China's Microdrama Market: Growth Structure, Platform Competition, and Value Distribution

Evidence cutoff: 9 November 2026. Principal scope: mainland China. Currency: renminbi (RMB), unless stated otherwise.

China's microdrama business is expanding most visibly in the frequency of viewing, the adoption of dedicated applications, and the ability of platforms to turn continuing attention into advertising inventory. These developments do not establish an equally rapid increase in unique national viewers or industry profit. CNNIC's national survey estimates were almost unchanged between the end of 2024 and the end of 2025, while QuestMobile recorded strong growth in dedicated applications during 2026. The two observations describe different populations and behaviors. Together they suggest that distribution and engagement deserve at least as much attention as first-time adoption. CNNIC, 55th report, PDF pp. 7, 45, 61–62; CNNIC, 57th report, PDF pp. 9, 40–41, 73–75; QuestMobile, autumn 2026, app-growth chart

The commercial consequence is a shift in bargaining power toward businesses that can repeatedly bring audiences back, sell their attention, and organize a dependable supply of rights-cleared content. Hongguo demonstrates the scale possible with an integrated free-viewing application. WeChat's aggregation of third-party dramas shows a competing route through an existing social application. Kuaishou's advertising disclosures show that a distributor can benefit from content suppliers' marketing expenditure even when those suppliers' own returns remain unknown. Producers therefore need to evaluate the revenue base, recoupment order, cash schedule, and usable rights behind a proposed sharing percentage. A large audience or a successful title alone answers none of those questions. QuestMobile, February 2026, pp. 6, 11; Beijing Daily's WeChat reporting, 15 October 2026; Kuaishou, Q2 2026 results, online marketing services

The National Radio and Television Administration, or NRTA, defines microdramas in its rules effective 1 September 2026 as series with episodes shorter than 20 minutes, a clear central theme, continuous and complete storytelling, and distinct characters. The rules span internet services and television distribution and explicitly address AI-generated productions. This report includes scripted live-action and animated microdramas distributed domestically, in vertical or horizontal formats. Promotional clips are acquisition material, rather than additional completed dramas. General short videos, livestream commerce, and 20–35-minute “mid-length dramas” do not enter the content total merely because they appear in the same platform presentation. NRTA Development and Management Measures, Articles 2 and 34; 2026 platform-policy speeches, Mango TV section

In practice, formats differ. QuestMobile describes vertical live-action and AI animation commonly running 1–3 minutes per episode, and horizontal live-action productions commonly running 5–15 minutes. A mini-program is a lightweight application operating inside a larger application such as WeChat or Douyin. It can handle playback, purchases, and advertisements without a separate app installation. An independent application has its own destination and return habit, even when it receives substantial traffic from an affiliated short-video platform. These distribution forms do not determine the payment model: each can support paid viewing, advertising, or a combination. QuestMobile, February 2026, pp. 4, 8; Douyin microdrama solution, introduction and monetization

The following structure uses one economic perimeter throughout: domestic consumption of microdrama content and the advertising or commissioning services attached to it. Production, licensing, distribution, and media buying are activities within that chain. Their receipts explain value distribution; they must not be summed as independent additions to final demand.

This table is an analytical accounting framework, grounded in the distribution and contract mechanisms in the source set. It is not a reconstruction of any publisher's proprietary market-size methodology. A top-up is cash entering a wallet; it may still represent an obligation to deliver episodes. Gross transaction value records a sale before some deductions. A platform's recognized revenue depends on the service it provides and whether it acts as principal or agent. A producer's recognized licensing revenue may arrive only after delivery or settlement confirmation. Net revenue is consequently unusable as a comparison label unless the deductions are stated. iReader's audit procedures explicitly distinguish top-ups, virtual-currency consumption, and reported revenue, while COL's accounting notes distinguish delivery and settlement conditions. iReader 2025 annual report, pp. 78–79; COL 2024 annual report, pp. 137–138

The frozen materials do not support a single fully reconciled monetary market total. COL's 2024 annual report cites an external estimate of RMB50.5 billion for 2024 and a then-forecast RMB63.43 billion for 2025. A Peking University National School of Development report presents RMB37.39 billion for 2023, RMB50.4 billion for 2024, and about RMB100 billion for 2025, based on leading companies' transaction flows and third-party data. It forecasts that 2026 could reach RMB200 billion. Separately, a Beijing regulator's June 2026 release calls the 2025 national total RMB100 billion of output value, with Beijing contributing RMB58.57 billion. These are not interchangeable audited final-expenditure totals. COL 2024 annual report, p. 19; PKU industry and employment report, pp. 2–3; Beijing regulator's report release, 25 June 2026

The RMB50.4 billion versus RMB50.5 billion difference is small relative to the much larger definitional uncertainty. Averaging the RMB63.43 billion forecast and RMB100 billion later estimate would create a number with no measurement basis. Likewise, agreement between two RMB100 billion headlines does not show independent verification when their constructions and underlying data overlap or remain unspecified. The defensible conclusion is rapid commercial expansion in the cited industry series, accompanied by a lack of sufficiently disclosed accounting to value the whole domestic market on one clean expenditure basis. PKU report, p. 3; 2025 association white paper, pp. 4, 14

CNNIC estimated 662 million microdrama users at December 2024 and 664 million at December 2025. Their shares of internet users were 59.7% and 59.0%. The reported counts imply only about 0.3% growth, while internet users increased. Both surveys cover mainland residents aged six and above reached through fixed and mobile telephone sampling frames, with weighting to population characteristics. The estimates support a picture of broadly stable measured national reach over that year. They do not establish an exact zero-growth population, and the reports' general sampling-error descriptions cannot be mechanically converted into a significance test for this particular two-wave difference. CNNIC, 55th report, PDF pp. 7, 61–62; CNNIC, 57th report, PDF pp. 9, 73–75

