How to run an RPG campaign (and keep it organised)

Running one great session is a skill. Running twenty sessions that remember each other is a different skill, and it's the one nobody warns you about. A campaign isn't a long session. It's a small world that has to stay consistent while you improvise inside it.

The story usually isn't the problem. The bookkeeping is.

You can run a brilliant single session on charm and a half-page of notes. A campaign quietly raises the stakes, and it does it on a curve. At session three you're fine; nobody forgets anything that recent. By session thirteen it's a rodeo: forty NPCs, three factions with grudges, a debt to a duke, a poisoner nobody caught, and a player who very reasonably asks "wait, didn't we promise the smith something?", and the table goes silent.

That silence isn't a failure of imagination. It's a failure of continuity. The story didn't get worse; the shared memory of it did. Most campaigns that fizzle don't end on a bad session. They end on the slow drift where the world stops feeling solid because nobody can quite remember what's true. And here's the part that stings: only what actually happened at the table counts. The clever subplot you prepped but never surfaced doesn't exist in the campaign. Your players never saw it, so it was never canon, just an hour of your evening that paid out nothing.

A campaign is continuity that happens to be fun. Lose the continuity and you've got a series of unrelated one-shots wearing the same hats.

Start with a session zero

The most useful session of an RPG campaign is the one before it starts, and it's the one most groups skip because they're impatient to roll dice. A session zero is a conversation, not a game. You sit down, build characters together, and agree out loud on the things that quietly sink campaigns when they're left unsaid.

What's actually worth covering:

• The premise. One or two sentences everyone buys into: "frontier mercenaries scraping a living between three squabbling lords." If players can't repeat the premise back to you, the campaign has no centre of gravity yet.
• Tone. Grim and political, or breezy and heroic? A table running a heist comedy and a table running a tragedy will read the exact same scene completely differently, so agree which one you're at.
• Lines and veils. Lines are content the table won't touch at all; veils are content that happens, but off-screen. Naming them isn't fragile. It's what lets people relax and actually lean into the dark parts of the story, because the genuinely off-limits ones are fenced off.
• Buy-in and logistics. How often you play, how you handle a missing player, whether characters are expected to cooperate. Boring questions, but campaign-ending if they surface as resentment in month three instead.

You don't need a contract. You need ten minutes of honesty before anyone's invested twelve sessions in a tone half the table didn't want. The session-zero notes are also your first chronicle entry: the premise and the cast, written down before the world starts moving.

Prep situations, not plots

Almost every new GM makes the same mistake, and plenty of veterans never stop making it: prepping the plot. The villain reveals himself in scene four, the betrayal lands in scene seven, the players confront the duke in the finale. It reads beautifully in your notes. It dies the instant a player decides to rob the duke in scene two instead.

The fix is to prep the board state, not the moves. Don't write what happens. Write what's true, and let the players' choices collide with it. That's the whole argument behind the Alexandrian's much-cited "Don't Prep Plots": prep situations and the tools to run them, never a script the players are obliged to follow. A situation preps itself. Give the world standing forces and pressures, then ask what they do in response to the party.

Tools for prepping a situation

• Factions. Two or three groups with goals that can't all win. The thieves' guild, the church, the new governor. The campaign's energy comes from them grinding against each other with the party caught in the gears.
• Fronts. A front is a looming bad thing with a face (a cult, a plague, an advancing army) and a rough sense of what it's working toward. It exists whether or not the party engages, which is what makes the world feel alive.
• Clocks. A clock is a countdown to something concrete: the ritual completes, the siege breaks, winter arrives. Advance it when the party dawdles. Clocks turn "the world reacts" from a vague promise into a thing you can actually track on paper.

Prep this way and you stop fearing improvisation, because you're never inventing a world live. You're reading the board and asking the question that matters: given what's true right now, what happens next?

Prep the parts that react, not a script

The tempting trap is to over-prepare: write the scenes, the dialogue, the clever twist. Players then walk left when you prepped right, and three hours of work evaporates. The fix isn't less prep. It's prepping the durable things, the parts that survive improvisation because they react instead of dictating. Sly Flourish's eight-step "lazy" prep boils this down to a slogan worth taping to your screen: prepare less, get more.

• NPCs with wants. An NPC who wants something will generate plot in any scene you drop them into. An NPC with scripted lines only works in the one scene you imagined.
• Factions in tension. Two or three groups pulling against each other give the world its own momentum. The players become the variable that tips a balance you already built.
• Open questions and threats. What's the clock ticking in the background? What does the villain do if the party does nothing? Prep the world's intentions, not the party's path.

Do this and you can improvise a believable session on the spot, because you're not inventing a world live. You're just asking what your prepared pieces would do.

What to track between sessions

Between sessions, a campaign is held together by a handful of things that keep coming back. Most of them never make it into traditional notes, because notes record what happened, and continuity is really about what's still pending. Lose track of any one and the cracks show fast.

