Start, sustain, and finish

I have an idea for a book. Now what?

August 16, 2026 · The StoryProp team

You have the idea. It has been with you for months, maybe years — surfacing in the shower, rearranging itself on long drives, quietly growing new rooms when you are not looking at it directly. Maybe you have told one person about it; maybe no one. And every time you sit down to begin, the idea that felt so alive an hour ago goes silent, and you end up tidying the desk instead. So — now what?

Two things, and both fit on a single page. Write down what your idea promises a reader, in one or two sentences, and treat those sentences as decided. Then put a person in a concrete place doing one small, specific thing, and write it. That is the entire starting kit — two sentences and a scene, set against a book that will not even count as a novel until it passes 40,000 words. Not an outline, not character sheets, not a world bible; those come later, and front-loading them is the most reliable way to never start.

That 40,000 is the one hard, published line anyone in this business has actually drawn. SFWA sets it in the Nebula Awards rules, effective January 2025: a novel is 40,000 words or more, a novella 17,500 to 40,000, a novelette 7,500 to 17,500, a short story anything under 7,500. Trade publishing asks considerably more of a debut — The Manuscript Academy's 2026 word-count guide, sourced to agent Noelle Falcis Math of Transatlantic Literary Agency, puts debut literary fiction at 80,000 to 95,000 words and fantasy at 90,000 to 120,000 — and at the Steve Laube Agency's working conversion of 300 words to a printed page, those come out around 267 to 400 pages.

None of those numbers is today's problem. They are the size of the object, and it is worth seeing the size before you agree to it rather than eight months in. Today's work is still two sentences and a scene.

The gap between having an idea for a book and writing its first page is where most books end — invisibly, before a single sentence exists. An unwritten idea is perfect. It holds every possible version of itself at once: the literary one, the propulsive one, the one your favorite author would admire. A first sentence collapses that. It commits you to one version and forecloses the others, and some part of you knows it. The paralysis is not laziness. It is protectiveness.

Can anyone write a book? Yes — there is no permission slip

Listen to what the question actually asks: am I allowed? It is a request for a permission slip, and there is no office that issues them. Nobody licensed the writers you admire. They started unqualified, because everyone starts unqualified; the qualification is conferred by the writing itself, in arrears.

The question that does have work in it is the next one: are you curious enough about your idea to find out what it actually is? A book is not the transcription of a finished object in your head. It is the discovery of what the object in your head turns out to be, and the result routinely surprises its author. If you can describe your idea to a friend and feel the temperature of the room change — even slightly, even once — you have enough to begin. That reaction, not a credential, is the raw material.

Develop the idea by asking what it promises

Most advice on how to develop a book idea points you outward: research the market, study comparable titles, validate the concept. Point inward first. An idea becomes a premise the moment you can say what it promises a reader — what experience they are being offered in exchange for their attention. Until you can say it, you have material for a book rather than a book.

Take an example worth staying with. “A woman returns to her hometown after her mother dies” is a situation, not a premise; it promises nothing. Now try: “Mara comes home to sell her mother's house and finds the kitchen exactly as it was, down to a chipped cup she watched her mother throw away thirty years ago — and the question of who kept it becomes the question of who her mother actually was.” That version makes an offer. Something hidden will surface. A daughter's picture of her mother will be revised. A reader who begins that book knows roughly what they were promised — and the writer knows what she owes.

Here is the test: say the promise out loud, in one or two sentences, to a person or a wall. If what comes out is a list of themes — grief, memory, small towns — you have an atmosphere, not yet a premise. Keep pressing on the same three joints: who is this about, what do they want, what stands in the way. When the answer holds its shape spoken aloud twice, you are done developing. Really. Everything past this point is stalling dressed as diligence.

The only two things you need to start

The first is a locked premise — those one or two sentences, written at the top of the page and treated as decided. Locked does not mean the book cannot surprise you; it will, constantly, at the level of scenes and people. Locked means you stop renegotiating the foundation. Every morning spent reopening the premise — should it be sisters instead, should it be a hotel, should it be present tense — is a morning the book does not get written. What a working writer has that a stalled one lacks is not confidence. It is fidelity to a decision.

