Are Short Dramas Getting Longer? The Trend Line Is Real, the Trend Is Not
Group 3,385 official uploads by publication month and the median runtime falls from 61 min 44 sec to 16 min 13 sec in six months. Split the same data by channel and the four with the longest histories move between minus 10 and plus 9 percent. The decline was our sample changing shape, not the format changing length.
We cannot tell you, and the reason is worth more than a yes
We tried to answer this from our own measurements and failed. This page is that failure written out, because how it failed is the useful part.
We read the duration of 3,385 videos on the ten official YouTube channels these platforms run, snapshot taken 1 September 2026, and grouped every video by the month it went up. The line that comes out falls hard — a median of 61 minutes 44 seconds in December 2025, down to 16 minutes 13 seconds by June 2026, then back up to 20 minutes 28 seconds in August. There is a headline sitting right there: short dramas got roughly four times shorter across the first half of 2026, then started growing again.
That headline is not true, and the same data shows why. Split those months by channel and nearly all the movement disappears: the four channels we can see furthest back changed by between minus 10 percent and plus 9 percent over eight to thirteen months, rising in about as many months as they fell. The collapse in the pooled line is our sample changing shape, not the format changing length.
So: we do not know whether short dramas are getting longer. Neither does any page quoting a chart built the way ours was.
The chart that looked convincing
Here is the pooled series as it came out. Median is the duration of a single YouTube upload; we take the upper median so these figures reproduce against our other monthly work.
| Month | Uploads | Channels visible | Median upload |
|---|---|---|---|
| December 2025 | 38 | 2 | 61 min 44 sec |
| January 2026 | 101 | 4 | 26 min 01 sec |
| February 2026 | 125 | 4 | 45 min 56 sec |
| March 2026 | 166 | 5 | 26 min 45 sec |
| April 2026 | 212 | 5 | 21 min 49 sec |
| May 2026 | 275 | 7 | 18 min 15 sec |
| June 2026 | 537 | 8 | 16 min 13 sec |
| July 2026 | 704 | 9 | 19 min 20 sec |
| August 2026 | 837 | 9 | 20 min 28 sec |
Read the first and last columns and you get a clean story. Read the middle two and it falls apart: uploads climb from 38 to 837, and so does the number of channels supplying them, from 2 to 9.
Nine channels did not go from 38 uploads to 837 in eight months. Two channels were visible in December and nine in August. That is not a production boom; that is our own corpus filling in.
Why the early months are nearly empty
Each channel’s history is cut off at a different date, because we hold roughly the most recent few hundred uploads per channel — and a channel that posts fast burns through that budget in weeks.
| Official channel | Earliest upload we hold | Channel median upload |
|---|---|---|
| Pine Drama | 28 May 2022 | 0 min 33 sec |
| FlexTV English | 13 Aug 2025 | 69 min 11 sec |
| GoodShort Moments | 26 Oct 2025 | 20 min 34 sec |
| MoboReels | 12 Jan 2026 | 18 min 11 sec |
| ShortMax (Watch Dramas & Show) | 25 Jan 2026 | 14 min 23 sec |
| GoodShort Episodes | 21 Mar 2026 | 18 min 22 sec |
| ShortMax Multilingual | 17 May 2026 | 1 min 6 sec |
| DramaBox | 23 May 2026 | 21 min 37 sec |
| ReelShort | 1 Jun 2026 | 1 min 22 sec |
| Drama Box English | 8 Jul 2026 | 41 min 36 sec |
Nothing in that middle column is a launch date. Drama Box English did not start posting on 8 July 2026; that is only the oldest upload we still hold, because it posts fast enough that everything earlier fell off the end of our window. Pine Drama reaches back to 2022 for the opposite reason — it stopped posting, so nothing pushed its old videos out.
Now read the right-hand column against the left. The two channels visible back in 2025 are FlexTV English at 69 minutes and GoodShort Moments at 20 minutes, both long-form; the shortest channels still posting arrive only in 2026 — ShortMax Multilingual in May, ReelShort in June. Pine Drama is shorter than either and older than all of them, but it stopped posting in November 2024, so it never reaches the months this page is about. December 2025 is a month made of long-form channels not because December was long-form, but because the short-form channels are invisible that far back.
A trend line drawn across that is measuring our collection method.
The proof: two channels, neither of which changed
The obvious fix is to freeze the channel list, so we did the strictest version: only the two channels present in December 2025, followed through to August 2026.
| FlexTV English | GoodShort Moments | Both pooled | |
|---|---|---|---|
| December 2025 uploads | 24 | 14 | 38 |
| December 2025 median | 66 min 25 sec | 20 min 45 sec | 61 min 44 sec |
| August 2026 uploads | 30 | 85 | 115 |
| August 2026 median | 70 min 27 sec | 20 min 05 sec | 22 min 02 sec |
| Change | plus 4 min | minus 40 sec | minus 39 min 42 sec |
FlexTV English got about four minutes longer, GoodShort Moments forty seconds shorter. Pooled, the two fell by forty minutes.
None of that is a paradox once you see the mechanism. In December, FlexTV supplied 24 of the 38 uploads, so the pooled median was essentially FlexTV’s number. By August, GoodShort Moments was posting 85 against FlexTV’s 30, so the pooled median became GoodShort’s number instead. The median migrated between channels, and the arithmetic recorded that as a collapse in runtime.
A pooled median across channels of different lengths measures who posted most, not how long anything is. That holds even with the channel list frozen, and gets worse as the list grows — ours grows from two to nine. For scale: 86 percent of the August 2026 uploads come from channels with no December history in our corpus at all.