• Open threads. The questions you left hanging on purpose: who poisoned the well, where the missing heir went, why the ferryman flinched at the duke's name. A thread you wrote down is a plot hook waiting to pay off; a thread you forgot is just a hole the players fall into.
• NPC goals. Not just who the NPCs are, but what each one is doing right now while the party isn't looking. The duke is consolidating debts. The cult is recruiting. Track the wants, and every NPC walks back into the next session already in motion.
• Debts and promises. The party owes the smith a favour; the governor owes them a pardon; someone swore an oath nobody's collected on yet. These are the strongest hooks you'll ever have, because the players built them. Forgetting a promise the table made is the fastest way to make a world feel weightless.
• Faction clocks. Where each front stands. Did the siege advance? Did the ritual gain a step? Even when the party did nothing this session, the clocks should have moved, and you need to know by how much before you sit down again.

The trick is that none of this can live in the GM's head alone. Put it somewhere the whole table can open, even a shared Google Doc or a spreadsheet pinned in your Discord, so the canon isn't yours to remember but everyone's to check. The notes for any single game are one problem. There's a separate guide on taking session notes for that, and a recap template for opening the next one. A campaign is the harder problem: making those notes connect, session after session, into something the whole table can rely on, which is what a campaign wiki is for.

Pacing across many sessions

A single session has a natural shape: a hook, some rising trouble, a payoff. A campaign has a shape too, but it plays out over months, and it's much easier to lose. Good campaign pacing is mostly about three rhythms running at once.

Spotlight

Over any few sessions, every player should get a stretch where the story is about them: their backstory, their rival, their hard choice. Spotlight isn't fairness for its own sake; it's how you keep five people invested in one shared story. The player who hasn't had a moment in six sessions is the player who starts checking their phone. Tracking each character's open threads is how you spot whose turn is overdue.

Escalation and downtime

Tension can't only climb. A campaign that's all rising stakes exhausts everyone; a campaign that's all wandering loses its spine. Alternate. Let a tense arc resolve, then give the party downtime, a quieter session to spend rewards, heal, scheme, and breathe. Downtime isn't filler. It's where players reconnect with the world, where you plant the next hook, and where the contrast that makes the next escalation land gets built.

The long arc

Underneath the episodes, something larger should be slowly turning: the front advancing, the conspiracy widening, the cost of an early choice coming due ten sessions later. You don't need it fully mapped. You need to feel it moving, so that when the campaign finally pays off, players recognise that the ending was implicit in the beginning. That recognition is only possible if the early threads are still on record when the late ones arrive.

Common campaign-running mistakes

Almost every campaign that quietly dies hits one of the same few rocks. None of them are dramatic, which is exactly why they're dangerous. They erode the game slowly enough that nobody calls a meeting about it.

• Over-prepping. Pouring hours into scripted scenes the players route around, then feeling crushed when the work goes unused. Symptoms: you steer the table back toward "your" content, and the game starts to feel like it's on rails. Cure: prep situations, not plots, as above.
• Railroading. The flip side of over-prepping. When the only good ending is the one you imagined, players feel it instantly, even when they can't name it. Choices that don't matter aren't choices, and a table that senses its choices are fake stops making interesting ones.
• Forgetting threads. The party did something memorable in session three; in session nine it's vanished without a trace. Each dropped thread teaches the table that nothing they do persists, and a world where nothing persists isn't worth investing in.
• No continuity. The slow killer. Names drift, an NPC dies and reappears, the geography reshuffles between sessions. No single slip is fatal, but together they tell the players the world isn't solid, and a world that isn't solid can't be taken seriously, no matter how good your improvisation is.

Notice that the last two are the same mistake at different scales: a failure to remember. And memory, conveniently, is the one campaign problem you can fully automate.

Let the chronicle carry the continuity

The quiet rule under all of this is that continuity isn't memory, it's shared memory. The moment the canon lives only in your GM binder, every recall becomes a negotiation, and negotiations are where campaigns lose their solidity.

So give the table one place to look. A running chronicle of what actually happened, plus living entries (cards) for the people, places, quests and factions that recur, updated as the world changes. Re-read the last entry before each game, open with a thirty-second recap, and the cursed-sword argument simply never happens. It's written down. Everyone can check.

The catch, of course, is that maintaining all of this is itself a job, the bookkeeping that always loses to actually running the game. And your prep time is finite. Every hour you spend re-reading old notes and updating trackers is an hour stolen from the things players actually feel: the next location described, the NPC voiced, the map sketched, the twist set up. So the move isn't trying harder to keep notes. It's making the chronicle write itself, so that every session you start knowing exactly where you left off (the open threads still flagged, the NPC goals still tracked, the faction clocks still ticking where you parked them) without spending an evening on it. That's the part worth handing off, and the reward is the time you get back.

More on the per-session side: how to take RPG session notes. Browse the rest of the guides.