CSM's January 2026 survey, published later in 2026, reports that 70.2% of internet users aged 18 and above watched microdramas by the end of 2025. Its age restriction differs from CNNIC's. CSM also reports 84.2% willing to pay in the future and 71.8% who had watched paid microdramas in the past year. Neither establishes the current share of active users actually paying a measured amount: watching paid content, willingness to pay, and verified payment conversion are separate events. The public findings do not supply enough sample and weighting detail to reconcile them numerically with CNNIC. CSM annual user findings, 11 August 2026

QuestMobile measures channel activity. Its February 2026 report gives 335 million monthly active users for dedicated apps, 427 million for WeChat mini-programs, and a deduplicated union of 718 million. The app series selects apps with more than 100,000 MAU in February 2026 and looks back at those apps' February 2025 activity. The mini-program series instead applies a more-than-one-million-MAU threshold separately in each February. Its reported year-on-year growth, 74.3% for apps and 59.9% for WeChat mini-programs, therefore includes different selection mechanics. The union is an application-channel measure, not a directly comparable replacement for CNNIC's national viewer estimate. QuestMobile, February 2026, p. 6 and footnotes

The same February numbers imply approximately 44 million people active in both measured channel groups: 335 + 427 − 718. Adding the two channel totals would overstate their union by that amount. May's chart broadens the union to include Douyin mini-programs: 851 million across apps, WeChat mini-programs, and Douyin mini-programs, compared with individual totals of 402 million, 508 million, and 14 million. Their sum exceeds the union by 73 million channel memberships; with three groups this excess is not itself a unique count of overlapping people. The chart separately records 472 million users who browsed microdrama-related content on the main Douyin and Kuaishou apps. Those are content-reach users, outside the 851 million application's union; adding them would count an unknown overlap and confuse browsing related clips with viewing a complete drama. QuestMobile marketing report, May 2026 channel chart

By September, QuestMobile's dedicated microdrama-video app category had reached 487 million MAU. Hongguo's live-action app recorded 429.69 million, equivalent to roughly 88.2% penetration within the category, not an exclusive audience or revenue share. The September app count is a current channel indicator. It is not a 2026 completed annual result or proof that 487 million people are new to the format. The same report finds 380 million Hongguo users also using the main Douyin app, 88.5% of Hongguo's users. The overlap strongly supports an affiliated distribution advantage and continued co-use; it does not prove that every overlapping user was acquired by Douyin or that Douyin lost no viewing time. QuestMobile autumn report, September app-category, ranking, and overlap charts

QuestMobile's January 2025–May 2026 chart shows monthly app use per user rising from 13.0 to 26.9 hours, and usage occasions rising from 62.0 to 125.6. Executed arithmetic gives increases of 106.9% and 102.6%, respectively. Dividing hours by occasions yields about 12.6 versus 12.9 minutes per use, only 2.1% higher. Thus the measured increase in app intensity is predominantly associated with more occasions, rather than a doubling of duration per occasion. These are aggregated usage ratios, not evidence that each individual doubled their visits; changes in the user mix can also move them. They are not episode completion rates. QuestMobile marketing report, time and frequency chart

This changes the product question. A compelling first cliffhanger can sell a single unlock, but an advertising business needs users to return repeatedly and continue far enough to generate useful inventory. Search, saved viewing progress, series discovery, reminders, and a reliable supply of relevant titles can make that return less dependent on purchasing the next impression. Douyin's documented return mechanisms include feed cards, a sidebar, search, and subscription messages. The platform also exposes attributed advertising-yield data to help operators estimate acquired users' lifetime value. Those tools connect distribution decisions with monetization; they do not guarantee favorable economics. Douyin microdrama solution, return traffic and advertising data

A useful decomposition is:

User payments = active viewers × paying conversion × paid consumption per payer, after specifying the window and adjusting for refunds and unused credit.

Advertising receipts = active viewers × return frequency × viewing depth × eligible ad opportunities × fill rate × realized price, with intermediaries' deductions identified separately.

The factors interact. A paywall can reduce viewing depth while increasing revenue from the people who continue. More ads can raise immediate yield while reducing future visits. A higher production budget helps only if the resulting improvement in conversion, retention, sponsor value, or licensing exceeds its cost. The source set measures several audience and use variables but lacks a common national series for realized payment conversion, ad fill, advertising price, or cohort retention. It therefore supports a causal decomposition and explicit scenarios, rather than a numerical attribution of every renminbi of sector growth to those factors.

The association's 2025 white paper reports more than 140,000 vertical-drama launches across short-video, mini-program, and app channels in January–August 2025, but explains that re-editing and repeated launches inflated this count. It estimates roughly 40,000 independently completed vertical productions for the full year. Ordinary live-action production budgets were reportedly rising toward RMB400,000–700,000 per title, with more productions exceeding RMB1 million. These findings undermine the assumption that low entry costs necessarily mean falling costs for a differentiated finished product. 2025 association white paper, pp. 6–8

IP owners can use web novels as a pool of tested premises and serial characters. “IP” in this industry often means a commercially adaptable story, character, or fictional world, together with the relevant intellectual-property rights. Reading popularity may reduce uncertainty about a premise; it does not establish demand for its screen adaptation. AI expands the number of feasible productions and can help visualize fantasy settings, but greater supply also increases the number of titles competing for attention. Kuaishou reports an increase of more than 500% in its supply of live-action and AI dramas between January and June 2026; its Q2 microdrama advertising expenditure more than doubled year on year. The former describes supply and the latter marketers' expenditure on Kuaishou, not a measured doubling of audiences, consumer receipts, or producers' profit. QuestMobile, February 2026, pp. 4, 19; Kuaishou Q2 2026 results, retained official-release text

