A microdrama script runs on a fixed rhythm: hook the viewer inside the first three seconds, turn the scene once, and cut on a cliffhanger. As of October 2026, a season is roughly 60 to 100 vertical episodes of about 90 seconds each, with the paywall placed around episode 8 to 15 so the strongest cliffhanger lands just before it. Get that skeleton right and the writing becomes repeatable across the whole season; get it wrong and no amount of production saves the retention. This guide lays out the episode skeleton, the season beat sheet, where the paywall goes, the tropes that convert, and a prompt-ready scene format you can hand straight to generation.
The 90-second episode skeleton
Every episode is the same shape underneath, which is what makes a season writable at all. The viewer arrives mid-situation, something shifts once, and the episode ends on a question the next one answers. A short film script format teaches you scenes and sequences; a microdrama needs something tighter, because you have about ninety seconds and a vertical frame, and the viewer's thumb is the edit.
Here is the skeleton as a timing table. Treat the seconds as a rhythm, not a stopwatch.
| Beat | Seconds | What happens | Why it is there |
|---|---|---|---|
| Hook | 0 to 3 | Drop the viewer into conflict already in motion | The scroll stops here or not at all |
| Setup | 3 to 25 | Make the stakes legible fast | The viewer needs to know what is at risk |
| Turn | 25 to 65 | One thing changes: a reveal, a betrayal, a choice | A scene without a turn is a clip |
| Escalation | 65 to 82 | The turn makes everything worse | Raise the cost before the cut |
| Cliffhanger | 82 to 90 | Cut on the unanswered question | This is what pulls the next tap |
The two beats that matter most are the first and the last. The hook earns the watch; the cliffhanger earns the next episode. A common failure in short drama scripts for students and first-time writers is spending the opening seconds on establishing instead of conflict. Start inside the argument, the discovery, or the lie, and backfill context in the setup once the viewer is already leaning in.
Season beat sheet: planning 60 episodes in 5 batches
You do not write 60 episodes in order from a blank page. You write a season beat sheet first: the premise in a line, the cast, the central lie or secret that drives the whole run, and the point where it finally breaks. Then you divide the season into five batches of roughly 12 episodes and give each batch one job.
A workable five-batch arc looks like this. Batch one establishes the world and plants the secret, and ends on the reveal that makes the viewer commit. Batch two raises the stakes and introduces the opposing force. Batch three is the midpoint reversal, where the protagonist's plan fails. Batch four is the long consequence, the run of episodes where things get worse before they get better. Batch five pays off the secret and resolves the central relationship. Planning in batches keeps each twelve-episode block coherent and gives you a natural place to assess retention before writing the next one.
Knowing how to write a micro drama series at this altitude is mostly discipline: one turn per episode, one secret per season, and a reason to tap "next" at the end of every single cut. When you move from the beat sheet into production, the full AI microdrama walkthrough takes one episode from beat sheet to finished cut, and the season planner covers scheduling the batches.
Where the paywall goes and why your strongest cliffhanger sits before it
On most vertical-drama apps the first several episodes are free and the rest are paid, unlocked per episode or by subscription. The paywall usually sits somewhere between episode 8 and 15. That placement is a storytelling decision as much as a commercial one.
The rule is simple: your single strongest cliffhanger goes on the last free episode, immediately before the paywall. The free run exists to get the viewer invested enough that paying feels like continuing rather than starting. So the free episodes front-load the hook, establish the cast quickly, and build to a reveal or a reversal that the viewer cannot leave unresolved. Everything before the wall is written to make the wall feel like the wrong place to stop.
Plan the paywall position while you write the batch-one beat sheet, not after. If your biggest free-run turn lands on episode 6, either the free run is too short or the turn is in the wrong place. Move one so they meet.
A vertical 9:16 microdrama scene, generated from a shot description rather than filmed
Genre tropes that convert
Microdrama is a genre engine, and the audience comes for the tropes. Leaning into them is not lazy writing; it is the contract. The archetypes that reliably convert are the hidden-power reversal (the overlooked person is secretly wealthy or powerful), the revenge arc (a wronged protagonist returns to settle the score), the secret-identity romance (one partner is not who they claim to be), and the supernatural-partner story (a human bound to a shapeshifter or a creature of folklore).
Write the archetype as a frame, then make the specifics yours. The hidden-power reversal works because of the moment the truth lands in front of the people who dismissed the protagonist, so write toward that moment and place it just before a paywall. The revenge arc works on escalation, so each batch should raise what the antagonist stands to lose. Keep the archetypes generic and the characters invented; the freshness comes from the specific lie, the specific cast, and the specific turn, not from the label on the genre.
Writing for AI generation: prompt-ready scene format
If the script is going straight into AI video, write each scene so it is already a generation brief. That means naming, per shot, the framing, the subject and action, the light and time of day, the camera move, and the emotional read of the line, exactly as you would brief a crew. A vertical drama script example that is prompt-ready reads less like a stage play and more like a shot list with dialogue attached.
