Most advice about this is encouragement, and encouragement is not the constraint. People who do not finish books rarely lack motivation at the start; they lose it at a predictable point, for a predictable reason, and the fix is structural rather than emotional.
Decide the length first
"A book" is not a target. A word count is.
| Kind | Words | Pages, roughly |
|---|---|---|
| Lead magnet or short guide | 5,000–15,000 | 20–60 |
| Practical non-fiction | 40,000–60,000 | 160–240 |
| Business or self-help | 50,000–70,000 | 200–280 |
| Novel | 70,000–100,000 | 280–400 |
| Memoir | 70,000–90,000 | 280–360 |
Words to pages is the conversion in detail, and it matters here for one reason: a number turns an indefinite project into a finite one. 60,000 words at 500 words a day is four months. That is a schedule; "write a book" is not.
Pick the low end. A shorter book that exists beats a longer one that does not, and first books are almost always too long.
Write the shape down before the prose
This is the single largest difference between finished and abandoned books, and it is why the middle third is where they die: with no map, the middle is just a long way from anywhere.
For non-fiction, the shape is the argument:
- One sentence saying what the reader can do afterwards that they cannot do now.
- Eight to fifteen chapter titles as statements, not topics. "Price is positioning" rather than "Pricing".
- Three or four beats under each, one line each.
- Check the order. Each chapter should need the one before it.
For fiction, the shape is what changes: where the character starts, what forces the change, what it costs, where they end. Scene cards rather than an outline, if that suits you better — Scrivener and its alternatives exist mostly for people who work this way.
The outline is not a contract. It is permission to stop deciding what comes next every morning, which is what actually drains people.
The session, not the day
Books get written in repeated sessions, and the properties that matter are boring:
- Same time. Decision fatigue is real, and "when I find time" means never.
- A small quota. 500 words is a real book in four months. 2,000 is a heroic week and then nothing.
- Stop mid-sentence. The hardest part of a session is starting, so leave yourself somewhere obvious to resume.
- Track the count. Not for motivation — so you know whether the schedule is true.
Do not edit while drafting. They are different jobs using different attention, and doing both is why chapter one gets rewritten eleven times while chapter nine does not exist. Write forward. Leave a bracketed note where you know something is wrong and keep going.
The three places books die
The middle third. The novelty has gone and the end is not visible. The outline is the defence; so is accepting that this stretch feels bad for everyone and is not a signal about the book.
The first read-back. You finish, read it, and it is worse than you hoped. This is universal and it is not evidence. A first draft is raw material, not a book — its job was to exist.
The formatting. A surprising number of finished manuscripts never become books, because the author hits the gap between a document and a publishable file and stops. That gap is solvable and it is worth knowing it exists before you reach it: how to self-publish a book is the whole sequence after the draft.
Where to write it
Less important than it feels, and worth deciding once so it stops being a question.
- Word or Google Docs if other people will touch the manuscript. Google Docs vs Markdown.
- Scrivener if the structure will genuinely move.
- A plain-text editor if you want the file to outlive the application — writing a book in VS Code, Obsidian, Notion.
Best book writing software compares them properly. The only choice with long consequences is whether the manuscript ends up locked inside an application, because every later export depends on that.
On AI: useful for structure, research and getting unstuck, and a fast route to a book that reads like nobody wrote it. Writing a book with AI is the honest version of where it helps.
Then the second draft
Finish first, then fix — in this order, because fixing sentences in a chapter you later cut is the most expensive work in publishing:
- Structure. Is anything missing, repeated, or in the wrong place? Cut whole chapters here.
- Chapters. Does each one do one job? How long should a chapter be.
- Line. Now the sentences.
- Proof. Last.
Self-editing your book covers doing the first three yourself, which removes most of what a paid editor would otherwise bill for.
What comes after the words
The manuscript is the hard part and it is not the last part. A finished draft still needs a title, a description, an interior and a cover before anyone can read it — how to title a book, writing a book description, and then the production itself.
That production step is where plain text pays off: bukpress takes a Markdown manuscript and exports a typeset PDF and a reflowable EPUB from the same file, so the draft you wrote is the thing that becomes the book rather than the input to a second project. Self-publishing from Markdown is that route, and what it costs is the budget for the rest.
Frequently asked questions
How long does it take to write a book? At 500 words a day, a 60,000-word book is about four months of drafting and a similar period of revision. The variable is consistency, not speed.
How many words should my first book be? The low end of its category. Shorter books get finished, and a finished 40,000-word book is worth more than an abandoned 90,000-word one.
Should I outline or just write? Outline, at least loosely. Discovery writing works for some people and is the main reason books stall in the middle — and even discovery writers usually outline after the first act.
Do I need to write every day? No. You need a schedule you actually keep. Three fixed sessions a week beats an intention to write daily.
When should I show it to anyone? After the second draft. A first draft invites feedback on things you already planned to fix, which wastes the reader's one fresh reading of your book.