Start, sustain, and finish

From years of notes to a manuscript

August 16, 2026 · The StoryProp team

Somewhere in your house there is a drawer or a folder tree that holds the book you have not written: notebooks with corners gone soft, voice memos recorded at stoplights, interview transcripts, at least one abandoned opening chapter. “I have years of notes,” you tell people, and it sounds like a head start. Then you sit down to write and discover the strangest thing about a full archive: it is harder to start from than a blank page. Part of that is arithmetic — the agent-stated targets in the Manuscript Academy's 2026 word-count guide put memoir and narrative nonfiction at 70,000 to 90,000 words, and almost none of those words are in the drawer. Writing a book from existing notes is a different problem from writing a book, and it has its own four-step method: inventory, spine, map, fresh draft.

The archive is an asset — and an anchor

More notes make starting harder, and it helps to understand why. Every note is a small promise you made to yourself — this matters, use this someday — and a manuscript that omits it feels like a promise broken. So the archive accrues a kind of moral weight: you can't start writing until you've reread everything, and every rereading session ends in filing rather than prose. Writers with a shelf of notebooks routinely feel further from their book than writers with a single index card, because the blank page asks only that you begin, while the archive asks that you do it justice.

Here is the reframe that unsticks people. Years of notes are not a draft; they are evidence — proof that this subject has survived your attention longer than any passing enthusiasm, which is the best sign a book is actually in there. But the notes were written by a dozen past versions of you, each capturing a moment, none of them addressing a reader. Treat the archive as testimony about your subject, not as fragments of the manuscript, and its weight starts converting back into value.

Take inventory: what do you actually have?

Before deciding anything about the book, spend a few sessions on a shallow, fast inventory pass. The rule: label, don't read. Go through the notebooks, files, and memos and record only what each item is and roughly what it touches — “Orchard — interview with Dad — grafting, the bad winter,” “voice memo, ten minutes, mostly the packing shed.” Give each item about a minute, which turns a shelf of forty notebooks into one evening rather than one season. The moment you catch yourself rereading with pleasure, stop; that is writing energy leaking into filing. What you want at the end is a list you can see in one sitting: these notebooks, those interviews, that stack of half-drafts, on these subjects.

The inventory tells you things the notes themselves can't. Clusters appear: our orchard writer discovers that a startling share of everything she has recorded circles one bad season and its aftermath, though she always assumed the book was about the whole family history. Recurrence is signal — the thing you keep returning to without planning to is usually the book's center of gravity. The gaps are signal too: eleven notebooks and not one page from the person the story keeps blaming.

Voice memos and interview audio deserve a special word at this stage: get them transcribed, roughly, so they can be scanned and labeled like everything else. Machine transcription has turned this from a project into a queue you leave running. Whatever you transcribe with, ask for timestamps and speaker labels, because that is what makes a transcript searchable and citable later. And be clear about what the result is. It is an inventory item — raw material with a speaker and a date attached — not prose, and not close to prose. We'll come back to why.

Find the book underneath the notes

Now comes the tempting move: sort the notes into folders, promote the folders to chapters, call the result a book. It fails, reliably, because notes and books do different work. A note captures; a book argues or narrates. A pile of true observations is not an argument, and a stack of remembered scenes is not a story — organizing ideas for a book is not the same as organizing the notes, because the book's real structure lives one level below the material.

So put the archive down and ask the underneath question. For nonfiction: if the whole archive burned tonight, what claim would you still stand up and defend — and what would a reader see differently after believing it? For memoir and narrative: who changes across this material, and what forces the change? If your notes are professional — talks, client work, a long-running newsletter — the path from expertise to a claim worth a book is its own craft, and turning your expertise into a book walks through it. Either way, the answer will be a sentence or two, and it will feel almost insultingly small next to the archive. That smallness is correct.

A reliable test: explain the book to a patient friend without opening a single note, and record yourself doing it. What you say unprompted, and the order you say it in, is the book's spine; your memory has already done years of selection that no filing system can match. Just as telling is what you leave out — the clever material you forgot to mention was probably filed under obligation, not conviction.