Growth is also uneven. Hongguo's September MAU was up 81.8% year on year, while Hema Theater was down 40.3% and Xingya Free Dramas down 76.8%. Xifan grew 123.9% year on year to 8.34 million, yet remained below its 11.018 million May reading. That combination illustrates why a favorable year-on-year rate can coexist with a more recent decline. The data show app-level adjustment, not confirmed company exits. The autumn report also describes short operating lives for many apps and mini-programs. A winner's growth cannot safely stand in for the experience of the median distributor. QuestMobile, May ranking; QuestMobile, September ranking and section 2.7

Demand does not unambiguously favor the cheapest or most repetitive output. CSM reports that more than 35% of surveyed users criticized homogenized plots, unrealistic stories, or insufficient originality; only 11.8% preferred a fixed studio or account's output. These results suggest room to build recognizable series and production brands, but weak existing studio loyalty. Their value must be demonstrated through return viewing and licensing demand rather than assumed from the word “IP.” CSM annual user findings

The most useful competitive grouping separates the operator of the audience relationship from the supplier of a title. Douyin can sell promotion to a drama distributor, host a mini-program, and send audiences toward Hongguo, a dedicated free-viewing app in the same group. WeChat hosts many third-party operators and also offers an official aggregation destination. Kuaishou combines native drama programming, advertising distribution, and its Xifan application. Tencent Video, iQIYI, and Youku bring subscription and horizontal-drama relationships. COL and iReader are relevant because they combine IP, production, and distribution, even though their listed-company revenue is not equivalent to revenue of a domestic consumer platform. Douyin solution; QuestMobile February report; WeChat reporting; COL 2025 annual report, pp. 30–32; iReader 2025 annual report, pp. 13, 17

The table compares mechanisms, not matched experimental results. The sources do not report equivalent cohorts, genres, acquisition campaigns, and contract costs across platforms. Audience composition differs: QuestMobile's February sample has 51-plus users at 31.2% of WeChat mini-program users versus 13.9% of app users. The CSM adult survey, the Beijing consumer survey, and QuestMobile's device/channel data also describe different populations. A higher completion rate or lower acquisition cost on one platform could therefore reflect the selected audience, genre, or launch stage as well as a platform effect. QuestMobile, February 2026, p. 7; CSM findings; Beijing report release

Douyin's documentation makes three sources of traffic explicit: organic links from videos and livestreams, commercial promotion, and return traffic. “Touliu,” a common Chinese industry term, means spending on media distribution, often optimized against conversions or expected lifetime revenue. It is not a production expense and does not necessarily mean that every user is purchased. Organic distribution still requires creative labor, accounts, rights clearance, and continued eligibility for recommendations. Search is commercially different again: a viewer seeking a named title or actor may already have intent that a purchased impression must first create. Douyin microdrama solution, traffic and promotion sections

The affiliation with Hongguo offers a place to retain the viewing relationship beyond a single promotional clip. The September overlap figures show extensive use of both applications, while the dedicated app's scale creates a large market for content suppliers. This is a plausible mechanism for distribution advantage, but the source set does not reveal a transfer price for internally supplied traffic or a counterfactual cost of acquiring those users elsewhere. Investors should not treat affiliated traffic as costless simply because the cash does not leave the corporate group. QuestMobile September overlap and ranking charts

Hongguo also helps organize production. The PKU study describes commissioned work, copyright partnerships, and completed-title submissions, with differing levels of platform involvement and guarantees. Its discussion of seven-day exclusive first-release windows describes a reported cooperation practice, not a universal contract. For a producer, a guarantee transfers some financing risk to the buyer but may restrict rights and upside; a completed-title submission preserves independence while leaving production risk with the supplier. Exclusivity can make a platform's catalog more distinctive, while non-exclusive rights permit additional recovery windows. The right comparison is the expected recovery across all permitted windows, less incremental delivery and marketing costs. PKU report, p. 6

In March 2026 Hongguo announced an expectation of more than 40% growth in its annual content budget and plans for greater sharing transparency. In April, Douyin Group announced RMB500 million of targeted support for live-action drama innovation and realistic subject matter. These announcements are evidence of procurement priorities and intended spending, not proof of the amounts ultimately disbursed. Nor does a dashboard displaying a settlement amount necessarily reveal the coefficient connecting viewing to that amount. October reporting contains a producer's complaint about precisely that distinction. The interview identifies a contract-information problem; it cannot establish that every supplier lacks transparency. Platform-policy speeches, Hongguo section; China Business Journal, April announcement; October reporting on Hongguo

Hongguo's proposed withdrawal of viewing rewards creates a particularly useful test of the business. According to the October report, its 21 October notice planned to stop offering coins to newly activated users from 1 November and stop issuance to all users before the end of December, preserving existing balances and providing redemption arrangements. By this report's cutoff, the first planned date had passed but the transition was incomplete; the frozen sources do not measure realized retention or revenue after the change. Removing a subsidy can improve contribution if ad receipts hold up, or reduce it if lost viewing exceeds the saving. Hongguo also denied a circulating claim that annualized revenue was approaching RMB40 billion. That rumor supplies no valuation input. October Hongguo reporting, notice and denial

