A microdrama episode should be 60 to 120 seconds long, with the major platforms clustering around 90 seconds, seasons running 60 to 100 episodes, and total series runtime landing near one to two feature films. Within that envelope, the format imposes harder rules than length alone: an emotional hook inside the first 15 to 30 seconds, a cliffhanger at every cut, and a free-to-paid episode wall after roughly episodes 5 to 10. These are not stylistic conventions. As of 2026 they are conversion mechanics validated by an $11 billion market, and breaking them costs measurable revenue.
The episode length standard, and why it exists
Across ReelShort, DramaBox, ShortMax, and the Chinese duanju ecosystem, episodes run between 60 and 120 seconds, with outliers from 30 seconds up to three minutes. The number is not arbitrary. It maps onto three forces at once. First, attention: average sustained attention in mobile contexts is estimated around 8 seconds, and a 90 second unit lets a story deliver a complete tension-and-turn cycle inside one ambient viewing moment, a queue, a commute stop, an elevator ride. Second, session math: viewers commonly watch 20 plus episodes per sitting in 3 to 10 minute bursts several times a day, so episode length determines how many monetizable decision points fit into a session. Third, the unlock economy: with episodes priced at $0.50 to $1.00, the unit of content must feel small enough to buy impulsively and complete instantly.
Practical guidance for 2026: write to 90 seconds as the default, allow pilot episodes up to two minutes when the premise needs one extra beat of setup, and resist three minute episodes unless your platform's data supports them. Longer episodes reduce the cliffhanger frequency that drives the format's economics.
Season size: why 60 to 100 episodes
Standard seasons run 60 to 100 episodes, with high-volume platforms going to 150 or more. The total runtime, 90 to 150 minutes, mirrors a film, but the segmentation is the business model. Each episode boundary is a potential payment, a potential ad view, and a potential push notification re-engagement. Sixty episodes means fifty plus monetization events per completed viewer, which is why a fan finishing one series commonly spends $20 to $40. Shorter seasons under about 40 episodes struggle to recover acquisition costs; in a market where return on ad spend swings between 0.7 and 1.6 per title, episode count is part of the unit economics, not just the story.
The internal clock of a 90 second episode
The strongest vertical episodes follow a consistent internal rhythm. Seconds 0 to 5: cold open mid-action, no establishing shot, no recap beyond a single line. Seconds 5 to 30: the episode's emotional spike, a slap, a reveal, a betrayal, the beat that would be the act break in television. Seconds 30 to 75: escalation, one or two turns deepening the situation. Seconds 75 to 90: the cliffhanger cut, ideally mid-revelation rather than after it, because the unfinished beat is what the next unlock sells. Producers in the Chinese system describe placing a turn near the 60 second mark explicitly as the purchase trigger. Every beat either deepens the binge or monetizes it; anything that does neither gets cut.
Structural rules beyond length
The hook rule. Episode one is your ad creative. Platforms test opening hooks as paid social material before committing to a series, and 68 percent of US microdrama ad spend goes to social networks, so the first 30 seconds must work out of context, in a feed, with sound off and subtitles on.
The wall rule. The free-to-paid boundary sits after episode 5 to 10 on most platforms. The biggest twist of act one belongs immediately after the wall, not before it; the wall converts on anticipation, and the post-wall payoff justifies the spend and sets up the next stretch.
The subtitle rule. Burned-in subtitles are mandatory in practice. Ambient viewing means sound is situational, and a vertical episode must read silently.
The recap rule. There is no room for recaps. Continuity is carried by visual consistency, the same faces, costumes, and locations, which is why character lock matters so much in AI production.
How AI production changes the format math
For live-action crews, episode length is a cost lever: every extra 30 seconds across 80 episodes is 40 minutes of additional shoot, which is why budgets cluster under $300,000 and shoots wrap in 7 to 12 days. AI pipelines change the lever. Generation cost scales with output seconds, so the discipline shifts from shoot logistics to shot economy: how few generated shots can carry a 90 second episode at full emotional density. In a node-based pipeline like MinionArts Vertex, the beat sheet structure described above becomes literal workflow input, with hook, escalation, and cliffhanger beats mapped to shot generation nodes, so format rules are enforced by the template rather than remembered by the editor. Teams iterating this way can also A/B episode lengths cheaply, testing 60 versus 90 second cuts of the same beats, something physically impractical in live production.
The 2026 answer in one paragraph
Default to 90 second episodes, 60 to 80 episode seasons, a hook by second 30, a turn near second 60, a cliffhanger cut at the end of every episode, and your strongest act one twist right behind a wall at episode 6 to 8. Treat any deviation as an experiment to be measured, not a creative preference. The format is young enough that the rules will keep evolving, and AI production makes testing them nearly free. For the structural why behind these numbers, our upcoming science series covers the attention and reward mechanics in depth, and the MinionArts Vertex template library bakes these format rules into ready-to-run episode workflows.
