Worldbuilding should usually come last, after the characters and the central conflict have shown you which parts of the world actually need to exist. The writers I trust most build out their settings in the wake of their stories rather than ahead of them, because a setting that exists to serve a specific scene tends to be more alive than one that exists in a binder.
This is one of those craft arguments where everything I am about to say has been, and has been, successfully contradicted by working writers, and where the right answer for any particular book depends on the particular writer, so take all of this with the understanding that there are no universal rules. I am not telling you to throw out your worldbuilding documents. I am telling you, in my view, that they should usually be written second, in the wake of the story, rather than first, in advance of it. That is a different claim, and I think it is the one most aspiring fantasy and web fiction writers need to hear.
Why worldbuilding-first traps so many fantasy writers
Worldbuilding-first feels productive because it produces concrete output without forcing you to commit to a scene. You finish an entry on the seven cardinal directions of the elven priesthood and you can mark it as done, and the document gets longer in a way the novel never does. The trouble is that none of it is load-bearing yet. A worldbuilding document is, almost by definition, a list of answers to questions no character has asked, and once you have a hundred of them, the book has to either honor all of them (which makes the prose feel like an audit) or quietly ignore most of them (in which case the time spent was sunk). I have watched writers I respect spend three years on what they called the bible and another four months on what turned out to be the first three chapters, and the chapters were the part that taught them what the book actually wanted to be about. The worldbuilding, in retrospect, was a very elaborate form of not writing.
There is also a quieter trap, which is that worldbuilding-first tends to make the world feel airtight in a way that is bad for fiction. Lord of the Rings feels infinite partly because Tolkien wrote himself into corners and left them visible, the way an old city has streets that don't quite line up because someone built the wrong wall in 1380. A world that was assembled by committee in a Google Doc before chapter one tends to feel like an airport: serviceable, internally consistent, and totally without history. My feeling about this is that history is what we mean when we say a setting feels real, and history, on the page, comes from the writer not knowing where she is going and improvising.
What characters and conflict tell you about the world you actually need
Start instead with the people, and the world you actually need will reveal itself. If your protagonist is a third son of a minor noble, you suddenly need inheritance law, and you need exactly as much of it as the plot will touch and not one footnote more. If she is a magistrate in a port city, you need the shipping economy and the gang that runs the docks, but you do not need to know the foreign policy of the empire on the other side of the continent unless a letter from there changes the plot. The clean rule, in my view, is that worldbuilding should be downstream of the scene that needs it. Then you build out, because then you know what is load-bearing.
This is also how working novelists actually work, even the ones who are famous for their worldbuilding. Robert Jordan filled notebooks with cosmology, but the notebooks came after he knew he was writing about Rand, Mat, and Perrin leaving the Two Rivers. Joe Abercrombie has said in interviews that his settings tend to emerge from the characters' jobs and grievances, not from a prior atlas. Even Ursula K. Le Guin, who built one of the most coherent fantasy worlds in the genre, started with Ged and the goats. The pattern is consistent, and I think it is the pattern most beginners would benefit from copying, because the alternative (atlas first, characters fitted in) almost never produces characters who feel like they actually belong where they live.
The web fiction authors who built their worlds backwards
In web fiction the case is sharper, because the format itself punishes worldbuilding-first. Royal Road readers vote with their thumbs on chapter three, not chapter thirty, and they vote on whether they care about the protagonist's next move. Matt Dinniman has said in podcasts that Dungeon Crawler Carl started as Carl, a cat, and an apocalypse, and the byzantine dungeon AI cosmology that readers now love was discovered in the act of writing, not designed before it. Andrew Rowe's Sufficiently Advanced Magic opens with a single boy entering a single tower, and the larger Arcane Ascension setting unfolded across later books only because the first one earned the room. Travis Bagwell's Awaken Online gives you exactly the worldbuilding you need to follow Jason into the game and then trusts the rest to be filled in as combat reveals it.
The contrast that proves the rule is the writer who shows up on Royal Road with twelve thousand words of glossary, three maps, and a prologue narrated by a god the reader has not yet met. I read those openings every week, and they almost always lose me by chapter two, not because the worldbuilding is bad but because there is no one to care about yet. The lore is sitting there in a vault waiting to be relevant, and the protagonist, when he finally walks on stage, is the least developed entity in his own scene. 'In the third age of the empire, before the breaking of the third seal, in the city of Kor-eth-thaal where the Iron Bishopric ruled by sword and prayer,' and so on. None of that is wrong. It just isn't yet anyone's problem, and a story is, finally, somebody's problem.
