The short answer to planning your scenes
A twelve-chapter novel with two scenes per chapter needs twenty-four scenes, and from an eighty-thousand-word draft each one carries about three thousand three hundred words. Every scene should arrive with three assignments already decided: a purpose that states what the scene accomplishes, a conflict shape that names the two pressures pushing against each other, and a turn type that describes how the situation ends differently than it began. Enter your chapter count, scenes per chapter, genre, story engine, and target length above, and the planner lays out the full numbered list with chapter slots, point-of-view slots, and word budgets per scene. What you receive is scaffolding to draft from, never finished sentences.
That answer sounds plain, and plain is the point. Writers stall most often not because the story lacks promise but because a blank chapter hides dozens of undecided small choices. Naming the purpose, the opposing pressures, and the closing shift before drafting converts one large vague task into twenty or thirty small defined tasks. You still invent everything that matters — voices, images, dialogue, surprises — but you invent it inside a frame that already knows where each scene starts, what it must do, and where it must leave the reader.
What each scene on the list carries
The planner gives every scene six fields. Scene number and chapter slot fix the order. Point-of-view slot tells you whose eyes the scene belongs to. Purpose, conflict shape, and turn type tell you what the scene is for. Together they form a contract: if a draft fulfills all three, the scene earns its pages, and if it fulfills none, no amount of polished wording will save it.
Purpose is the job. Open a new threat, test a close bond, plant a fresh clue — each purpose names a single accomplishment. One scene, one job. Writers who assign two or three jobs to a single scene usually finish none of them, because the pages split their attention and the reader feels the blur. When a scene feels flabby in revision, the first question is whether it carries one clear purpose or several half-purposes competing for space.
Conflict shape is the resistance. Every entry pairs two opposed forces: duty against desire, honesty against pride, evidence against alibi. Notice the pairing matters more than the topic. A scene where everyone agrees moves information but creates no tension, while a scene where two genuine pressures collide forces somebody to choose, hide, or pay. The planner rotates conflict shapes from your genre so the resistance feels native to the kind of book you are writing rather than imported from elsewhere.
Turn type is the change. Reversal, revelation, decision, setback, discovery, escalation — six ways for a scene to end somewhere new. A scene without a turn is a postcard: pleasant, static, forgettable. Readers keep turning pages because each scene rearranges what they believed at its start. The rotation of turn types across your list exists to stop every scene from ending the same way, which is one of the fastest ways a manuscript starts to feel repetitive around the middle.
Point-of-view slot deserves its own note because beginners treat it as decoration. With one scene per chapter, every scene belongs to the lead and the book gains steadiness at the cost of surprise. With two per chapter, the list alternates lead and second voice, which lets you show events the lead cannot witness and keeps secondary characters load-bearing instead of ornamental. With three per chapter, an antagonist slot appears, and that slot carries a duty: every antagonist appearance must advance the opposition visibly, not merely remind readers that opposition exists. Choose the slot pattern that matches how many viewpoints your story can actually feed.
How the arithmetic works, with the numbers shown
The planner performs two divisions and three rotations. Understanding them lets you trust the output and adjust the inputs with intent.
First division: total scenes. Chapters multiplied by scenes per chapter. Twelve chapters at two scenes each gives twenty-four scenes. Ten chapters at three each gives thirty. Sixteen chapters at one each gives sixteen. Nothing hidden, nothing estimated — the count follows directly from the shape you chose.
Second division: word budgets. Target manuscript length divided by total scenes gives words per scene; divided by chapters gives words per chapter. Eighty thousand words over twenty-four scenes is three thousand three hundred thirty-three per scene, rounded to three thousand three hundred thirty-three on the page, with six thousand six hundred sixty-seven per chapter. The same eighty thousand over thirty scenes is two thousand six hundred sixty-seven per scene and eight thousand per chapter across ten chapters. Ninety thousand over sixteen single-scene chapters is five thousand six hundred twenty-five per scene, which is identical to per chapter since each chapter holds one scene.
Then the rotations. Purposes cycle from your story engine: an action-driven engine rotates through opening threats, raising stakes, forcing choices, staging confrontations, showing costs, and closing escapes. A character-driven engine rotates through exposing wounds, testing bonds, offering belonging, demanding honesty, breaking habits, and asking forgiveness. A mystery engine rotates through planting clues, deepening suspicion, ruling out suspects, hiding facts, sharpening the central question, and turning evidence over. Conflicts cycle from your genre in groups of four, and turn types cycle through the six in order. Scene seven therefore reuses the purpose of scene one but pairs it with a different conflict and a different turn, because seven divided across a six-item purpose list wraps around while the four-item conflict list sits at a different offset. That offset drift is deliberate: it recombines familiar jobs with fresh resistance so the list never reads as six scenes copied four times.
