How to Write a D&D Adventure (A Practical Framework for Scenario Design)

Most first-time DMs try to write a D&D adventure the way they'd write a short story: beginning, middle, end, with the players cast as actors reading lines. Six sessions later, they're exhausted, improvising frantically, and the "adventure" has almost nothing to do with what they wrote. What they wrote is sitting unused in a notebook.

The fix isn't to write less. It's to write differently. An adventure isn't a script — it's a situation: a set of pieces on a board, a handful of NPCs with agendas, a clock ticking in the background, and several ways it can resolve. The DM's job is to build the situation with enough detail to be discoverable, and enough space for the players to make it theirs.

This guide walks through the framework experienced designers actually use. It works for one-shots, multi-session arcs, and full campaign chapters. Once you've run it two or three times, you'll stop writing "adventures" as linear plots and start writing them as playable systems.

The Foundation: Situation, Not Story

Before you write a single NPC name, decide what the situation is. A situation has four parts:

• An inciting pressure — something is wrong, something is about to happen, or something has just changed. A village is being raided. A prince has disappeared. A comet will hit in seven days.
• An agent causing or exploiting the pressure — a villain, a faction, a natural force, a misunderstanding.
• Stakes if nobody intervenes — what gets destroyed, lost, or corrupted if the players do nothing.
• A hook that puts the PCs in position — why are these characters, at this time, in a place to act?

Write these four things in one paragraph. If you can't, you don't have an adventure yet — you have a mood. Don't proceed until the one-paragraph pitch is concrete.

Example: "The fishing town of Vell is losing one person per moonless night. The disappearances are caused by a sea-hag bound by ancient treaty to take one life a month; the treaty was broken three weeks ago when the town's mayor killed her familiar. If nobody intervenes, she'll escalate to a mass drowning when the next tide comes in. The party is hired by the dead familiar's witch-sister, who wants the mayor punished but the hag destroyed."

That's a full adventure in 80 words. Everything else is detail in service of that paragraph.

The Three Layers of an Adventure

Every runnable adventure exists on three layers at once:

Layer 1: The Clock

Something is getting worse. Every session, something ticks forward whether the party acts or not. This is the engine that makes the world feel alive and keeps the adventure from stalling.

Common clocks:

• A ritual counting down to completion
• An army on the march
• A disease spreading
• A political alliance hardening
• Resources running out (food, water, magic)

Write the clock as a list of events. "Day 1: the festival opens. Day 3: the cult's ritual begins. Day 5: the ritual reaches phase two, the first sacrifice. Day 7: the sacrifice is complete and the god wakes." If the party does nothing, this is what happens. Their job is to intervene at some point along the clock; the consequences differ depending on when.

Layer 2: The Web

Who's in the story, what do they want, and how are they connected? Draw it literally — boxes with lines between them. For a small adventure, five to eight NPCs. For a longer one, ten to fifteen.

Each NPC gets:

• A name and one-line description
• What they want (the thing driving their actions)
• What they know that the party doesn't
• Who they trust, fear, or hate

This is where most adventures either come alive or flatten. A web of NPCs with agendas creates emergent story; a cast list of characters with "dialogue" creates a script. Investing 30 minutes in the web saves hours of improvisation later.

If you already have a campaign setting built, pull NPCs from your existing wiki. Recurring faces carry emotional weight that brand-new NPCs can't match. A wiki with relationship tracking makes this especially valuable — you can see which of your existing NPCs already know each other and build connections into the adventure naturally.

Layer 3: The Scenes

Now — and only now — you write scenes. These are the specific moments where the party interacts with the clock and the web. They're the parts the players will experience directly.

Each scene should have:

• A purpose — what it's supposed to reveal, change, or test
• A setting — where it takes place, one vivid detail
• Participants — which NPCs are involved
• Variable outcomes — at least three ways it could resolve

Crucially, scenes are modular. They don't have to happen in a fixed order. The party might visit the witch before the tavern, or after, or not at all. If your scenes require a specific sequence, you've written a railroad, not an adventure.

The Hook

The hook is how the party gets involved. Weak hooks are the number-one reason adventures fizzle in the first 20 minutes. Characteristics of a strong hook:

• Personal, not hired. "A stranger offers you gold to investigate" is weak. "Your old mentor's son has gone missing" is strong. Tie the hook to a PC's backstory, a faction they care about, or a place they've been.
• Immediate, not hypothetical. The thing is happening now, or about to. Don't open with "you hear rumors of" — open with "as you enter the town square, a man collapses at your feet, bleeding, clutching a sealed letter."
• Revealing, not just activating. A good hook introduces the core tension, not just the job. The collapsing man's letter contains the first clue, the first NPC name, the first threat.

Write three hooks per adventure, not one. Sometimes the party bites immediately; sometimes they resist and need a second approach. Having backups keeps you from railroading when they don't engage with your first idea.

