Non-fiction is sold on a proposal rather than a finished manuscript, which surprises people who assume they have to write the whole thing first. You do not. What you have to do is make a case, and the case is commercial before it is literary.
Fiction works the other way: novels are sold complete, on a query letter and the pages. Everything below is about non-fiction.
What the proposal is really being read for
An editor is deciding whether to commit an advance, a design budget, a print run and eighteen months of a team's attention. They are reading for three things, and only one of them is the writing.
- Is there an audience, and can it be reached? Not "is this interesting" — who specifically buys it, and how do they find out it exists.
- Can this person write the book? The sample chapters answer this and nothing else does.
- Can this person help sell it? Which is the platform section, and the reason it is uncomfortable to write.
The eight sections
| Section | Length | Job |
|---|---|---|
| Overview | 1–3 pages | The book, the reader, the argument |
| The market | 1–2 pages | Who buys it, and how many of them exist |
| Comparable titles | 1–2 pages | Where it sits on a shelf |
| About the author | 1 page | Why you, and what you can reach |
| Chapter outline | 3–8 pages | The whole book, a paragraph each |
| Sample chapters | 1–2 chapters | Proof you can write it |
| Specifications | Half a page | Length, delivery date, illustrations |
| Covering letter | 1 page | Why this agent, why now |
Thirty to fifty pages in total is typical. Agents' submission pages vary and their instructions override anything here — read them, and follow them exactly, because not following them is the cheapest possible reason to be rejected.
The overview
The hardest page, and the one written last. It has to state what the book is, who it is for and what it argues, in prose good enough to demonstrate the voice while doing all three.
- Open with the hook, not with context. The first sentence should be the most interesting thing in the proposal.
- Name the reader explicitly. "This book is for X who needs to Y." Vagueness here is read, correctly, as not knowing.
- State the argument in one sentence. If you cannot, the book does not have one yet.
- Say what the reader can do afterwards that they could not before.
- Keep it under three pages. An overview that runs to six is an overview that has not decided.
This is the same act of compression as a cover blurb, at a slightly greater length — and you will reuse it. How to write a book description is the shorter version you will need whichever route the book takes.
Comparable titles, and what they are testing
The section people get most wrong. Comparable titles are not flattery and not competitors to dismiss — they are evidence that a market exists and proof that you have read it.
| Do | Avoid | Because |
|---|---|---|
| Four to six titles | One, or twenty | Too few looks unread; too many looks unsorted |
| Published in the last 3–5 years | A classic from 1997 | The market it proves no longer exists |
| Comparable in size, not aspirational | Sapiens, Atomic Habits | Naming a mega-seller reads as naïve |
| A sentence on what each does | A list of titles | The analysis is the point |
| A sentence on what yours does differently | "Nothing like this exists" | That claim means no market |
That last row is worth dwelling on. "There is no book like this" is intended as a strength and reads as a warning: either the market has been tested and found empty, or you have not looked. The answer an editor wants is "books like this sell, and here is the gap in them".
The platform section
Uncomfortable to write and impossible to skip, because it is the section that most often decides the outcome for a non-fiction book. It asks: when this publishes, who already listens to you?
- Give numbers, not adjectives. Mailing list size, monthly readers, talks given, clients, followers. Specific and checkable.
- Be honest about small numbers. An engaged list of 1,200 in a professional niche is worth more than 40,000 general followers, and saying so with the context is credible.
- Include professional standing — the role, the practice, the credential. For a professional readership this outweighs social media entirely.
- Describe what you will do, concretely: the newsletter, the podcast circuit, the conference, the client base you can email.
- Never inflate. Every number is checkable in ten seconds, and one inflated figure discredits the whole document.
The chapter outline and samples
The outline is the whole book in miniature: every chapter, a paragraph each, in order. It proves the book has a structure rather than a subject, which is the difference between a book and a long article.
Each paragraph should say what the chapter covers and what it establishes for the next one. Chapters you cannot summarise in a paragraph are chapters doing two jobs — the same test as the structural pass in how to edit your own book, applied before the writing rather than after.
For the samples, send chapter one and one chapter from the middle. Chapter one shows the opening; a middle chapter shows the book at cruising speed, which is what the other forty thousand words will actually read like. Send your best work, not your first draft.
How it should look
The proposal is the first thing you have made that the agent will see, and it is competing with a full inbox. It does not need design; it needs to be effortless to read.
- One PDF, unless the submission page asks otherwise. Not a zip of six documents.
- A serif face at 11 or 12 point, generous line height, and margins wide enough to read comfortably on a screen — page margins and line length.
- Page numbers and a running header with your name and the title, because pages get printed and separated.
- A contents page if it runs past about twenty-five pages.
- Each section starting on a new page, which is a
---page break in the manuscript: page breaks in Markdown. - No cover design, no illustrations, no colour. They signal effort spent on the wrong thing.
Writing it in plain text and exporting the PDF is the path of least resistance here for a reason that goes beyond tidiness: the proposal is a document you will send to twenty agents over a year, revising it each time, and each revision has to come out looking identical to the last. Markdown to PDF is that export.
What happens next
Slowly, and mostly silence. Responses take weeks to months, many never come, and a rejection is usually a form. Query in small batches of five to eight so you can revise between rounds, and keep a record of who has it and since when.
Two outcomes are worth planning for before you start. If it sells, you will write the book to a deadline against an outline you have already committed to. If it does not, you have a fully structured non-fiction book planned, a chapter outline, two chapters written and a description drafted — which is a considerable head start on publishing it yourself, and the reason the work is not wasted either way.