How to Write a Book, Step by Step: The Nine Stages That Actually Finish One

So this guide is a sequence, not a list of tips. Nine stages, each with the decision it exists to force and the specific way it goes wrong.

Key takeaways

• A book is a sequence problem, not a talent problem. Every stage below fails when it is attempted before the one above it.
• The premise sentence is a gate, not a formality. If you cannot say what the book is about in one sentence, the second act will not hold.
• Choose a structure before drafting. Not to be enslaved by it, but because the structures encode where the energy has to change.
• The midpoint has to reverse something. This single mechanical fact prevents most sagging middles.
• Set the daily target at a normal day, not a good one. Adherence beats intensity across a project measured in months.
• Revision is a sequence too — structure, then scene, then line. Doing them in the wrong order wastes months polishing prose you will cut.

Stage 1 — Earn the premise sentence

Before anything else, write one sentence that says what the book is about. Not a theme, not a vibe, not a paragraph. One sentence containing a protagonist, a want, and the thing standing in the way.

A grieving cartographer discovers her dead husband mapped a coastline that does not exist, and following it means admitting he had a life she was never in.

Being unable to write that sentence is not a warm-up problem. It is diagnostic. A book whose premise cannot be stated in one line usually has no engine, and the absence shows up 30,000 words later as "I don't know what happens next" — which is almost never a failure of imagination, and almost always a premise that never contained a conflict in the first place.

How this stage fails: writers substitute a situation for a premise. "A woman moves to a coastal town after her husband dies" is a situation. Nothing in it opposes anything. Add the opposition and it becomes a book.

The test: say the sentence out loud to someone who does not read your genre. If they ask "and then what?", it works. If they say "that sounds nice", it does not.

Stage 2 — Decide who this is for, and what you are promising them

Every genre is a contract. Romance readers arrive expecting two people to end up together and will not forgive you for a clever subversion. Mystery readers expect a solution they could have reached from the clues on the page. Thriller readers expect a clock they can hear ticking.

This is not commercial cynicism. It is the difference between surprising a reader and cheating them.

Write down, explicitly: who reads this, and what do they turn up expecting? Then keep it somewhere you will see it during revision, because the promise you make in chapter one is the thing you will be graded on in chapter thirty.

How this stage fails: the book quietly becomes two genres at once and satisfies neither. A literary novel with a murder in it must decide, by the midpoint, whether the murder is the engine or the wallpaper.

Stage 3 — Pick a structure, then stop arguing about structures

Story structures are not formulas. They are maps of where the energy has to change in a story that holds attention for 90,000 words, derived by different people looking at the same phenomenon from different angles.

The main ones, and what each is actually good at:

Structure	Beats	Best for	Its one big idea
Three-Act	11	Anyone, any genre, first books	Setup, escalation, resolution — the irreducible shape
Save the Cat	15	Commercial fiction, plotters	Exact percentage markers; the most precise map available
Hero's Journey	12	Fantasy, adventure, myth	Departure and return; the protagonist comes back changed
Seven-Point	7	Thrillers, plot-driven work	Start from your ending, build the book backwards
Fichtean Curve	4+	Horror, suspense, literary	No slow opening — begin in crisis, escalate crisis by crisis
Freytag's Pyramid	5	Tragedy, classical drama	The fall matters as much as the rise
Story Circle	8	Memoir, character-driven fiction	You, need, go, search, find, take, return, change
Snowflake Method	10 steps	Planners who freeze at a blank outline	One sentence expands to a scene list in ten passes

If you are writing a first book and have no strong instinct, use Three-Act. It is free of ideology, it underlies most of the others, and it will not make you feel like you are filling in a form.

We compared these properly, including where each one is the wrong tool, in which story structure you should actually use. And if you want the full beat sheets — every beat, the job it does, the word count it lands at — CipherWrite publishes all seventeen of its book outline templates as readable pages, no account required to read them.

How this stage fails: structure tourism. Reading nine structure books is not outlining. Pick one in an afternoon.

Stage 4 — Outline to a spine, not to a blueprint

Here is the compromise that survives contact with reality. Before drafting, know:

1. The ending. Not in detail — in outcome. Who wins what, and at what cost.
2. The midpoint reversal. What changes at the halfway mark such that the second half is a different pursuit.
3. The first three scenes. Concretely enough to write tomorrow morning.

That is the spine. Everything else can stay loose, and for many writers should.

The reason the midpoint gets singled out: the sagging middle is the single most common reason books die, and its mechanical cause is nearly always a midpoint that reverses nothing. The protagonist reaches the halfway mark wanting the same thing, pursuing it the same way, knowing the same facts. Of course it sags. Nothing has happened.

Give the midpoint a job — a revelation that reframes the goal, a victory that turns out to cost more than it gained, a false defeat that forces a new method — and the second act stops being a corridor.

How this stage fails: the outline becomes the book. Some writers outline for eleven months because outlining feels like progress and cannot be judged. Cap it. Two weeks, then draft.

Stage 5 — Build the smallest story bible that stops you stalling

You do not need a 40-page world encyclopaedia. You need the things whose absence will stop you mid-sentence at 11pm:

• Characters: name, what they want, what they need (not the same thing), one physical detail you will otherwise contradict.
• Places: enough to be consistent.
• Rules: if there is magic, technology or a legal system doing plot work, write down what it costs. Powers without costs produce plots without tension.
• Timeline: a plain list of what happens on what day. This is boring and it saves weeks in revision.