Map the material, not the other way around

With a spine in hand, build the map: a chapter list or scene list that comes from the argument or the story, with the inventory assigned to it — under each chapter, the handful of notes that support its claim; under each scene, the memos and transcripts that feed it. Give the map a size while you are at it, because an unbounded map is another way of not starting. At the 2,000-to-4,000-words-a-chapter working range Reedsy publishes, a memoir at the 70,000-word end of the agent-stated target comes to roughly 18 to 35 chapters — a countable number of slots to fill and to check your inventory against. The direction matters. Notes serve chapters; chapters are not containers invented to house notes. The discipline of holding a structure firmly without welding it shut is the subject of outlining without killing the book, and nearly all of it applies to nonfiction as well.

Two things fall out of the map immediately. First, gaps become visible and turn into tasks: an interview you still need to conduct, a scene you will have to draft from memory, a claim with nothing under it yet. Second, a large share of the inventory will support nothing on the map — set it in a parking lot without shame, and do not let it lobby for a chapter of its own. From now on you open the archive the way a researcher does: with a question, looking for something specific, instead of wandering it hoping the book will assemble itself.

Draft fresh — collage reads like collage

When drafting begins, the notes-rich writer meets one more temptation: assemble. Take the best paragraph from the old opening, the vivid passage from the journal, the good quote from the third interview; paste them in map order; smooth the joins. Resist it. Collage reads like collage — the seams show in register, in tense, in distance from the material, because a paragraph written in grief and a paragraph written in analysis do not sit together just because both are good. Notes were written toward the moment; a book is written toward the reader; and readers feel the splice even when they can't name it.

Draft fresh instead. For each chapter, read its assigned notes with full attention — then close them and write the chapter from the map, in your voice, now. The note's job is to load you, not to be pasted. You will still reach back into the archive deliberately, for the irreplaceable specifics: the exact phrase your father used for the frost, the figure on the crumpled invoice, the detail memory had smoothed away. Those go in as chosen citations inside prose that has one voice, which is the difference between a book built from notes and a scrapbook with transitions.

A map with a size also gives you a schedule, and there is one drafting rate documented precisely enough to plan against. Anthony Trollope wrote with a watch on the desk and published the numbers in his Autobiography: three hours from 5:30 a.m., the first half hour spent re-reading yesterday's pages, then 250 words every quarter of an hour — a self-imposed average of forty 250-word pages, or 10,000 words, a week, kept up while he held a full-time job at the Post Office. The arithmetic is unglamorous and clarifying: at that rate a 70,000-word memoir draft is seven weeks of mornings. Your rate will be your own, and lower is fine. What matters is that the map converts into a count of mornings, because an archive will happily absorb an unlimited quantity of preparation, and a schedule is the thing that ends the preparation.

Interviews and voice memos need this discipline doubled, because transcription is raw ore, not prose. Speech leans on repetition, backtracking, gesture, and shared context, and on the page it reads either flat or chaotic — a faithful transcript is routinely unfaithful to how a person actually came across. Turning interviews into a book means quoting only the sentences no one but that speaker could have said, and paraphrasing the rest into your own narration. Turning voice notes into a book means the same thing at lower stakes: mine the ramble for the one live phrase, then let the rest of the memo stay what it was — a message from you to you, already delivered.

Keep what you don't use

Most of the archive will not appear in the manuscript, and this needs saying plainly: that is what selection looks like from the inside. The book is not the notes reorganized; it is a claim or a story the notes made possible, and the unused material did its work by teaching you what the used material means. Keep all of it, and keep sources and dates attached — who said it, when, in what context — because attribution you shrug off now is attribution you cannot reconstruct later. A project like this wants book-writing software that stores research with its sources and dates in a place of its own, separate from the draft it informs. That is how StoryProp holds an archive: close at hand, and not pasted into the prose.

And the parking lot may not stay parked. Our orchard writer's unused cluster — everything about the farm co-op two properties over, which no amount of arranging could tie to the bad winter — is not waste; it is the visible beginning of a second book she does not have to write yet. Years of notes often hold more than one book, which is precisely why forcing them into one produces none.

What changes, in the end, is posture. The collector serves the archive; the author makes the archive serve a manuscript. Your notes did their job — they kept the subject alive, sometimes for a decade, until you were ready to decide what it all argues or tells. The book starts on the day the notes stop being the destination and become the ground under one.

A map of which note serves which chapter only pays off if the thing drafting beside you actually consults it. In StoryProp, the decisions you settle are promoted to records the agent reads before writing, and research keeps its sources and dates in a place of its own — so you can draft the chapters one at a time with years of notes standing behind every one of them, and none of them pasted in.

← All posts