The verdict logic watches one number: words per scene. Between one thousand two hundred and four thousand five hundred, the plan passes. Above four thousand five hundred it warns that scenes run long, because a single purpose rarely stretches across that many pages without sagging in the middle. Below one thousand two hundred it warns that scenes run short, because turns arrive before pressure has time to build. These are craft conventions, not physical laws — some published books run shorter or longer — but they mark the range where most developing writers do their best work, and the warning tells you exactly which lever to pull.
Worked check on the defaults: twelve chapters, two per chapter, fantasy, action-driven, eighty thousand words. Total scenes twenty-four. Words per scene three thousand three hundred thirty-three. Words per chapter six thousand six hundred sixty-seven. Twenty-four sits inside the healthy band, so the verdict passes and the table lists scene one in chapter one from the lead opening a new threat with oath against survival ending in reversal, scene two in chapter one from the second voice raising the stakes with duty against desire ending in revelation, and so on through scene twenty-four. Each chapter pairs two viewpoints, each purpose appears four times across the book at different offsets of conflict and turn, and every row stays near its three-thousand-word budget.
Worked check on a denser shape: ten chapters, three per chapter, mystery genre, mystery engine, eighty thousand words. Thirty scenes at two thousand six hundred sixty-seven words each, eight thousand per chapter. The verdict passes. The antagonist slot now appears in every chapter, which suits mystery: the opposition needs regular visible pressure so suspicion has somewhere to land. Per-scene budgets near two thousand seven hundred words favor clue-planting rhythms — short scenes, frequent turns, steady accumulation.
Worked check on a sparse shape: eight chapters, one per chapter, romance, character-driven, eighty thousand words. Eight scenes at ten thousand words each. The verdict warns long, correctly: ten thousand words is far past the point where one purpose — say, demanding an honest answer — can hold attention without subdividing. The fix the planner suggests is structural, not stylistic: move to two scenes per chapter for sixteen scenes near five thousand words, or to twelve chapters at two per chapter for twenty-four scenes near three thousand three hundred. Same story, same total length, scenes sized so that each purpose owns a manageable span.
Illustrative hypothetical example: a fantasy quest on an action engine
Consider an illustrative hypothetical case. A writer plans a twelve-chapter fantasy at two scenes per chapter from an eighty-thousand-word target, action-driven throughout. The planner returns twenty-four scenes near three thousand three hundred words each, purposes cycling through threat, stakes, choice, confrontation, cost, and escape, conflicts cycling through oath against survival, duty against desire, power against its price, and loyalty against truth.
Scene one, chapter one, lead: open a new threat, oath against survival, reversal. In the writer's hands this becomes a border captain receiving orders to seal the mountain pass, which would strand her own village on the wrong side — the oath says obey, survival says refuse, and the reversal is her deciding to delay the order rather than obey or defy outright.
Scene two, chapter one, second voice: raise the stakes, duty against desire, revelation. The captain's younger brother, serving as a courier, learns the pass order came early because someone at court moved the timetable. Duty says deliver the sealed orders unopened; desire says warn his sister. The revelation — the timetable moved — raises the stakes without repeating scene one's choice.
Scene three, chapter two, lead: force a costly choice, power against its price, decision. The pattern continues, each scene inheriting its three assignments and the writer supplying everything else. The value of the list here is not that it invents the captain or the pass; it is that across twenty-four rows the writer never faces a morning wondering what a scene is for. The jobs rotate, the pressures vary by genre, the turns alternate, and the manuscript accumulates at roughly three thousand three hundred words per unit.
Illustrative hypothetical example: a romance duet on a character engine
Take a second illustrative hypothetical case with different DNA. A writer plans ten chapters at two scenes per chapter, romance genre, character-driven engine, seventy-five thousand words. Twenty scenes at three thousand seven hundred fifty words each, seven thousand five hundred per chapter. Purposes cycle through wounds, bonds, belonging, honesty, habit, and forgiveness; conflicts cycle through closeness against fear, honesty against pride, commitment against freedom, and need against independence.
