Boss fights are hard. Most editions of D&D really aren’t designed to handle them, so there’s no simple math you can do to turn a fight into a boss. You could just apply a template to a monster or give the players an enemy X levels above them or give it Y minions, but none of those methods work on their face because they’re all intended to make the fight harder. They don’t make it more interesting. A boss should be difficult, and we’ll talk about that soon, but more imperatively it should be an interesting encounter. The players should remember the boss fight more than they remember the mooks leading up to it, and that generally means it shouldn’t be simply a scarier version of those mooks. The fight needs something to make it fun, and that responsibility falls squarely on the DM.
The most important thing in a boss encounter is what the boss does. Monsters are differentiated by their powers, far more than by their ability scores or equipment or strategy, and a boss should have boss-level powers. Coming up with those powers is an art, especially in 5E where monsters tend to work a lot like each other and Challenge Ratings are more the results of guesswork than usable tools for estimating encounter difficulty. And since we don’t have a good system for creating boss powers, and thus boss monsters, from whole cloth, we have to rely on a tried-and-true method that has existed since the beginning of the game: stealing them.
Thousands of videos games exist with boss battles. Unlike television or movies or other media, video games bosses already have numbers and behavior that can translate to D&D. Just by looking at different genres you can find a wide swath of boss concepts and powers, from a platformer’s projectile-shooting glass cannons to a RPG’s multi-phase monstrosities to a fighting game’s ordinary people with screen-filling punches. By picking bits and pieces out of these bosses, or ripping them off wholesale and giving them a fresh coat of paint, you can give your encounters completely different feels without writing an entirely new Monster Manual (which I’ve done—stealing is much faster). All you have to do is find something you like, translate it to D&D, and adjust its power level using existing monsters as a baseline.
It helps if you already know how to strip something down to its constituent parts, as discussed in an earlier post. You don’t need to steal every single bit of a boss encounter to have something interesting. Look at Bowser from Super Mario Brothers. He jumps, spits fire, and throws hammers, he’s weak to fire, and he can be killed by destroying the platform on which he stands. That’s five unrelated traits. For your purposes, you might decide you only care about the hammers and the jumping. That is, you have a boss who’s constantly on the move, hopping from place to place in his arena and pelting the party with weapons. It’s wildly different from a normal “get into base-to-base with the enemy, smack it around, and heal when necessary” fight, and it frustrates the players into trying a new strategy but not so much that it’s actually frustrating.
You can do this with pretty much any game that has bosses and with most bosses in those games. For example, consider The Legend of Zelda: a Link to the Past, a game one of my DMs references frequently in his campaigns. We can take parts of each boss and turn it into something unique and challenging for D&D:
Armos Knights: Six mundane enemies, fairly boring except for their number. They may even be weaker than other enemies the party fought recently. But once five of them die, the last living enemy suddenly becomes much stronger. Just when the party thinks they’re at the end of the fight, they have to reassess their situation and adapt to a completely different encounter.
Lanmolas: Enemies who spent part of their time fighting the party and the other part hiding (buried underground, popping in and out of walls, jumping between planes, etc.) The fight is longer because the players can’t really focus-fire one enemy before dealing with the others. The most common player strategy I’ve seen in encounters like this is to have everybody ready an action to obliterate one creature the instant it appears, so it may help to have something like a turret or terrain effect constantly harassing the party even when the enemies aren’t accessible.
Moldorm: Most obviously, this boss tries to knock the party out of the arena, dealing falling damage and forcing delays as characters work their way back up. But often a single boss can lead to multiple interesting powers or traits. Here, the boss is also constantly moving, so it could do something to players as it goes. It could trample, or its abilities could change depending on how far it moved that turn, or it could punish attacks of opportunity. The challenge then is restraining the boss or forcing it into a particular place where the party has an advantage. And feel free to take a boss’ actual statistics if that inspires you. For example, the moldorm is immune to piercing damage.
Agahnim: Some abilities seem neat on the surface, but don’t work as well in-game. This boss’ signature trait is that he can only be damaged by reflecting his attacks back at him. In D&D, that may mean readying actions to parry the attack by meeting a set AC or beating the boss’ roll. If so, it means the entire party is waiting around hoping the character with the highest attack roll gets targeted while everybody else does nothing. Instead, look deeper at the boss and see what else it does. Agahnim also teleports and casts three different spells with different hit boxes and behavior, so fighting him might be more a battle of Spellcraft checks and battlefield positioning than simply playing tennis.
The game has nine other bosses, but I don’t think we need to go over them exhaustively here. If anybody’s interested in seeing a game taken apart like this, it can be a separate post.
The secret here is that the players don’t actually care where your boss ideas come from. They want to have fun and most of them want a challenge, and you’re providing that. The sort of player who dismisses a boss because it’s a simple rehash of something he’s seen before in a different medium is not the sort of player who contributes to the game. In my experience you’re less likely to find somebody who says “ugh, this is just Yunalesca” than somebody who says “oh, this is Yunalesca! Nice!”*. And given how many video games there are in the world, you may even have players who guess the source incorrectly. It doesn’t matter. The point is that they get an interesting encounter, even if it’s not totally original, and that’s just fine.
Obviously you can take boss ideas from anything. I’m talking about video games here because that’s my area of expertise and because each one tends to have more boss encounters than the average movie or book. Even an simple, straightforward game like Doom has wildly different boss fights at the end of each chapter and several difficult set pieces you might consider mini-bosses. And even something objectively bad like that Mortal Kombat RPG/adventure thing had creative applications of fighting-game mechanics. It’s not the source that matters, it’s what you do with the ideas you get from that source. Anything and everything is an opportunity for a fun encounter; all you have to do is balance it.
And how do you balance it? Well, that’s harder.
* — They may be more likely to say “oh, this is Yunalesca! Nooooooo!”. Either way, they’re invested emotionally.