How to Build a Campaign Bible: Running a Tabletop World You Can Actually Hold in Your Head

Your prep notes are not a campaign bible

Almost every GM already has a pile of documents, and almost none of them constitute a bible. The pile is prep: session outlines, encounter tables, a dungeon map, a page of names to grab when someone talks to a stranger. Prep is forward-looking and disposable. It describes a version of next Thursday that will not survive contact with the party.

A campaign bible looks backwards and does not expire. It records what turned out to be true — not what you planned. The priest in the river town is only in the bible because the party met him and he mattered; the eleven other NPCs you statted for that session and never used are not. This is why a bible built by filing your old prep never works. Prep is mostly about things that did not happen.

The practical test is a question: if you deleted this document, would anything you said at the table become unverifiable? If yes, it is bible content. If no, it is prep, and it should be allowed to die quietly after the session it was written for.

Keeping these separate is what stops the record from bloating. The most common reason GMs abandon their campaign bible is that it grew to eighty pages of mixed prep and canon and became slower to search than their own memory — at which point it has negative value, because maintaining it costs time and consulting it costs more.

The related discipline of turning what happens into what is written down is covered in session prep for game masters; this pillar is about the record that prep should be feeding.

What actually belongs in it

A campaign bible holds constraints, not colour. A constraint is any fact that limits what can happen next — and those are the only facts you can contradict. You will never break a campaign by misremembering what a city smelled like. You will break it by misremembering who rules the city.

Five categories carry almost all the weight. **Named NPCs**: who they are, what they want, their current status — alive, dead, ally, hostile, missing — and who they are connected to. Status is the field people forget, and it is the one that causes the most damage, because a dead NPC turning up is the most memorable kind of continuity break there is.

**Locations**, recorded for what they constrain rather than how they look: who controls it, how you get in, what is forbidden there, what the party has already done to it. A location the party burned down is a different location, and that fact belongs in the record before it belongs in your memory.

**Factions**, with their goal, their current standing with the party, and their standing with each other. Faction state is the fastest-changing thing in most campaigns and the least often written down, which is why the question "wait, are we still on good terms with the Guild?" so often gets a shrug. Managing this properly is its own topic — see managing NPCs and factions.

**The setting's hard rules**: what magic can and cannot do, what the gods intervene in, what technology exists, what happens when you die. These are the ones players stress-test hardest, because a rule is an invitation to find its edge. If it is not written down, you will rule inconsistently on it, and a table will notice inconsistency in a rule far faster than inconsistency in a person.

**A dated timeline** of events that actually occurred at the table. Not a history of the setting — a log of the campaign. Session eleven: the bridge burns. Session fourteen: the envoy arrives, three weeks later in-world. In-world dates are what let you answer "how long ago was that?" without guessing, and guessing at elapsed time is how travel, healing, and pregnancy-length continuity errors get in. The general technique is covered in building a story timeline.

What to deliberately leave out

The instinct with a world record is to be thorough, and it is the wrong instinct. Every entry you add is an entry you have to keep current, and a bible with stale entries is worse than one with fewer entries — because you will trust it, and it will be wrong.

Leave out anything you are happy to reinvent. Minor NPCs the party spoke to once and will never see again do not need to persist; if they come back, you can invent them fresh and nobody will know. Leave out description that does not constrain: a paragraph on the architecture of the capital is worldbuilding for your own pleasure, which is legitimate, but it is not bible content and mixing it in makes the bible slower to use.

Leave out anything that has not happened yet. The plot you intend to run is prep, however confident you are. The moment it lands in the bible you start defending a future that the party has every right to derail, and GMs who write their intended plot into their canon record are measurably more likely to railroad — because now it is written down and contradicting it feels like an error rather than a Tuesday.

The exception worth making is for secrets, and it is important enough to be its own practice. What the villain is actually doing, the twist about the party's patron, the reason the gate really opens — these are true, they constrain everything, and they must never leak. They belong in the record, marked as hidden. In CanonBoard that is what the Locked Vault is for: entries the engine keeps out of conflict checks and the crowd never sees. It is a Core-tier feature; the underlying discipline — keep secrets in the canon, marked, rather than outside it — costs nothing and works with a folder and a naming convention.

This is the difference between a homebrew setting that stays runnable and one that becomes a second job. Building a homebrew campaign setting without drowning in lore covers the design side of the same restraint.

Your players are writing canon whether you like it or not

This is the one thing that makes a campaign bible structurally different from a story bible, and most GM advice underrates it. In a novel, canon is authored. At a table, canon is negotiated in real time by people who are not consulting your notes.

It happens constantly and mostly invisibly. A player says "my character would know this district — she grew up here," and you say yes, and a fact about your city now exists that you did not write. A player misremembers something you said three sessions ago, states it confidently, and the rest of the table accepts it. If you do not correct it in the moment — and often you should not, because stopping the game to litigate a detail costs more than the detail is worth — it is canon now. It is in four people's memories. Yours is the only version that disagrees.

