How to find the shape before you draft

1 The bottom line

Find your piece's shape before you draft: name what it is about, choose a spine to carry it, and keep the plan in pieces you can move. A shape you can bend saves you from drafting the wrong piece at full length.

2 Key takeaways

• Say what the piece is about in one sentence or question before you draft. Every later choice, what to keep and what to cut, gets a test.
  • Andrew Stanton9:01 took about three years to land on the line for Finding Nemo.
  • Nick Bilton24:59 boiled his Apple Vision Pro cover story down to "the future is coming whether we like it or not."
• Pick a spine that gives each part a reason to come next. A spine such as a clock, a chain of events or a rising tension stops the piece from becoming a heap.
  • In Tinkers, Paul Harding44:43 made each section a day before the man dies, and the first sentence announces he will die in eight days.
  • Richard Powers1:03:49 builds the long middle so that each solved problem creates a bigger instability.
• Know where the piece ends, and leave the route open. A fixed destination lets you wander without feeling lost.
  • Harlan Coben4:27 has written 38 novels knowing the ending at the start.
  • Mitch Albom2:23 had the ending of Twice, the hero's love mistake, in mind from page one.
• Keep your material in pieces you can handle and rearrange. Moving pieces by hand shows you what fits with what.
  • Susan Orlean1:38 used about 700 index cards each for Rin Tin Tin and The Library Book.
  • Yann Martel1:59 printed about 400 pages of Life of Pi notes, cut them with scissors and filed the fragments in labeled envelopes from start to end.
• Treat the shape as a hypothesis you will revise. What the work is really about often shows up halfway through.
  • Wright Thompson31:00 outlined The Barn only broadly, wrote a 280,000 word first draft and cut about 173,000 words.
  • Robert Macfarlane42:55 works on books for three to eight years, so the plan has to survive a lot of change.

I can't write a book unless I see at least the skeleton of it

Daniel Pink, Drive

2.1 Name what the piece is about

Before you pick a structure, write the one sentence or question every part must serve.

• Write one sentence naming character, conflict and conclusion, and keep rewriting it. Andrew Stanton9:01 borrows Lajos Egri's idea of a premise that tells you the structure.
  • Keep a page open and write sentence after sentence. Treat each as a litmus test of what you think you have made, and do not sweat it when one is wrong.
  • Once the sentence is right, it tells you what to keep and cut in everything written or still to write.
  • His Finding Nemo sentence came from his own overprotective weekends with his young son.
• Put the core question at the top of the draft. Michael Dean36:45 says a fuzzy first draft means a fuzzy question, and redrafting sharpens it.
  • After drafting, reverse outline. Mark which paragraphs answer which sub-question, double down on the ones that matter and cut the ones you do not want to discuss.
  • For a book, write out the questions you want to answer and pin them where you will see them daily Johann Hari28:04.
  • Dean finds a hidden hierarchy in Wallace's "Consider the Lobster": one title question, three section questions, and a question behind each group of paragraphs.
  • Hari's questions for Stolen Focus hung on his door for three years, so he could kill anything that did not serve them Hari28:04.
• Name the theme in plain words, not the events. Plot is stuff happening, and story is the emotional layer underneath Michael Jamin24:48.
  • Say it the way you would in an elevator pitch. Bilton24:59 will not start until he can, and every scene then loops back to the theme like a braid.
  • Ferris Bueller looks like a day of hooky, but it is about a best friend trying to save his friend's life, and the film says so in its second scene.
  • Bilton's Elon Musk feature ran on "he needs a problem to solve and if there's not one there, he'll create one" Bilton24:59.
• Narrow the scope until one thread is left. The cutting starts before the draft.
  • For a memoir, decide which slice of life it covers. Tucker Max17:10 says that answer almost always settles the structure, and most great memoirists write three to five books.
  • Tie each story to one or two themes. Rob Henderson5:24 chose family instability, foster care and social class, and cut a dozen drinking anecdotes down to two.
  • Henry Shukman58:20 decided his memoir was not about his life but about his journey into Zen, and only material serving that thread was allowed back in.
  • A 500 to 800 word column must organize around a single idea, not one and a half, or the structure collapses Fareed Zakaria13:24.

2.2 Pick a spine

A spine is the line that carries the reader from one part to the next, and your material hangs on it.

