The Sequence Approach: Why Eight Units Beat Three Acts

The framework has a specific lineage. Frank Daniel developed the sequence approach at Columbia Film School, and Paul Gulino documented it in Screenwriting: The Sequence Approach (2004). In its native film form each sequence runs 10 to 15 minutes and operates as a mini-story: its own dramatic question, its own goal, escalating conflict, and a resolution. The approach does not replace three-act thinking; it fills in what three-act thinking leaves blank. Gulino’s classic mapping distributes the eight sequences 2–4–2 across the acts: Sequences 1 and 2 are Act One, ending when the protagonist is locked into the main conflict; Sequences 3 through 6 are Act Two, with the midpoint at the Sequence 4/5 seam; Sequences 7 and 8 are Act Three, confrontation and resolution. Published accounts vary slightly in where they pin the low point — the classic film texts close it at the end of Sequence 6, while this library’s model runs the trough through Sequence 7’s dark night — but the proportions and the jobs are the same either way, and so is the diagnosis the mapping enables: Act Two stops feeling formless once you can see four sequences inside it, each with its own arc, because the writer always has a local target.

The most useful move the approach makes is also its most counterintuitive: a sequence is not a fraction of an act. It is its own complete dramatic unit — with its own internal question, its own small arc, and its own resolution. Sequence 1 asks who is this person and what is their world and answers it. Sequence 5 asks was the protagonist’s whole approach wrong and answers it. When a sequence does its one job completely, the next sequence has something solid to push against. When it doesn’t, the seams show as drag.

The spine nests at three resolutions:

Work at whichever resolution the problem lives at. A saggy middle is usually a major-sequence problem; a scene that won’t land is usually a section-level one.

What unifies a sequence is its dramatic question: a cluster of scenes bound by a single question that opens at the sequence’s start and closes — answered or deflected — at its end. The sequence sits between the scene and the act in the structural hierarchy, larger than a scene and smaller than an act, and its question is subordinate to the story’s central dramatic question. In North by Northwest (Alfred Hitchcock, 1959) the central question — will Thornhill survive and expose the spy ring? — persists across the entire film, while each sequence opens a narrower one: can he escape the UN building without being arrested? Can he survive the crop-duster attack? Can he reach Eve in time? Each subordinate question gets answered, and each answer pushes the story toward or away from the central resolution.

This two-level structure is the mechanism that makes Act Two manageable. The central question creates the through-line the audience tracks across the whole story; the sequence questions create local urgency in every unit. It also supplies the working definition of filler: a scene that contributes to neither level has nowhere to stand.

A sequence question can close three ways. Yes — the protagonist gets what they wanted, and the achievement opens a harder problem. No — the protagonist fails and must adjust. Or deflected — the question becomes impossible to answer as originally posed, usually because the situation has changed. Deflection is the most interesting of the three: it tends to generate the strongest Act Two pivots, because it means the protagonist’s frame of reference turns out to be wrong.

Read top to bottom, the eight units describe a single causal chain — each one creating the conditions the next requires.

The chain is the argument: every sequence is the setup whose payoff is the next. That is why a weak Sequence 1 produces a weak Sequence 2, and why a midpoint that doesn’t truly break the wrong strategy leaves the dark night with nothing to confront.

Rear Window (Alfred Hitchcock, 1954) — a photographer confined to his apartment with a broken leg comes to suspect a neighbor of murder — makes the sequence structure visible. Sequence 1 establishes Jefferies, his apartment, and the courtyard, and opens the question of how he will deal with his confinement. Sequence 2 introduces the neighbors systematically and ends when Jefferies first suspects the salesman; the central question locks in: did Thorwald kill his wife? Sequence 3 pursues early evidence to its first dead end. By Sequence 4, Jefferies has convinced Lisa and Stella — but has no proof. Each sequence runs 10 to 15 minutes, and each ends on an answer that generates a harder problem. None of it feels like filler because none of it is: every sequence is answering something, even when the answer is "no, that approach doesn’t work."

Chinatown (Roman Polanski, 1974; written by Robert Towne) — a private detective’s routine adultery case opens into a water-rights conspiracy — is the contrast case: a film whose sequences feel unusual precisely because they keep defeating the protagonist’s frame. Sequence 1 introduces Jake Gittes as a competent detective in his element. Sequence 2 launches what looks like a routine investigation — is Hollis Mulwray having an affair? — and answers it by the sequence’s end: yes, apparently. The answer is immediately invalidated when the real Evelyn Mulwray appears and Gittes realizes he has been set up; the story’s central question is replaced before Act One is over. This is the deflected question in action: what seemed central turns out to be a decoy, and the true question — who is manipulating Gittes, and why, and at what cost? — opens in its place. Every subsequent sequence refines and darkens it. By Sequence 5 Gittes believes he understands the water conspiracy; by Sequence 6 he believes he understands the personal stakes; both beliefs are wrong. The sequences do not so much answer the central question as reveal how much more complex it is than he thought. Chinatown’s sequences are shorter and more irregular than the classical model — Polanski and Towne work closer to a five-act shape, with the midpoint falling when Gittes finds the bifocals in the pond and the film’s world rotates entirely — but the sequence logic still operates: every cluster of scenes answers a specific question, and every answer opens a worse one.