How this stage fails: world-building becomes a hobby that replaces writing. If you have a language and no chapter three, that is what has happened.

Stage 6 — Draft badly, on a schedule you can actually keep

This is the longest stage and the one everything else exists to protect.

Set your daily target at a normal day, not your best day. The writer who commits to 2,000 words a day writes 2,000 words for nine days and then nothing for three weeks, because the target has become a thing to fail. The writer who commits to 400 finishes.

Anchor the session to a real cue, not a time. Psychologist Peter Gollwitzer's work on implementation intentions is unusually robust for behavioural research, and the finding is blunt: "I'll write more" fails; "if it is 7am and the coffee is made, then I write 200 words" succeeds substantially more often. The cue has to be something that already happens in your day without effort.

Never end a session at a natural stopping point. Stop mid-scene, and write the next scene's name at the bottom of the file — "Tomorrow: Maya finds the letter, 400 words." An unfinished sentence is a much easier thing to return to than a blank page, and the open loop does real cognitive work overnight.

Ban editing during the draft. Not because prose does not matter, but because you cannot polish something whose shape is not settled, and rewriting chapter one for the ninth time is the most common way a book dies while its author believes they are working on it.

How this stage fails: four days of silence. That is the number worth memorising. Two days off is a normal life. Four days is where the momentum breaks, the guilt starts, and the file stops being opened. If something is going to intervene, that is when.

That specific problem — the silence, not the craft — is what CipherWrite's accountability agent The Finisher was built for, and we wrote up the whole behavioural argument in why most novels die in the middle.

Stage 7 — Diagnose the stall before you treat it

When you do stall — and you will — the instinct is to reach for motivation. This is nearly always the wrong medicine, because writers systematically misname their own problem. "I'm blocked" is a feeling. Underneath it is a mechanism, and the mechanisms have different treatments:

What it feels like	What it usually is	What actually fixes it
"I don't know what happens next"	The midpoint never reversed	Plot the next three scenes only — not the outline
"I hate everything I've written"	Editing during drafting	Ban revision for one week; the draft is allowed to be bad
"I've lost who this person is"	Character has a want but no need	Write their worst memory; find the lie they believe
"Life got heavy"	Target set for a good day	Shrink to 50 words a day and pause the deadline out loud
"I'm afraid of the ending"	Ending is undecided, not unwritten	Write the ending now, out of order
"I don't know why I'm doing this"	Lost the original reason	Reread what you wrote on day one about why this book mattered

Notice that only one of those is solved by trying harder.

Stage 8 — Revise in the right order: structure, scene, line

Revision has a sequence too, and doing it out of order is how people spend two years on a second draft.

Pass one — structure. Reverse-outline the draft: one line per scene, what changes in it, whose scene it is. Then look at the list, not the prose. Scenes where nothing changes get cut or merged. This pass will feel brutal and it will save the book.

Pass two — scene. Every scene needs a goal, an obstacle and an outcome that differs from the goal. Scenes that end where they began are transport, not story.

Pass three — line. Now, and only now, the prose. Filter words ("she saw", "he felt", "she realised") pushed out of deep POV. Dialogue doing work rather than exchanging information. Adverbs earning their place.

Pass four — consistency. Eye colours, day counts, whether the storm was Tuesday. Tedious, mandatory, and the thing human readers are worst at across 90,000 words they have already read forty times.

How this stage fails: starting at pass three. Polishing sentences in scenes you are going to delete is the single largest waste of hours in amateur revision.

Stage 9 — Package it for whoever decides next

Whatever the route — agent, small press, self-publishing — the last stage is turning a manuscript into a submission. A query letter that states the premise, the stakes and the word count without coyness. A synopsis that gives away the ending, because that is what a synopsis is for. Comparable titles from the last three years.

If you are self-publishing, the equivalent is the cover, the blurb and the first ten pages, because those are the only three things a browsing reader will ever see.

The shortest possible version

1. One sentence that contains a conflict.
2. Know your reader's expectations, and honour them.
3. Pick a structure in an afternoon.
4. Know the ending, the midpoint reversal, and the next three scenes.
5. Write down only what will otherwise stop you at 11pm.
6. Small daily target, anchored to a real cue, stopping mid-scene.
7. When you stall, name the mechanism before reaching for motivation.
8. Revise structure → scene → line → consistency.
9. Package it honestly.

Where software helps, and where it doesn't

Software cannot make you write. It can remove three specific frictions: losing the thread of your own outline, losing the story bible, and losing the habit after a bad week.

If you want those handled in one place, CipherWrite holds the outline in a Blueprint separate from the manuscript, keeps a story bible that fills itself as you work, and encrypts all of it in your browser so the draft is not sitting in plaintext on somebody's server. Its AI Writing Coach runs the stage-7 diagnosis against your actual pages — the full root-cause read and your first graded exercise are free, which is enough to judge it.

One caveat worth stating plainly, because we would rather you learn it here than feel misled later: the encryption claim applies to the editor, the outline and the story bible. The AI coaching tools necessarily send the text you submit to a model, and CipherWrite says so on the page rather than blurring the two.

For a wider comparison of what to write in, see our ranked list of the best writing apps for authors. Coming off Scrivener in particular? Seven alternatives, and when to keep it.

Nothing above requires talent you either have or don't. It requires a sequence, a small target, and the willingness to be told which of the six stalls you are actually having. That is a much more tractable problem than "am I a writer", and it has the advantage of being true.