An early scene might read: test a close bond, honesty against pride, decision. The writer stages a catering tasting where one partner discovers the other already changed the venue without discussion — honesty demands a confrontation at the table, pride demands smiling through it, and the decision lands on postponing the fight until the drive home, which itself becomes the next scene's fuel. A later scene reuses the same purpose with a new pairing: test a close bond, commitment against freedom, setback. Same job, new pressure, new ending shape. Readers experience variety while the writer works a controlled set of materials.
Notice the budgets doing quiet work. Three thousand seven hundred fifty words gives a romance scene room for the slow middle — the meal, the drive, the unsaid thing — that the genre's emotional payoffs require. Had this writer chosen three scenes per chapter, each unit would shrink to two thousand five hundred words and those slow middles would compress into summary. Shape is pacing, and the planner makes the trade visible before drafting begins.
Illustrative hypothetical example: a short mystery that warns correctly
A third illustrative hypothetical case shows the warning system earning its keep. A writer plans six chapters at one scene per chapter, mystery genre, mystery engine, eighty thousand words. Six scenes at thirteen thousand three hundred thirty-three words each. The verdict warns long, and rightly so: a thirteen-thousand-word scene tasked with planting a fresh clue would drown the clue in incident, and readers would finish unsure what they were meant to carry forward.
The writer follows the planner's suggestion and moves to twelve chapters at two per chapter. Twenty-four scenes near three thousand three hundred thirty-three words each. Now each clue gets its own container sized to hold exactly one discovery plus its misdirection. The mystery engine's rotation — plant, deepen, rule out, hide, sharpen, turn over — maps cleanly onto investigation rhythm, and the antagonist-adjacent suspicion has somewhere to breathe because the second viewpoint slot appears in every chapter. Nothing about the story changed except the subdivision, yet the book went from unwritable to plannable in one input change. That is the whole argument for planning scenes as countable units before writing them as prose.
Reading your result screen
The verdict block headlines your scene count and per-scene budget in one line, then explains the shape in two or three sentences. When it passes, it confirms the chapter math and reminds you the table is scaffolding rather than writing. When it warns, it names the direction — long or short — and prescribes the structural correction: add or merge scenes, add or cut chapters. Treat warnings as measurements, not grades. A flagged plan is a useful plan, because it tells you the fix before you spend weeks drafting at the wrong subdivision.
The metrics block states four figures: total scenes with the chapter-times-density derivation, words per scene with its division, words per chapter with its division, and chapter count with genre and engine for context. Screenshot these or copy them onto a note above your desk. During drafting they answer the two questions you will ask weekly: how long should this unit be, and how many units remain.
The table is the product. Scene number, chapter, point-of-view slot, purpose, conflict shape, turn type — row after row in drafting order. Read it once top to bottom for rhythm: do turns alternate or cluster, do purposes distribute or bunch, does each chapter contain a genuine shift. Then read it once per viewpoint, following only the lead rows, then only the second rows, then any antagonist rows, checking that each voice owns a coherent thread rather than scattered errands. Problems found at this stage cost minutes; the same problems found at page two hundred cost weeks.
The recommendations block closes with use instructions tuned to your verdict: draft top to bottom honoring each row's three assignments, split scenes that overrun at their turn instead of thinning the conflict, and keep the list visible while drafting so drift gets caught early. Every recommendation ends where the work continues — with the artifact in hand, moving toward pages.
Edge cases and failure modes worth knowing
Very small plans behave oddly. One chapter at one scene per chapter from a modest target yields a single scene carrying the entire manuscript — a warning fires, correctly, because one purpose cannot span a book. Treat single-scene outputs as outlines of outlines: useful for a flash piece, useless for a novel. If your book genuinely needs few chapters, raise the density to three per chapter before concluding the tool dislikes your structure.
Very large plans behave oddly in the other direction. Forty chapters at three per chapter yields one hundred twenty scenes, and even a three-hundred-thousand-word epic gives each only two thousand five hundred words. The verdict may still pass on budget while the practical problem becomes management: one hundred twenty rows exceed what most writers can hold in memory, and purpose repetition across twenty cycles dulls the sense of progress. For very long books, consider planning in parts — generate the list for part one, draft it, then generate part two with the engine that part requires.