Design for Agency, Not Coverage

Bad adventures try to cover every possibility. Good adventures concentrate on a few strong choices. The question for every scene: does this give the party a meaningful decision?

Meaningful decisions have three properties:

• More than one option (obvious)
• All options have costs or tradeoffs
• The party can't predict the consequences perfectly

"Do you fight the goblins or sneak past them?" is a decision only if fighting means lost resources and sneaking means risking detection later. "Do you trust the informant?" is a decision only if both trust and distrust lead somewhere interesting.

Audit your adventure: how many genuine decisions does it offer? If the answer is "one — the final boss fight," your adventure is a corridor. Add branches. Make information cost something. Make safe paths longer. Make fast paths dangerous. Every genuine decision is where the adventure comes alive.

The Scene Budget for a Four-Hour Session

A typical session runs three to four hours of actual play. In that time, most groups handle about:

• Three to four roleplay-heavy scenes (conversations, investigation, social tension)
• One or two combat encounters
• One or two exploration or puzzle moments
• A session-ending cliffhanger or revelation

If your one-shot has twelve scenes in it, you're writing a two-session adventure and calling it one. Cut ruthlessly. A six-scene adventure with three strong decisions beats a twelve-scene adventure with one decision every time.

The Information Economy

An adventure is partly a mystery: there are things the party doesn't know, and discovering those things is half the play. Design what the party learns, when, and how.

For each piece of important information, answer:

• Who in the web knows this?
• How does the party get it from them — conversation, observation, action?
• What happens if they never learn it?

Apply the three-clue rule: any fact essential to the adventure should be discoverable in at least three ways. If only one NPC knows it and the party doesn't find that NPC, the adventure breaks. Redundancy isn't lazy — it's how you let the party miss obvious clues without the plot collapsing.

Stakes That Scale

"The world will end" is a bad stake for most adventures. It's too abstract; players can't feel it. Better stakes are specific, visible, and connected to people or places the party has met.

• Not "the kingdom falls" but "this specific town, which the party has visited twice and knows three people in, is destroyed."
• Not "magic dies" but "the mentor NPC the party loves loses her ability to protect them."
• Not "chaos reigns" but "the merchant who patronizes them is ruined, and their home base of operations is gone."

Stakes are a feeling, not a scale. A 2nd-level adventure with genuine, specific stakes is more gripping than a 15th-level adventure where abstract "world-ending" language has become noise.

The Villain in an Adventure

Every adventure needs opposition. Not necessarily a classic villain — could be a faction, a natural phenomenon, a rival party, a moral dilemma. But there has to be something pushing back.

For adventure-length villains (as opposed to campaign-length), you can write them faster than a full-campaign antagonist. You still need: a specific grievance, a motivation that makes sense to them, a flaw the party can exploit, and a presence that makes them more than a statblock. Our fantasy villain guide goes deeper if this is where your adventures typically fall flat.

Ending States

How does this adventure end? Write at least three resolutions before running it:

• The expected resolution — what you think will probably happen. Party fights the villain, wins or flees, story wraps.
• The alternate resolution — they bypass combat, negotiate, or find a creative exit. What does that look like?
• The failure resolution — they lose. What happens? Does the clock tick all the way? Do they escape with consequences? Failure shouldn't mean the campaign ends; it should mean the world changes.

Running with three endings in mind keeps the game flexible. If the party tries something you didn't anticipate, you can often map it to one of the three branches and keep going without panicking.

What You Don't Need to Write

Cut:

• Detailed read-aloud text (you'll speak naturally in character; prewritten text sounds stiff)
• Minor NPC names (generate on the fly with an NPC generator)
• Exact dialogue (know what the NPC wants, not what they say)
• Full battle maps for every possible fight (sketch the important ones, improvise the rest)
• Loot tables (a loot generator handles non-critical treasure)

Focus your writing on the clock, the web, and the scenes. Everything else is runtime work.

The Template You Can Reuse

Every adventure you write can use the same template:

1. One-paragraph pitch (situation, agent, stakes, hook)
2. The clock (what happens on days 1, 3, 5, 7 if nothing intervenes)
3. The web (8 NPCs, their wants, their connections, diagram it)
4. Three hooks to get the party in
5. Six to eight modular scenes, each with purpose and branches
6. Three key information beats and how they surface
7. Three possible ending states

Two to four hours of focused prep. One adventure ready to run. Over time, your adventures stop feeling like narrow scripts you're defending and start feeling like playgrounds you're opening up. That's when the party starts writing the best parts for you — and that's what every memorable session has in common.

Steal techniques. Run them. Iterate. The framework gets faster every time, and the adventures get sharper. Before long you'll be writing a session's worth of material in a single evening — and your players will be telling each other "no, remember when we did X" for years.