Prose writers cannot count minutes, so the 10-to-15-minute metric does not transfer; the underlying concept does. Ask of any cluster of scenes: what dramatic question is driving it? When does the reader first feel that question? When does it close — and does the answer force a harder question open? In practice a sequence often corresponds to a chapter group or a named section of the novel. The question does not have to be explicit in the text; it has to be operating. If you cannot identify what question a stretch of chapters is answering, that is itself diagnostic: the section probably lacks internal shape.

Two hazards are specific to prose. First, pacing: film enforces sequence length by clock time, while a novel’s sequence can bloat to a hundred pages or compress to twenty without the writer noticing. Track shape instead of length: the question opens (the reader feels what is at stake), it escalates (the obstacle intensifies before the answer arrives), and it turns (the closing forces the next question open). If all three are present, the sequence has shape regardless of length. Second, interiority: film sequences are almost entirely external, but a prose sequence’s question can operate at two levels at once — the external question (will she leave the marriage?) and the internal one (can she stop believing she deserves this?). The two need not close at the same moment, but both must be felt throughout the sequence. A sequence that answers only the external question reads plot-mechanical; one that answers only the internal question reads static. The strongest sequences in literary prose resolve both, usually through a single action that carries both weights at once.

The approach operates at a different level than most structural frameworks, which is why it works alongside them rather than instead of them. Save the Cat (Blake Snyder, 2005) operates at the story level: fifteen named beats with page positions, telling you the midpoint arrives around page 55 of a 110-page script. The sequence approach explains why the midpoint has the function it does: it closes Sequence 4’s question and opens Sequence 5’s. The beat sheet locates the beats; the sequence framework explains what the beats are supposed to accomplish. Three-act structure answers macro shape — beginning, middle, end, in what proportions, with what turning points — while the sequence approach answers meso shape: the working units inside that macro structure and how they generate forward motion. Three-act thinking without the sequence breakdown is what produces the saggy middle, a single Act Two beat dividing fifty pages that feel shapeless on both sides of it. The Hero’s Journey, in Christopher Vogler’s craft adaptation, operates primarily at the character-arc level: what the protagonist is learning and the stages that learning moves through. The sequence approach is agnostic about character arc; it cares about dramatic questions and answers. Used together, the Hero’s Journey maps the inner arc and the sequence approach maps the outer one, and the relationship between the two arcs is where the richest structural analysis happens.

The framework reads existing drafts as well as it plans new ones, and the protocol is descriptive, not prescriptive: do not use the eight-sequence template as a target. Read the draft and mark where dramatic questions open and close; a sequence boundary is wherever one question closes and another opens. Some drafts divide cleanly into eight sequences; most do not, and the irregularities are where to look. Three problems account for most findings.

A sequence with no clear question is the most common. Scenes accumulate — things happen, characters talk, plot advances — but there is no specific tension the reader is tracking. The stretch feels like padding even when its individual scenes are good, because good scenes without a unifying question do not add up to momentum. The fix is to identify what this section of the story could be asking — what outcome is in doubt, what goal is being pursued — and rebuild the scenes around making that question felt.

A question answered too early is the second. The question closes at scene three of a twelve-scene sequence, leaving nine scenes with nothing to do, and the writer fills them with side plots, character business, and thematic dialogue that feel disconnected from the main story. The fix is to delay the answer or, better, to complicate it: the protagonist thinks the question is answered, then finds out it is not.

A question too large is the third, and it usually appears in Act Two. "Will Jake solve the case?" runs for forty pages without internal shape because it is not a sequence question at all: it is a story question masquerading as one. A sequence question must be answerable inside its unit; if it cannot be, a more specific local question needs to run underneath it. For draft diagnosis across all the structural positions, see Structural Diagnosis, which applies the same logic to every major position.

Structure is only one of the four dimensions of story, and the spine is where the other three attach. Genre fills each position with its own tropes — what Sequence 2 looks like is not the same disruption in a romance, a thriller, and a horror, even though the structural job is identical. Character arc bends the inner journey across the same eight units: the wound planted in Sequence 1 is what the dark night finally confronts in Sequence 7. Technique — point of view, scene, prose — is how each beat is rendered on the page.

Master the eight jobs and the contradictory advice resolves itself: not because one school was right and the others wrong, but because each was describing a different position on the same spine.