Screenwriting and television

How to create a beat sheet (without a template)

August 16, 2026 · The StoryProp team

A beat sheet is the shortest useful description of a story: a numbered list of moments where something changes. Not a synopsis, not an outline with scene headings and weather — a list of turns. Most advice on the subject begins with a published template. The best known is Blake Snyder's Save the Cat! beat sheet, published in 2005, which pins fifteen named beats to specific pages of a 110-page screenplay: the catalyst at page 12, the break into act two at 25, the midpoint at 55. Filled in cold, templates like that produce a story-shaped list with nothing alive inside it. The order of operations that works is the reverse: build the sheet from your own material first, and bring the published sheets in afterward, as inspection equipment.

What follows is the practice — how to create a beat sheet from nothing, how big a beat should be, and what to do with the sheet once it exists. The theory of the frameworks themselves, and the argument for treating them as diagnostic lenses rather than blueprints, lives in a companion piece on story structure frameworks. This piece stays at the workbench.

A beat is a change, not a scene

The most common failure in beat sheets is granularity. Writers list scenes: “Mara and Daniel argue in the kitchen.” That is a location and an activity, and it could sit in the story anywhere, which is exactly the problem — nothing in the sentence says what is different afterward. A beat records a change of state: Mara stops believing Daniel's account of the night. Now the line can only sit in one place, after she believed him and before whatever her disbelief causes.

The test for any candidate beat is blunt: name something that is true after this moment that was not true before it. If you can't, you have written down a scene you are fond of, not a beat. The changes worth listing come in a few recurring kinds — knowledge (someone learns), intention (someone decides), relationship (trust or power shifts between people), and circumstance (the world opens a door or closes one). A useful sheet is mostly verbs of this sort: learns, decides, loses, refuses, admits, breaks.

Beats and scenes are different units, and they do not map one to one. A single scene can carry two beats; one beat can take three scenes to land. In The Quiet House, Mara finds a chipped cup in a kitchen that was supposed to have stood untouched for a decade. The scene is small — a woman washing up. The beat is seismic: Mara now knows someone has been living in the house. Two pages later, still in the kitchen, Daniel explains the cup a little too smoothly, and a second change lands: Mara stops trusting the explainer. One room, one activity, two beats.

Granularity has a second axis: size. “Mara doubts Daniel” is too small if it happens in a glance and changes nothing downstream; “Mara learns the truth about the house” is too big, because it is four changes wearing one sentence. The right size is the smallest change that forces another change. When a candidate beat feels padded, split it and see whether both halves force something; when three beats feel thin, merge them and see what single turn they were circling.

How to create a beat sheet from a blank page

Start with the two states you already know something about: how things stand when the story opens, and how they stand when it ends. Write each as a short stack of true sentences. Opening: Mara believes the house has stood empty for ten years; Daniel is the caretaker she has never met; a chipped cup sits drying by the sink. Ending: Mara knows who has been living there; Daniel's account of the decade is in ruins; the cup has an owner. The distance between the two stacks is the story. Every beat is a step across that distance.

Then interrogate the gap from both ends. Of the ending, ask: what must change for this to be possible, and what must change for it to be earned? Mara can't confront the truth of the house until she stops taking Daniel's word for things, so somewhere a beat breaks her trust. Of the opening, ask: what does this state make inevitable? A woman who believes a house is empty will treat every sound as settling timber — until one beat won't let her. Each answer goes on the pile as a candidate. Don't sequence yet. Gather.

When the pile stops growing, sequence it — by consequence, not by the order the ideas arrived. Each beat should be more possible, or more costly, because of the beats before it. Anywhere two adjacent beats could swap places without damage, at least one of them isn't doing causal work yet. As for the count: Snyder's sheet names fifteen beats, but that is a tally of named positions, and yours is a list of changes. The number is a result, not a target. The sheet is done when the opening state, pushed through every listed change, produces the ending state without a leap.

Map it onto the famous sheets to find the holes

This is where the published beat sheets earn their keep. Lay your template-free list beside a known one and read the disagreements. The page numbers are reference marks, not a contract — but they encode where audiences have been trained to expect pressure to change, and a mismatch is worth a question. Why does nothing turn for forty pages here? Why does my biggest reversal arrive three beats before the end, with nothing left to spend it on?

Use three sheets, because they cut at different angles. The Blake Snyder Beat Sheet is the most granular of the three: on a 110-page script it wants the opening image on page 1, the theme stated around page 5, the set-up through page 10, the catalyst at 12, the debate from 12 to 25, the break into two at 25, the B story at 30, fun and games from 30 to 55, the midpoint at 55, bad guys close in from 55 to 75, all is lost at 75, the dark night of the soul to 85, the break into three at 85, the finale from 85 to 110, and the final image on the last page. Lay a list of changes against that and your dead stretches show up as gaps between page numbers.

