imprisoned Gargoyle in the basement demanding to be set free (the players passed on this) and a swamp outside one entrance with a being known as ‘Mister Mouthy’ living in it. He ended up being used to dispose the main villain by an unexpected combination or throwing axes and flight potions.

But since we’re using this to pit the players against higher level villainy than the Manual recommends, we must…

Give the Players An Exit

If things are getting overwhelming, the players should never be unable to leave in such a way that it feels like an entertaining tactical and temporary withdrawal rather than failure. An exit should always lead to further complications and opportunities. Keep exit scenarios in mind in case you need to use them, if they aren’t used, they’ll keep.

Players leaping out of windows may find themselves in a moat containing an alternative entrance to the Dungeon. Probably the one with the goblin mentioned above if the players hadn’t found that by other means.
While the Villains are looking for the misplaced players, they can sneak back into the game, but the players have new pursuers to add to their troubles later.
They’re hiding in a Priests Hole when the encounter the Priest, a pythonesque character capable of giving the players important information, but is similarly insisting on shouting and singing and attracting the guards.

The players will find their way

Players will find their preferred solution to a problem: some will want to fight, some to sneak and some to roleplay their way into the castle, so it’s worth wile considering at least one ‘scene’ of each. Make them work for their preference and mix it in with the other things:

Players who want to Sneak can find the secret entrance, but also encounter That Priest or That Goblin meaning that they have to either Fight or Roleplay in order to maintain their sneakyness.
Players who want to Fight can find the fight, but will find themselves outclassed and find themselves having to explore to find tactical advantage, or Roleplay with other entities for support.
Players who want to Roleplay can find the tense negotiations, but should probably have to do it while hiding behind a barrel of explosives, or during which they witness a mugging through the window in the background.

The characters choose what success and failure means.

Part of the benefits of throwing a lot of stuff in is that it always gives the players (and their characters) choices: about how to prioritise threats and benefits, and about how to judge their own success.

OK we didn’t rescue the prisoner, but he tried to kill us, and anyway look at all this magical loot.
OK we are now wanted for the crimes of our enemies, but we freed the goblin slaves and our conscience is clear.
OK we had all our gear stolen, but we have all these barrels of expensive booze.

Pirate Town — A town harbouring Pirates, who prefer to remain secret, but with a vested interest against securing the town against the Mayors necromancer daughter, who had escaped from her ‘private residence’. It wasn’t a bad setup, fight was fun and the players had to decide what to do with the Daughter, who could be viewed as ‘innocent’ on some levels… But it was confused by having multiple factions of Pirates whose motives and power game in the town was vaguely explained and distracted from the meat of the situation.

Some mistakes I make:

Don’t overdo it and confuse the players, but should have maybe two immediate concerns from the three pillars during any scene, and maybe up to half a dozen such things going on in a bigger picture.
The players should largely be able to distinguish between the local story and the Campaign Story. The Campaign story should probably be pretty simple and easy to grasp: if you do have a grand conspiracy, interest in smoke and mirrors will wane pretty quick of you don’t start being decisive about giving players answers.
Aside from the Campaign Theme, don’t keep a concern live for more than two to three sessions otherwise either the players will bet bored with it, or conflate it with a bigger plot when you don’t want them to. So: tidy up none Campaign plot lines reasonably fast and have one closing as the next opens. After introducing an assassin, the players should start the next Dungeon being pursued but have the opportunity to confront and defeat them within the next two Dungeons.

Conclusion

As I got better I tended to make lots of modular bits and pieces: characters, objects, sets and combining these, ‘scenes’.

I should say that I always have a reasonably well sketched out situation: the players would feel as cheated if I was making it up as I went along as if I was rigid in the outcomes. This is hugely difficult to get right.

I usually have had some idea how these are to be strung together, but I’m usually quite happy to leave them out of a scenario for later use if they don’t occur. I’ll throw the band of drunken Dwarves in in when they’ll add complexity to the situation in response to a particular player action rather than necessarily that it has to happen at this particular point in the story.

By complexity, I sometimes mean threat, but more often than not I’m motivated by what seems funny and upsets the apple cart if the players are too relaxed about the setting: they should feel their characters are competent but always off balance.

The pacing of adding and resolving these multiple layers in a Scene or Dungeon is very difficult to get right. Finding a Rhythm is key. Some Inspiration:

Jackie Chan, Pirates of the Caribbean, the Left 4 Dead Director, Pixies

Dungeons, Dragons, Scenes and Complexity