When worldbuilding really does need to come first
I want to concede the strongest version of the opposing view. There is a class of stories where some worldbuilding genuinely has to come first, and those are the ones whose plots depend on a hard system that the reader has to be able to game alongside the protagonist. A LitRPG where the leveling math doesn't work out, a hard-magic mystery where the trick the wizard pulls in chapter twenty has to be foreshadowed in chapter two, a time-loop where the loop's rules dictate every decision: in cases like these, you cannot get away with discovery-drafting the system. You need to know, before you write, that mana costs scale quadratically, or that the loop resets at noon, or that a soul-bonded weapon cannot be wielded by anyone else. Mother of Learning by nobody103 is the cleanest example I know of in serial fiction. The loop mechanics are obviously designed up front because the entire structural payoff depends on it.
So here is the qualifier. If your story's central pleasure is the system, build the system first, but build only the system, not the whole world. The empire's coinage and the kingdom's wine regions can still come later. Brandon Sanderson has said many times that he designs the magic before the plot, but he is also a generally honest reporter of his own process and admits that the rest of the cosmere has been retrofitted around the books one by one. Even in the strongest case for worldbuilding-first, what comes first is a single system, not a setting. That is a much smaller commitment than the worldbuilding-first crowd usually defends.
A practical order of operations for web fiction writers
A workable order of operations for most web fiction writers is character sketch first, situation second, two or three weird-world details third, and then chapter one. I would say: write a one-page sketch of your protagonist and what they want; write a one-page sketch of the situation that frustrates the want; write a paragraph about the world's two or three most unusual features, the ones that make this story not possible anywhere else; and then write chapter one. Do not write any more worldbuilding documents until you have finished a draft of the first three chapters, and when you do go back to the document, write only entries about things that have already shown up on the page. 'The Greycoat Watch enforces the harbor curfew between bells nine and four; their captain takes a cut of every smuggler's haul; the protagonist's uncle was hanged by them in 1487.' That is a useful worldbuilding entry, because every fact in it was earned by something already in the manuscript. Compare it to the same entry written cold, before the book exists, where the smuggler cut and the year would be arbitrary, and you would have no way of knowing whether either of them was the interesting part.
This isn't a rule, and it isn't going to be right for everyone. Some writers really do need the atlas to feel anchored, and if that is you, I'm not going to argue you out of your own process. But if you have spent a year on your world and zero months on the actual story, I think the most useful thing you can do for the project, today, is close the bible and write a scene where the protagonist wants something he cannot have. The world will know, immediately, which parts of itself need to exist, and the rest of it will turn out to have been optional all along.
Common questions about worldbuilding order
Should worldbuilding come before or after writing the story?
For most writers, the story should come first and the worldbuilding should follow in service to it. A small starting sketch of the world is fine, even necessary, but the deep work of cosmology, languages, dynasties, and magic-system edge cases tends to pay off far more when you do it after you know which scenes you are actually writing toward. Otherwise you end up with a world that has answers to questions the book will never ask.
How much worldbuilding do you need before you start writing fantasy?
I usually recommend the absolute minimum: the place where the opening scene happens, the rough shape of the conflict, one or two cultural details that matter for the protagonist, and a vague sense of the magic system's costs. Everything beyond that is best discovered while drafting. Andrew Rowe, Will Wight, and Matt Dinniman have all talked publicly about starting with less than readers assume.
Doesn't hard magic require worldbuilding first?
Hard magic is the strongest case for some upfront worldbuilding, because the rules need to be self-consistent before the reader can enjoy them being broken. Even then, my feeling is that you only need the rules that matter for the first arc. Brandon Sanderson tends to publish the impression that his magic systems are fully designed up front, but his own essays often describe filling in the bottom 80% of the rules during the drafting itself.
What is the most common worldbuilding mistake new fantasy writers make?
The most common mistake I see is treating the worldbuilding bible as a substitute for the novel. It feels productive, it feels safe, and it never ends, which is why some writers spend years on it without ever writing chapter one. The fix is uncomfortable but simple: write a real opening scene before you write another lore document, and let the scene tell you which lore matters.
Can a great world save a story with weak characters?
In my experience, no, or at least not for long. A vivid setting can carry a reader through chapter one and maybe chapter five, but if the protagonist is flat or the central conflict is uninteresting the reader will close the tab. The reverse is much more forgiving, though. Strong characters in a generic world tend to read fine, while strong worlds with thin characters tend to feel like museum tours.
by Jacob Tam · May 19, 2026
I run IlorisNovel, a platform for writers of web fiction. If this kind of craft writing is your thing, the rest of the blog lives here.