Most abandoned books die around chapter twelve, and almost none of them die from bad prose. They die because the writer no longer knows what the next chapter is for.
That is a planning failure, but not the one people assume. The fix is not a bigger outline — plenty of thirty-page outlines have produced nothing. The fix is planning the right layer: a handful of questions the book is obliged to answer, in an order that makes each one urgent. Scenes are cheap and replaceable. Questions are the load-bearing wall.
Why detailed outlines fail as often as no outline
A scene list is a set of answers. You will change half of them by chapter twenty, because drafting teaches you things planning could not. When the answers change, an answer-shaped plan collapses and takes your confidence with it. A question-shaped plan survives, because the questions were never the part that needed to be right.
A plan made of scenes
"Ch. 4: Layla finds the letter in the attic and confronts her brother at dinner." Cut the attic and four later chapters stop making sense.
A plan made of questions
"Ch. 4: the reader must learn the brother lied, and must begin to doubt Layla's version of the fire." The attic is now one option among ten.
The same distinction runs through nonfiction. "Chapter 4: three case studies" is brittle. "Chapter 4: convince a sceptical reader that the obvious explanation is wrong" survives losing a case study.
Five decisions that make a plan
Work through these in order. Each one takes an hour at most, and you will have a plan by the end of a long afternoon.
The promise
One sentence, told to a stranger, that would make them want the book. Not a summary — a promise. "A widow discovers her husband's second family, and decides to protect them." "Why smart teams keep shipping the wrong product."
If you cannot say it in a sentence, you do not yet have a book; you have a subject. Every later decision gets measured against this line.
The reader
Name one real person who should read this. What do they already believe? What would make them put the book down at page thirty?
This decides your vocabulary, your pace, how much you explain and what you can assume. "Everyone" is not an answer; books written for everyone are written for no one in particular and read like it.
The spine
Five to nine beats, no more. For a novel: what changes, and what forces each change. For nonfiction: the argument's steps, in the order a doubtful reader will accept them.
E. M. Forster's old distinction is the test. "The king died and then the queen died" is a list. "The king died, and then the queen died of grief" is a spine. If your beats are joined by and then rather than therefore or but, you have a sequence, not a structure.
The chapter jobs
Now expand each beat into chapters, and give every chapter one job in one line: what the reader knows, feels or believes at the end that they did not at the start.
This is the whole method in one habit. A chapter with no job is the chapter you will be unable to write in four months, and the chapter a reader skims.
The end condition
Write the last page's state, not its wording. What is true then that was not true at the start? What has the reader given up believing?
Knowing the destination is what lets you cut. Without it, everything looks potentially relevant and the draft swells to 140,000 words of maybe.
The one-page plan
Keep this page open beside the draft. Reread it before every session; it takes twenty seconds and prevents the drift that kills chapter twelve.
Promise
One sentence. The reason a stranger would pick this up.
Reader
One named person. What they believe now. What would lose them.
Spine
Five to nine beats, each joined to the next by therefore or but.
Chapter jobs
One line per chapter: what changes in the reader by the end of it.
End condition
What is true on the last page that was not true on the first.
Out of scope
Three things this book will deliberately not cover. This section saves more time than the rest combined.
If you want a named method
The five decisions above are deliberately method-agnostic. If you prefer a system, these four are the ones working writers actually use — and each one is really a different answer to the same question of how much to decide before drafting.
The Snowflake Method
Randy Ingermanson's approach: one sentence, expanded to a paragraph, then a page, then scenes. Best if you like growing a plan by controlled steps rather than staring at a blank outline.
Save the Cat
Blake Snyder's beat sheet, borrowed from screenwriting into fiction. Highly prescriptive, which is its strength for commercial plotting and its weakness if your book resists a standard shape.
Truby's twenty-two steps
John Truby builds structure from the protagonist's moral need rather than from plot slots. Slower to learn, better at producing a story that means something.
Discovery writing, with a safety net
Plenty of good novelists plan almost nothing. If that is you, keep the promise and the end condition anyway — two lines. They are what let a discovered draft be revised rather than restarted.
Choose one and stop reading about the others. Comparing methods is the most comfortable way to not write a book.
What to do when the plan breaks
It will, and that is not a failure. Drafting is how you discover what the book is about; a plan that survives untouched usually means you were not paying attention. Three common breakages and their fixes:
A character refuses the plot
Usually the plot is wrong, not the character. Keep the beat's job and find a different event that does it.
The middle sags
Look for consecutive chapters whose jobs are near-duplicates. Sagging middles are almost always repetition wearing different clothes. Merge them.
The promise changed
If chapter fifteen is more alive than your one-sentence promise, rewrite the promise and re-derive the remaining chapter jobs from it. Do it deliberately, in the plan, rather than drifting.
Related reading: why structure must come before detail, and how to write a nonfiction book if your project is argument-driven.
A plan you can see is a plan you follow
Plans die in a separate document you stop opening. In Writer Atlas the outline is the book: the same tree you plan in is the one you write in.
Outline and draft in one tree
Write the chapter jobs first, then fill each one in place.
Reorder without breaking anything
When the spine changes, drag the chapters to match it.
Characters and research beside the text
The facts of your world stay one click from the paragraph that needs them.
Version history
Change the plan boldly; the earlier shape is still there if you were wrong.
Questions about planning a book
How do I plan a book before writing it? +
Make five decisions rather than a scene list: the one-sentence promise, the specific reader, a spine of five to nine beats joined by cause rather than sequence, a one-line job for every chapter, and the end condition. That fits on one page and takes an afternoon. Keep the page beside the draft and reread it before each session.
How long should a book outline be? +
One page for the plan, plus one line per chapter. Longer outlines are not more rigorous; they are more brittle, because they commit to answers the drafting process will overturn. If you find yourself writing scene-level prose in the outline, you are writing the book in the wrong document — start the draft.
Should I outline or discovery-write? +
Both work, and published novelists are split roughly down the middle. The real difference is when you pay the cost: outliners pay before the draft, discovery writers pay in revision. Pick by which you tolerate better. Even committed discovery writers benefit from two lines — the promise and the end condition — which turn a shapeless draft into a revisable one.
How many chapters should a book have? +
There is no correct number, but there is a useful derivation: chapters follow from jobs. If your spine has seven beats and each needs three or four moves, you have a book of roughly 25 chapters. Working from a target number backwards produces padding, which readers feel as slowness even when they cannot name it.
What if I abandon the plan halfway through? +
Then the draft taught you something, which is what drafts are for. Do not keep writing against a dead plan and do not start over: update the promise, re-derive the remaining chapter jobs from it, and continue from where you are. A revised plan takes an hour; a restarted book takes a year.
Write the one page today
Promise, reader, spine, chapter jobs, end condition, out of scope. Then start chapter one tomorrow.
Start your book free