Most chapters that don't work fail at the scene level. The chapter has prose, dialogue, a setting, sometimes a feeling. What it doesn't have is a scene. The unit of meaning that moves the story.
Short version, before the long one. A working scene has three load-bearing parts: the protagonist enters with a goal, faces escalating conflict that tries to stop them, and exits with a disaster (a setback, a reversal, or a "yes, but" complication) that changes the story's state. Without those three, you have a passage. With them, you have a scene.
This post walks through scene mechanics, the most useful templates, and the most common failure modes.
What a scene is, structurally
A scene is the smallest unit of story that changes the story's state. After a scene, something is different. A character knows something they didn't, has decided something they hadn't, has succeeded or failed at something.
If your prose passage doesn't change the state, it's not a scene. It might be exposition. It might be transition. It might be a sequel (more on that below). It's not a scene.
Most working scene-craft frameworks trace back to Dwight Swain's Techniques of the Selling Writer (1965), which formalized the "scene and sequel" structure that genre fiction has run on for sixty years. Swain's structure is what most modern writing teachers teach when they teach scene craft.
The three load-bearing parts
1. Goal
The scene opens with the protagonist (or POV character) having a specific, articulable goal in the scene. Not the book's overall goal. The scene's goal.
Examples of scene goals:
- Convince her father to let her enlist in the war academy.
- Get into the morgue without anyone noticing.
- Tell him she's pregnant.
- Find out whether the artifact is real.
- Avoid the conversation entirely.
The goal must be specific (not "have a good night"), concrete (not "feel better"), and timeline-bound to the scene itself.
If you can't name the protagonist's scene goal in one sentence, you don't have a scene. You have a passage where things happen.
2. Conflict
Something resists the goal. The conflict escalates across the scene; each beat raises the cost or the stakes.
Conflict types, in order of strength:
- Direct human opposition. Someone wants the opposite of what your protagonist wants. Best conflict; produces dialogue.
- Internal opposition. The protagonist wants the goal but also doesn't want it. Second best.
- Environmental opposition. The setting fights the protagonist (storm, mountain, ticking clock).
- Information opposition. The protagonist needs to find something, and the search is the conflict.
Many scenes layer two or three of these. A protagonist convincing her father (direct human opposition) while doubting her own choice (internal opposition) while a storm closes the roads (environmental) is a richer scene than any one of those alone.
The escalation matters. Conflict at the start of the scene should be lower than conflict at the end. Beats stack: she asks; he refuses; she pushes; he gets angry; she reveals a secret; he reverses; she presses for what she actually wants. Each beat raises the temperature.
3. Disaster
The scene ends with a state-change. The most useful framing comes from Randy Ingermanson and Dwight Swain: scenes end with one of three outcomes.
- Yes. The protagonist gets exactly what they wanted. (Rare. Use sparingly. Too many "yes" endings = no story momentum.)
- No. The protagonist fails. They wanted X; they got nothing or got blocked.
- Yes, but. They got what they wanted, but it costs them something. Or they got it, but it isn't what they thought. This is the most useful outcome; it gives momentum without flat failure.
- No, and furthermore. They failed, and made it worse. A common variant of "no" that increases stakes.
"Yes, but" and "no, and furthermore" are the workhorses. They keep the story moving while preventing the protagonist from being either too lucky or too pathetic.
The sequel (the scene's quieter twin)
Swain's other contribution was the sequel. A sequel is what happens after a scene. The protagonist processes the disaster, has an emotional reaction, considers options, makes a decision about what to do next.
A sequel has its own three parts:
- Reaction. Emotional response to the disaster.
- Dilemma. The protagonist faces a choice between bad options.
- Decision. They pick a path, which becomes the goal of the next scene.
Sequels can be one paragraph or one chapter. They're how internal life happens in genre fiction. Skip them and your novel feels like a sequence of action set-pieces with no protagonist.
The pacing rhythm of a working novel alternates:
scene → sequel → scene → sequel → scene → sequel
Action and processing. Thriller writers often run scenes back to back to back; literary writers often run sequels back to back to back. Most novels live in the middle.
The Scene-First Workflow
Before you draft a chapter, name the three load-bearing parts on a single line.
Scene 14: Maren needs to convince Captain Avery to let her test for the dragon-rider trials. Avery refuses on regulation grounds. Maren reveals her grandfather's medal, which Avery served under. Avery agrees, but says he'll personally fail her on day one. Yes, but.
That's enough to draft from. You know what you're writing toward. The actual scene fleshes it out with prose, dialogue, setting, and voice, but the structural decisions are made.
If you can't write that line, don't draft yet. Drafting an unscened-out scene almost always produces a meandering passage that you have to cut or restructure.
Common scene failures
The scene with no goal
The protagonist enters a scene with no articulable agenda. Things happen to them. They react. They leave.
The reader's experience: nothing was at stake; nothing changed; this could be cut.
