The usual advice on how to start worldbuilding for a novel is to build the world: a map, a history, languages, a pantheon, a calendar, then write the book that happens inside it. I think that is the wrong order for most novelists, and I say that as someone who wrote a fantasy novel, Medalion, and later built Lokee, a tool whose whole purpose is keeping a story's world straight. The world that a reader meets is the part a story needs. The rest is pleasant to make and expensive to maintain.
The order that works, in my experience, is the reverse: start from the story's questions, decide the handful of things the plot cannot happen without, and leave the rest fuzzy until a scene asks for it. This guide walks through those decisions, gives you a first-session worksheet with concrete prompts, and shows what to write down and what to leave open. Most of it works with a notebook. The last part shows how the same decisions map onto a Lokee vault, if you want the record to stay checkable as it grows.
The running example is The Ashen Circle, the demo story you can open in Lokee: a rebel network in the walled city of Duskhollow, with Rowan Ashe, his sister Mira Voss, Lady Iseult Varnhold and Magistrate Ophelia Crane. Beside it I'll use an invented setting, Thornwater, to show the same decisions made from nothing.
Why an encyclopedia is the wrong first move
A world has no edges. You can always add another province, another century, another god. A story has a definite shape: a protagonist, a want, an obstacle and a sequence of scenes. If you build the world first, nothing tells you when to stop, so you stop when you are tired, usually with a long document and no first chapter.
Facts decided before the plot exists are also facts the plot must now respect. Decide in week one that Duskhollow has a standing army of ten thousand, and in week six, when the story needs a city that cannot stop a rebellion, you either rewrite the number or bend the plot around it.
The picture to keep in mind is an iceberg. What the reader sees is a few scenes, a handful of names and the rules that are actually in play. What you know sits below the waterline and surfaces only when a scene needs it, and the submerged part is allowed to be unfinished. A reader never sees the parts you have not decided; they only see contradictions between parts you have.
How to start worldbuilding for a novel: from the story's questions
Before deciding anything about the world, write down the questions the story will ask of it. Not themes. Questions, in the form a scene would ask them.
For The Ashen Circle the first few look like this:
- How does a scout like Rowan get across the city at night without being seen?
- Why can the Watch not simply arrest everyone in the lower city?
- What does the Varnhold household have that a rebellion needs?
- What happens to Mira's son Finn if she is caught?
Each question points at a piece of the world. The first wants rooftops, a curfew and a patrol pattern. The second wants a reason the crown's power is limited. The third wants grain, money and a forged signature. The fourth wants a law about what happens to the families of traitors, probably the most important worldbuilding fact in the book, because it is what makes Mira's choice cost something.
Notice what did not come up: the founding of the kingdom, the name of the river god, the currency. None of it is needed to write the scene in the Gilded Lantern.
Do the same for your book; ten questions is plenty. If you cannot think of ten, write the first scene anyway and the questions will arrive.
The five decisions that pay off first
Of everything a first session could decide, five decisions pay for themselves fastest. They are the ones a first scene needs and the ones that are most painful to change after chapter ten.
Setting scale: a room, a city, a continent
Decide how much ground the story covers. One city, with its districts, is enough for an entire novel, and it is far easier to keep consistent than a continent. The Ashen Circle is Duskhollow: a walled capital with a lower city (Tarnside), a manor, a watch house by the East Gate and tunnels beneath the old quarter. Everything happens within a night's walk.
The scale decides what you will have to track. A city needs districts, a few named buildings and the routes between them. A continent needs travel times, borders and who holds what, the hardest kind of fact to keep straight, as the borders and rulers guide shows. Start smaller than feels generous; you can add the continent when a character leaves the city.
For Thornwater, the invented example: one fishing town of perhaps four thousand behind a sea-wall, open sea on one side and salt marsh on the other. No other nations yet.
What is different from our world, and what it costs
A fantasy or science-fiction world usually has one or two things our world does not. Name them precisely and decide what they cost, because a difference with no price is decoration.
In Duskhollow the difference is political rather than magical: the crown is suppressing the old provinces, and an underground network funded by a noble house is resisting. The price is that loyalty cannot be read from rank. A duke's daughter funds the rebels; a watch captain is an old friend of an informant. That one difference generates most of the plot.
In Thornwater, the sea rises a hand's width every year and the town survives only because every citizen owes one year of labour on the wall. The cost is a year of your life, and a town that cannot afford to lose anyone. A protagonist who wants to leave is in conflict with the setting before anyone has said a word.
If your world has a magic system, this is where you decide its one hard limit, not its full ruleset. "Healers can close a wound but cannot restore what was lost" is a limit; a table of spell costs is an encyclopedia entry.