Genre and engine mismatches produce strange but instructive rows. A mystery engine inside a romance genre plants clues in a love story, which can work brilliantly for romantic suspense but reads as clutter in a straight courtship narrative. An action engine inside literary pacing stages confrontations the surrounding style may not support. The planner assembles deterministically and will not stop you; the combinations are your responsibility. When a row feels alien, change the engine first — purpose rotation shapes the book more than conflict flavoring does.
The drift failure deserves attention. Writers generate a careful list, draft three chapters, then abandon the assignments when a shiny detour appears. Detours are allowed; unlogged detours are the failure. If scene nine's decision becomes a different decision mid-draft, update the row's turn and let the downstream rows absorb the change. A plan you revise stays a plan; a plan you silently discard becomes guilt.
The budget failure is subtler. Writers set a target length to please the tool rather than the genre — declaring sixty thousand words for an epic fantasy so the per-scene numbers look tidy, then drafting one hundred twenty thousand anyway. Every budget downstream of a false target is fiction. Set the manuscript length your category actually expects, accept the scene count that follows, and let the warning tell you the truth about subdivision.
Finally, the completeness failure: treating the list as the writing. Scaffolding tells you what each scene must accomplish and how it turns; it cannot supply voice, sensory detail, comic timing, or the exact line of dialogue that makes a decision hurt. Writers who draft directly from the rows without inventing inside them produce competent, forgettable manuscripts. The list removes structural dread so your invention has somewhere to go — it never replaces the invention.
Genre-specific guidance
Fantasy lists run on obligation and price. The conflict rotation — oath against survival, duty against desire, power against its price, loyalty against truth — assumes characters bound by vows, ranks, and bargains. Give those bindings text early: the oath spoken aloud, the price stated before it is paid, the duty with a name attached. Fantasy scenes fail most often when the magical solution dissolves the conflict instead of sharpening it, so draft each row with the rule that power must raise the cost inside the same scene it displays the wonder. Budgets near three thousand to four thousand words suit the genre's need for travel, council, and ritual without letting any single ceremony consume a chapter.
Romance lists run on vulnerability scheduled against pride. Closeness against fear governs the early book, honesty against pride the middle, commitment against freedom the late crisis, need against independence the resolution. Map your rows to that progression rather than letting the rotation fall where it may: if the planner assigns commitment against freedom to scene three, consider swapping conflicts between rows so the deep commitment pressure lands after the bond exists. Romance scenes need slow middles — meals, drives, work shifts, waiting rooms — so favor densities that keep per-scene budgets above two thousand five hundred words, and protect the second viewpoint, because a love story told entirely from one side halves its evidence.
Thriller lists run on time. Speed against safety, trust against doubt, control against chaos, truth against the cover story — every pairing implies a clock. State the clock in each scene: the flight boards, the file deletes, the source stops answering. Thriller scenes tolerate the shortest budgets on this page; two scenes or three per chapter keep units near two thousand to three thousand words, which matches the genre's quick-cut promise. The antagonist slot at three per chapter is native here — use it for visible opposition moves, not brooding, and make each one change the lead's options.
Mystery lists run on information discipline. Evidence against alibi, suspicion against loyalty, curiosity against danger, logic against misdirection — each scene should move exactly one salient fact across the table while misdirecting about one other. Track facts per row in a separate column of your own: what the reader learns, what the detective learns, what neither learns yet. One-scene-per-chapter densities endanger this discipline because long scenes leak extra facts; two per chapter keeps each discovery container-sized. The mystery engine's rule-out-a-suspect purpose deserves protection — never let it share a scene with planting a fresh clue, or the clearing and the implicating cancel each other.
Science fiction lists run on consequence. Progress against caution, orders against conscience, discovery against containment, crew against mission — the genre asks what a novum costs the people touching it. State the speculative rule once per scene at most, then spend the pages on who pays for it. Science fiction tolerates both long and short scenes depending on subgenre — idea-driven books breathe at three thousand plus, action-driven ones cut faster — so set density from your nearest shelf neighbors rather than from default habit, and use the second viewpoint for the crew-level cost of bridge-level decisions.
Story engines and point of view, chosen deliberately
The action-driven engine suits books where events pressure people: quests, chases, heists, survival, war. Its purposes escalate outward — each accomplishment makes the external situation worse or narrower — which carries readers who want momentum. Pair it with one or two scenes per chapter; three can fragment action into beats too small to stage. Watch for the engine's characteristic failure, the all-threat manuscript where confrontations stack without cost scenes between them. If your table shows confrontation purposes in adjacent rows, separate them with a cost or choice row so victories and losses register.