Use a consistent per-shot block so the whole season is formatted the same way:
EP 07 / SHOT 3
Framing: tight vertical close-up, 9:16, eyes upper third
Subject: LENA realizes the letter is a forgery
Action: she lowers the paper slowly, jaw tightening
Light: cold window light from frame left, dusk
Camera: slow push in, 2 seconds
Line (saved voice: LENA): "You wrote this. You wrote all of them."
Read: quiet, not shouting, the anger underneath
Writing to this format does two things. It keeps the prose out of the generator, so the model gets direction instead of narration, and it makes the script reusable: each shot block becomes an input you swap into a saved episode workflow rather than a scene you re-explain every time. It also forces the discipline of one clear action per shot, which is what keeps a generated scene from reading as generated. For holding the same faces and world steady across every one of these shots, the series bible template is where you lock the references the prompts point back to.
Full example: three consecutive episode outlines
Here is an original, invented three-episode run to show the skeleton in motion. The series premise: a line cook at a failing restaurant discovers the new investor is the sibling who abandoned the family years ago.
Episode 6 (free). Hook: the investor walks into the kitchen and the cook drops a pan. Setup: the staff fawn over the money; the cook says nothing. Turn: in a private moment the investor uses a childhood nickname only family would know. Escalation: the cook realizes the sibling does not recognize her. Cliffhanger: she picks up the staff roster and signs her real surname where the sibling will see it.
Episode 7 (free, pre-paywall). Hook: the signed roster is on the investor's desk, circled. Setup: a tense service while the investor watches the cook work. Turn: the investor pulls her aside and admits the recognition was immediate, that the money is an apology. Escalation: the cook refuses it, because the apology arrived only after it was profitable. Cliffhanger: the investor says the restaurant closes in thirty days unless she agrees to something, and the episode cuts before the condition.
Episode 8 (first paid episode). Hook: "The condition is you let me tell them who I am." Setup: the cook weighs exposing the family story against saving the staff's jobs. Turn: she agrees, but on her own terms. Escalation: the staff overhear a fragment and the rumor starts wrong. Cliffhanger: the oldest staff member, who knew both siblings as children, steps forward and says the version nobody expected.
Notice the pattern. Each episode turns once, each ends on a question, and the strongest free-run cliffhanger, the thirty-day ultimatum, sits on episode 7 directly against the wall. The first paid episode opens by answering the question the free run left, which rewards the viewer for paying within the first three seconds.
Build it in Morphic
There is no template file to download here, and you do not need one. The structure above is the template: a season beat sheet in five batches, a 90-second episode skeleton, a paywall-aware free run, and a prompt-ready per-shot block. Hold those four in a single working document and the whole season stays consistent.
The faster path is to build the script directly where it becomes video. In Morphic, Copilot can read your beat sheet and shot blocks and generate against your saved character references, and you assemble the episodes on the same timeline, so a scripted shot and a generated scene live in one project with nothing exported between them. Save the sequence you run for episode one as a reusable workflow and each later episode is duplicate, swap the shot blocks, run. Start a script in Morphic at studio.morphic.com and turn the first scene into a shot the same day.
Preguntas frecuentes
On a five-beat skeleton inside about ninety seconds: a hook in the first three seconds that drops the viewer into conflict already in motion, a quick setup that makes the stakes legible, one turn where a single thing changes, a short escalation that raises the cost, and a cliffhanger on the cut. The hook earns the watch and the cliffhanger earns the next episode, so those two beats get the most attention.
Most seasons run roughly 60 to 100 vertical episodes of about 90 seconds each. Plan them in five batches of around a dozen episodes, each batch with one job: establish and plant the secret, raise the stakes, reverse at the midpoint, run the long consequence, then pay off. Writing in batches keeps each block coherent and gives you a place to check retention before committing to the next one.
Usually between episode 8 and 15, with the rest unlocked per episode or by subscription. Put your single strongest cliffhanger on the last free episode, immediately before the wall, so paying feels like continuing rather than starting. Write the free run to build to a reveal the viewer cannot leave unresolved, and decide the paywall position while you write the first batch, not after.
It is the habit of ending every episode on an unanswered question rather than a resolution. The escalation makes the situation worse, then the cut lands eight to ten seconds later on a reveal, a decision, or a line that demands an answer. The next episode opens by answering it within the first few seconds, which rewards the tap and sets up the next cliffhanger, so the formula is really a loop.
Per episode, a script is short: about ninety seconds of screen time, which is roughly one page of prompt-ready shot blocks. A full season script is that same page repeated across 60 to 100 episodes, held together by one season beat sheet. Because each episode follows the same skeleton, the length is predictable and the writing scales by swapping specifics into a fixed structure rather than composing each episode from scratch.