How to write a memoir
· The StoryProp team
Most people who set out to write a memoir begin with a pile of material and a sense of obligation. Something happened, it mattered, and the story keeps getting told badly at dinner parties. The instinct is to fix that by writing it all down in order. Give that instinct up first. A memoir is not the record of a life but a shaped account of part of one, told by someone standing at a distance from it. Everything difficult about the form follows from those two facts. The hard part is not the sentences. It is the judgment about which sentences belong.
Memoir is a slice; autobiography is a span
The distinction is not fussy, and getting it wrong is the most common structural failure in a first draft. An autobiography covers a life, roughly end to end, organized by chronology: this happened, then this. It presumes the reader already has a reason to care. A memoir covers a subject, and the life supplies the material. Its logic is thematic, and it earns attention from the writing rather than from the byline.
Look at how the good ones hold their edges. Cheryl Strayed's Wild is a single season on the Pacific Crest Trail; her mother's death and the end of her marriage enter only where they bear on the walking. Joan Didion's The Year of Magical Thinking keeps to the year after her husband collapsed at their dinner table, and a long, decorated writing career stays almost entirely outside the frame. Practically, for you: a memoir about caring for a parent through a long illness does not need your childhood unless your childhood explains the caretaking. A memoir about leaving a religious community owes the reader nothing about your later career, unless that career is what the book is about. The test for any chunk of material is not whether it happened to you but whether it belongs to the subject. Most of your life will fail that test for any given book, and that is a feature.
Try naming your subject in a sentence with no proper nouns in it. A memoir about learning that competence is not the same as safety. A memoir about the year you stopped being the reliable one. If you cannot manage it, you may have an autobiography, or three memoirs stacked on top of each other. Mary Karr got three books out of one life — The Liars' Club, Cherry and Lit — by giving each of them its own subject instead of trying to write the single volume that covered her. Stacking is not solved by drafting more pages; it is solved by deciding which book you are writing.
Truth versus accuracy: the honest reconstruction problem
Here is the problem plainly. You are going to write a scene from a Tuesday in a year you can barely place. You remember the room, roughly. You remember the sentence that ended the conversation — or you remember what it did to you, which is not the same. Nobody recorded it. And a memoir written entirely in summary reads like a deposition and moves like one: we argued, and afterward things were different. True, and unreadable.
The workable position is that you may reconstruct and you may not invent. Reconstruction means rendering in scene an event you are confident occurred, using the best evidence you have and your honest memory of its texture. Invention means adding an event, a person, or a revelation that did not happen because the chapter would be shapelier if it had. The line is real, and you know when you are crossing it. James Frey's A Million Little Pieces is the case everyone reaches for, and it is the right case: what the reporting exposed was invented events and invented consequences, not imperfectly remembered dialogue. The distinction the whole form runs on is exactly the one he crossed.
Some rules that hold up. Dialogue is reconstructed to the sense, not the syllable: write it if you remember the substance and the register, and never put a sentence in someone's mouth that changes who they were. Compress time openly rather than secretly — that spring is honest about three conversations that felt like one, where an invented Tuesday is not. Do not present a composite as a person without saying you did; Dave Eggers put the method at the front of A Heartbreaking Work of Staggering Genius, cataloguing in his own preface what he had compressed, renamed and reconstructed. A reader told up front will forgive a great deal. A reader who finds out afterward forgives nothing. And where memory is uncertain, put the uncertainty on the page: in the version I believe today, she was already standing when I came in. What you owe the reader is not a transcript but the assurance that you will name your doubts.
Selection is the craft
A life is not a story until you choose. Nothing that happened arrives pre-sorted into important and unimportant; you do the sorting, and the sorting is the book. Two writers with an identical set of events will produce entirely different memoirs, and the difference is not style. It is what each of them decided the events were about.
For every scene you are considering, ask what it does that no other scene does. If the answer is that it shows your mother was controlling, and four other scenes show the same thing, you have one scene and three deletions — or, better, one scene and three chances to show what the first could not. Repetition is the tax memoir pays for being true: it happened many times, so the writer writes it many times. The reader needs it to land once.
Selection also means letting genuinely significant events go. You may have to cut a death, a move, a job that ate four years, because the book is about something else. That feels like a betrayal and is not: you are declining to make the event part of this account, not denying it. Write the table of contents before the chapters and give each entry a job rather than a title: the reader stops trusting my judgment about him. If a chapter has no job, it is material, not structure. Sorting journals and letters into usable material is a separate discipline, handled in turning notes into a manuscript.