The October Beijing Daily report describes Lüpao Pao, literally “Green Bubble,” as a free service inside WeChat, with inserted advertisements, an immersive feed, friends' viewing recommendations, regional popularity cues, and small-window playback while using messaging. The quoted user agreement identifies Tencent as the service counterparty and third parties as the drama suppliers. This division is economically consequential: the aggregator can improve discovery and monetization without purchasing every production outright, while suppliers gain access but may lose some direct recognition by the audience. Beijing Daily/Xinhua, 15 October 2026

The report also observes the same titles appearing on Hongguo and Lüpao Pao. That is concrete evidence of catalog overlap, not a measure of the proportion of content licensed non-exclusively. Such overlap allows competition over recommendation, convenience, ad load, and relationships even when viewers see some of the same stories. It also gives a rights holder an alternative outlet, subject to the actual territorial and window restrictions. WeChat's social signals may extend a title's discovery beyond its initial launch, but this remains a mechanism to test rather than a demonstrated universal retention advantage. Beijing Daily/Xinhua

QuestMobile's September ranking already records Lüpao Pao at 156.35 million MAU, while the October article frames it as a recent launch and describes a prior official mini-program whose entrance could no longer be found. Those observations do not provide a precise public launch or renaming chronology. The ranking's own footnote says missing year-on-year comparisons can reflect new launches, renaming, or a previously very small user base. It is safer to use September as the measured activity date and October as the interface-observation date, rather than infer that all 156 million users were acquired after an October launch. QuestMobile September ranking chart; Beijing Daily/Xinhua

The opportunity for third-party mini-program operators is therefore mixed. The host reduces installation friction, but a centralized free destination may compete for the same viewers and suppliers. A producer with transferable rights may welcome that additional buyer. An operator whose principal asset is a paid-acquisition funnel may face a harder comparison when a viewer can find a similar story free within the host. The source set does not establish how far such substitution has progressed, but May's Douyin mini-program activity was down 83.3% year on year even while dedicated apps and WeChat mini-programs grew. There is no single “mini-program model” moving in one direction. QuestMobile May channel chart

Kuaishou's Q2 statement links increased short-drama supply to increased marketing demand. This is an important counterweight to the idea that platforms prosper only by taking a share of viewer payments: a media seller can be paid when a supplier launches a campaign, before that supplier knows its final content return. The statement does not disclose the supplier's realized acquisition cost, profit, or retention. Xifan's September 8.34 million MAU and 6.32 million overlap with the main Kuaishou app show that an affiliated route exists, but its scale remains far below Hongguo's. Corporate traffic resources alone do not guarantee equivalent app adoption. Kuaishou Q2 results; QuestMobile September ranking and overlap charts

Subscription-oriented video platforms seek content that contributes to their larger service. In March 2026 Tencent described unified sharing rules for exclusive horizontal long-, mid-, and short-format projects: under pure sharing, the portion above RMB1 million earns an additional 15% incentive; eligible high-performing projects can receive longer sharing periods; series incentives depend on specified popularity, timing, and continuity of creative team and fictional world. This makes continued exploitation of a successful title more valuable and can reward repeatable production relationships. It also means the “box office” named in a sharing policy is a contractual settlement measure, not cinema ticket sales or necessarily the title's raw consumer collections. Tencent policy speech

QuestMobile's April compilation describes iQIYI's use of platform revenue multiplied by a sharing percentage, with different terms for exclusive first releases and other works, and Youku's emphasis on effective playback and new-member contribution. It is a research institute's summary of public policies, rather than an executed agreement for every title. The percentage alone cannot be ranked against Hongguo's settlement amounts or a mini-program commission: the attributable revenue, eligibility, content budget, period, deductions, and guarantees all differ. Similarly, the iQIYI speech's cost-recovery arrangement for long dramas, and Mango's 20–35-minute procurement terms, must not be quoted as standard microdrama offers. QuestMobile, February report published April, p. 18; 2026 platform speeches

Free viewing consequently has a strong demonstrated distribution position, but the evidence does not establish its inevitable dominance in profit or every audience segment. CSM respondents favored platform membership among paid options and also expressed interest in paying to remove ads; roughly four in ten expressed interest in ad-supported or sharing-based free access. A hybrid can let price-sensitive users fund viewing through advertisements while higher-value users buy convenience. Its success depends on whether subscription receipts exceed displaced advertising and added service costs, not just whether both payment buttons and ads are present. CSM findings

“Fenzhang” means revenue sharing or settlement under a contractual formula. “Baodi” is a minimum guarantee, whose cash timing and recoupability must be read from the contract. Neither term tells the reader whether the amount includes advertising, membership attribution, refunds, or acquisition expenses. A producer may receive a fixed fee while the commissioning platform takes the demand risk; a rights holder may receive a royalty ahead of profit; and a distributor may finance promotion and recover it before the producer receives a residual. These structures can give participants very different returns on the same successful title. PKU report, p. 6; Tencent policy speech; Douyin solution

The table describes mechanisms rather than a universal set of contracts. Douyin provides a concrete example of why the timing and recipient matter: its promotion program lets developers choose a 1%–29% payment-sharing commission for participating creators and a 0–3-day maximum period after which the platform automatically initiates that sharing. The separate transaction system specifies settlement at T+7 after payment. The first range is not Douyin's overall fee, and the shorter period does not mean the developer receives all consumer cash in three days. Douyin solution, promotion tasks and transaction system; Douyin operating rules, payment requirements

Two different advertisements can appear in the same economic chain. One is bought by a distributor to bring a user into a drama; another is shown to that user to finance free viewing. For the media seller, both may be revenue from selling advertising. For a study of final demand for microdrama viewing, acquisition is an input cost funded from the operator's capital or eventual receipts. Adding the user's payment, the distributor's acquisition bill, the platform's settlement, and the producer's revenue would count the same flow several times. Conversely, an external brand's separately purchased campaign represents a distinct demand source, although a sponsorship bundle still needs to be separated from any distribution expenditure already included in its price.

