Understand the format before you write a single scene
If a film is a sightseeing trip, short drama is a roller coaster. A film has two hours to build atmosphere. Short drama has about ninety seconds to make someone say "fine, one more episode". Every strange rule in this format comes out of that difference.
This course takes the format apart down to what can actually be measured: what has to happen at which second, where viewers usually leave, why a series runs eighty episodes instead of one feature, and why platforms care about watch time far more than about beautiful dialogue.
You do not have to buy anything. The full material is open on this page. Enrolling only saves your progress and issues a completion certificate at the end.
After this course you can:
Built by the Microdrama AI team from patterns that repeat across thousands of episodes produced and distributed through this platform.
For anyone new to vertical short drama: writers, creators, marketers, or anyone who wants to understand why the format works before helping make it. No technical background and no production experience required.
Vertical short drama is serialised fiction where each episode runs about 60 to 90 seconds, framed upright at 9:16, released as one long series of 60 to 100 episodes. The form appeared in China around 2021 under the name duanju, spread into English-speaking markets through apps like ReelShort and DramaBox, and is now growing quickly across Southeast Asia.
Vertical is not just the camera turned ninety degrees. It changes what can be told at all. A phone screen is narrow, and the more people you put in one frame, the smaller their faces get. Faces are this format's main language.
| Works well upright | Works poorly |
|---|---|
| Faces and held stares | Large crowds |
| A slap, a shove, a gripped hand | Wide fight scenes |
| Small objects: a ring, a letter, a phone, a photo | Long landscapes |
| Two people facing each other | Four people talking in turn |
Film tells its story through landscape. Short drama tells it through faces.
The format gets dismissed because its stories repeat and its conflicts run hot. The repetition is real. But it is not an accident, it is the product of very hard selection by data. The patterns that survive are the ones proven again and again to hold people.
Looking down on it will not make you write better. Understanding why a pattern survives is what lets you use it deliberately, and one day break it for a reason.
This is the most expensive mistake a new writer makes: assuming the platform sells stories. What a platform sells is watch time. Story is the instrument for getting it.
The longer people stay, the higher the chance they see an ad, unlock a paid episode, or subscribe. That is why the algorithm loves stories that make people say "just one more", until suddenly they are thirty episodes deep.
So the early episodes are not merely an opening. They are the sales floor. If the sales floor fails, the product never sells, however good episode forty turns out to be.
Think of the viewer's attention as a balance in a wallet. Every boring second spends it, and it cannot be topped up. The four parts below are how to spend it in an order that has already been proven.
| Part | Length | Its job |
|---|---|---|
| Pickup | 0 to 5 seconds | Close the previous hook, as fast as possible |
| Hook | first 3 seconds | Give a reason not to swipe |
| Body | 15 to 70 seconds | One event happens, not three |
| Turn | about 10 seconds | The situation changes direction |
| Cliffhanger | last 3 to 5 seconds | A new question opens, deliberately unanswered |
In the first three seconds the viewer is not judging quality. They are answering something much blunter: is there any reason not to swipe. Everything there has to serve that question.
Now the viewer asks what is actually going on. Here you supply context, but only as much as needed. Context delivered through action always lands faster than context delivered through explanation.
The viewer has to lose their footing. They must feel they did not get the answer. A good cliffhanger is not a scene cut mid-motion, it is a door opening: someone walks in, a message arrives, a name is spoken.
A waitress, RINA, sets down a plate of rice.
An older man throws a handful of coins on the floor.
OLDER MAN
Pick them up. That is what your dignity is worth.
The whole room turns. Rina holds back tears.
A young man walks in and picks up the coins one by one.
Everyone assumes he is about to join in on the humiliation.
He slips the coins into the older man's pocket.
YOUNG MAN
If you need them that badly.
Plot how many viewers are still watching at each second and you do not get a gentle slope. It falls very sharply at the start, flattens, then falls again at places you can predict.
The exact numbers differ by series and by platform. What does not change is the shape. Your job is not to stop everyone from leaving, which is impossible. Your job is to slow the fall.
| Point | Most common cause | What to fix |
|---|---|---|
| Seconds 0 to 3 | No conflict yet, only mood | Start from consequence, explain the cause later |
| Seconds 5 to 15 | Explanation runs too long | Slip the information in through action |
| Seconds 20 to 40 | The conflict does not develop | Move the win-lose or safe-threatened position |
| Near the end | No promise for the next episode | Close on a question, not an answer |
| Episodes 2 to 5 | Episode 1's promise went unpaid | Pay a little, then open a larger debt |
This is the simplest tool for deciding whether an episode deserves to exist. Compare the character's state at the start and at the end. If nothing changed, the episode can be deleted with nothing lost, and the viewer will notice that before you do.
| Start of episode | End of episode |
|---|---|
| Safe | Threatened |
| Humiliated | Striking back |
| Suspected | Trusted |
| Trusting | Betrayed |
| Unaware | Holding a secret |
| Broke | Given hope |
| Winning | Losing on a new condition |
A SECURITY GUARD blocks MAYA at the turnstile.