The character-driven engine suits books where people pressure events: family sagas, romances, coming-of-age, workplace dramas. Its purposes escalate inward — each accomplishment exposes more of the wound or bond — which carries readers who want recognition. This engine needs the longest scenes, because honesty and forgiveness cannot be rushed without reading as summary. Prefer one or two scenes per chapter, guard per-scene budgets above three thousand, and ensure the second viewpoint belongs to someone with genuine standing to demand change, not a witness who merely observes it.
The mystery engine suits books where knowledge is the terrain: cozies, procedurals, psychological suspense, investigative memoir shapes. Its purposes manage the reader's certainty dial — plant, deepen, rule out, hide, sharpen, turn over — which requires strict sequencing. Draft these rows in order and resist reordering on instinct, because the engine's rotation encodes a workable revelation curve. If you must reorder, move pairs of rows together so each planted clue keeps its eventual turnover within a reasonable distance.
Point-of-view density follows from the above. One per chapter gives steadiness and depth per voice but surrenders surprise and parallel action. Two per chapter is the default for good reason: alternating voices double the angles without doubling the cast management, and most commercial genres accept the rhythm. Three per chapter buys pace and opposition visibility at the price of fragmentation and a demanding antagonist thread. Choose by asking what the story withholds: if it withholds the opposition's moves, you need the third slot; if it withholds interior growth, you need longer single slots; otherwise take two and spend your energy on the rows themselves.
Questions writers ask when the list sits in front of them
How fixed is the order once generated. Fixed enough to draft from, loose enough to revise deliberately. The rotations encode a tested rhythm — purposes distributed, turns alternating — so reordering casually usually clusters similar endings together. When revision demands a move, relocate the purpose, conflict, and turn as a unit rather than swapping single columns between rows, and recheck the turn alternation afterward. Order is an instrument, not a monument, but retune it consciously.
Whether every scene truly needs all three assignments. In practice the scenes readers remember have all three, and the scenes writers cut in revision lack at least one. Purpose without conflict reads as errand; conflict without turn reads as argument that resolves nothing; turn without purpose reads as twist for its own sake. If a planned scene resists one of the three, that resistance is information — either the scene wants to merge with a neighbor or the missing element wants inventing before drafting starts.
What to do when two rows feel identical. They share a purpose by design, since six purposes cycle across twenty-plus scenes, but identical feeling means the conflicts or turns aligned too closely. Change the turn type of one, or swap its conflict with a distant row, and the doubling usually dissolves. Repetition in a plan is cheap to fix; repetition in finished pages is not, which is exactly why the table exists before the manuscript.
How the list relates to chapters as readers experience them. Readers feel chapters; writers manage scenes. A chapter break is a rest stop the list does not dictate — the planner groups scenes into chapters for budgeting and viewpoint rhythm, but you decide where the reader breathes. End chapters on turns that promise rather than resolve: decisions and discoveries close chapters well, while full reversals sometimes serve better mid-chapter where the fallout can play out immediately.
Whether short books and long books use the tool differently. Short books live or die on purpose discipline, because twenty scenes leave no room for a decorative one. Long books live or die on rotation management, because sixty scenes will repeat every purpose ten times and only conflict variety and turn alternation keep the repetition musical rather than mechanical. Set density so the per-scene budget matches your genre's patience, then audit the table for the failure your length invites.
How much invention the table leaves you. All of it that readers pay for. The planner never names a character, places a city, writes a line, or chooses an image. Two writers generating identical inputs receive identical scaffolding and produce unrecognizably different books, because purpose, conflict, and turn are the skeleton and every reader-facing pleasure is flesh the writer grows onto it. If generating the list felt like the creative work, that feeling passes with the first drafted scene — which is where the actual book begins.
From list to manuscript
A finished table is a promise, not a manuscript. Read it whole, revise the rows that feel doubled, confirm each viewpoint thread coheres, and check the budgets against your genre's patience. Then drafting starts, one scene per sitting where life allows, each sitting aimed at a single row's three assignments and word budget rather than at the abstract mountain of a book.
When the list is settled, open Co-Writer at /write and draft the scenes in order, keeping each scene's purpose, conflict shape, and turn type visible beside the draft. Work one scene at a time to its word budget before moving on, and let each turn close its own pages before the next purpose opens. The list stays the plan while the manuscript grows scene by scene, and the book that results carries the structure readers feel but never see.