• Hang your information on a narrative line. Michael Pollan35:20 calls it the laundry line.
  • Pick a sequence of events, such as the life of one steer from birth to slaughter, and hang your information on it: corn versus grass, drugs, pollution.
  • Space the exposition out, because clumped together it sags, and readers follow what happens next over explanation.
  • His editor Jerry Marzorati suggested the approach, and the Times Magazine cover story hung six themes on it.
  • If explanation stalls the action, Nat Eliason23:53 moved it into short between-chapter pieces after three drafts. About half of readers skip them and stay with the story.
• Let a mystery force the order. Because the next clue has to arrive, the story cannot stall Gillian Flynn1:12:13.
  • Attach your themes to the mystery. That frees them to be larger and less obvious.
  • Her early Sharp Objects draft had characters talking, then going to another room to attack a trapped bird. Dennis Lehane's Mystic River showed her the way.
• Make each event the direct result of the one before. Andy Weir50:30 builds plots as chains, so nothing feels like coincidence.
  • In The Martian, to survive the hero has to do one thing, which forces another, and so on.
  • Readers stay immersed because they never have to accept a lucky break.
• Borrow a proven arc when the piece is a story. Outline the change in the character, not just the events.
  • Will Storr22:28 runs five acts: this is me and it's not working, is there another way, I transform at the midpoint, can I take the pain of change, and a final showdown that confirms the change.
  • Or use ABDCE: action early, then background, development, climax and ending. The climax needs a kind of killing, which may be of an illusion or a prejudice Anne Lamott1:03:02.
  • In Jaws the midpoint is Brody deciding to go into the ocean Storr22:28.
  • Put a false victory near the middle, such as the happy lovers just before the iceberg in Titanic David Gelb32:22.

2.3 Fix the destination

Decide where the piece lands, and the middle gets easier to find.

• Know the ending before you start. Coben4:27 compares a novel to driving from New Jersey to Los Angeles with the destination in the GPS.
  • With the ending known, you can detour to Mount Rushmore without panic, because a quiet voice pulls you back to the main road.
  • In columns, the first and last paragraphs work together, so you sail toward the last one Albom2:23.
  • Without the ending, off-road scenes feel like being lost.
  • Albom knows the ending of a novel but not necessarily the middle, and sometimes rewrites the beginning Albom2:23.
• Write the ending first and build backward. Tish Rabe5:05 learned it from Sesame Street writers, who wrote comic skits punch line first.
  • Write the last page before the first, so you know what the story has to prove. Then ask what must happen in the middle to earn that ending.
  • In her Rotary book about Amani and the zebras, the lion in the middle came from the ending she needed.
• Let the early turning point hint at the climax. At the end of act one, give the character a call that lets the audience flash forward to the final confrontation Steven Pressfield4:32.
  • In Rocky, Apollo Creed picks Rocky from a book of local fighters, and the whole story then points at the fight.
• If the middle is a blob, plan only a few steps ahead. Eric Roth53:17 knows the opening and the end.
  • Outline the next five scenes with a single word each, and the story starts taking hold. Then build tension toward the last act.
  • His example list reads "wedding, shootout" and so on.

2.4 Build a system you can rearrange

Cards, tables and beat lists turn structure into something you can see and move.

• Put the material on cards and move them by hand. Handling cards keeps reminding you what you have and which pieces fit together Orlean1:38.
  • Transfer only the material that matters to 5 by 8 index cards, then lay them out on a big open floor.
  • For stories, use one scene per 3x5 card, about 60 for a movie, sorted by act on a wall Pressfield1:12:05.
  • Start with a clothesline. Hang the few scenes you know, like the ending, then ask what has to happen in the gaps and at the start. Swap out cards that bore you or repeat each other Pressfield1:12:05.
  • For nonfiction, sort the cards by theme and let the themes become chapters. Robert Greene41:30 started his current book with twenty or so chapters and whittled them to twelve.
  • Pressfield's Moby Dick example: Ahab nearly reaches the whale, the whale escapes, and you fill in around that Pressfield1:12:05.
• Use a table you can scan. A table of contents or a spreadsheet shows gaps and slack at a glance.
  • Outline the whole book through the table of contents and expect to redo it. Alex Hormozi1:58 calls it the hardest thing he works on and rewrote it from the top whenever he tired of a version.
  • Make the chapter headings descriptive enough that a reader can decide whether to read the book from the contents page alone Hormozi1:58.
  • Log each scene in Excel: the day, who appears, how they interact, what the reader learns and where the protagonist is personally. Then check that each scene earns its place and pushes into the next Andrew Hunter Murray15:57.
  • For $100M Leads, pages of messy outlines and discarded structures narrowed to one clean table of contents, and the writing was then easy Hormozi1:58.
  • Murray asks of any scene, even one set in a fireworks factory, whether we learned enough Murray15:57.
• Look for the skeleton while you are still gathering. Do not wait until the research is done.
  • While reporting, jot eight or nine beats: the start, the end, the big moments and where you want to flip the reader's belief. Research then fills in beneath them Patrick Radden Keefe32:16.
  • Put a whiteboard or big post-its behind your chair after months of research. Turn around and stare until the structure appears, and only then draft Pink6:06.
  • The beats limit Keefe's research: he can cut a cast of fifty possible characters down to about five Keefe32:16.
• Order the pieces by how they will make the reader feel. Henrik Karlsson6:28 treats an essay as units whose order puts the reader in a state.
  • Lie on the sofa, picture each cluster as a colour and a feeling, and swap the order until you feel energy released.
  • A story placed before an argument can leave the reader emotionally ready for Aristotle.

