Dungeon Replayability

19 May, 2025

A common way of playing long-term dungeon adventures takes the same format of the West Marches campaign: go down an entrance at the start of the session, do as much dungeoneering as you can and make it back to base by the end of the session, and repeat this each week until the dragon is slain or whatever. It’s so common, various games have rules in place to facilitate this (Five Torches Deep for example), and for good reason: it has all the benefits of the West Marches – a seamless combination of episodic one-shot adventures with an overarching plot, easy exchanging of characters if somebody gets bored or killed, facilitated scheduling – on top of the benefits of the dungeoncrawl, and you don’t need me to tell you how great a dungeon is.

There’s an issue with this though, and it’s an issue of prep efficiency. A thing I love about sandbox campaigns in general is how often players can interact with the same pieces of content over and over again, because the context around them has changed: think of the players committing a slight against an NPC, then that same NPC trying to take vengeance against them, then the players going to foil the NPC’s plans. That’s three whole encounters all involving the same one piece of content. This technique of remixing a piece of content by changing what it does in response to the players, or by presenting it in combination with other pieces of content, or even by hiding/revealing information about it, works for many kinds of things (mostly NPCs, factions, population centers – can you tell I like city campaigns?) but one thing it doesn’t work well with as a baseline is the dungeon room.

At its inception in The Early Days of the hobby, the dungeon room had one purpose: to be cleared, and then forgotten about. Gamers of that era (as Matt Colville once told) might remember a time when progress was measured by the number of rooms cleared. The dungeon room, in its simplest form, can only be “consumed” once – kill the monster, solve the puzzle, avoid the trap, get the treasure, and you’re on your way. I find this terrible design ethic! It would be like turning every NPC, with all their multifaceted personalities, connections to other parts of the world, and narrative potential, into a combat encounter that always ends in them dying, never to be seen again!

The Techniques

So how do we go about making dungeons replayable? There’s various tricks to employ.

0. The Restocking Procedure. It’s a practice as old as time, going back to the Early Days of the hobby, but I’m putting this outside the numbering order because I don’t really find it satisfactory. Taking a room that has been freshly emptied by the players and filling it back up with “stuff” that was just as meaningless as whatever you had there beforehand doesn’t solve the problem, but just patches it up. Sure, it’s a new, stocked dungeon room, but it’s also a different encounter entirely. Most of the info the players had about the room will be lost anyways. The following tricks, more modern in design sensibility, instead focus on reusing the same key feature of the room multiple times, or on modifying other existing content.

1. Information-keys. Rooms that, beyond the basic interaction their contents provide, can do something else if the players have a specific piece of info, that can be retrieved elsewhere in the dungeon. For instance: a coffin leaning against a wall contains a mummy that will attack the players if they open it. If they find the ghost of whoever the mummy was in life, it can tell them that the coffin’s bottom can be opened for access to a secret tunnel.

2. Actual keys, or key items. Zelda style. Use doodad X on thing Y and Z happens, and the only way to make Z happen is to use X on Y. Pretty standard. Keep in mind that tricks 1 and 2 can apply to any kind of reward: treasure, access to more of the dungeon, pathways to ease traversal, or even more keys or information. Less subtle than 1, but clearer.

3. New angles. You enter a room from one point and you approach whatever’s inside from one angle, but you also spot another entrance that lets you approach the challenge from a different – maybe easier- angle. Think of the Asylum Demon from Dark Souls: you can enter floor-level to face the demon head-on (and get slammed), or enter from the balcony above the demon and plunge-attack on it. ¹

4. Monster placement and chase states. Enemies can act as a sort of movable barrier that can dynamically respond to whatever the players are doing. You interact differently with a space when you’re being chased by an owlbear: you might not have time to climb walls to get to high places, and conversely you’ll favor passages the owlbear can’t follow you in, like tight tunnels. Being chased by something can force you to rethink dungeon traversal, and the nature of what you’re being chased by affects how you think: you’ll dive underwater in response to flying enemies, you’ll climb high structures in response to grounded ones.² Additionally, if the players exit the dungeon after a chase, the monster will probably stay around there when they come back, making for a dynamic “lock” that modifies the dungeon’s layout in response to the player’s mistakes or deliberate actions: the players might get chased to lure a foe out of a zone and go through it later unimpeded.

5. Secret connections. Rather than the much more boring option of having secret rooms be extra treasure, I like it better when they are extra hallways that connect different parts of the dungeon, allowing players to skip dangerous zones.

6. One-way doors. Another design cue from Dark Souls. Some doors can be opened only when going upwards towards the exit, and stay open when the players go back down. It modifies the path of descent to the lower levels, and provide an incentive for returning to a point: see a locked door on the way down and come back to it on the way up to open it, and the reward for the extra work is that next time they get to the juicy bits of the dungeon sooner.

All these boil down to two key points: Incentive and Dynamism. Players will go somewhere if they have a reason (an incentive) to go there, and making sure the incentive is not exhausted after the first encounter with a given room is the setup to make the players engage with it multiple times. Dynamism instead makes sure that the encounter is fun each time: by altering the situation in which the players engage with the room (by having them be chased by a monster, or enter from a new angle, or with new info, or using any of the above tricks), they can act upon the knowledge they collected on the first pass-through, but in a different framework. The content is the same, but the encounter isn’t: it’s fun for the players since they don’t feel like they wasted an encounter’s worth on intel, and it’s easier on the GM’s side too.

But all this isn’t just for long term campaigns; it can be used just as easily to make one-shots out of very few components, or to make a one-shot’s worth of content into two- or three-shot games. Additionally, it makes it feel much more cohesive: all time spent with an encounter feels like it matters, since it will probably come back, and the info they gained will turn out useful in the future.

Well, what now? I won’t hold you further. Go have fun making dungeons, but most of all: hirelings of the world, UNITE!

Leave a comment on Bluesky

#theory