Your story document is 38 pages long. Your sprint board has one card on it that says "implement story." Somewhere between those two artifacts a game is supposed to happen — and it usually doesn't, because nobody can estimate a card like that, nobody knows which scene blocks which level, and the writer keeps adding beats that need a character who was cut in month four.
The fix is not a better outline. It is a different unit of work. Prose outlines describe what happens; chapters describe what the team builds. Here is a workflow for getting from one to the other, and for catching structural problems while they are still cheap.
Why game stories break at the chapter line
A film script is one timeline, executed by one department. A game story must survive being interrupted, repeated, failed and skipped. Three things break when a prose outline goes straight to production:
- Beats have no cost. "She confronts her brother at the harbour" could be a 20-second exchange or a new level, a new NPC, a new animation set and an hour of content. Nothing in the sentence says which.
- Beats have no gate. Prose assumes the reader turns the page. A game has to let the player reach the next scene — a new ability, an opened door, a resolved objective. Leave the gate unwritten and it gets invented late and badly: an invisible wall and an NPC saying "not yet."
- Beats hide dependencies. Chapter 7 quietly needs a boat, a night lighting setup and a second voice actor. You find out in month nine.
Chapters fix this by forcing every story moment to declare what it costs, what unlocks it and what it needs.
Step 1: Write the beat sheet, one line per beat
Before anything is a chapter, get the whole story onto one page as 10 to 14 one-line beats. If a beat needs a paragraph, it is two beats or it is backstory.
A worked example for a hypothetical 2D exploration game, Saltwake, about a diver salvaging a flooded town:
- Mira dives the shallows for scrap; the water is wrong.
- She finds her mother's dive log in a drowned house.
- The harbourmaster offers fuel in exchange for silence.
- She reaches the sunken chapel and hears the bell ring underwater.
- The log's last page names a place that is not on any chart.
- Her brother Teo blocks the lock gates, and he has reasons.
- The deep trench opens; her gear is not rated for it.
- She finds the town's old pumping station, still running.
- Mira chooses: drain the bay, or let it stay drowned.
Nine beats. No dialogue, no adjectives, no lore. You are checking one thing: does each beat change what the player wants? If two consecutive beats leave the player with the same goal, one of them is decoration.
Step 2: Turn beats into chapter cards
Now give every beat a card with fixed fields. The fields are the trick: they make omissions visible. A card that cannot name its new verb will play as a corridor with text on it.
| Field | Saltwake, Chapter 4 |
|---|---|
| Beat | She reaches the sunken chapel and hears the bell |
| Player goal | Get inside a building that is fully submerged |
| New verb / upgrade | Air bladder: refill oxygen at trapped-air pockets |
| Gate (what unlocked this) | Fuel from Ch. 3 opens the east channel |
| Delivery | Environmental: no cutscene, one radio line |
| Characters on screen | Mira, harbourmaster (radio only) |
| New assets | Chapel interior kit, bell SFX, 2 air-pocket props |
| Est. playtime | 18 min |
| Status | Blocked on air-bladder prototype |
Nine of those cards and you have something a producer can read. In GameDesignerX these live in the Story Chapters module, linked to the characters and locations each one touches, so changing Teo's motivation shows you the four chapters that reference him instead of making you remember them.
Step 3: Gate with a mechanic, not a cutscene
The strongest structural rule in game narrative: every chapter transition is paid for by player capability. The player gains a verb, a number or a piece of knowledge, and the next chapter is where that thing belongs.
This shows up reliably in games people cite as well-paced. Portal teaches one portal rule at a time, then puts you in the room that needs it. Metroid-style structure is the pure form: the gate is a door your current weapon cannot open. Outer Wilds swaps ability for information — the gate is what you now know — which is why its chapters play in almost any order. Hades gates narrative on runs, advancing the story through repetition.
Pick a gate type per chapter and write it on the card. Three cover nearly everything:
- Ability gate — a new traversal or combat verb (air bladder, grapple, double jump).
- Resource gate — a threshold to reach (fuel, reputation, 3 of 5 artefacts).
- Knowledge gate — a code, a name, a place the player must have seen.
Three chapters in a row on the same gate type feel flat no matter how good the writing is. Saltwake runs ability, resource, knowledge, ability, knowledge — and its one repeated pair (Ch. 7–8, both ability gates) is deliberate, because that is the stretch where the diving gear is the story.
Step 4: Budget the chapters against the calendar
Add up the playtime column, then the new-asset rows. This is where the outline collides with the schedule, and it should happen in week two, not month nine.
For Saltwake: nine chapters averaging 16 minutes is about 2.5 hours of story content — a reasonable target for a small team's first commercial game. But the asset column lists six distinct environment kits, and on a two-artist team six kits is the shape of the project. The cut is obvious once it sits in a column: merge the chapel and the pumping station into one architectural family, go from six kits to four, keep all nine beats.
That is the practical value of the cards: scope decisions become asset decisions instead of story decisions. You rarely need to cut a beat. You need to cut the number of places beats happen in.
Two rules worth stealing:
- Reuse rule: every environment kit serves at least two chapters, ideally in different states (flooded/drained, day/night, intact/collapsed). A kit used once is the most expensive content you will ship.
- One-new-thing rule: no chapter gets more than one new system and one new location. Break it and that is the chapter that slips twice.
Step 5: Three revision passes before you lock it
Run these separately; mixing them means rewriting prose instead of fixing structure.
The motivation pass. One sentence per chapter: why does the player press forward right now? Not the character — the player. Curiosity, a named goal, a threat, an unresolved promise. A chapter whose answer is "because it's next" is where people put the controller down.
The redundancy pass. Tag each chapter with the single emotional note it delivers (dread, relief, betrayal, awe, competence). Two neighbours sharing a note means one is doing no work. In Saltwake, Ch. 2 and Ch. 5 both landed on "discovery," so Ch. 2 moved toward unease: the log is found, but a page has already been torn out.
The cut pass. If this chapter were deleted, which later chapter breaks? If the answer is "none," it is either an optional side chapter — label it so — or dead weight. A story where every chapter is load-bearing is a story you can finish.
Before you lock the outline
- 10–14 one-line beats, each changing what the player wants
- Every chapter has a gate type written down
- No two neighbouring chapters share an emotional note
- No chapter adds more than one new system and one new location
- Every environment kit serves two or more chapters
- Playtime total matches your real production window
- Each chapter names the characters and locations it depends on
- Every chapter is load-bearing or explicitly optional
Do this and the 38-page document becomes nine cards, each estimable, each with a visible dependency. Lay them beside your milestones on a timeline and "implement story" stops being one card: it becomes nine, in order, with the blockers named.
The outline will still change. It always does. But it changes one card at a time, with the cost showing.