How to Outline a Novel: 5 Methods Compared (Snowflake, Save the Cat, Three-Act, Character Arcs, Pantsing-Plus)

I have used all five below on real manuscripts. To make the comparison fair, I will run each on the same premise:

Mara, a royal archivist who has spent ten years erasing her family's name, must forge a succession document before the coronation, but the only person who can authenticate it is the brother she betrayed.

The comparison at a glance

1. The Snowflake Method

Randy Ingermanson's method: start with one sentence and expand it in steps, each step doubling the detail. Sentence, then paragraph, then a page per character, then a page per paragraph-sentence, then a scene list.

Mara in snowflake form.

Step 1, one sentence: An archivist must forge a royal document with help from the brother she destroyed.

Step 2, one paragraph (five sentences: setup, three disasters, ending): Mara has spent a decade hiding her family's treason inside the archive she now runs. When the heir dies, the regent orders her to forge a succession document, and the only authenticating seal belongs to her exiled brother Teo. She finds Teo dying and unwilling. Forced to choose between saving the crown and saving him, she realises the regent intends to burn the archive either way. She authenticates the truth instead of the forgery, losing the archive and her anonymity, and keeps her brother.

Step 3, a page per major character: their goal, motivation, conflict, epiphany, and a paragraph of their storyline. Teo's page reveals that he does not want revenge; he wants her to stop hiding, which reshapes the ending.

Step 4 expands each sentence of step 2 into a paragraph. Step 5 is a scene list with POV and a one-line "what changes." Step 6 is the draft.

What it is good at: you cannot get to step 5 with a hole in the plot; the expansion exposes holes. Ensemble books benefit because every character gets a step-3 page.

What it is bad at: books whose engine is voice or atmosphere. Snowflaking a quiet novel produces a plot it does not want. It also front-loads weeks of work before you write a sentence of prose, which for some writers is a way of not writing.

2. Save the Cat (the novel version)

Blake Snyder's screenwriting beat sheet, adapted by Jessica Brody for novels. Fifteen beats at set percentages of the manuscript: Opening Image (0–1%), Theme Stated (5%), Setup, Catalyst (10%), Debate, Break into Two (20%), B Story, Fun and Games, Midpoint (50%), Bad Guys Close In, All Is Lost (75%), Dark Night of the Soul, Break into Three (80%), Finale, Final Image.

Mara's beat sheet, abbreviated:

• Catalyst (10%): the heir dies; the regent's order arrives.
• Break into Two (20%): Mara leaves the archive for the first time in ten years to find Teo.
• Midpoint (50%): false victory: Teo agrees, and the seal works. Then she learns the regent plans to burn the archive.
• All Is Lost (75%): Teo collapses; the coronation is tomorrow; the forgery is in the regent's hands.
• Break into Three (80%): Mara understands that the archive she has protected is the evidence that could stop the regent. She has been hiding the weapon.
• Finale: she authenticates the truth publicly.

What it is good at: pacing. If your midpoint is at 30%, you will feel it in the beat sheet before your readers feel it in the book. Commercial fiction readers have been trained on this rhythm and it satisfies them.

What it is bad at: it tempts you to hit beats rather than earn them. "Theme Stated at 5%" produces a lot of wise bartenders. And the percentages assume a single-protagonist, three-act shape; multi-POV epics and serials have to bend it hard.

3. Three-act structure

The oldest and loosest. Act one (roughly the first quarter) establishes the world and ends with a point of no return. Act two (the middle half) escalates, with a midpoint reversal. Act three resolves.

For outlining, I add two things to make it work: a midpoint that reverses the protagonist's understanding, and a pinch in each half of act two where the antagonist's pressure is felt directly.

Mara in three acts:

• Act one: the archive, the death, the order. Point of no return: she takes the seal-box and leaves.
• Act two, first half: the journey, the checkpoint, the brother's house. Pinch: the regent's men search the archive in her absence. Midpoint: Teo agrees; the seal works; the regent's plan for the archive is revealed.
• Act two, second half: Mara tries to have it both ways. Pinch: the regent lets her know he knows about the family name. Low point: Teo collapses; she has the forgery and no leverage.
• Act three: coronation. The truth over the forgery. Cost and reward.

Then I break each act into 8 to 15 scenes with a scene card each. That card format is the one that matters for any method, so here it is:

Scene card: POV / place / time. Goal. Obstacle. Turn (what changes). Exit state. New fact for the bible.