The functional response is to treat the table as a co-author with commit rights, and to reconcile after rather than gatekeep during. Capture what got established, including the things you did not plan to establish, and write them into the record the same way you would write your own inventions. A campaign bible that only contains the GM's contributions is not a record of the campaign; it is a record of half of it, and it will disagree with the players' half at the worst possible moment.

There is a real craft to accepting player-invented canon without losing the shape of your world, and it is covered in player-driven canon and improvisation. The record-keeping consequence is simply this: the bible must be easy enough to edit that updating it after a session is a ten-minute job. If it is hard to edit, you will not do it, and the version in your players' heads will drift permanently away from the version in your notes.

The ten minutes that keep it alive

Every campaign bible fails the same way: not at creation, but at maintenance. The initial build is fun — it is worldbuilding, and worldbuilding is the part GMs like. Updating it after session twenty-three, at eleven at night, when the game just ended and everyone is tired, is not fun. That is the moment the system either holds or does not.

The only approach that reliably survives is to split capture from filing. During the session, keep one rough running list. Names you invented on the spot, promises your NPCs made, facts a player established, anything that changed status. Do not organize it — organizing while running is how you end up doing neither well. One line each, ugly, fast.

Then, before the next session rather than immediately after, spend ten minutes moving that list into the bible properly. Doing it later rather than at midnight matters more than it sounds: you will remember the important things and forget the trivial ones, which is exactly the filter you want. Perfect recall would have you writing down too much.

Ten minutes a session is about two hours a year for a fortnightly game. That is the entire cost. Set against it is the thing every long-running GM eventually values more than any other: arriving at session forty and still being able to introduce a character from session six by name, with their history intact, because it is written down. Players notice that. It is most of what makes a long campaign feel like a world rather than a series of rooms.

The failure modes of long campaigns specifically — and how they compound — are covered in keeping a long campaign consistent.

Structure it so you can find things mid-session

A campaign bible is consulted under time pressure, in front of an audience, while you are also running a game. That single constraint should drive its structure more than any organizational theory.

The practical requirement is that any entry is reachable in under fifteen seconds. That rules out long prose documents, however well written — scanning three pages to find a name while your table waits is exactly the friction that makes GMs stop consulting the record and start improvising over it. It favours short, discrete, named entries: one NPC per entry, one location per entry, searchable by name.

It also favours explicit connections over implied ones. The question you actually ask mid-session is rarely "what do I know about this person" — it is "who is this person to the party, and who else does that drag in." A record where the priest's entry names the faction he serves, the town he is in, and the party member he owes a favour to answers that in one lookup. A record where those facts live in four separate documents does not.

This is why a spatial, connected canvas works better for campaign material than a document tree: characters, locations, factions and rules as typed cards you can lay out the way the world actually fits together, with labelled relationship lines between them, so pulling on one node shows you everything it touches. It is the same argument made at length in why a world bible belongs on a canvas, and it applies with more force to tabletop than to prose, because prose writers consult their bible at leisure and GMs consult theirs with four people watching.

Whatever form you choose, the test is the same: can you answer a player's question about session nine without breaking the flow of session forty? If not, the structure is wrong, no matter how complete the contents are.

Finding the contradictions before your players do

A campaign bible does not prevent contradictions. It makes them findable, which is a different and more achievable goal. The contradictions that damage a campaign are almost never the ones you noticed — they are the ones that sat in the record for four months, unread, until a player built a plan around the wrong half.

Most of them fall into a few recognisable shapes. Status contradictions: someone recorded as dead appearing, or an ally who was never formally moved to hostile after they betrayed the party. Rule contradictions: a ruling you made in session eight that you unknowingly reversed in session twenty-two, which players track far more carefully than GMs do. Timeline contradictions: two things placed in the same in-world week that cannot both have happened, usually because travel time was never counted. Location contradictions: a place described one way early and another way later, most often after the party changed it and the description never got updated.

Reading for these by hand works, and it is worth doing before a big arc — but it scales badly, because the number of pairs of facts that could contradict grows much faster than the number of facts. This is the specific job CanonBoard's Conflict Detection does: it reads the whole canvas and surfaces contradictions in plain English, with an explanation of what conflicts with what. It is a Core-and-above feature and costs one credit per scan, and it reads the cards on your canvas — never anything you have written in the Sandbox, which stays yours until you decide it is canon.

The general taxonomy of how worlds break, and which failures are catchable at all, is set out in common continuity errors.

Starting one twenty sessions in

Most GMs who need a campaign bible do not have one and are well past the point where starting would have been easy. The advice to start early is true and useless to them. So: how do you build one for a campaign already in progress, without spending a month on archaeology?