What each channel actually did
Here is the decomposition, one row per channel across its own visible window. We count only months where the channel posted at least eight videos. Pine Drama is not in the table: its qualifying months all fall in 2022 and 2023, and it has posted nothing since November 2024, so it never overlaps the period in question.
| Official channel | Months counted | First month | Last month | Net change | Months up / down |
|---|---|---|---|---|---|
| FlexTV English | 13 | 72 min 05 sec | 70 min 27 sec | minus 2% | 4 / 8 |
| GoodShort Moments | 9 | 20 min 45 sec | 20 min 05 sec | minus 3% | 4 / 4 |
| MoboReels | 8 | 18 min 03 sec | 19 min 37 sec | plus 9% | 5 / 2 |
| ShortMax (Watch Dramas & Show) | 8 | 20 min 04 sec | 18 min 03 sec | minus 10% | 4 / 3 |
| GoodShort Episodes | 6 | 22 min 42 sec | 16 min 37 sec | minus 27% | 1 / 4 |
| DramaBox | 4 | 25 min 01 sec | 20 min 39 sec | minus 17% | 0 / 3 |
| ShortMax Multilingual | 4 | 1 min 23 sec | 0 min 54 sec | minus 35% | 1 / 2 |
| ReelShort | 3 | 14 min 11 sec | 0 min 55 sec | minus 94% | 1 / 1 |
| Drama Box English | 2 | 41 min 21 sec | 42 min 42 sec | plus 3% | 1 / 0 |
Three things fall out.
The four channels with real history show nothing. FlexTV English, GoodShort Moments, MoboReels and ShortMax (Watch Dramas & Show) have eight to thirteen qualifying months each, and land at minus 10, minus 3, minus 2 and plus 9 percent, rising in about as many months as they fall. Over more than a year, FlexTV English moved by under two minutes.
Every channel that does trend down has almost no history. GoodShort Episodes, DramaBox and ShortMax Multilingual come in at minus 27, minus 17 and minus 35 percent, on six, four and four months. DramaBox’s whole case is three consecutive falling months, which a coin produces one time in eight. And minus 35 percent for ShortMax Multilingual is a move from 1 min 23 sec to 0 min 54 sec. Twenty-nine seconds.
ReelShort’s minus 94 percent is real and still not a trend. Its median fell from about 14 minutes to 55 seconds between July and August, and the cause is this page’s own arithmetic running inside a single channel. ReelShort posts long uploads and short clips side by side; clips under three minutes were 40 percent of its June uploads, 34 percent of its July, and 69 percent of its August. Once clips pass half of what a channel posts, the median crosses over to them. That is one scheduling decision, not a drift. We wrote it up where it belongs.
One channel moved 27-fold in a month
How much noise lives inside a single channel: ShortMax (Watch Dramas & Show), month by month.
| Month | Uploads | Median upload |
|---|---|---|
| January 2026 | 15 | 20 min 04 sec |
| February 2026 | 29 | 84 min 47 sec |
| March 2026 | 53 | 54 min 35 sec |
| April 2026 | 61 | 2 min 01 sec |
| May 2026 | 38 | 15 min 20 sec |
| June 2026 | 49 | 10 min 03 sec |
| July 2026 | 37 | 11 min 57 sec |
| August 2026 | 31 | 18 min 03 sec |
March to April is a 27-fold drop, on 53 and 61 uploads — samples large enough that this really is what the channel posted. Then it climbs back, and eight months later it sits within two minutes of where it started.
This channel posts complete dramas as single videos alongside short promotional cuts, and whichever of the two it favours in a month sets its median. In March, cuts under three minutes were 40 percent of what it posted; in April, 51 percent. Eleven points of mix moved the median 27-fold, because a median only reports which side of that split the middle video fell on. That column holds no runtime trend, only a scheduling one.
What would actually answer the question
- Complete channel histories, not a rolling window. Every upload each channel has ever made, so each month is compared against the same publishers. What we hold cuts every channel off at a different date, and those dates line up with runtime — the long-form channels are the ones we see furthest back.
- A per-channel series long enough to beat the noise. One channel swings 27-fold inside a month. Detecting a genuine drift under that needs years, not the four to thirteen months we have.
- In-app episode length, which nobody publishes. Every figure here is the length of a video on YouTube, and a platform’s channel is marketing: clips, trailers, stitched compilations, sometimes complete films. None of that is the same measurement as an episode inside the app. Even a flawless YouTube trend would not tell you whether the product got longer.
Until those three exist, the question stays open — and we would rather say so.
What to do instead
- Distrust any runtime trend that pools platforms together. Ask how many publishers are in each month. If the early months have fewer, the slope describes the dataset, not the category. Ours did exactly that, and we built it ourselves.
- Check the length of the specific thing you are about to watch. YouTube shows a runtime before you click; an app shows an episode count before you start. Either beats a category average, which spans a 126-fold range across these ten channels.
- For a signal of change, watch how much gets posted rather than how long it runs — volume needs no duration model, though it carries the same window problem. We measured publishing pace separately.
- Treat a channel’s format as this month’s decision. ReelShort went from about 14 minutes to under a minute in one month. Whatever a channel is doing now, it chose recently and can unchoose.
Durations and publication dates read from the official YouTube channels of ten short-drama platforms, 3,385 videos, snapshot 1 September 2026. Durations come from the videos themselves, not from titles. Monthly medians are upper medians; per-channel figures count only months with at least eight uploads from that channel. Every channel still posting has its history truncated by our collection window at a different date — the subject of this page. We hold no data on in-app episode length and have not estimated one.