Who holds power, and over whom
A story needs someone who can say no to the protagonist and make it stick. Decide who that is and what their power rests on. Then decide one way it is limited, because unlimited power makes the plot a foregone conclusion.
In The Ashen Circle: the crown, through the Crown Watch and its magistrate, Ophelia Crane. Its limit is information. The Watch can question suspects at the Watch House and double the eastern patrol, but it cannot see inside Varnhold Manor's wine cellar, and it does not know what Rowan has become. The whole book is a race between the magistrate's reports and the Circle's secrecy.
In Thornwater: the Wardens of the Wall, who assign the year of labour and keep the ledger of who has served. Their limit is that the town is small enough that everyone knows who has been excused. Favouritism shows.
What the protagonist wants, and what stands in the way
This is a story decision, but the world is what makes the obstacle concrete. Write one sentence: who wants what, and what stops them. Then ask which world facts the sentence relies on.
Rowan wants to keep his sister and her son safe while serving the Circle. What stands in the way is that the two aims are incompatible: the better he serves, the closer the magistrate gets to Mira's dye shop. The world facts this relies on are the law on traitors' families, the fact that Mira's shop is a known message drop, and the fact that the Watch has started looking at the lower city.
In Thornwater: Elin Varr, a wall clerk, wants to leave the town before her year of service. What stops her is that leaving before service is desertion, and deserters' families serve the missing year in their place. That one law is now the spine of the book.
The one rule the plot bends around
Every story has one world rule that the climax depends on. It is the rule the reader must understand by the midpoint, and the one you must never break by accident. Decide it now and write it as a single sentence with no exceptions you have not also written down.
For The Ashen Circle it is: the Circle's safety depends on nobody in the crown's service knowing a real name. Sera Nightingale is "Kestrel" to nearly everyone; Rowan earned the real name "the hard way". The moment a real name reaches the Watch House, the network is finished.
For Thornwater: the wall's accounts must balance, because the Wardens' authority rests on the ledger being true. Elin's discovery that it does not balance is the plot.
Once you have this rule, you know what to protect. A story bible exists mostly to keep it, and the five or six facts that support it, from drifting while you draft.
The first session: a worksheet you can finish in an evening
Here is the exercise I would do before writing a word of chapter one. It takes one sitting. Write answers by hand or in a plain document; the point is to decide, not to format.
1. Ten questions. Write ten questions a scene in your book will ask of the world. Phrase them as a scene would: "How does X get into Y?", "Why can't Z just…?"
2. Scale in one line. "The story happens in: ___." A building, a town, a city, a region. Add the number of named places you expect to need in the first act. Keep it under eight.
3. The difference and its price. "In this world, unlike ours, ___. The cost of that is ___." One or two of these at most.
4. Power in three lines. Who holds it. What it rests on (law, force, money, information, belief). One way it is limited.
5. The sentence. "___ wants ___, but ___." Then list the two or three world facts that sentence needs to be true.
6. The rule. One sentence, present tense, no exceptions. Then one line: what happens at the climax if the rule holds, and what happens if it breaks.
7. Six names. The protagonist, the person who can say no to them, one ally, one place where the first scene happens, one place the protagonist cannot easily go, and one group or institution. Six is enough for the first three chapters.
8. The not-yet list. Everything you were tempted to decide tonight and did not. Write it down and leave it, so you stop worrying that you will forget it.
Here is the sheet filled in for Thornwater, compressed:
Scale: one town behind a sea-wall, four thousand people, the marsh behind and the sea in front. Places: the Wall, the Wardens' Hall, the salt-houses, Elin's family's net-loft, the marsh causeway (the only road out). Difference: the sea rises a hand a year; the wall is kept by one year of labour from everyone. Cost: a year of your life, and a town that cannot spare anyone. Power: the Wardens, who assign service and keep the ledger; limited because the town is small and favouritism shows. Sentence: Elin, a wall clerk, wants to leave before her year, but deserters' families serve in their place. Rule: the ledger must balance, because the Wardens' authority rests on it. Not yet: who lives beyond the marsh, religion, what the town traded before the sea rose, whether the rising can be stopped.
That is a complete starting world in under two hundred words. It is enough to draft three chapters without stopping, which is the only test that matters.
What to write down and what to leave fuzzy
The worksheet gives you the minimum. What else deserves a line in the bible? My rule: write down anything the reader could catch you contradicting; leave open anything only you would notice.
| Write it down now | Leave it fuzzy for now |
|---|---|
| Names and spellings of people and places that appear in act one | The history of anything older than living memory |
| Who is related to whom, and who knows whose secret | Second cousins, dynasties, succession rules (until a saga needs them) |
| The rule the plot bends around, and its stated exceptions | The full magic or technology system |
| Which places are inside which, and roughly how far apart | Exact distances, travel times, a drawn map |
| The one law or custom that makes the protagonist's choice costly | The whole legal code |
| What the difference from our world costs | Where the difference came from |
| Ages of the few characters whose ages matter to the plot | Everyone else's birth date |
Fuzzy is not the same as careless. A fuzzy fact is one you have deliberately not decided, and the discipline is to write "not decided" rather than making something up in passing and forgetting you did. The invented-in-passing facts are the ones that come back in chapter eighteen.
