Vertical drama economics: what the numbers say and don't

Every post about this format quotes a market size. Almost none of them say which numbers are measured, which are modelled, and which are neither.

What's actually measured?

App store revenue and downloads, because the app stores report them.

That's the reason category figures are reliable and production figures aren't. In-app purchase revenue passes through two payment processors that publish aggregate data, and analytics firms model from that. Nobody publishes what a series cost to make.

So the numbers divide cleanly. Anything about consumer spending inside an app is well evidenced. Anything about what happens before the app is an estimate.

Consumer spending is measured. Production cost is estimated. Most writing about this format doesn't distinguish them.

What do the category numbers show?

Three things, all from named research firms.

That third figure is the one worth sitting with. It isn't a revenue number, it's an attention number, and attention is what precedes revenue in every format that has ever worked.

Omdia's same Q4 2025 data put microdrama app downloads at 733 million against 658 million for long-form streamers. So on both engagement and acquisition, the format is no longer competing for scraps.

Growth is steep too. Q1 2026 brought more than 850 million downloads, up around 140% year on year.

Where this falls short. Every figure above is category-level. None of it tells you whether a series you make will earn anything, because none of it is broken down to the level a producer works at.

Where is the money concentrated?

Two platforms, and this matters more than the growth rate.

ReelShort and DramaBox each generated close to $140 million in in-app purchase revenue in Q1 2026, out of a category total around $750 million. Both crossed 50 million monthly active users in 2025, and together they accounted for roughly 70% of global short-drama in-app purchase revenue in early 2025.

ReelShort took around $1.2 billion in gross consumer spending across 2025.

The two run different models. ReelShort monetises through coin unlocks — free episodes, then micro-payments at cliffhangers. DramaBox runs a subscription tier alongside in-app purchases, which retains longer and extracts less per user.

What don't the numbers tell you?

Four things, and each is a genuine gap rather than something I couldn't find.

Production cost per episode. Estimates in circulation vary by roughly 25 times depending on geography and quality tier, and they come from single-analyst surveys rather than disclosed figures. A range that wide isn't a number, it's an acknowledgement that nobody knows.

Customer acquisition economics. The widely repeated claim that most apps spend around $5 per user and recover around $3 is one firm's estimate. It may well be directionally right. It isn't reported data and shouldn't be quoted as though it were.

What an individual creator earns. Day rates and per-series budgets circulate freely and trace to content sites rather than to anyone's accounts. We're not repeating them.

Whether an individual series pays back. No aggregate exists. The platforms know; nobody else does.

A range that spans 25 times isn't a number. It's a way of saying nobody knows.

Which figures should you stop quoting?

Three, and one of them is in almost every article about this format.

"A $14 billion market." It appears constantly. It's an estimate, and the analysts publishing it note that it's heavily skewed by China's domestic ecosystem — which is a different market with different platforms, different economics and no relevance to a US creator's decision. Quoting it without that qualification overstates the addressable opportunity substantially.

"$150,000 to $300,000 per series." Traceable to content sites rather than to disclosed production accounts, and inconsistent with the 25x range other estimates give.

Per-day actor rates. Single-sourced, and irrelevant to an AI-assisted production in any case.

Where this falls short. None of those is necessarily wrong. They're just unverifiable, and a number you can't source is a number that damages you when someone checks it.

What does the honest picture look like?

Strong category demand, concentrated distribution, and unknown unit economics.

That's a genuinely good position for a creator and a difficult one for a platform. Demand is evidenced. Attention is evidenced. What isn't evidenced is whether running an app pays — and analysts consistently expect consolidation, with most non-China apps unprofitable.

For someone making content rather than running a platform, the acquisition problem is somebody else's. What matters is that the buyers have money and are short on supply rather than demand.

What's knowable on your own side?

Your own costs, and they're the only unit economics you'll get accurate numbers for.

Credit consumption is visible before you generate. On Hexcoded, cost scales with clip length and resolution tier, and generation is priced per second at each tier — so an episode's cost is calculable in advance rather than discovered afterwards. Failed renders return credits automatically, which matters when you're iterating.

Plan First generates a storyboard and script preview without consuming credits at all, which makes the planning stage free.

What we haven't done is publish a cost-per-finished-minute figure, because that requires a controlled test across models, resolutions and re-roll rates that we haven't run. When we do, it'll say what was measured and how.

Category figures are as reported by the named research firms in the periods stated and are not independently verified here. Production cost and unit economics figures described as estimates are exactly that. Current as of the publication date.

More on workflow, continuity and shot planning in Production.