Do not reconstruct. The instinct is to go back through every session and recover everything, and it is the reason most retroactive bibles are abandoned in week two. The history is large, mostly irrelevant, and painful to extract.

Instead, build forward from the present state. Write down what is true *now*: who is alive, who is hostile, where the party is, what they owe and are owed, which factions are in play and where they stand, and the rules you have actually ruled on. That is a couple of hours and it is immediately useful, because current state is what you consult ninety percent of the time.

Then let history accrete on demand. The next time a player asks about something from session nine and you have to dig, write down what you find. Within a few months you will have recovered exactly the parts of the past that turned out to matter, and none of the parts that did not — which is a better bible than the exhaustive one you would have built, and it cost you nothing you were not already spending.

If your existing notes are substantial, the recovery can be faster: CanonBoard's Smart Import reads a document you already have and builds the cards out of it — characters, locations, rules, lore, a starting timeline — so a long-running campaign's notes become a structured record without retyping. It is a Core-and-above feature and costs a small number of credits depending on length. It organizes what you wrote; it does not invent anything you did not.

What a campaign bible actually buys you

The obvious benefit is consistency, and it is real but oversold. Perfectly consistent campaigns are not automatically good ones, and tables forgive a surprising amount of drift when the game is fun.

The benefit that actually changes how you run is ambition. A GM who trusts their record introduces more: more named NPCs, because they will not evaporate; more factions, because their standing will still be tracked in six months; more consequences, because consequences require remembering what was done. A GM without a record learns, usually without noticing, to keep the world small enough to hold in their head. The campaign does not fail. It just becomes less than it was going to be.

The second benefit is that the world outlives the campaign. A setting with a real record can be run again with a different party, handed to another GM, expanded into something you write rather than play, or opened up to more people than fit around a table. That last one is worth saying plainly: a world with a solid canon behind it is a world other people can contribute to without wrecking it, which is the whole premise of opening a project to an audience.

And the third is the least glamorous and probably the truest: it makes prep faster. Most session prep time is not invention — it is re-establishing what is already true so you can build on it. A current bible turns that from twenty minutes of scrolling into one lookup, and gives the time back to the part of the job you actually enjoy.

Start with the five categories. Keep it to constraints. Spend the ten minutes after each session. Everything else is refinement, and none of it matters if the record is not current.

Frequently asked questions

What is a campaign bible?

A campaign bible is the single authoritative record of everything true in your tabletop world: the people, the places, the factions and their standing, the rules the setting runs on, and the dated events that have actually happened at the table. It is not your prep notes and it is not your setting document. Prep is disposable — it describes what you hope will happen next session. A campaign bible describes what is true regardless of what happens next, and it is the thing you check when a player says "but you told us the gate only opens at low tide." The distinction matters because the two have opposite lifespans: prep is worthless a week later, and the bible is worth more every month you keep it.

What should go in a campaign bible?

Only what you would be embarrassed to contradict. In practice that is five things: named NPCs with their status and allegiance; locations with the details that constrain what can happen there; factions with what they want and who they currently hate; the setting's hard rules — how magic works, what the gods can and cannot do, what technology exists; and a dated timeline of events the party witnessed or caused. Everything else is flavour and can be reinvented freely. The most common mistake is writing down atmosphere instead of constraints. Nobody has ever run into trouble because they forgot how a tavern smelled.

How do you keep a campaign bible up to date during play?

By capturing at the table and organizing after, never both at once. Running a game and maintaining a record are different mental modes, and trying to do them simultaneously means doing both badly. Keep a rough running log during the session — names invented, promises made, facts a player established that you accepted — then spend ten minutes afterwards moving those into the bible properly. Ten minutes after each session is roughly two hours a year, and it is the difference between a world you can run from memory in session forty and one you have to reconstruct from Discord scrollback.

Is a campaign bible the same as a story bible?

Structurally, almost. Both are the authoritative record of a fictional world, and the categories are near-identical — characters, locations, lore, rules, timeline. The difference is who writes it. A [story bible](/blog/what-is-a-story-bible) is authored: one person or one room decides what is true. A campaign bible is co-authored under pressure, because your players invent canon at the table faster than you can approve it, and a good campaign absorbs that rather than overruling it. That single difference changes how you build it — you need a record that expects to be edited by events, not one that assumes it was right the first time.

Do you need a campaign bible for a short campaign?

For a one-shot or a three-session arc, no — you can hold it in your head, and building infrastructure for it is wasted effort. The threshold is roughly the point where you start forgetting names you invented, which for most tables is somewhere between session six and session ten. The useful rule is that you should start the bible one session before you need it, because the cost of building it grows with the size of the backlog. Ten sessions in, it is an afternoon of reconstruction. Forty sessions in, most GMs simply never do it.