Novel craft and story design

How to outline a novel without killing it

August 16, 2026 · The StoryProp team

Outline a novel by writing down only the decisions you would defend, and leaving everything else on a separate list of wishes. Scale is what makes the distinction necessary: adult commercial fiction runs 80,000 to 100,000 words in the agent-stated targets The Manuscript Academy published for 2026, which at the 2,000-to-4,000-word chapter range Reedsy publishes as the working convention is somewhere between 20 and 50 chapters. No document written before the draft can honestly specify thirty chapters. It can specify the dozen or so facts every one of those chapters will have to obey — and that shorter document is the one that survives contact with the pages.

The argument against outlining is usually framed as a matter of temperament: planners plan, discoverers discover, and forcing either to work the other way kills the book. But most outlines that die were not killed by temperament. They died of a category error. The writer recorded the wrong things — guesses dressed as commitments — and then had to choose between obeying a document that was mostly guesswork, or abandoning it the first time the draft disagreed. The alternative is to change what an outline is for. An outline is not a summary of the novel you hope to write; it is a record of the decisions you have actually made. Learn to tell those two things apart and the question of how to outline a book stops being a fight about method and becomes bookkeeping — a practice planners and discovery writers can run in opposite directions and still get right.

Why outlines fail: filling in a template is not deciding

The common way to outline is to take a named scaffold and fill in the blanks: the three-act structure, the fifteen beats of Blake Snyder's Save the Cat sheet, the expanding summaries of the Snowflake Method, the departure-and-return arc of the Hero's Journey — surveyed side by side in story structure frameworks. The scaffold is rarely the problem, and it helps to see plainly what each one actually hands you. Save the Cat hands you fifteen labelled slots in a fixed running order. The Snowflake hands you a procedure for expanding one sentence into a paragraph, then a page, then a synopsis. Three acts hand you two turning points and a rough sense of where they fall. Every one of those is a shape for answers rather than an answer. The filling is the problem.

A blank demands an answer, so you invent one: something plausible, something that sounds like a novel. Suppose the book you're outlining is The Quiet House: Mara comes back to her late mother's house in the small town she swore off, and Daniel has been keeping the place and its accounts. Your firm decision that Daniel has known about the mother's debt since before the funeral sits on the same line, in the same font, as a midpoint you invented on a Tuesday to complete the form.

Then drafting begins, and the draft — which is smarter than the outline, because it has to survive scene by scene — exposes the invented answers. At that point most writers do one of two bad things. They obey, grinding out scenes that exist only because a slot said so, which is where outlines get their reputation for embalming. Or they revolt and throw out the whole document, including the two or three real decisions buried in it, and the book quietly loses its spine. Either way, the outline did not kill the book. The indistinguishability did.

A wishlist is not a checklist

The repair is a single distinction. Sort every line of your outline into one of two piles. The checklist holds what is settled: decisions you have made and will protect, facts other scenes are already leaning on. The wishlist holds everything you merely like: images, maybes, scenes you haven't committed to, endings you're auditioning. In The Quiet House, the checklist might read: Mara returns because of Daniel's letter; the chipped cup was her mother's, and Mara knows it; Daniel has known about the debt since before the funeral. The wishlist might read: maybe a storm during the confrontation; maybe the neighbor saw more than she admits; maybe the last chapter is a sale notice.

The difference is obligation. A checklist item obligates every scene that touches it: if Daniel has known for years, his kindness in chapter two reads as management, and the confrontation must be about concealment rather than discovery. A wishlist item obligates nothing. It waits, costs nothing, and can be discarded without a ripple. Trouble starts when the two piles share a page with identical formatting, because then the draft cannot tell a law from a mood.

Much advice about how to organize ideas for a book treats the problem as sorting — by theme, by character, by chapter. Sort however you like; the sort that matters is by status: settled or suggested. Keep the piles visibly different — two documents, two colors, a mark in the margin — and promote by one test: would you revise finished scenes to protect this line? If yes, checklist. If not — however much you love it — it's a wish, and it should look like one. Run that test down a filled-in beat sheet and most of the fifteen entries will demote to wishes, which is useful news, not bad news.

The minimal outline: premise, load-bearing facts, an ending's rule

Once decisions are the unit, a working outline turns out to be small. Three components cover most novels. First, a premise sentence — who wants what, against what — because it is the one decision everything else consults. Second, a handful of load-bearing facts: the facts multiple scenes stand on. These are usually facts of timeline (the funeral was in March; Mara left eleven years ago), of knowledge (Daniel knows about the debt; Mara doesn't, yet), and of objects (the chipped cup survives every purge of the house, and the book must eventually say why).