The eight-sequence structure cuts coarser and asks a different question. It comes from Frank Daniel's teaching and was set down in print by Paul Gulino in Screenwriting: The Sequence Approach — a feature read as eight sequences of roughly ten to fifteen minutes each, about the length of a reel of film in the era the form was set, every sequence carrying its own question and delivering its own answer. Where Snyder asks what happens on page 55, the sequence approach asks what the audience is holding for the next fifteen minutes and whether it gets resolved before a new question opens. Sheets that satisfy Snyder often fail here: the beats are all present, but four of them serve one unanswered question, which is what a sagging second act actually is.

The third lens is circular rather than linear. Dan Harmon's story circle compresses the hero's journey — twelve stages in Christopher Vogler's screenwriting adaptation of Joseph Campbell, The Writer's Journey — into eight movements: a character is in a zone of comfort, wants something, enters an unfamiliar situation, adapts to it, gets what they wanted, pays a heavy price for it, returns to where they started, and is changed. Run your sheet through those eight and you are testing something the page-count sheets cannot see: whether the person going through the changes ends up different, or merely somewhere else.

Two findings come up again and again. Missing beats: the template expects a change your sheet doesn't have — often the quiet ones, the recommitment after a defeat, the cost paid for the midpoint's gain. Dead stretches: a run of scenes where your sheet shows no change at all, which is where drafts sag. Resist the urge to fill a hole by pasting in the template's label. Writing “all is lost” on a card is not a beat; it's a labeled hole. Ask instead what specific thing — a belief, an alliance, an option — could break there, given everything upstream. The three sheets disagree with each other, and the disagreements are the useful part: your beats first, the published sheet second, every mismatch a question rather than a correction.

Beats are decisions you can lock

A beat sheet is not a list of ideas; it is a list of decisions at different stages of commitment. Some lines you know are true — the story doesn't exist without them. Mark those as settled and stop relitigating them every session. Others are placeholders, your current best guess at how Mara gets from doubt to proof. Mark those too, honestly. The worst drafting weeks come from treating the whole sheet as equally firm: you defend beats you never actually chose, and you waver on beats you settled months ago.

This distinction matters double when software is drafting alongside you. In StoryProp the difference has a mechanism: a beat you confirm is promoted to the project's records, and the agent reads those records before it writes anything, so a locked beat stays load-bearing from session to session. A beat you merely floated stays a suggestion. What you agree to gets written down; what you don't, doesn't harden behind your back — decided and drifting stay visibly separate for as long as the project runs.

What an AI beat sheet generator can't decide

Ask any capable AI for a beat sheet and you'll get one in seconds — fluent, proportioned, plausible. This is the appeal of an AI beat sheet generator, and also the trap. The problem isn't quality; it's ownership. A beat is a decision, and a generated sheet is a stack of suggestions wearing the formatting of decisions. If you paste it in and start drafting, you are executing choices nobody made. The sag shows up later, in chapters that are competent and inert, because no one ever wanted anything in particular to change.

Generation is far more useful as interrogation. Ask for five versions of the one beat you're unsure about, and notice which one you argue with — the argument is information. Ask what would have to change between beat nine and beat ten for the ending to feel earned. Ask it to attack your midpoint. Every answer is raw material for a decision that is still yours to make.

If your beats are headed for the screen, the sheet eventually has to survive translation into scenes, sequences and pages — a trip that goes smoother in AI screenwriting software that treats those as structural units rather than formatting chores. The sheet itself stays plain either way; its power is exactly its poverty, too short to hide in and too plain to mistake mood for progress. And the distinction it enforces — a change in what happens, not a change in how a line sounds — is the same one the screen trade writes into its contracts. The WGA's Minimum Basic Agreement defines a rewrite as significant changes to plot, story line or character interrelationship, and a polish as changes to dialogue, narration or action only — and sets its minimum for the structural work at almost exactly twice its minimum for the surface work.

The reason to do any of this is asymmetry. A line on a beat sheet costs a minute to change; the chapters built on it cost weeks. Ten minutes spent discovering that Mara's disbelief needs to arrive earlier is a month of revision you don't do later. Keep the sheet to one page, keep it visible, and let the draft talk back to it — when a scene surprises you into a better change, promote the surprise to the sheet and demote the plan. The sheet serves the story, and it serves best when every line on it is a change somebody actually chose.

A beat sheet only pays off if its decisions survive into the draft. StoryProp is built around that survival: beats you confirm become project records the agent reads before it writes, so what you decided on the sheet stays decided in the chapters.

← All posts