Vertical Short Drama Script Handbook: Structure, Rhythm, Hooks, and Format

Maosika Editorial | Last updated

Vertical short drama scripts live or die in the first 3 seconds. This handbook lays out the structure, rhythm, hook placement, and formatting rules that production teams use to keep audiences watching episode after episode.

What makes a vertical short drama script different

A vertical short drama (also called micro-drama) is not a TV script cropped to 9:16. It is a format engineered around the feed: viewers hold the phone in one hand, scroll with the thumb, and decide to stay or leave in roughly three seconds. Every structural choice below is a response to that constraint.

A vertical short drama script is a beat-driven, hook-forward document designed for 1–2 minute episodes (occasionally stretching to 3–4), where each episode opens with a grab, escalates a single line of conflict, and ends on a cliffhanger that forces the next tap.

That definition is worth holding onto. If your script reads like a short film with long establishing shots, internal monologue, and a slow burn, it is not wrong as art — it is wrong for this format.

The one-episode skeleton

Every episode in a vertical drama follows the same three-part spine. Deviations are allowed; abandoning the spine is not.

PartPositionJobCommon failure
Opening hookScene 1Force the viewer to stop scrollingSlow world-building, narration, backstory dump
EscalationMiddle scenesRaise the stakes, land a small payoffScenes that don't change the situation
CliffhangerFinal sceneMake the next episode the only reasonable choiceSoft landing, resolution, "to be continued" text

The middle section is where most scripts fall apart. A scene that does not move the conflict, reveal new information, or pay off a promise from earlier is a scene the thumb will scroll past. Cut it.

The Golden 3 Second rule

The first scene of every episode must open on strong conflict or strong suspense. No exceptions. The audience does not owe you setup time.

Good opening images:

  • A slap landing across a face.
  • A character reading a will that names them the heir — in front of the family that despises them.
  • A door kicked open mid-secret.
  • A line of dialogue that reverses what the last episode ended on.

Bad opening images:

  • A city skyline.
  • A character walking down a hallway thinking.
  • Two minutes of "Previously on…" explanation.
  • A narrator explaining who everyone is.

Beat sheet before dialogue

Before writing a single line of a new episode, write the beat sheet: a short list of what changes, beat by beat, and what the cliffhanger is. Then write the scenes against that list. Writing dialogue without a beat sheet is how you end up with 90 seconds of people talking in a room and nothing happening.

A useful beat sheet for a standard episode looks like this:

  1. Hook beat — the image or line that stops the scroll.
  2. Escalation beat A — the conflict gets worse, or a new fact lands.
  3. Small payoff — a mini-satisfaction (a comeback, a reveal, a piece of evidence).
  4. Escalation beat B — the thing the viewer thought was stable breaks.
  5. Cliffhanger beat — a new threat, a reversal, or an identity hint that demands the next episode.

The small payoff matters. Vertical drama runs on a rhythm of promise and partial delivery. If you only ever delay satisfaction, viewers learn to stop caring. Give them a win every episode — a jab landed, a mask cracked, a clue found — and then rip the floor out from under them at the end.

Rhythm: payoff density across the season

Rhythm in vertical drama is not about individual episodes; it is about the cadence of payoffs across a run. The pattern that holds attention is:

  • Every episode: at least one small payoff (a comeback, a reversal, an identity hint, evidence secured).
  • Every 3–5 episodes: one large payoff (a major identity reveal, a public humiliation reversed, a villain cornered, a long-buried truth surfaced).
  • Across the season: a clear escalation curve — stakes go up, not sideways.

Think of it as a stairway, not a flat line. Each large payoff is a landing. The small payoffs are the steps between landings. If you stretch more than a handful of episodes without a landing, viewers drop off because they no longer believe the show will deliver.

The cliffhanger taxonomy

Not all cliffhangers are equal. Some earn the next tap; some annoy the viewer into leaving.

Cliffhanger typeExampleStrength
Identity reveal cutaway"So you're the one who…" — door slamsStrong, when earned
Reversal of powerThe character who had the upper hand loses it in the final secondsStrong
Physical dangerA car, a hand, a fall — cut to blackModerate; overuse numbs the audience
Information cliffhangerA document, a photo, a text message shown but not readStrong when tied to a long-running question
Fake tensionA character gasps at something the viewer never seesWeak; trains the audience to distrust you

The worst cliffhanger is the placeholder: "To be continued" typed at the end of a scene where nothing actually changed. That is not a cliffhanger. That is a missing scene.

Dialogue rules for the vertical format

Vertical drama dialogue is not how people talk in real life and it is not how people talk in novels. It is engineered to be read off a phone screen in real time.

  1. Short sentences. Most lines should land under roughly 20 characters in the source language; in English, that means tight, single-beat lines. If a character needs a paragraph, they probably need two scenes.
  2. One thought per line. Do not stack argument, explanation, and emotional turn into one speech.
  3. No essay voice. Characters do not lecture each other about themes. If a theme needs to land, put it in a choice, not a monologue.
  4. No narrator crutch. Voiceover that explains what the audience just saw is a sign the scene is not working.
  5. Subtext is fine; confusion is not. A viewer watching on a bus with one headphone in should still follow who wants what and why.