Third, the ending's rule, if you know it. Not the ending scene — the rule the ending will enforce. You may have no idea whether the last chapter happens in the kitchen or out on the county road, and still know the law it must obey: the truth about the house costs Mara the house. Rules travel; scenes don't. An ending staged a year early is almost always wrong in its particulars, but a rule can be honored by a dozen different stagings.

Everything else — the storm, the neighbor, the sale notice — stays on the wishlist, growing as fast as you can dream. That is the trick of outlining without killing anything: a short checklist held firmly, a long wishlist held loosely. And the disproportion is the point. Ten settled facts have to hold up a book of the size agents actually buy.

How to plan a novel: up front, as you go, or both

If you plan before drafting, keep the habit — just run the promotion test on the result. Fill in your template if it helps you think, then ask, line by line, which entries you would actually protect. Most will demote to wishes; that is the correct outcome, not a failure of planning. What survives is your real outline — shorter and stronger than the document you started with.

If you discover as you draft, outline behind yourself. At the end of a session, ask what the pages settled — Mara lies easily to strangers and badly to Daniel; the house carries a second mortgage — and write it down as checklist. This is outlining as a ledger rather than a blueprint: by the middle of the book you hold exactly the outline you needed, made entirely of decisions the draft has already tested. And if you are earlier than any of this — an idea, no pages — that is a different problem, covered in what to do when you have an idea for a book.

When the outline should change mid-draft — and what it costs

Sooner or later the draft contradicts the checklist, and this is not failure; it is the system working. Say the outline holds Daniel's secret until the last third, and in chapter four he simply tells her — the scene went there, and it is the best scene you've written. The draft may well be right. But “the draft felt right” is the beginning of the decision, not the end of it.

You have two honest options, and each has a price. Keep the draft's version, and you must reprice everything downstream that leaned on the secret: scenes premised on Mara's ignorance, tension the concealment was paying for, a confrontation that no longer contains a revelation. Or keep the outline: cut the confession and work out what the scene was actually reaching for, because a scene that good is usually right about something even when it is wrong about timing. The cost, either way, is the audit. What kills novels is the third option — keeping both, deciding nothing, and letting manuscript and outline drift apart until neither is telling the truth.

Keeping the outline honest with the manuscript

An outline is only useful while it agrees with the pages. Left alone, it decays: the manuscript moves, the document doesn't, and after enough divergence you are consulting fiction about your fiction. The maintenance is small if it is regular. End each session with two questions: did anything become settled, and did anything settled get contradicted? Promote the first kind. For the second, don't silently delete the old decision — mark it superseded, because scenes written under the old rule still carry its assumptions.

The habit long predates the software. Anthony Trollope's An Autobiography, published in 1883, records the tally he kept week by week: a self-imposed average of forty manuscript pages of 250 words — 10,000 words a week — written in the three hours before his Post Office day began, with a recorded range from 20 pages in his slowest week to 112 in his fastest. Of Lady Anna, drafted at sea at 66 pages a week for eight weeks, he wrote that every word was counted. The counting was the discipline, and it is the same discipline applied to decisions rather than to words.

Bookkeeping is what software should be for. An AI outline generator for novels will hand you a chapter list on request, and there is nothing wrong with generated beats, provided everyone knows which pile they belong in: suggestions you haven't agreed to. StoryProp, an AI writing agent built for novel-length work, holds that line. The project conversation is permanent and ordered, so a suggestion stays visible without becoming binding, and when you confirm a decision it is promoted to a record — canon, characters, story shape — that the agent reads before drafting anything. When a new passage contradicts a settled fact, the contradiction is surfaced rather than silently smoothed over. What you agree to gets written down; what you don't stays a suggestion.

Nothing here asks you to convert from one camp to the other. Your decisions and your wishes only have to stop wearing the same clothes. Keep the checklist short and defended, the wishlist long and cheap, reconcile outline with manuscript as you go, and the document stops being the thing that kills the book. It becomes what it should have been from the start: the novel's memory, not its cage.

An outline that records decisions needs a place where decisions stay recorded. StoryProp gives them one — confirm a choice and it becomes part of the project's record, read before every draft; float an idea and it stays safely a suggestion until you say otherwise.

← All posts