GUARD
Fake invitation. Out.
Maya holds up a company badge. The guard laughs.
GUARD
Anyone can print one of those.
The lift opens. Every employee bows.
The CEO walks over. The room braces for Maya to be thrown out.
CEO
Why is my wife standing outside?
The lobby freezes.
Notice the resolution does not close the story. That is the pattern: a small repayment that opens a larger debt. A series that avenges its lead completely in episode five has nothing left by episode six.
If you are arriving from one of these forms, some of your habits will help and some will sink you. This table is for telling them apart.
| Aspect | Film | Soap opera | TV movie | Short drama |
|---|---|---|---|---|
| Main goal | A complete experience | Filling a broadcast slot | Resolved in one sitting | Getting to the next episode |
| Length | 90 to 150 minutes | Hundreds of episodes | 90 to 120 minutes | 60 to 90 seconds |
| Tempo | Flexible | Slow | Medium | Very fast |
| Hook | May be slow | Slow | Medium | Required within 3 seconds |
| Cliffhanger | Sometimes | Not always | Rarely | Almost every episode |
| Dialogue | Natural | Explanatory | Mixed | Very dense |
From film, the habit of withholding information to build mystery. Here that is dangerous. A confused viewer leaves, a curious viewer stays, and the two feel similar in a writer's head while looking nothing alike in the data. Short drama tension comes from consequence, not confusion.
From soap, the habit of letting a scene run three minutes. On a phone, three minutes without a change of direction is a very long time.
From the TV movie, the habit of resolving everything at the end. In serial form, resolving without opening something new is the same as saying goodbye.
The biggest difference between a film writer and a short drama writer is not skill. It is the question they ask themselves while staring at the page.
| Stop asking | Ask instead |
|---|---|
| Is this scene good? | If I were the viewer, would I stop scrolling? |
| Is this line beautiful? | Does this line make me curious? |
| Is this character realistic? | Is this character instantly legible? |
| What theme am I expressing? | Why can the viewer not stop right now? |
| How do I close this episode? | What question am I leaving behind? |
None of this is an invitation to abandon craft. The opposite: a ninety second limit forces clarity, and clarity is a far harder skill than beauty. A writer who can make someone understand and care within three seconds can write anything afterwards.
The episode opens on a city skyline for ten seconds, then the character walks into the office.
The episode opens as the guard pulls the character's access card: "As of today you are fired."
The viewer does not care about that city yet. They care about a problem. Atmosphere can be built after they decide to stay.
The character introduces herself: "Hi, my name is Rina, I have worked here three years."
Rina is fired in front of the whole staff, holding a box of her things.
That one scene tells us who she is, what her problem is, and why we side with her. A verbal introduction supplies information without supplying a reason to care.
The episode ends once every question inside it is answered.
The episode ends just before the answer arrives: a courier hands over an envelope marked DNA results.
Answers close curiosity. Questions move it to the next episode. That is the only reason anyone taps the next one.
Five characters take turns speaking in one boardroom frame.
Two characters face each other; the other three are reactions with no lines.
On an upright screen, more people in frame means smaller faces. Expression is this format's main language, so splitting it five ways is the same as throwing it away.
Episode 12 has the character waiting for news, growing anxious, then going to bed.
Episode 12 has the character waiting for news, receiving the wrong news, then acting on it.
The first has no value change: the end state matches the start. The second moves the character from unaware to wrongly informed, and that fuels the next episode.
A five row table with columns: episode, second of first event, value change, cliffhanger.
A three column table: start state, end state, triggering event.
A one to two page script with a time marker every ten seconds.
You are ready to move on when all of these hold:
No. The entire material is open on this page with no enrollment and no payment. Enrolling only saves your progress and issues a completion certificate at the end.
Yes, and this is the right entry point. This course teaches you to read the format. Writing it is covered in Writing Hooks That Hold and Short Drama Story Structure.
Exact figures vary by series and platform, so what is shown is the shape of the curve rather than a benchmark. What matters is not the numbers but where the drops occur.
