English
Subscribe

Editor's notes

ReelShort spends 55% of its revenue on marketing

We counted all 227 players on the microdrama map. Of the 18 that lead with revenue, 11 are AI tool companies; among the 50 microdrama apps, only 2 do. This piece follows where the money goes once it comes in.

ReelShort spends 55% of its revenue on marketing

China's microdrama market has been counted as larger than the domestic film box office for two straight years, 2024 and 2025. Company accounts tell a different story. ReelShort, the leader outside China, reported $785M in revenue for 2025 and a net loss of about $12M (△). This piece asks one question: where does the money that enters microdrama go? The map covers 227 players along the whole chain of apps, studios, source IP, advertising and AI tools, with research as of October 14, 2026.

This time we did not pick standout cases. We counted the whole map by place on the chain, headline figure and cost field, and three things came out. Of the 227 players, 18 use revenue as their headline figure, and 11 of those are AI tool companies. Among the 50 microdrama apps, only 2 use revenue. And of the 53 players whose cost field is filled in, 11 mention marketing or ad buying, while the cost field reads unverified for 174.

More than half leaves first

  • ReelShort's marketing is 55% of revenue (△). On in-app payments, the app store can take up to another 30%. Most of the revenue seems spoken for before anyone decides how much to spend on the next title.
  • Companies with audited numbers look the same. COL Group's 2025 selling expenses were RMB 953M (+105%), or 57.5% of its RMB 1.657B revenue (✔). Short dramas and IP derivatives grew to RMB 896M (+125%), but total revenue rose only +42.9%. Selling expenses grew faster, and the year ended with a net loss of RMB 671M.
  • China's paid titles lean even further. Press reports put ad buying (投流) at RMB 800K–900K out of RMB 1M in gross revenue, with a break-even ROI of about 1.2x and only 1 in 10 titles turning a profit (△). The map describes paid microdramas as "effectively an ad-arbitrage business."

Where the money lands

  • Few players collect the ad spend. China's 2025 ad buying came to RMB 42.3B (+28% · △), and Douyin's Ocean Engine captures 70–80% of it. ReelShort's 2025 ad spend was split among Facebook 25%, TikTok 19%, Snapchat 16% and Instagram 8%. In effect, a content company's selling expense becomes a platform's ad revenue.
  • Going free does not close this channel. In China, ad buying for free apps surpassed paid apps in 2025 (△). Even when viewers pay nothing, the cost of buying those viewers appears to keep running.
  • Studios' shares depend on platform policy. Hongguo's monthly payouts grew from RMB 100M (Jun 2024) to RMB 500M+ (Mar–Apr 2025), but its rev-share formula is undisclosed. Early 2026 brought reports of canceled small-team guarantees and 3-month settlement cycles (△). In May that year Douyin answered with RMB 1.5B+ in guarantees and a rev-share raised from 70% to 80%.
  • Even large companies report thin margins. Dianzhong, with about RMB 14.5B in 2025 revenue (△), reportedly said its company-wide gross margin is about 10% and its short-drama net margin is under 1% (△).

Count the whole map and app revenue barely shows

  • The 50 apps lead with usage. Of the 50 apps (25 global-facing, 25 domestic), only DramaWave and FlareFlow use revenue as their headline figure. 10 use MAU, 9 use downloads and 27 have no confirmed headline figure. Only 4 apps have a ✔ headline figure.
  • Costs are hidden even more. For 43 of the 50 apps the cost field reads unverified, and only 3 list marketing spend. ReelShort's 55% is itself an MPA estimate, not a company filing. Across the whole map, only 53 players have a cost field filled in, and 11 of those mention marketing or ad buying.
  • Three ways of charging overlap. Of the 50 apps, 30 carry a per-episode coin tag, 27 a weekly or monthly subscription tag and 18 a free, ad-supported tag. 25 of the 50 use two or more at once, so these counts alone cannot tell you the split between paid and free.

The revenue that shows sits with tool sellers

  • 11 of the 18 are AI. The 18 players with revenue as their headline figure are 11 AI companies, 3 source-IP and web-novel companies, 2 global-facing apps, 1 studio and 1 legacy entrant. 9 of them are ✔.
  • Video-generation and dubbing tools post big numbers. Kuaishou's Kling AI is at about $500M ARR (Mar 2026 · ✔) and HeyGen at $200M+ ARR (Jun 2026 · ✔). ElevenLabs is listed at $500M+ ARR (Q1 2026 · △). The more microdramas get made, the more the tools that make them get used.
  • But this is not microdrama money alone. These are company-wide figures that mix in other uses such as advertising, education and corporate video. No company on the map breaks out its microdrama share. Do not read them as "tool makers earn this much from microdrama."