The second is an opening worth reacting to. Not a good opening — nobody has one of those yet — but a concrete one: a person, a place, an object, a small motion. Mara in her mother's kitchen, washing that cup even though it has sat in a closed cupboard for a year. Concreteness is what makes reaction possible. You can have opinions about a woman washing a cup — wrong room, wrong object, she would never wash it, she would wrap it in newspaper — and opinions are how a book begins. That is the whole kit. Everything else can be picked up while moving.

What not to prepare

Most lists of steps to writing a book front-load preparation: complete an outline, fill in character questionnaires, build the world bible, and then — somewhere around step seven — write. For a first book, that ordering is where the project usually dies, because preparation feels like progress while risking nothing. Every document you produce about the book postpones the moment a real sentence can disappoint you. Prepare less than feels responsible.

You do not need a full outline. Whatever outlining will eventually do for you — and it can do a great deal — it operates on decisions you have not made yet. The load-bearing choices of a first draft are discovered in scenes, not planned above them. A paragraph of intentions is plenty for now.

You do not need character sheets. A questionnaire will yield your protagonist's eye color, blood type, and favorite breakfast, and none of it will survive contact with a scene. Character is what a person does under pressure, and there is no pressure in a form. Daniel — the man Mara left behind in that small town — becomes a person the moment he hands her a box of her mother's unopened mail and does not ask a single question. No sheet predicts that. A scene produces it.

You do not need world documents either. If your idea arrives with a world attached, build exactly what the opening scene requires and stop; a binder on currencies and dynasties is a comfortable place to hide from sentences. The working rule for all preparation is the same: provision the first scene, nothing further. If Mara walks into the kitchen, you need the kitchen, the cup, and whatever she is avoiding by washing dishes. You do not need her transcript.

Write the first true sentences

When people ask how to start writing a book, what they usually mean is: how do I produce an opening as good as the openings I love? Wrong target. Those openings are late drafts, rewritten after their books taught their authors what they were about. Your target is smaller and more honest: true sentences. A sentence is true when it is specific and accurate about a person in a place — when it reports something instead of gesturing at something.

“Mara had complicated feelings about the house” is not a true sentence; it is a memo about one you have not written. “Mara washed the cup twice, though it had sat in a closed cupboard for a year” is true — and true sentences are the only kind that accumulate into a book. So give yourself the wrong opening on purpose. First pages are nearly always replaced once a book reveals its actual subject; the pressure on this one is imaginary. Count it, if counting helps: at the 250 words of a standard double-spaced manuscript page — the unit SCBWI measures book length in — an 80,000-word debut runs about 320 pages, and today you are being asked for the single page most likely to be thrown away. Start anywhere concrete, start in motion, and let the sentences be plain. Plain and true beats brilliant and vague every time.

Structure can wait — decisions cannot

Once pages exist, a new anxiety arrives: is this structured correctly? Should something have incited by now? Set it down. Structural frameworks earn their keep later, as instruments for diagnosing a draft that exists; held up to page three, they are just another way to stop. The middle of the book will have problems. The middle is not today's problem.

What cannot wait is a record of what you decide. From the first session onward you will settle load-bearing facts — the year, the month the house went quiet, what Daniel knows and when he learned it — and a few weeks later you will not remember which version you committed to. Write decisions down where you will actually reread them: a running page at the top of the manuscript, a notebook, or, if you draft with an AI partner, software built for novel-length work that treats settled choices as canon to consult rather than suggestions to drift from. Working that way has a method of its own, and writing a novel with AI without losing your voice walks through it. The premise you locked is the first entry in the record. Guard it.

So: you have an idea for a book — now what? Say its promise out loud until it holds in two sentences. Write those sentences at the top of a blank page and agree not to reopen them. Put a person in a room with an object that means more than it should, and write plain, true sentences about what she does next. Everyone who has ever done this started exactly this unready. Not permission, not preparation — a premise you have agreed to keep, and a small scene you can react to.

The fragile part of starting a book is not the first page; it is holding your premise steady through the months that follow, while everything else stays usefully loose. StoryProp is organized around that exact line — what you agree to gets written down and consulted before every draft, and what you don't stays a suggestion — so the promise you began with is still the one you are keeping in chapter twelve.

← All posts