COL, the listed company known in Chinese as Zhongwen Zaixian or Chinese Online, and iReader, known as Zhangyue, provide useful evidence because their accounts expose costs and cash flows. They are diversified groups operating across formats and geographies. Their results cannot establish a domestic short-drama industry margin.

Sources: COL 2025 annual report, p. 14; COL Q3 2026, pp. 1–2, 11; iReader 2025 annual report, pp. 6–7; iReader Q3 2026, pp. 1, 13. Annual and nine-month rows are deliberately not treated as equal-duration growth observations.

COL's short-drama and IP-derivatives segment generated RMB896.0 million in 2025, up 125.02% on the reclassified 2024 comparison of RMB398.2 million. Its 38.41% gross margin was down 8.95 percentage points. The 2024 original report used a different “IP derivative development products” category, with RMB454.9 million of revenue. Substituting that older category into the new growth calculation would splice unlike definitions. At group level, 2025 gross profit was RMB599.3 million and selling expense RMB953.2 million, leaving a RMB353.9 million deficit even before other operating and non-operating items. COL attributes the increase in selling expense primarily to overseas promotion, and foreign revenue was 40.98% of its total. The figures show a capital-allocation and acquisition challenge; they do not show that its domestic short-drama segment lost that amount. COL 2025 annual report, pp. 31–35; COL 2024 annual report, pp. 29–30

The ownership boundary also matters. Crazy Maple Studio, associated with ReelShort, ceased being consolidated by COL from May 2023. COL's 2023 report describes a 49.16% ownership interest but only 47.81% voting rights and three of seven board nominations following the governance change. It records a RMB55.9237 million non-recurring profit effect from deconsolidation. Ownership, control, equity-accounted profit, and consolidated sales are different measures. ReelShort's platform receipts therefore cannot simply be added to COL's consolidated revenue or classified as Chinese domestic sales. COL's own FlareFlow expansion is relevant to where the group deploys promotion and production resources, but its overseas registrations and receipts remain outside the domestic audience and spending totals. COL 2023 annual report, pp. 23, 92; COL 2025 annual report, pp. 30–35, 42

iReader's derivatives segment, which includes short dramas, grew to RMB1,854.9 million in 2025 and reported a 68.00% gross margin. Yet group selling expense was RMB2,078.6 million, of which RMB1,985.6 million was promotion. Group gross profit less selling expense left only RMB162.2 million before management, research and development, and other items. The segment's reported margin thus cannot be used as the distributor's return after acquisition. Management also identifies the decline in digital-reading revenue and relatively rigid associated costs as a drag. Several businesses contribute to the group result; allocating all promotion or all losses to microdramas would be as misleading as ignoring those expenses. iReader 2025 annual report, pp. 13, 16–19, 78–79, 212

The latest figures show adjustment rather than a uniform collapse. In the first half of 2026 COL's short-drama and IP-derivatives revenue rose 108.72% to RMB411.3 million with a 48.62% gross margin; group net loss narrowed to RMB43.0 million and operating cash flow turned positive at RMB126.8 million. The company attributes improved cash inflow mainly to overseas short dramas. Through September, however, group revenue was down 17.64% year on year, and Q3 alone was down 43.95%; the nine-month operating cash flow was only RMB42.8 million. Subtracting the half-year amount implies a Q3 operating cash outflow of RMB84.0 million. This is a period calculation, not a disclosure of domestic title-level cash recovery. COL H1 2026, pp. 8, 21–22; COL Q3 2026, pp. 2, 11

iReader's derivatives revenue reached RMB1,185.1 million in H1 2026, up 41.45%; group loss narrowed, although operating cash flow remained negative at RMB116.5 million. Its nine-month operating cash flow then turned positive at RMB52.7 million, implying a Q3 inflow of RMB169.3 million. Its latest revenue was still growing, up 12.73% year on year for the nine months and 5.96% in Q3. Different recovery paths within two expanding content businesses argue for company-specific assessment of acquisition and settlement, rather than a single sector profitability narrative. iReader H1 2026, pp. 6, 9–10; iReader Q3 2026, pp. 1, 10, 13

COL's annual-report summary describes its domestic series Yipin Buyi as a major Hongguo success, reporting more than three billion first-season plays, followed by further seasons. These are company-reported play counts, not unique viewers or profit. October reporting also cites a title, The Young Madam Comes from Northeast China, with more than RMB10 million of March sharing receipts, and over 200 live-action titles with monthly settlements exceeding RMB1 million across Douyin Group platforms. These examples establish the possibility of large title-level rewards. They supply neither a denominator of all released and abandoned projects nor the costs and participating rights behind the settlements. COL 2025 annual-report summary, domestic short-drama section; October Hongguo reporting

A repeatable series can nevertheless matter economically. It may give viewers a name to search, reuse a fictional world and production resources, and improve negotiating options across releases. Tencent's conditional series incentives make that route explicit. The counterweight is concentration: commissions, successful sequels, and unusually large settlements can be overrepresented in promotional disclosures. The appropriate investment unit is the slate of projects whose development, failures, and overhead must all be financed. No frozen source permits an industry-wide hit rate or net margin to be calculated from these selected successes. Tencent policy speech; 2025 association white paper, pp. 8–9

