Guide
How to Write a Micro Drama in 7 Steps
A practical guide to writing a vertical micro drama: build a renewable premise, map a 10-episode arc, write 90-second episode cards, and end on a question.
Updated November 13, 2026
A micro drama is not a short film with more episodes. It is a serial built for a phone: vertical, 60 to 120 seconds per episode, and designed so the next episode starts before the viewer decides to leave. That changes what you write first. You do not start with a screenplay. You start with an engine that can produce a new problem every 90 seconds.
This guide walks through the seven steps we use when turning one twist into a 10-episode outline. It works for ReelShort, DramaBox, GoodShort, TikTok, and any other 9:16 series.
What makes a micro drama different
Three things separate a micro drama from a traditional pilot:
- The episode is the unit. A scene can end inside an episode, but the episode must end on a change. If the start and end states are the same, the episode is an excerpt, not a chapter.
- The first three seconds carry the episode. Viewers arrive mid-scroll and leave with one swipe. Every episode needs a visible reason to stay before anyone speaks.
- The season is a chain of questions. A premise gives you one twist. A series needs a machine that turns that twist into new pressure again and again.
Step 1: Build a premise with a renewable engine
Most first drafts fail here. The writer has a strong opening reveal, then spends episodes repeating it. A reveal is a scene. An engine is a sentence you can keep mining.
Write the engine in this shape:
Each time [protagonist] tries to [recurring action], [opposing force] makes them choose between [two costs], while [larger threat] moves closer to [irreversible result].
For a contract marriage story, that might be: each time a forensic accountant tries to prove the marriage is fake, the CEO gives her a real reason to stay, while his family closes in on the clause that would void it.
Test the engine with four questions:
- What do the two leads have to do together, repeatedly, in public?
- Who benefits from the lie, and who is trying to expose it?
- What changes the balance of power between them?
- Which answer creates a worse question instead of closing the story?
If the twist only works once, you have a scene. If it creates a recurring obligation, a public performance, and a private suspicion, you have a season.
Step 2: Give the protagonist a want, a wound, and a lie
Vertical drama plays in close-up, so the audience needs to understand the protagonist fast. Give them three things:
- A want the camera can show: get the money, find the sister, keep the child, clear the name.
- A wound from before episode one: the betrayal, the disappearance, the night they did not speak up.
- A lie they believe about themselves: I am only valuable if I am useful; I cannot trust anyone who has power over me; I deserve the punishment.
The antagonist should have a competing goal, not a personality disorder. The best micro drama villains are people who want the same thing as the protagonist, by a method the protagonist cannot accept.
Keep the cast small. Four to six speaking roles is enough for a 10-episode starter arc. In 9:16, a new face costs screen time; a new location costs production time. Reuse both.
Step 3: Map a 10-episode starter arc
A full micro drama season often runs 40 to 100 episodes. But a 10-episode arc is the right unit to outline first: it is long enough to prove the engine and short enough to see the whole shape. You can stack another 10 episodes on top once it works.
| Episode | Function |
|---|---|
| 1 | Hook, inciting lie, and the public commitment that traps both leads. |
| 2 | Rules of the new arrangement. First ally, first enemy, first small win. |
| 3 | First real reveal. The protagonist learns the cost of staying. |
| 4 | Power shift. Someone else discovers the secret and wants a piece. |
| 5 | Midpoint betrayal. The alliance was never what it looked like. |
| 6 | Fallout. The protagonist loses the thing they were protecting. |
| 7 | Lowest point. A public humiliation or an arrest, not a private sulk. |
| 8 | Counterattack. The protagonist uses the enemy's rules against them. |
| 9 | Climax. The choice between the original want and the new relationship. |
| 10 | Resolution plus a next-season hook that changes the rules again. |
This is a shape, not a formula. The useful part is the escalation logic: each episode should make the previous solution unavailable.
Step 4: Write one episode card per episode
Before dialogue, write a card for each episode. A card is faster to fix than a script and exposes a dead episode immediately.
| Field | Question |
|---|---|
| Start state | Who holds power when the episode begins? |
| Carried question | What did the previous cliffhanger make the viewer ask? |
| Objective | What does the protagonist try to do in this 90 seconds? |
| Obstacle | Who or what blocks it, visibly? |
| Turn | What new fact, choice, or consequence changes the scene? |
| End state | Who holds power now? Which option disappeared? |
| Exit beat | What is the final image, line, or action before the cut? |
| Continuity | Props, injuries, time of day, and location carried forward. |
One turn per episode. If the card has three turns, you have three episodes. If the card has none, you have a recap.
