Deciding who pays and who gives something up

Free, advertising-funded reading apps changed what Chinese readers expect to pay. I do not treat that as a reason to make everything free. I separate the catalogue by what a reader is buying: the next chapter of a serial still being written, or a finished book discovered later. New serials carry an urgency readers will pay for; completed backlist mostly does not. I model each tier with three groups in view: paying members who might downgrade, new readers who would never have paid, and authors whose income depends on the split. Advertising revenue is volatile and sensitive to rules on minors, so I never let it carry the whole plan. The mistake I watch for is counting a rise in readers as success while authors quietly stop writing for us.

Short-drama producers, audio platforms, games studios and television companies all want our stories, and they differ in more than price. Private producers move quickly, offer larger advances and sometimes ask for exclusivity across a whole catalogue. State-owned broadcasters and their production companies move more slowly, bring stronger review discipline and often prefer historical or socially themed titles. I compare offers by what the company still owns afterwards: reversion dates, format limits, the author's approval of the adaptation outline and who is responsible for the regulatory filing. Since 2024 the broadcasting regulator has tightened review and filing for online short dramas, so a licence must also state what happens when a production fails review. I prefer several non-exclusive relationships to one dependent one.

Synthetic voices cut the cost of a finished audio hour from about RMB1,600 to about RMB140 for us, which makes them hard to ignore. I treat the decision as a rights question first. The narrator whose voice trains a model has to agree separately and be able to withdraw; Chinese personal-information and deep-synthesis rules point the same way. The author's contract may not cover a machine reading, so we ask. Listeners must be told, both in the recording and in the app, and the Cyberspace Administration's labelling measures for AI-generated content that took effect in September 2025 made that a formal obligation. I also decide where synthetic voices should not be used, which for us still means new serials and flagship titles.

Every reading business in China has carried some legacy channel: operator billing, set-top box packages or prepaid reading cards. When cable and IPTV households began to cancel, our set-top income fell from RMB38 million in 2017 to RMB9 million in 2020. The temptation was to cut prices to slow the loss. I asked instead what the channel cost to serve, which partners depended on it and where its staff could go. We set an end date, told the operators early, honoured content commitments and moved 26 people into audio production and partner support. A declining channel can still fund the next one for a while, provided nobody pretends it is stable and the board can see the date it will end. Partners deserve that date too, in writing and early.

Chinese web fiction has readers across Southeast Asia, and every few years someone proposes our own overseas app. I compare three routes: licensing translation rights, co-funding translations with a local platform, or operating directly. Operating gives a direct reader relationship but requires local content rules, payments, marketing and, if we bring readers' data home, the personal-information law's requirements for transfers abroad. Licensing earns less per reader but keeps capital and data obligations small. In 2023 I recommended licensing with Serambika Reads and a Thai publisher rather than an RMB85 million launch. That choice gave up a direct overseas audience, and I review it each year against what the partners' readers actually buy. Translation quality, reversion after five years and the partner's own content rules are written into every licence, because a weak translation damages an author's name in a market we cannot see directly.