What it is good at: everything, adequately. It is a skeleton you cannot really get wrong, and it imposes no genre rhythm.

What it is bad at: the middle. "Act two: things get worse" is where first novels sag, because the method does not tell you how. The pinches and midpoint above are my patch; Save the Cat is essentially a more prescriptive patch for the same problem.

4. Character-arc outlining

Outline the internal change first, then find the external events that force it. K.M. Weiland's approach is the fullest treatment. The core: the protagonist begins believing a Lie about herself or the world, the plot tests that Lie, and she ends by accepting a Truth (positive arc), rejecting it (negative arc), or having it confirmed (flat arc, where she changes others).

Mara's arc:

• Lie: "If no one knows who I am, no one can hurt me." (Her ten years of erasure.)
• Truth: "Hiding is what let them hurt everyone else."
• Ghost (the wound): she gave evidence against her brother to save herself.
• Plot points are chosen to press on the Lie: each act-two scene either rewards the hiding (short-term) or punishes it (long-term), so the pressure builds toward a choice.

Then, and only then, the external outline: what events would force an archivist who hides to stand up in a coronation hall and say her name?

What it is good at: endings that land. Because the climax is defined as the moment the Lie breaks, you always know what the last chapter must do. Literary and romance writers, whose plots are often internal, get a proper structure without a car chase.

What it is bad at: mystery and plot-puzzle books, where the engine is information rather than transformation. And it can produce books where the theme is visible from space, so keep the Lie off the page.

5. Pantsing-plus (discovery writing with a net)

Pure discovery writing, "pantsing," means starting with a character and a situation and finding out what happens. It produces vivid first drafts and a lot of abandoned manuscripts. Pantsing-plus is the compromise I recommend to every discovery writer who has not finished a book:

1. Write the ending first: one paragraph. It will change, but you need somewhere to aim.
2. Write the next three scenes as scene cards. Only three.
3. Draft them.
4. Update the bible. Write the next three cards.

That is the whole method. Never more than three scenes ahead, never zero. The ending paragraph is the net.

Mara, pantsing-plus, day one:

• Ending: Mara says her real name in the coronation hall. Teo is alive to hear it. The archive burns anyway.
• Next three: (1) Mara in the archive, the courier at the door. (2) The regent's order; she opens the seal-box and it is empty. (3) The river road at dusk.

Day four, after drafting: the empty seal-box turned out to be the whole engine, and Teo has the seal. Next three cards. And so on.

What it is good at: momentum and surprise. Discovery writers produce their best material when they do not know what is coming. This method keeps that while removing the "I have 60,000 words and no idea what this is" failure.

What it is bad at: mysteries and any plot that depends on a planted clue paying off. You can retrofit clues in revision, but it is expensive. Also weak for serials with a fixed release cadence, unless you keep the three-card buffer well ahead of publication.

Choosing, and combining

Most working novelists I know run a hybrid, and the hybrid is usually: character arc for the spine, three-act for the shape, scene cards three to five ahead during drafting. Snowflake and Save the Cat are what you reach for when a book is not working and you need to find out why.

A few honest matches:

• First novel, no idea where to start: three-act with the pinches and midpoint, plus the scene card format.
• Thriller, romance, cozy mystery, anything shelved by genre: Save the Cat, then loosen it.
• Big cast, big plot, secondary world: Snowflake, and keep the step-3 character pages in your story bible.
• Quiet literary, romance, coming-of-age: character arc.
• You have never finished anything and you hate outlines: pantsing-plus. Non-negotiable ending paragraph.
• Web serial: pantsing-plus with a bigger buffer (ten cards, not three), because you cannot revise chapter 12 after it is published. See the serial guide.

Where AI helps and where it does not

An AI is a competent reader of outlines and a poor author of them. What it does well: paste your scene list and ask which scenes lack a turn, where the protagonist stops choosing, which character vanishes for fifty pages, where the midpoint actually falls. What it does badly: generating the outline from the premise, which tends to produce the average outline of every book with that premise. Use it as a structural editor after you have made the decisions.

Try it in SparkLore

If you want to see the "competent reader" mode in action before committing to a method, the free AI plot generator will take a premise and return a three-act skeleton with a midpoint and pinches, which is a decent starting draft to argue with. Inside a SparkLore project, the outline lives beside the bible as scene cards, and each card can be drafted from with the relevant characters attached. Create a project and write the ending paragraph first.