Step 5: Put the hook in the first three seconds
Open on a visible problem, not a mood. A woman crying in a car is a mood. A woman crying in a car while a wedding invitation with her own name on it sits on the passenger seat is a problem.
Useful hook shapes:
- A key that will not turn while someone is inside.
- A bride who is calm at her own wedding, for the wrong reason.
- A doctor who recognizes a scar they should not know.
- A contract with one clause circled in red.
- A stranger who calls the protagonist by a name they buried.
Then re-hook every episode. Assume the viewer paused, got distracted, or swiped in mid-season. The first three seconds of episode two are not a recap; they are a new disturbance that only makes sense because of the recap.
Step 6: End on a question, not a resolution
A cliffhanger is not a cut in the middle of a sentence. It is new information that changes the meaning of what the viewer just watched. The best ones force a reappraisal: the thing you thought was protection was a trap; the person you thought was dead sent the message.
Reliable cliffhanger moves:
- Arrival: someone who should not be here walks in.
- Reveal: a document, scar, or recording contradicts the scene.
- Reversal: the person with power is the one asking for help.
- Choice: the protagonist agrees to something they cannot undo.
- Recognition: a look that says I know you, before the name is spoken.
Do not fake it with a dream, a new character with no setup, or a gun that appears from nowhere. A cheat teaches the viewer not to trust the next episode.
Step 7: Format for the shoot, then read it aloud
Write the season in one document with a clear episode heading for each installment. Use standard scene headings, action, character names, and dialogue. Skip camera directions unless the shot is the story. For the full layout, see our micro drama script format guide.
Then read it aloud with a timer. A 90-second episode is roughly one to one and a half formatted pages. If the read runs long, cut explanation before you cut conflict. If it runs short, do not add a new subplot; add a consequence.
Finally, block for vertical. Put one face in the foreground, stack subjects in depth, and keep important action in the center safe zone. Write dialogue that can survive subtitles: short lines, one thought each, no speeches.
A 90-second example
Here is the shape on the page. The hook is visible before the first line; the turn arrives at 60 seconds; the cliffhanger changes what the hook meant.
EPISODE 01 — THE OTHER BRIDE
INT. WEDDING VENUE, BRIDAL SUITE - DAY
A wedding dress hangs on the door. NOOR (28) sits in front of it,
still in her coat, holding an invitation with her own name printed
as the bride.
She did not send it.
NOOR
(quiet)
Whose wedding am I supposed to be at?
The door opens. HANA (30), her sister, wearing the same dress.
HANA
Mine. You're the witness.
Noor looks at the invitation. Then at Hana's wrist: a hospital band,
cut off, still in her sleeve.
NOOR
You were discharged this morning.
HANA
I checked myself out.
Hana takes the invitation and turns it over. On the back, in their
father's handwriting: "If she remembers, the deal is off."
Noor's phone buzzes. A message from their father, who has been dead
for six years: "Don't let her say I do."
CUT TO BLACK.Common mistakes
- Writing a synopsis instead of a beat sheet. A synopsis says what happens. A beat sheet says when it happens and what changes.
- Three turns in one episode. Pick the turn that costs the protagonist the most and move the other two to later episodes.
- Explaining the twist in dialogue. Let the viewer see the scar, the document, the wrong name. Explanation is the fallback, not the reveal.
- Too many people and places. A new character should either block the want or reveal a secret. Otherwise cut them.
- Resolving the episode. If the last line answers the question the episode asked, the viewer has no reason to stay.
FAQ
How long is a micro drama episode?
Most episodes run 60 to 120 seconds, with 90 seconds a common target. The exact length should follow the platform brief and a timed read, not a page count.
How many episodes should I plan first?
Outline 10. It is enough to prove the engine and short enough to rewrite. Once the first 10 work, extend the same escalation logic into 40 to 100 episodes.
Do I need Final Draft or special software?
No. A standard screenplay layout with a clear episode heading is enough. Final Draft added micro drama and vertical templates in 2026, but the format is not proprietary.
Can AI write the whole thing?
AI can draft the structure, beats, and dialogue fast. You still have to choose the engine, decide what the twist costs, and cut anything that does not change the pressure. Treat the first output as an outline to edit, not a finished script.