The following scenarios are constructed, not estimates of any named platform or an average industry margin. All commercial parameters are explicit assumptions. RMB600,000 is used as a production budget to illustrate a point within the association's reported RMB400,000–700,000 range for ordinary 2025 productions; the chosen conversion, advertising, sharing, and acquisition inputs are not taken from that survey. The paid and free cases use different assumed acquisition costs and contracts and therefore cannot rank the models' actual profitability. They show which variables a participant needs to negotiate or measure. The arithmetic was executed in the accompanying calculation file. Production-budget context: association white paper, p. 7

Assume one million acquired users start viewing a title. Four percent pay, averaging RMB75 of gross purchases per payer, for RMB3 million. Assume 5% is refunded and all remaining purchases are consumed during the example's window; there is no unspent wallet balance. A channel/payment provider takes 10% of consumption after refunds. A rights holder then receives 10% of the remaining settlement. The distributor spends RMB1.20 per acquired starting viewer, or RMB1.2 million. The producer finances RMB600,000. Under the assumed contract, acquisition expenditure is recovered before production, the producer next recovers its budget, and any surplus is split equally between producer and distributor.

These 10% assumptions are not Douyin's documented creator-promotion commission or a claim about any platform's payment charge. The example deliberately identifies each recipient and base.

The producer's RMB854,250 receipt is not its profit: RMB600,000 financed the production. Similarly, the channel's and media seller's receipts still have their own costs. The project surplus of RMB508,500 is before unspecified corporate overhead, tax, and financing. An accounting entity acting as principal might recognize the fulfilled service revenue and separately expense commissions; an agent could recognize its fee. The table is a cash allocation and does not impose one revenue-recognition policy on all participants.

The same assumptions produce the following combined producer/distributor surplus after rights, acquisition, and production, in RMB thousands:

At RMB1.20 acquisition cost, break-even conversion is approximately 3.12%, from (1.2m + 0.6m) / (1m × 75 × 0.95 × 0.90 × 0.90). At 2.5% conversion the residual for production recovery is only RMB242,812.50; the producer loses RMB357,187.50 while the distributor recovers its media bill and receives no surplus. This asymmetry follows from the assumed waterfall. It is not an inherent feature of paid viewing: a different guarantee or loss-sharing arrangement would move the exposure. In a more severe failure, receipts may not even recover media spend.

Holding conversion and acquisition fixed at the base case, increasing the production budget from RMB600,000 to RMB1 million cuts project surplus from RMB508,500 to RMB108,500. Increasing refunds from 5% to 12% reduces it to RMB338,400. A higher-quality production can still be worthwhile if it improves conversion or later licensing sufficiently, but the required improvement must be specified. An acquisition dashboard reporting gross purchases divided by media spend would miss both production recovery and the refund/channel/rights deductions.

For a separate assumed cohort of one million viewers, use 14 monetizable active days over the recovery window. Each active day offers four early ad opportunities and eight additional opportunities available to viewers who reach a later episode block. Assume 60% reach that later block, giving 4 + 8 × 0.60 = 8.8 opportunities per active day. An 80% fill rate and RMB20 of realized advertiser expenditure per thousand filled impressions produce:

1m viewers × 14 days × 8.8 opportunities × 80% × RMB20 / 1,000 = RMB1,971,200.

Here, “completion” means reaching the defined later content block, not completion of an advertisement or the whole series. Actual operators would measure the episode curve rather than assume this two-block structure. The model separates within-visit continuation from return days so that the two effects are visible.

Assume an ad network retains 20%, leaving RMB1,576,960 for the publisher. A 35% content share pays the producer RMB551,936. The platform keeps RMB1,025,024 before its assumed RMB700,000 acquisition cost and RMB100,000 operating cost. Platform contribution is RMB225,024, while the producer's RMB600,000 production cost produces a RMB48,064 loss. Their joint surplus is RMB176,960. The advertiser's original RMB1,971,200 reconciles to network receipts of RMB394,240, content receipts of RMB551,936, and the platform's pre-cost allocation of RMB1,025,024.

At the base completion assumption, the platform breaks even at approximately 10.93 active days, but the producer needs 15.22. Changing the content share from 35% to 45% would instead give the producer RMB109,632 and the platform RMB67,328 at 14 days. Joint surplus remains RMB176,960. Sharing creates no additional revenue by itself; it determines who can keep supplying content. A higher guarantee might improve the producer's financing position while increasing the platform's exposure if audience response disappoints.

Advertising price and fill multiply the same inventory. A 20% decline in either one, with other inputs fixed, cuts the revenue pool 20%. The assumption that the supply of episodes can rise rapidly does not imply that advertiser demand or price rises equally. Conversely, superior targeting can raise realized yield without increasing the number of viewers. This is why a time-share chart cannot be read as a revenue-share chart.

The model also clarifies the possible withdrawal of viewing incentives. If RMB60,000 of the assumed RMB100,000 operating cost were rewards that could be removed, the platform's saving would be erased by losing approximately 0.82 active days per viewer, or 5.85% of the 14-day base, assuming other factors remain constant. This is a sensitivity threshold, not a prediction of Hongguo's coin withdrawal. Changes in ad yield, user mix, and content cost would change it. The real decision needs retained-cohort monetization after incentives disappear, not only a reduction in reward expense.

Suppose 50,000 people in that free cohort choose an ad-free pass priced at an assumed RMB7, with 5% refunds and a 10% payment deduction. Net payment receipts would be RMB299,250. If those buyers otherwise generated average advertising revenue, lost net ad receipts would be RMB78,848, leaving incremental publisher receipts of RMB220,402 before extra service, sales, and content-sharing costs. If they were twice as valuable to advertisers as the average viewer, the increment falls to RMB141,554. Counting the full payment while retaining all the displaced ads would overstate the hybrid's value.