The retrospective narrator: two selves in one voice
The most underused tool in memoir is the fact that you are two people. There is the person the events happened to, who did not know what was coming, and the person writing, who knows how it turned out. This split has names worth knowing. Vivian Gornick, in The Situation and the Story, calls the events the situation and calls what the writer has made of them the story, with the narrator as the instrument that converts one into the other. Sue William Silverman, in Fearless Confessions, divides the same labour between a Voice of Innocence, which lives inside the experience and does not know better, and a Voice of Experience, which does. Weak memoir collapses the two into one voice, usually the present one, explaining events as they occur. That drains the scenes of suspense and turns the book into a long verdict.
Keep them apart and let them talk. The then-self gets the scene: her confusion, her bad reasoning, her conviction that the plan was working. The now-self gets the frame: what she could not see, what the moment turned out to mean, and where she still does not know. That last part is what separates a memoir from a lesson. A narrator who has processed everything has nothing left to find out on the page.
Didion does this on almost every page of The Year of Magical Thinking. The woman in the scenes is consenting to an autopsy and refusing to give away her husband's shoes because he will need shoes when he comes back; the woman writing knows precisely what that behaviour was, names it, and declines to sneer at the person doing it. The same split works on the smallest image. In the demo manuscript we use around here, Mara returns to a house she left and finds a chipped cup still in the cupboard, in the same position. As fiction that is an image. As memoir it splits: the then-self sees a cup and feels something she cannot name, and the now-self writes, I want to say I understood, standing there, that Daniel had changed nothing because changing anything would have meant admitting I was gone. What I felt was mostly hunger.
Writing the real people in it
Everyone in your memoir is a person who did not agree to appear, and the way through is a distinction you can apply case by case: writing about someone is not the same as exposing them. You can render a person fully, including badly, while restricting yourself to what belongs to your subject. A father's temper, in a book about growing up in that house, belongs. His affair, if it has nothing to do with the subject, does not — even if true, even if you are angry about it.
Give people their own logic. A portrait in which someone exists only as their effect on you is a portrait the reader will not believe, and disbelief spreads to the parts you most want believed. Write the version of your sister that your sister would recognize as an honest attempt, even if she would dispute your conclusions. That is the difference between a book with an antagonist and a book with a grudge.
Know what changing a name does and does not do. It is a courtesy, and it is not a shield. Two working assumptions, offered as rules of thumb rather than as law: assume a person is identified if the readers who know them can recognize them, so the aunt every cousin will place is identified no matter what you call her on the page; and assume your position is strongest when the account is true and when opinion reads as opinion rather than as an assertion of fact. A legal read before publication is what turns those assumptions into answers about your particular manuscript, and it is worth arranging early enough that what it turns up can still change the book rather than only delay it.
Then handle the rest of the practical layer deliberately and early. Decide which names are changed and stay consistent. Decide who you will tell before publication and who you will not. Decide what you would cut if the person asked, and whether you should cut it regardless. Making these calls consciously lets you draft honestly; the alternative is negotiating with an imagined jury of relatives in the middle of every sentence, which is where a great many memoirs quietly die. And be hardest on yourself without performance — self-flagellation is evasion wearing accountability's coat.
Where to begin this week
Write the subject sentence. Then list twelve scenes you are certain belong, one line each, in any order, and name the job each one does. You will find duplicates; keep the sharpest and note what the others were reaching for, because those notes often become the retrospective passages later. This gives you a spine before you have a style, which is the right order — voice in memoir tends to arrive once you know what you are claiming.
Then draft one scene at full strength, dialogue and all, with a single retrospective sentence at the end and no more. See whether it survives without the explanation. If it does, you have found the voice. If it does not, the scene is usually the wrong scene, and no framing will rescue it. Finishing matters more here than planning does: what you are finally offering anyone, editor or reader, is the writing itself, so the writing is what has to exist. Memoir is unusual in that the research is already done and the hard part is judgment: what mattered, what it meant, and how much you will admit you still do not know.
A memoir runs on decisions you make once and then have to keep: which names are changed, which version of a contested scene you trust, what belongs to the subject and what does not. StoryProp holds those as records the writing agent reads before it drafts, so the call you made in chapter two is still in force in chapter eleven.