A quick litmus test: read the scene out loud while scrolling it on a phone. If your eye skips lines, cut them. If a line needs to be re-read to make sense, rewrite it.

Format: the script shape the pipeline expects

Whether you are writing for a human crew or feeding an AI production pipeline, the script needs a consistent shape so that downstream roles — casting, art, camera, edit — can parse it without guessing.

A workable per-episode format:

  • Episode title and number at the top.
  • Character list for the episode.
  • Scene headings that carry interior/exterior, location, and time of day (INT./EXT., location, DAY/NIGHT).
  • Action lines that describe only what the camera can show.
  • Character names above dialogue, consistent across the whole season.
  • One clear cliffhanger in the final scene.

Two format failures that break every downstream step:

  • A character called "the cop" in episode 2 and "Officer Li" in episode 7. Pick one name and lock it. AI pipelines and human script supervisors both fail on this.
  • Scene headings that omit time or location. "A room" is not a scene heading. The people (or systems) building the set need to know if it is day or night, inside or outside.

Continuity is a document, not a memory

Continuity bible is a structured record maintained across the entire season, tracking character identities, stable traits, current state (injuries, disguises, revealed secrets), relationships, open and resolved plot threads, episode-by-episode appearance tables, batch-level plot summaries, and prop visual descriptions.

You need one. Not because you are forgetful, but because at episode 60 no human — and no model — reliably remembers that the lead was stabbed in the left shoulder in episode 14, that the villain knows a secret the lead does not, or that a specific jade pendant was introduced as a Chekhov's gun three weeks ago.

Practical continuity habits:

  • Track every plot thread as open or resolved. An open thread that never resolves becomes a reason viewers stop trusting the show.
  • Record character current state separately from base personality. "Loyal friend" is a trait; "now knows the protagonist's secret" is a state.
  • Keep an appearance table — who is in which episode. A character who vanishes for 30 episodes and returns without explanation reads as an error, not a mystery.
  • Log props by their script name. If episode 3 calls it "the old locket," episode 30 must call it the same thing.

Writing in batches without losing the thread

Long-running vertical dramas are written in batches of several episodes, not one at a time and not all at once. The discipline that keeps a 100-episode season coherent is:

  1. Lock the creative brief before writing. Logline, central conflict, arc direction, ending direction, hook rhythm, platform and audience, episode-by-episode synopsis, and notes to the writer. The brief is the contract.
  2. Plan the batch before drafting it. Confirm four things: where this batch goes in the arc, which characters carry it, which threads advance or open, and what the batch-level hook pattern is.
  3. Write beat sheets first, then scenes.
  4. Run a rules check after each batch. Episode titles correct, scene counts match the plan, character labels present, dialogue not missing, no placeholder text.
  5. Update the continuity bible before starting the next batch. The next batch writes against the updated record, not against memory of what you meant.

Past the midpoint of the season, force the ending into view. A script that is still introducing new characters and new threads when it is more than halfway done is a script that will not land its ending.

Common structural traps

  • The endless setup. Ten episodes of "things are about to happen" with no payoff. Viewers leave.
  • The reset episode. After a major cliffhanger, the next episode opens as if nothing happened. This is the single fastest way to train your audience to stop caring about cliffhangers.
  • The villain who only monologues. A villain who never acts is not a threat; they are a speech-giver.
  • The identity reveal with no consequences. If the big reveal does not change what any character does next episode, it was not a reveal — it was a fact.
  • The爽点 without setup. A comeback or reversal that has not been earned lands as noise, not satisfaction.

Where AI helps, and where it does not

AI is useful for the parts of scriptwriting that are repetitive, easy to drift on, or easy to forget: generating a first draft from a locked beat sheet, checking format consistency, flagging missing character labels, surfacing open plot threads from the continuity record, and producing batch-level summaries so the next batch writes against a clean state.

AI is not useful for: deciding what the story is about, choosing which cliffhanger earns the tap, judging whether a line of dialogue lands as human, or knowing when a payoff has been properly set up. Those are editorial judgments, and they sit with the writer and the producer.

A healthy AI-assisted workflow treats the model as a first-draft and continuity engine, not as a showrunner. Lock the brief, lock the beat sheet, feed the continuity record, and then rewrite what comes back until a human would sign off on it.

---

*This handbook is part of Maosika's ongoing series on vertical short drama production. Maosika (猫斯卡) is an AI production operating system for vertical short dramas — it turns a locked brief and continuity record into a repeatable pipeline from script to shot, while keeping审美 and editorial judgment on the human side of the desk.*

About Maosika — Maosika · Professional AI Video Production System. It connects briefing, scripting, look development, shot prompts and delivery into one reviewable pipeline. www.maosika.com