How to write a novel

How do I write a novel? It is the oldest question in the craft, and the one this catalog has so far answered only at its edges: how to start, how long the book should be, whether to outline, which shape to build on. This guide is the front door those corner guides were missing. It walks the sequence of decisions every first novel actually demands and routes you to the page that goes deep on each one, instead of restating them.

Start from the definition, because it quietly sets the workload. Wikipedia opens its novel article with: a novel is "an extended work of narrative fiction usually written in prose and published as a book". The same article calls the novel today's longest genre of narrative prose fiction, followed by the novella, and notes that a precise length boundary between the forms is not possible. Extended is the operative word. A novel is the storytelling form whose defining cost is duration: not one good scene or one good chapter, but a chain of them that holds together across months of your attention.

What finishing a novel actually takes

What finishes a novel is rarely talent at the sentence level. Working writers' accounts agree on a less glamorous list:

• A decision you keep making. Every drafting session re-asks whether the book is worth continuing. A finished novel is mostly a long streak of yes.
• Tolerance for the bad middle. The stretch between the exciting opening and the ending you can see is where almost every abandoned novel dies. Expect it, plan for it, and do not read it as a verdict on the book.
• A record of what you established. At forty chapters, memory is not a filing system. More on this below.

The reference works frame the same demand at scale. The benchmarks are wide: 50,000 words is both the minimum E. M. Forster suggested and the threshold NaNoWriMo requires, while the novelist Jane Smiley lists novels as typically 100,000 to 175,000 words. The exact number matters less than accepting that the unit of work is months, not sessions.

The five decisions that shape a novel

Almost every stalled novel is stalled on one of five undecided things. Decide them in roughly this order:

• Genre and its promises. Pick the shelf the book sits on and read what that shelf owes: a mystery owes a fair solution, a romance owes the emotionally satisfying ending, a progression fantasy owes measurable growth. Genre is a contract you benefit from signing early, because it tells you which promises the ending must keep.
• A length target. Length is a structural decision, not a cosmetic one: it sets how many escalations the middle can afford and how much subplots can breathe. Set a working target early; how long is a novel? gives the numbers per genre.
• A process. How much you plan before drafting determines what kind of trouble you draft into. The trade is covered honestly in plotter vs pantser; the practical answer for most first novels is a light skeleton plus permission to deviate.
• A structure. Choose the shape you are building toward before you are thirty chapters deep. Three-act structure is the default skeleton; if the novel is or might become a serial, choosing a structure for a web serial explains what changes when the first draft publishes live.
• A working habit. The unglamorous fifth decision: when, where and how often you draft. It is the only one of the five you cannot decide once and file away; it renews daily, which is exactly why it decides most books.

The muddy middle

The middle of a novel fails in a specific way: not with collapse but with repetition. Events keep happening, the plot keeps advancing on the map, and none of it compounds. Two tools diagnose this from different directions. Story Grid's five commandments audit the scenes: if a scene's complications never compress into a real choice, it is marking time. And the want vs need pair says what the middle is for: the external pursuit gives scenes their spine, while the internal need surfaces through that pursuit, act by act. A middle that only advances the want is a travelogue of the plot; the need is what turns movement into story.

Keeping the book straight across drafts

Novels are drafted out of order, revised in passes and written over months, and every one of those processes is a chance to contradict yourself. The fix is boring and effective: keep a story bible from chapter one, recording names, rules, dates and who knows what. Treat the book's facts as canon: once established, they change only by deliberate decision, never by drift. Ten minutes of logging per session is the cheapest revision you will ever do, because an inconsistency caught at draft time is a line edit, and the same inconsistency caught after publication is an error.

Draft, then revise

Keep the two jobs separate. The draft's only job is to exist: complete, linear, end to end. Revision is a different activity with different standards, and it works best applied to a whole draft rather than a polished opening loop. The classic first-novel trap is rewriting chapter one for a month; the chapter cannot know yet what the book will be, so there is nothing to polish toward. Finish first, then revise in passes: structure first, then character and continuity, then prose.

Where to go next

The corner guides, in the order a first novel usually needs them: how to start a novel for the opening and its hooks, the length guide for your target, the plotter-pantser guide for your process, and the structure guides for the shape. If your idea is smaller than a novel, or you want a finished thing sooner while the novel grows, how to write a short story covers the single-sitting form; a novel is not a short story with more words, and the difference is worth learning from both directions.

The honest summary: a novel is the form where extension is the point. Choose deliberately, keep a record, finish the draft, and revise. The catalog handles each of those in depth; this page just holds the door open.