Studios earn from orders, not hits

  • 27 of 29 studios carry a production-fee or commissioned-work tag. Among the 29 studios, 9 carry IP licensing and rights, 5 platform ad-revenue share and 4 per-episode coin payments. 7 use title count as their headline figure, and for 27 the rev-share field reads unverified.
  • The price of a title varies widely. A top-tier Chinese title costs RMB 1.5M–3M, a lower-tier one RMB 100K–300K, and a ReelShort title about $200K with roughly one week of shooting (△). Among long-form studios, Linmon Media disclosed its short-drama revenue most clearly: RMB 129M in 2025 (141 titles · +110% · ✔).
  • Apps carry the hit risk. A studio working on commission is paid per title, while the app does the ad buying and absorbs the profit or loss. So the 55% and 80–90% above are numbers in the app's accounts, not the studio's.

Attempts to narrow the exit

  • Web billing and mixed-in ads. MPA forecasts $1.05B in revenue and $40M in net profit for ReelShort in 2026, and attributes half of the margin expansion to web billing and ads (△). It also sees marketing falling to 44% of revenue. All of this is forecast, and 44% is still close to half.
  • AI cuts a different line. Joey Jia has claimed "AI cuts production costs by ~90%," and more than 80% of DramaWave's new titles are AI-generated. But the big expense above is ads, not production. Kunlun Tech shows a gross margin of about 80%, yet the whole group booked RMB 4.182B in selling expenses (+81.5%) and a net loss of RMB 1.593B in 2025 (✔). Cheaper production alone does not seem to close the exit.
  • More titles may mean more ad competition. Over 95% of China's 128K new Q1 2026 titles used AI, and Kuaishou's AI comic-drama ad spend jumped 100x in the same quarter (△). The map does not claim causation, but if cheaply made titles still have to be sold through ads, it is too early to say AI reduces ad buying.

Four players closed, all with a format that did not fit

  • Quibi did the opposite of today's apps. It launched in April 2020 and shut down eight months later. With famous actors, big budgets and no per-episode payment, the map keeps it as a control case with the reverse of today's revenue model.
  • AI apps did not last long either. OpenAI's Sora 2 app launched in 2025-09, saw installs down 45% in 2026-01 and announced the end of the app and API on 2026-03-24. Pocket TV closed its video microdrama app in 2026-06 and went back to audio, and Korea's 72초TV was a leading studio of the early web-drama era.
  • Regulation targets how apps charge. Of the 4 entries marked "regulatory action," 2 are the US FTC's crackdown on auto-renewing subscriptions and the EU's principles on in-game virtual currency. No FTC action against a microdrama app has been confirmed yet (0 cases), but the map lists weekly-subscription apps as potential targets and calls it open to interpretation whether per-episode coins are covered.

Five questions to ask about a microdrama company

  • Is revenue filed or estimated? Only 4 of the 50 apps have a ✔ headline figure.
  • What share of revenue goes to marketing? Check whether it passes half, like ReelShort's 55% or COL Group's 57.5%.
  • Does it take payments outside the app store? Web billing is one of the few ways to cut a fee of up to 30%.
  • Who carries the hit risk? A studio paid by the order and an app that buys ads see different numbers on the same title.
  • Which line does AI cut? If production costs fall but ad spend stays, the bottom line barely changes.

How to read the numbers

  • Verification marks. ✔ = filings or primary source, △ = press, estimates or app analytics, ✘ = not true. Of the 227 entries, 183 have a △ headline figure and 44 have ✔. ReelShort's revenue and loss are MPA estimates (△), while COL Group and Kunlun Tech come from annual reports (✔), so they do not carry equal weight.
  • COL Group's loss is not ReelShort's loss. COL Group holds 49.24% (47.81% of votes) of Crazy Maple Studio, which runs ReelShort, uses the equity method and does not consolidate it. That same year its in-house FlareFlow had revenue of RMB 350M and a net loss of RMB 470M+ (✔), while Crazy Maple reported 2025 revenue of RMB 5.721B and a net loss of RMB 85.84M (✔).
  • The denominator on "80–90%" is gross revenue. The 80–90% for ad buying is measured against gross revenue, not production budgets. RMB 800K–900K of every RMB 1M in gross revenue goes out as ads, so it cannot be compared directly with production cost. It is an industry-press figure with mixed dates, so read it as △.
  • Profit rates differ by source. "Only 1 in 10 titles turns a profit" sits next to a Jiuzhou Culture insider's "~70% break even · 30% fail" (year unknown · △).
  • The box-office comparison does not use one yardstick. Box office is an official China Film Administration figure (✔), while microdrama "market size" is an estimate that blends ad buying, ad revenue and platform payouts (△). Ad spend may be sitting inside the market size itself.
  • Our own counts and the blanks. Player counts are our tally of the map data by each player's main place on the chain. A tag means that revenue source exists, not its share. 118 entries are graded "?", and 174 entries have an unverified cost field. The pattern above comes from the few companies whose accounts are visible, not from an industry average.

Money that enters microdrama shows up as revenue first at ad platforms and tool makers rather than in the titles themselves. And most apps' profit and loss is still not visible on its own.

Open the Microdrama Industry Landscape