Why Short Dramas Keep Reusing the Same Plot
We sorted 2,221 short-drama videos by genre and found real concentration — plus the most-produced genres post the lowest median views, not the highest.
Draft — not reviewed by a human editor. Automated fact-check returned
NEEDS_FIX.
Yes, the Plots Repeat — and We Can Show You By How Much
Scroll through the comments under almost any short-drama trailer and someone is asking a version of the same question: why is it always the same setup. We do have a classifier for exactly that phrasing — the same regex pass that powers the question-pattern numbers elsewhere on this site, matching things like “why is it always,” “why do they always,” “how many times.” Run against the 11,017 Latin-script comments in our 12,211-comment corpus (297 videos), it catches 7 comments — 0.1%. That is a real pattern, but it is rare, not common: the impression that “everyone in the comments says this” outruns what the data actually shows. Treat that figure as a floor, not an exact rate — it’s regex classification, not a human read of every comment, so misspelled or reworded versions of the same complaint get missed.
What we can hand you is a harder number, from the other side of the same dataset: exactly how concentrated the plots are, genre by genre, across every full-length video we’ve classified.
They are concentrated. Contract-marriage plots alone are 20.0% of the 2,221 full-length videos we sorted by title (444 videos). Billionaire and old-money plots add another 16.3% (363 videos). Those two genres alone are more than a third of the catalog. Add revenge-and-rebirth (11.4%), secret-baby (10.1%), and hidden-identity (10.0%) plots, and the top five genres cover roughly 68% of everything we looked at. At the other end, military and special-ops plots are 1.7% of the catalog (37 videos), genius/prodigy plots are 1.8% (39 videos), and campus/coming-of-age plots are 3.1% (69 videos). So the instinct readers keep voicing in the comments is correct, and it’s not a small effect — five genres do more than two-thirds of the work.
The Genres in the Catalog Most Are Not the Genres People Watch Most
Here’s the part nobody typing a reply in a comment section can see, because it takes the whole dataset to find: how much of each genre shows up in the catalog, set against how well each genre actually performs, view for view.
| Genre | Share of catalog | Median views |
|---|---|---|
| Mafia / power | 7.2% | 14,780 |
| Campus / coming-of-age | 3.1% | 12,725 |
| Military / special-ops | 1.7% | 10,280 |
| Forbidden relationships | 7.4% | 10,224 |
| Billionaire / old-money | 16.3% | 7,400 |
| Contract marriage | 20.0% | 6,671 |
| Revenge / rebirth | 11.4% | 5,273 |
The two genres that make up the most of the catalog — contract marriage and billionaire — sit at the bottom of this table. The genres that make up the least of it — military, campus, mafia — sit at the top. That’s a real mismatch between what fills the catalog and what performs, and it only shows up when you cross catalog share against per-video median views for the same corpus — which is exactly the kind of number a single trailer’s comment section can’t give you.
Our read: this is probably not because audiences only want contract-marriage and billionaire plots. It’s more likely that those two formats are the cheapest to keep turning out at volume — familiar beats, a stock premise, minimal setup. That’s an inference on our part, not something we measured directly. We don’t have production-cost or greenlighting data for this industry, and we’re not going to pretend we do.
What this table can’t tell you
Four things belong next to that table every time it’s cited, or it’s misleading on its own:
- This is title-regex classification, not human review. We sort videos into genres by matching keywords in their titles, and it has known false positives —
donmatchesdon't,captainmatches a hockey captain. Nobody has manually checked the results. - The tails are thin. Military has 37 videos, genius/prodigy has 39, and campus (69) isn’t a large sample either. That’s not enough to support a strong claim about any of the three — a handful of unusually popular uploads can move a median that small on their own.
- Median views aren’t directly comparable across channels, and channels don’t all post the same kind of thing. The ten channels we track have collection windows from 41 days to 917 days, and genres aren’t evenly spread across channels. Some channels post short vertical clips; others post uploads running well past half an hour that read closer to conventional TV episodes. A genre that clusters on one kind of channel isn’t being compared apples-to-apples against a genre that clusters on another.
- Views measure YouTube promotion, not paid viewing. A high view count means a trailer did well as an ad. It says nothing about how many people paid to keep watching inside the app — which is the number platforms actually decide what to make against, and which we cannot see.
One more caveat specific to this table: it has already been corrected once. Before we measured actual video duration, the same analysis put campus at 20,823 median views and military at 17,311. Once we filtered out uploads that turned out to be much shorter than a full episode, both numbers came down to what’s shown above. The mismatch is real, but it is smaller than our first pass suggested, and we’d rather undersell it than have you catch us rounding up. It’s also possible — we can’t rule this out — that what we’re picking up here is a gap in how genres get promoted on YouTube, not a gap in what audiences actually want inside the app. Those are two different claims, and our data only supports the first one.
Familiar Formulas Get More Comments, Not Fewer
If viewers were genuinely tired of the two biggest genres, their comment sections should go quieter. They don’t. Measured as comments per 1,000 views, werewolf/supernatural and contract marriage tie for the top at 0.25, with billionaire and revenge close behind at 0.23 each. Compare that to genius/prodigy at 0.09 and rags-to-riches at 0.14 — under half the engagement per view. That runs backward from what “sick of the same plot” would predict: the more recognizable the formula, the more people show up to type something. Familiarity doesn’t seem to be killing engagement here — if anything, knowing exactly what you’re getting seems to invite comments rather than suppress them. We’d guess this has as much to do with the size of the fan base a well-worn genre has already built as it does with any individual viewer’s patience for repetition, but that’s a guess; the numbers only show the correlation.
“Formula” Isn’t an Insult in This Format
Short vertical video lives or dies on the first few seconds. A genre label a viewer recognizes instantly — contract marriage, secret baby, hidden identity — does real work: it tells someone scrolling exactly what kind of hook they’re about to get before they’ve spent any time on it. That’s not laziness, it’s closer to the format’s basic grammar. A feature film can afford a slow burn. A vertical clip competing against the next thumbnail cannot.
Once you know the handful of genres that make up most of the catalog, you can use that instead of fighting it. Skip the trailer-after-trailer scroll and go straight to the pattern you actually want. That’s what our trope explorer is for — it’s sorted by exactly the categories in the table above.
The Bottom Line
The repetition readers keep noticing in the comments is real and it’s measurable: five genres cover about two-thirds of everything platforms upload as full-length video, and the two most-produced genres post some of the lowest median views in the catalog. We can’t fully explain why platforms keep making more of what performs worse on YouTube — the honest answer is that view counts measure promotion, not the in-app paid viewing that actually decides what gets made, and that number isn’t ours to see.
Data collected 2026-08-18 and rechecked 2026-08-19, covering 2,221 full-length videos across ten official platform YouTube channels. Catalog composition and view counts will shift as channels keep uploading — treat the shares above as a snapshot, not a constant.