One exception: if your plot depends on time, decide the calendar before you date anything. Not the month names, only the units. The demo's Ashen Calendar is nothing more than Era, Year and Day, and that is enough to say Rowan was born in Era 2, Year 44, Day 22 and know he is older than Iseult. The calendar guide explains why the units matter more than the names.
Mistakes that cost the most later
These are the starting mistakes that cost the most to undo, and what to do instead.
Building what you enjoy instead of what the story needs. Three evenings on heraldry, none on the law about traitors' families. Fix: keep the ten questions pinned where you can see them, and only build toward one of them.
Power without a limit. A ruler who can simply order the problem solved. Fix: write the limit on the same line as the power. If you cannot find one, the antagonist is too strong for the story you have.
A difference with no cost. Magic that is free, a technology with no downside. Fix: decide the price before the first scene uses the difference, because the price is what creates choices.
Naming everything. Every guard, every tavern, every street. Fix: name what recurs. A one-scene guard is "the guard"; when he comes back in chapter nine, name him then.
Inventing in passing. You give Mira a husband in a stray line of dialogue and never record it. Fix: whenever a sentence states a world fact for the first time, mark it, so it can be recorded at the end of the session. The companion guide on worldbuilding while you write turns that into a weekly habit.
How this maps onto a Lokee vault
Everything above works in a notebook. Here is what it looks like in Lokee, if you want the decisions to be checkable later rather than merely written down. A new vault needs nothing but a name, and I would set it up in this order, which is deliberately short.
1. Two or three entity types. Under Settings → Structure → Entity types, add Character, Location and Faction, each with its own colour, as the demo does. Add an Object type when an object matters to the plot, not before. The story bible guide covers types and fields in depth; for a first session, this is enough.
2. One or two fields. Use + Add field on Character to add a Place field called Location, so the vault knows where people are. If ages matter, define the calendar first under Settings → Structure → Calendar systems with its units in order, then add a Story date field such as Born. Skip everything else.
3. The six names from the worksheet. Create each one with Create entity (or right-click an empty spot on the Graph and choose Add entity). Give each a short Description of one or two lines, the way the demo describes Rowan as "a former city-guard scout turned informant for the Ashen Circle". Everything decided but not yet worth a field goes there.
4. The relationships you will ask about. Rowan Sibling Of Mira, Mira Parent Of Finn, Rowan Member Of the Ashen Circle. The demo's six relationship types (Knows, Sibling Of, Parent Of, Member Of, Allies With, Rival Of) are a sensible starting set.
5. The rule, where you will see it. Lokee has no "rules" object; a rule is prose. Put it in a note under Notes with + New note, title it with the rule itself, and link it with + Link to… to the entities it governs, so it shows up on their pages. The not-yet list goes in a second note.
6. One map or one timeline, not both. If the story is about movement or territory, start a map with + New map and draw the five places as areas; the atlas guide covers the tools. If the story is about sequence, open Timeline and under World chronology add the three or four points you already know with Add point. Order is what the Timeline trusts; dates are optional, as the three kinds of when guide explains.
That is a vault of perhaps ten entities, two notes and one map or timeline. It is small on purpose. The guide to organising worldbuilding notes is about what to do when it is no longer small, and the module-by-module tour is for deciding whether you want the rest at all.
What Lokee does not do here. It will not tell you what your story is about, which of the five decisions you made badly, or whether the rule you chose is interesting. It also does not stop you building the encyclopedia; an empty vault accepts a thousand entities as happily as ten. The discipline of starting small is yours. What the vault gives back is that every decision you record stays findable, and that when you change one later, you can record the change instead of overwriting it.
In five minutes, if you are starting today
- Write the sentence: who wants what, and what stands in the way.
- Write the one rule the plot bends around, with no exceptions.
- Name six things: protagonist, opponent, ally, first-scene place, forbidden place, one institution.
- Decide the scale in one line, smaller than feels generous.
- Start the not-yet list and put everything else on it.
Then write the first scene. The world will ask its next question before the scene is over.
Where to go next
The companion guide, worldbuilding while you write, takes over from here: answering the world's questions as scenes raise them, avoiding the research rabbit hole, and revising a rule when the plot changes. The help pages on vault setup and entities and relationships list every button the steps above use. If you would rather explore a finished small world first, the demo vault at /demo holds The Ashen Circle with its three chapters, seventeen entities and a city map.