2.5 Treat the shape as a hypothesis

Plan enough to start, then let what you learn rewrite the plan.

• Write the plan as a tentative letter to yourself. Macfarlane begins "Dear Rob, the future Rob" when he commits to a book.
  • Say what your heart hopes the book will be, its shape, its key metaphors and what you want it to do in the world. Keep it tentative, not a set of commandments Macfarlane42:55.
  • Reading it after finishing shows how far the book has mutated. Sometimes it is hugely discrepant, and other times surprisingly close Macfarlane42:55.
• Stay open to the work being about something else. AJR1:30:08 started the album Maybe Man with songs pointing in different directions.
  • When they wrote the title track, about not knowing who the real you is, they decided it was the arc.
  • By the final song, 2085, they had learned what matters at the end of a life, which they could not have known at the start. Their father died while they made the album.
• Outline loosely so you know what is load-bearing. Thompson wrote everything because he was unsure what mattered, and calls that a rookie inefficiency Thompson31:00.
  • On his next book, with six narrative arcs, he plans to storyboard much more so he does not lose the thread.
• Write what you can already see, in any order. Andrew Sean Greer4:20 compares it to filling in a crossword puzzle.
  • Write the scenes you can imagine, wherever they fall, and let each give clues to the others. This gives the book an organic feel and avoids a mid-book breakdown from a plot that drifted.
  • Or start with twenty short scenes without worrying whether they fit, then write the connective tissue between them Suleika Jaouad23:38.
  • Greer wrote a scene two thirds of the way through Villa Coco about finding a work of art without knowing where it led. Nabokov reportedly wrote on index cards this way.

2.6 What the books say

Barbara Minto's book says the shape of a piece is a pyramid: one main point on top, grouped support below, and you can build and test it before you write a full draft.

• Start by writing down the Subject, the Question in the reader's mind, and your Answer. Barbara Minto, The Minto Pyramid Principle
  • Test them by stating the Situation and the Complication. If the Complication does not raise your Question, change the Question or rethink the Complication.
  • A memo on a disk billing system described how it would work but never said whether it was a good idea. Rebuilt around the question, it answered yes with three reasons.
• Put the point that sums up the ideas before the ideas. Barbara Minto, The Minto Pyramid Principle
  • Readers take in one idea at a time and guess at the links you leave out. A speaker listed beards in Zurich, sideburns in New York and moustaches in London, and listeners guessed London was ahead.
  • State the introduction, main point and Key Line points up front, so the reader has your whole argument in 30 seconds.
• Build each group from ideas of one kind, and keep it to four or five. Barbara Minto, The Minto Pyramid Principle
  • Name the group with one plural noun, such as reasons, steps or problems. If no noun fits, regroup.
  • Give the group an order: time, structure or degree. If you cannot find one, the grouping is faulty, and the effort can show a missing step.
• Check that each summary says only what its group shows. Barbara Minto, The Minto Pyramid Principle
  • If the top point could cover other things, or claims more than the ideas show, lower its level of abstraction.
• Word headings as ideas, not as labels like Findings or Background. Barbara Minto, The Minto Pyramid Principle
  • Then the table of contents reads as a summary of the piece. Never use a single heading at a level, and keep headings in a group parallel.

2.7 Where writers disagree

Writers split on how much to plan, and the right amount depends on what the draft does to you and how long the piece is.

• Plan heavily before you draft. Some writers say the plan is where the work is.
  • Steven Levitan23:09 says his room spent weeks breaking stories before anyone wrote pages, and a story that truly works lets the script almost take care of itself.
  • Bill Browder8:05 writes a mini chapter for every chapter before the full text.
  • A 140,000 word book gets a mini book of about 50,000 words, with around 45 chapters of 10 pages at most Browder8:05.
• Plan little, and make the decisions inside the scene. Others find long plans hurt the work.
  • David Baldacci4:14 always throws outlines out and writes linearly, because inside a scene there is no kicking a decision down the road.
  • Jonathan Franzen34:28 wants a destination but no idea how he will reach it, so the work becomes making an unlikely sequence of events seem inevitable.
  • Franzen thinks a fully planned book reads like a fully planned book, perfect but sterile Franzen34:28.
  • Even Baldacci says to outline if it works for you, since no template fits everyone Baldacci4:14.
• Match the shape to the length. Short pieces want one idea, and long ones can braid several.
  • A column cannot wander, though an essay of about 2,000 words can Zakaria13:24.
  • Elif Shafak21:30 held three storylines at once and wrote them like a braid, which took a lot of thinking, notes and arrows. She still says to trust intuition.
  • Check that chapters two and three do something different from chapter one, or a perfect article becomes a padded book Dana Gioia1:40:45.