How to Cut Scenes That Don’t Belong

A scene that does not belong is rarely “badly written.” It is a scene doing a job the book already did, a job the book does not need, or a job in the wrong chapter. The Cut/Merge/Move Scene Test decides the scene’s fate before you spend a weekend polishing dialogue that should not survive.

The Cut/Merge/Move Scene Test

For each scene, write one sentence: Character wants X; by the end, Y is true that was not true before.

Then pick one verdict:

• Cut — Y is empty, or Y already happened
• Merge — Y is real, but another scene already produces a sibling of Y
• Move — Y is real, but it lands before the reader can care or after they have stopped needing it
• Keep — Y is new, costly, and in the right place

This test sits inside a first-draft revision after you know the spine. It is the practical sibling of fixing a pointless scene: pointless is a diagnosis; cut/merge/move is the treatment.

Cut: nothing irreversible happens

Irreversible does not mean explosions. It means the character cannot honestly return to the previous state of knowledge, alliance, or options.

Cut candidates:

• Travel that exists so geography feels “complete”
• A second dinner that restates the first dinner’s thesis
• Flashback that explains a motive the present scene already dramatizes
• Comic relief that does not change how anyone will act tomorrow

After a cut, read the join. Often you need one transitional clause, not a replacement set piece. If the join feels like a highlight reel of disconnected events, you have an episodic problem in the remaining spine—not a reason to restore the cut.

Merge: two scenes, one job

Merging is how you keep texture without paying twice.

Example — two exposition scenes. Scene A: the mentor explains the treaty in a classroom. Scene B: the protagonist overhears the same treaty in a tavern, with jokes. Merge: the tavern is the classroom. The jokes carry the rules, and a drunk official contradicts a clause, which gives the protagonist a question to act on. You keep the information and gain conflict. You lose a lectern.

Merging is not concatenation. Do not tape scene B under scene A. Steal the function of B (a contradiction, a sensory hook, a side character) and let A do both jobs in fewer pages.

Move: right scene, wrong hour

Some scenes are good and early. The reconciliation, the confession, the “I know who you are” beat—if it lands before the cost is visible, it reads as cheap. If it lands after the plot has already pivoted, it reads as leftover.

Example — repeated reconciliation scenes. A couple fights in chapter 6, makes up in chapter 7, fights in chapter 12, makes up in chapter 13. The second reconciliation is a move-or-cut. Move the first honest make-up to after the midpoint betrayal, and let chapter 7 end on an unresolved debt. The relationship now becomes plot instead of a loop.

Moves also rescue worldbuilding. A market tour that bores in chapter 2 can become the chapter 14 chase route—same stalls, now with a reason to run.

Keep: a high bar on purpose

Keep if the scene:

• Changes a goal or a plan
• Makes a secret expensive
• Closes an option the character wanted
• Forces the next scene to exist

If you are bargaining (“but the banter”), you are lobbying for merge, not keep.

How to cut without losing your nerve

Create cuts.md or a folder. Paste the scene with a one-line epitaph: “ch9 orchard — kept the frost image.” Grief is real; a shrine in the manuscript is optional.

Tell yourself the truth about length. Cutting 3,000 words from an 80k draft is not a failure of imagination. It is how the remaining 77k can be read in a sitting without a nap in the orchard.

Re-thread checklist

After each verdict:

1. Does the following scene still have a cause?
2. Does a character still refer to an event that no longer happens? (Search names, objects, “yesterday.”)
3. Did you just cut the only place a clue appeared? If yes, plant the clue in a kept scene—do not un-cut by panic.

Work one act at a time. Cutting across the whole book in one night invites you to delete the quiet scene that was actually the midpoint.

What cutting is not

It is not “make every chapter loud.” Quiet scenes belong when they change the inner contract: a character who can no longer pretend, a loyalty that costs a night of sleep. It is not “kill your darlings” as a personality. It is logistics. The darling can live in the scraps file and return in book two, or in a different scene that pays for it.

When the inventory is honest, prose work becomes possible. Until then, rewriting a scene that should be merged is overtime on the wrong job.

A worked morning (both examples in one pass)

Take a stack of ten scene cards.

Cards 2 and 3 are both “mentor explains the treaty.” Merge into the tavern. Keep the joke that makes the official angry; cut the chalkboard.

Cards 6 and 9 are both “couple makes up.” Move the honest make-up to after the betrayal; cut the first make-up’s kiss; keep one line of unfinished apology as a hook.

Card 7 is a market tour. Move it to the chase. Card 8 is comic relief at a well. Cut it; steal the well as a hiding place in the chase.

You have not “killed darlings.” You have assigned jobs. The frost image from the orchard still lives in scraps for a later winter chapter that actually changes a plan.

If you stall, you are probably bargaining with keep. Read the one-line Y. If Y is “we understand them better” with no new obligation, it is still a cut. Understanding that does not change tomorrow is a blog post, not a scene.

Tools can wait; this test is a pencil-and-list job, and PlotnAI does not get a vote on which rooms you demolish.