The scenario does not identify an optimal subscription price or conversion rate. It shows why payment willingness is potentially valuable without implying that a paid tier should replace free viewing. The practical question is whether net payment from the particular converting users exceeds their forgone advertising and incremental costs, and whether the offer harms other users' retention.

Cash timing adds a separate constraint to every scenario. Production can be paid before release, media expenditure before cohort recovery, and a sharing receivable before actual collection. Douyin's T+7 payment rule is one specific settlement step, not a guarantee that a producer's complete investment is recovered in seven days. Minimum guarantees, milestones, reconciliation rights, and refund reserves determine how much working capital each participant needs. A profitable expected slate can still exhaust cash if those obligations and receipts occur in the wrong order. Douyin transaction rules; COL H1 2026, receivables-risk discussion, p. 28

China's microdrama rules attach obligations to production, release, promotion, and continued distribution. “Beian” is regulatory filing or registration. It is a specific procedural step, not a general assurance that every version can be sold indefinitely on every platform. The important commercial question is which party must obtain which approval, for which version, before production or release, and who finances any revision or delay.

The chronology matters. The earlier NRTA notice required unapproved or unfiled microdramas to stay offline from 1 June 2024. Although the frozen webpage was published in February 2025, its document number, signature, and implementation provisions belong to 2024. Live-action investment thresholds were adjusted from 1 January 2026, and separate AI-drama thresholds took effect on 1 July. The comprehensive Development and Management Measures were published on 31 July and became effective on 1 September 2026. The following table combines those measures with the Nanjing regulator's September implementation explanation; it does not present the local explanation as the original national threshold circular. Earlier NRTA notice, sections III–IV and signature; Baoji official threshold notice; Xinhua's AI classification notice; NRTA measures; Nanjing implementation explanation

For category I, Article 20 sets a decision period of 20 days from acceptance of the application, including ten days for expert review. It also provides for written modification requests and resubmission. This is not a promise that the whole process from an unfinished script to release takes 20 days. For non-exclusive category III titles, each platform must use the number corresponding to its own reviewed version; rebroadcast titles must be reviewed again. Multi-channel distribution consequently carries version-control and compliance work as well as potential extra receipts. The rules also require conspicuous AI-production notices in each episode and apply filing and review provisions to dramas shot domestically for overseas release. Overseas allocation therefore does not automatically bypass domestic production obligations. NRTA measures, Articles 13–20, 33–35, 52

There are incentives as well as restrictions. The 2026 quality-production plan aims to support 1,000 selected works by year-end, with funding, creative support, and distribution resources. That is a program target, not an achieved count at the cutoff. It favors eligible, filed projects and specifies application and delivery arrangements. Alongside the measures' recommendation obligations, the plan means that “organic discovery” is influenced by editorial and policy priorities, not only a neutral ranking of immediate clicks. A producer may improve its access to funding or recommendations by meeting those priorities, but should budget only confirmed support and satisfy the associated delivery conditions. NRTA 2026 quality plan, sections III, V–VI; NRTA measures, Articles 30–31, 37

On 6 November the NRTA announced a two-month campaign running from 1 November through December, focusing on vulgar or repetitive content, unauthorized use of faces and voices in AI production, altered names used to recycle titles, and homogenized recommendations. Its statement that 75,000 works and more than 1,500 accounts had already been removed or dealt with refers to earlier actions during 2026. Those counts cannot be attributed to the completed results of a campaign that had just begun. Commercially, the campaign increases the relevance of identifiable source material, usable performer rights, distinct finished works, and reviewed promotional clips. A cheaply regenerated catalog can lose both its distribution and its expected recovery window. NRTA campaign announcement, 6 November 2026

The Consumer Rights Protection Law's implementing regulation, effective 1 July 2024, requires clear pricing and conspicuous notice before accepting an automatically renewed service and before renewal dates. The retained statutory text does not specify a universal five-day reminder period. It also restricts unfair standard terms and requires prepaid-service contracts to identify the service, price, refund arrangements, and liability. A wallet top-up therefore creates both liquidity and an obligation to supply the promised service. Implementing regulation, Articles 10, 17, 22, 53; retained official web text

Douyin's microdrama mini-program rules are more operationally specific. They require disclosure of the full-series price, stable token-to-cash conversion, consistent per-episode prices within a drama, and affirmative consent for renewal or automatic unlocking. A single membership card cannot run longer than a quarter. The rules require staffed customer support for at least ten hours a day, seven days a week, a response within 15 minutes during staffed hours, and complaint handling within 48 hours. These are operating costs and product constraints, rather than optional additions after launch. Douyin operating rules, experience and customer-service requirements

The platform's refund provisions cover unavailability for reasons outside the user's control, removed content, unused rights, and repeated purchases. When a minor buys virtual goods and requests a refund, those rules require a full refund regardless of fulfillment status. The Civil Code's general treatment is more differentiated: children under eight lack civil capacity, while minors aged eight and above generally have limited capacity, with validity depending on the transaction and guardian consent or ratification; there are also provisions for older self-supporting minors. It would be incorrect to turn the platform's categorical commercial rule into a claim that every transaction by every minor is automatically void under national law. Operators must satisfy the applicable host's contract as well as the law. Douyin operating rules, section 4.2.5; Civil Code, Articles 18–20 and 144–145

This makes gross top-up return on advertising spend an incomplete optimization target. A campaign that attracts mistaken purchases, unavailable-content claims, or minor purchases can have a high initial collection figure and poor retained receipts. Account sanctions can also interrupt settlement or access. Acquisition teams need to monitor refunds and complaint-adjusted recovery by campaign and creative, while producers need contracts that specify whether their royalty base falls when refunds occur. The model's refund sensitivity represents that economic channel; its assumed rates do not estimate current complaint prevalence. Douyin rules, sections 4–5