The fix: give the protagonist a goal, even a small one. Even "avoid running into Sarah at the coffee shop" is a goal. The conflict ("Sarah is there") writes itself.
The scene with no escalation
The protagonist has a goal. They face one obstacle. They overcome it. They leave.
The reader's experience: flat. No urgency.
The fix: each beat should raise the cost. If she asked once and he said no, escalate to her offering something, then to her threatening something, then to her revealing something. Five beats, each higher than the last.
The scene where they get exactly what they wanted
"Yes" endings are the most common failure in beginner drafts. The protagonist asks for X, gets X, exits the scene. The next scene is unmoored from consequence.
The fix: even when the protagonist "wins," cost them something. They got the information, but had to lie. They saved their friend, but their friend doesn't know yet. They escaped the trap, but lost the artifact.
The scene that's secretly two scenes
The protagonist enters with one goal, achieves it midway, then pursues a second goal in the same scene. The result feels long and unfocused.
The fix: break it into two scenes. Each with its own goal, conflict, disaster.
The scene with too many people
Scenes with five or more named characters speaking are hard to track. The reader loses who's said what.
The fix: most working scenes have two to three speakers. Move other characters to the background or off-stage.
The dialogue scene where nothing changes
Two characters talk for six pages. Their feelings are clearer, but the state of the story is identical to when the scene started.
The fix: dialogue is the conflict, not the scene. The scene still needs a goal and a disaster. If your characters are talking about their feelings, the goal might be "get them to admit something," and the disaster is whether they admit it, what it cost them to admit it, or what they admit instead.
Scene length
Working scene lengths in commercial fiction:
- Thriller, mystery, romance, romantasy: 1,500 to 3,500 words per scene. One scene per chapter is common.
- Literary fiction: 2,000 to 6,000 words per scene. Scenes are longer, dwell more.
- YA: 1,500 to 2,500 words per scene. Chapters are shorter.
- Fantasy, sci-fi: 2,500 to 5,000 words per scene.
If a single scene runs over 6,000 words, it's probably two scenes. If a scene is under 800 words, it's probably half a scene; consider if it's actually a sequel or transition.
Scene openings
Every scene opens in one of four registers:
- In medias res. Drop the reader into action mid-event. Most modern thriller and YA scenes open this way.
- Setting establishment. Where we are, briefly. Useful when the location is unfamiliar.
- Time/distance from previous scene. "Three hours later," "Across town." Useful when there's been a jump.
- Internal state. The protagonist's emotional state right now. Useful for sequels and contemplative scenes.
Avoid: lengthy weather description, prose-poem prologue, character-thinking-about-backstory-for-three-paragraphs. These are the most-skipped scene openings.
Scene endings
The strongest scene endings:
- A revealing line of dialogue. "I know what you did." End the chapter on the period.
- A decision. The protagonist commits to something. The reader has to turn the page to see them act on it.
- A reveal. A piece of information that recontextualizes everything before it.
- A physical action that promises consequences. She picked up the gun. He opened the letter. They kissed.
- A question the next scene has to answer. What happens now?
Avoid: the scene fading out on tired protagonist going to sleep, the scene resolving with characters agreeing about everything, the scene ending on a paragraph of philosophizing.
Read the last line of every scene in your draft in sequence. If each one isn't doing one of the five things above, your scenes are ending soft. Soft endings produce the "I put the book down and didn't pick it back up" feeling.
How to revise a flat scene
When you find a flat scene in revision, work through this checklist:
- What was the protagonist's goal in the scene? If you can't name it, give them one.
- What was the conflict? If there wasn't one, add one, even a small one.
- How did the scene end? If "yes" or "nothing changed," rework to a "yes, but" or "no, and furthermore."
- Did each beat raise the temperature? If not, reorder or escalate.
- Is the scene actually two scenes? If yes, split.
- Did the disaster set up the next scene's goal? If not, the scene isn't connected to the story's forward motion. Re-aim.
Most flat scenes need only one or two of these changes to come alive.
Where Inkett fits
Inkett is the writing stack for working novelists. The Co-Writer drafts in your voice, plan-aware; the /im-stuck slash command surfaces ideas for what the next beat could be when your scene flattens mid-draft. The Editor flags scenes that don't change state and chapters that end soft. The Planner gives you a node-based view where you can outline each scene's goal/conflict/disaster before you draft.
Worth pairing with: How to Write a First Chapter That Doesn't Get Skipped, How to Fix a Sagging Middle, and How to Fix Pacing Issues in a Novel for the chapter-level and book-level views of the same problem.
Scenes are the unit of story. Get the scene right and the chapter follows. Get most of your scenes right and the book follows.
Tags
The writing stack for novelists.
A developmental editor for your finished manuscript. A visual story planner. A pair-writing partner for your draft. A native publisher for your readers. The tools work in your voice. You stay the writer.