Let's suppose this was an eight-week campaign, and I was planning to devote the whole session to a climactic battle against this guy, and the first thing they did was they used this instakill on him. What do I do next?
There's a possible “next” there, but if you've come to this point you've already broken a cardinal rule and are naturally suffering the consequences. So the thing to really do is to back up, remember the GM's Agenda, and follow that agenda instead of following old D&D-conditioned DMing habits.
Specifically, with emphasis mine, your Agenda is expected by the game to be:
Your agenda makes up the things you aim to do at all times while GMing a game of Dungeon World:
- Portray a fantastic world
- Fill the characters’ lives with adventure
- Play to find out what happens
Everything you say and do at the table (and away from the table, too) exists to accomplish these three goals and no others. Things that aren’t on this list aren’t your goals.
By creating a climactic battle and planning for it to take up the session, you're departing from Dungeon World's rules framework and entering uncharted territory where all the rest of the game's rules will start falling apart or having undesirable results when followed (such as the Barbarian's Smash! move).
(You can tinker with Dungeon World's rules, but it's surprisingly easy to get a really unsatisfying game by doing so without first having a deep understanding of how they all interlock. Accidentally not following them is an even better way to discover ways it can work unsatisfyingly.)
Play to find out what happens
Dungeon World demands improvisation. It doesn't say that and then leave you hanging though: every single rule in the GMing chapter is part of a framework for that improvisation, so that you can improvise without ever running the risk of having no idea of how to improvise next when its your turn to do so.
To do this though, DW demands that GM let go of planning. A GM following the game's rules is literally not allowed to plan ahead. They may create possible situations, but never outcomes; brainstorm ideas, but never force them to happen; brainstorm new and interesting ways to use GM moves, but never use them when unsuited to the growing fiction.
You are even allowed to prep a whole situation… but you're explicitly not allowed to use it as a script! You are instead instructed to exploit your prep and mine it, dismember it, steal from it the pieces that end up fitting the game that unfolds at the table.
Improvising means being reactive (like a player!)
A Dungeon World GM is almost entirely reactive. This takes some time to get used to, and often takes some unlearning of GMing habits from other games.
For example: many of the examples of how you might answer this question about D&D are against the GM's rules in DW:
This is why you should never offer the players a single boss monster. Use two or three smaller ones. (But sometimes the plot requires a single creature.)
Yeah, sure! Sometimes you might be better off with a few different baddies than one singular one. But sometimes the fiction demands a single Big Bad, and it's a rule: Begin and end with the fiction.
Introduce some sort of other monster, like the boss monster's henchmen. Power them up until they're a worthy opponent for a boss fight. (Not awful, but this robs the players of their victory.)
Not allowed! The only time the GM can do this is when a PC gets a 6− or when they give you a Golden Opportunity.
Introduce some sort of environmental threat, like the portal to the chaos realms has gone unstable, or the dungeon is collapsing and the players have to get out. Let that be the final encounter. (What if I don't have one on hand?)
Also not allowed, for the same reason.
Improvise a GM Intrustion: politely ask the players to agree that the boss monster survives the attack, so that the final battle can be more satisfying. (Lame!)
Indeed, this is unfun! It's also not allowed, because it violates the GM's Principles to Address the characters, not the players, Be a fan of the characters, Begin and end with the fiction, and Think dangerous.
“But no, really, what should I do when that happens?”
Suppose it's the climactic fight, you've just started the session, and the Barbarian up and decapitates them just like that. Let's suppose that you've followed the rules and only prepped the potential situation, not planned it out.
What you do next is describe the world and say how the head comes off and lands at their feet / is raised up in the Barbarian's fist / whatever suits the details established already. Do that, then ask the question “What do you do?”
Then the players will tell you how the session continues. Respond with GM moves if and when the rules call for them, notice PC moves triggering and follow them, and keep playing to find out what happens. Let your players surprise you…
But keep thinking dangerously. Keep thinking offscreen too. Just because the Big Bad is dead doesn't mean the session or the campaign is done quite yet; it definitely doesn't mean they're out of danger and excitement. Play it out, find out together what happens based on what they do next, and know that a missed roll or a Golden Opportunity still lets you make a GM move. The game is still on!
Yes, the players might say “Uh, we go home now” when asked what they do, but just like at every other moment in the game they need to explain how they do that, so you should respond with “okay, how?” A player saying what their goal is never enough for their PC to accomplish it! They're still in the game, and have to say what immediately-possible action they do next, not just what their next goal is. And as soon as they start saying how, you're back into the fiction, into the action, and back to the possibility of triggering moves.
But more likely than not their answer won't be “Uh, we go home now” anyway. More likely it will be something like searching for the Big Bad's treasure vault (is there one? play to find out!), burning the place down, attempting to recruit the newly-jobless minions, taking over the base, turning the head into a grotesque trophy, or any of a thousand different things players pumped by a major victory tend to want to do. Going home might be their answer, but it's not the only or even most likely answer. So ask!
TL;DR: When the big boss is killed instantly is not a trigger for any GM move, so Dungeon World doesn't let you do anything to fix or mitigate it. Don't try to.
A sudden and early end to a big fight is not a real problem for the game engine, and doesn't need solving. The game's normal rules will continue to make the rest of the session interesting — possibly in unexpected ways that are more interesting than a the big fight would have been! (Who knew the Big Bad had a planar portal in their private sanctum? These things can just kinda… happen… when player moves and GM moves interact.*) Assuming that a quick end to the Big Bad is a problem that needs to be avoided or solved is a habit carried over from other games that can only interfere with letting the Dungeon World game engine work properly.
When the engine is allowed to run properly, one of Dungeon World's greatest strengths shows up organically: every unexpected event leads to things that are more timely and interesting than what most GMs could have planned for. Instead of panicking and assuming the session is ruined, just calmly keep following your Agenda and Principles, and (when allowed) keep making GM moves. The game will keep going, and what comes after the Big End Fight might even end up more important than the Big Bad.
* As an example, I once invented a bottomless pit caged by a carved stone screens, on the spot as the PCs entered a room. A bunch of player moves and GM moves later, the druid was thrown at it, turned into an elephant mid-air, and awoke a primordial being dwelling in the elemental darkness the pit lead to. (All of those stages involving player and GM moves, each one's consequences improvised on the spot, per the rules.) This ended up introducing what turned out to be in hindsight the main storyline of that entire game, and none of it was planned by me. They ended the game by raising a new demigod in the Forgotten Realms, which was much more epic than anything I would have planned for in a different game system. Bringing it back to the point: it was way more interesting than the fight that caused the druid to been thrown in the first place.