A court's February 2025 case introduction describes an unauthorized operator renaming a commissioned microdrama, directing viewers through short-video and WeChat channels, and charging for access. The rights holder received RMB46,000 in damages and reasonable enforcement expenses against a claim of RMB500,000. This is a court-published case summary, not the full judgment, and the amount is not a general tariff for piracy. It shows the practical importance of rights ownership and evidence when the same story can be repackaged and distributed quickly. Court's microdrama copyright case introduction

The economics of rights clearance extend beyond avoiding a claim. A producer selling an exclusive window must retain a reliable record of prior grants; a distributor needs rights covering its specific channels, territory, term, edits, and promotional use; an AI production needs usable rights to inputs and likenesses where relevant. These are contract-diligence implications of the documented disputes and current enforcement priorities. An attractive settlement offer is worth less if the promised version cannot stay available, and a multi-channel plan is worth less if the seller has already granted conflicting exclusivity. Court case; NRTA measures; November campaign

The strongest supported judgment is that repeat access and contractual control over revenue recovery have become central competitive advantages. Dedicated apps are growing quickly, free viewing has achieved substantial reach, and distribution is increasingly organized around large platforms. Yet national audience estimates, declining competing apps, producer-side losses in otherwise positive scenarios, and listed-company acquisition expenses all contradict a simple story in which larger reach guarantees attractive returns. The appropriate strategy changes with the participant's assets and financing obligations.

For a platform, Hongguo's scale supports taking free viewing seriously, but does not justify treating its rumored revenue as known or every overlapping Douyin user as a free acquisition. The immediate observable tests are how return frequency, fill, realized advertising price, and supplier retention respond to changes in rewards and sharing. If the platform retains most viewing after removing incentives, it may improve cash contribution without new users. If the lost high-frequency users generated disproportionate ad value, aggregate MAU can remain healthy while contribution deteriorates. The scenario gives the relevant trade-off: savings and lost retained advertising must be measured on the same cohort and time horizon.

For producers, the choice is partly a financing decision. A commissioned production can be attractive even with a limited upside when the buyer funds a predictable schedule and carries demand risk. A producer with financing capacity and a demonstrably reusable series may gain more from contingent sharing and later licensing. The material distinction is the promised revenue base and recovery order. In the paid example the producer bears the first loss after media recovery; in the free example the platform is profitable while the content supplier is not. A headline 35%, 45%, or 90% share has little meaning without the denominator and the costs that precede it.

For distributors, adding another outlet can diversify access while also duplicating acquisition spending on the same users. Channel activity should be deduplicated before presenting market reach, but allocation decisions require a further question: does the next campaign create additional profitable viewing or simply reacquire someone already reached elsewhere? Named-title search and repeat-series consumption may reduce that dependence. The evidence does not quantify the benefit, so investment should follow observed incremental recovery rather than presumed organic traffic.

For investors, the most useful contrary evidence is already visible in the accounts. iReader's strong derivatives growth coexists with a group loss and substantial promotion expense; COL's changing consolidation, overseas allocation, and uneven 2026 quarterly cash flow complicate any simple domestic-growth thesis. Those facts do not invalidate every microdrama investment. They establish the need to identify the asset being purchased: a production capability, a rights library, an audience relationship, an advertising business, or a capital-intensive distribution operation. Each has different control, margin, and collection risks.

The evidence would support a more optimistic sector-wide profit conclusion if comparable disclosures showed broad cash recovery after acquisition, unsuccessful projects, content commitments, and refunds, across a sufficiently representative set of domestic operators. It would support a stronger slowdown conclusion if consistent audience series, return frequency, advertising yield, and new-project recovery weakened together. At the cutoff, the evidence supports strong channel and engagement growth, concentrated platform advantages, and unresolved industry-wide profitability. Capital can be allocated on those narrower findings without pretending that an unreconciled market-size headline settles the decision.

All factual observations come from the 39 designated frozen source records and their supplied original attachments. Source identifiers in the accompanying registry distinguish records, not necessarily independent evidence. CP15 and CP17 represent the same QuestMobile research; MDX002 duplicates CP27; MDX001 reproduces the CNNIC release whose fuller report is CP19. The monetary estimates are quoted with their original periods and definitions, and no foreign receipts are added to domestic market totals.

The public CSM findings do not disclose enough survey design to reconcile their adult population estimate with CNNIC. QuestMobile supplies activity measures with channel-specific selection rules; the CP16 and CP24 charts specify the relevant channel populations, units and footnotes. Active users, browsing reach, play counts, registrations, and paying users remain separate measures. Later reports sometimes repeat earlier survey values; a repetition does not provide an additional independent sample.

CP10 and CP20 are retained outputs of prior web access to the official regulatory and investor-relations pages. Their relevant text is available, but they are not saved full original websites. CP23 contains only an access-failure record in the supplied material; no substantive claim in this report relies on it. Citations to long reports identify the relevant pages or passages. News accounts, management statements, research estimates, and statutory provisions retain their different evidential status.

The executed calculations.py and calculation-output.json separate source-derived arithmetic from assumed commercial scenarios. The scenarios are deterministic illustrations, with no fitted retention model or implied industry average. They exclude unspecified corporate overhead, financing costs, and tax; each named contribution measure states which included costs it deducts. Their purpose is to make the recovery conditions auditable and expose disagreements about assumptions, rather than supply a valuation disguised as observed performance.