Chief Strategy Officer / China
Three choices that each cost us something
- Company revenue, 2025
- RMB684m
- Paid reading, advertising, audio and rights licensing
- Monthly active readers
- 23m
- June 2026, free and paid tiers together
- Titles licensed for short drama
- 74
- To twelve producers, 2024–2025
Decisions & consequences
01Making older completed titles free while keeping new serials paid
By late 2021 free, advertising-funded reading apps had taken much of the casual readership in China. Our paying members had fallen from 2.31 million in 2019 to 2.05 million, and monthly active readers were 9.4 million. Jiarui Fan, our head of growth, proposed making the whole catalogue free, as several competitors had. Mingzhu Ge, our editor-in-chief, argued that any free tier would teach readers to wait and would cut authors' income.
Both were partly right. A fully free model would probably have grown readers fastest, but our authors earned most of their income from paid chapters of active serials, and many would have left. Staying paid meant losing readers who would never pay.
I recommended a split. From April 2022, completed titles more than two years past their final chapter, 21,400 in all, became free with advertising; serials still being written and recently completed titles stayed paid. Authors received half of the net advertising revenue attributed to their titles by reading time. Because nobody could forecast that income, we set aside RMB9 million to top up any author whose combined earnings fell below their previous year during the first twelve months. The board accepted lower profit: operating profit fell from RMB52 million in 2021 to RMB11 million in 2022.
The top-up paid out RMB6.3 million. Paying members fell to 1.71 million at the end of 2023, then recovered to 1.9 million by June 2026 as free readers moved into paid serials. Monthly active readers reached 23 million, advertising revenue was RMB176 million in 2025, and revenue rose from RMB571 million in 2021 to RMB684 million. Fewer authors left than in the previous two years.
What I would do differently: we set the free threshold by calendar age, which made some still-popular titles free too early. A threshold based on each title's paid reading would have protected those authors better. I would also publish the earnings formula before launch; publishing it only after authors asked cost us trust.
Licensing short-drama rights title by title instead of joining a venture
In early 2024 vertical short dramas adapted from web fiction were growing quickly, and producers were competing for source stories. Feiluanpai Studios, a private producer, offered a joint venture: 51 per cent theirs, 49 per cent ours, three years of exclusive first access to every adaptation right in our catalogue and a committed minimum of RMB60 million over that period. Weilong Sha, our chief financial officer, and the directors appointed by our two venture funds supported it because the income was certain.
I recommended against it. Exclusive first access would have tied 92,000 titles to one producer's capacity and review record just as the broadcasting regulator was tightening review and filing for online short dramas. Authors would have had no say over which producer adapted their work. The cost of saying no was a certain RMB60 million and an equity stake in a growing producer.
We built a rights register first, then licensed titles one at a time: an upfront fee of RMB150,000 to RMB600,000 per title, a revenue share of six to ten per cent, author approval of the adaptation outline and an obligation on the producer to complete the regulatory filing. We also agreed eight co-productions with the production company of Taolingxu Broadcasting Group, a state-owned provincial broadcaster, for historical titles where review demands were higher. Our rights team grew from eight people to fifteen.
By December 2025, 74 titles had been licensed to twelve producers, including Feiluanpai on standard terms, and short-drama licensing earned RMB46 million in 2025. Seven productions were shelved or did not complete review; under our licences the fees were kept and the rights returned to us. Three of the Taolingxu co-productions have been released. Growth is now slowing as producers consolidate, and 2026 income will be lower than 2025.
What I would do differently: I would cap the number of producers per genre from the start. Twelve relationships stretched our rights team, and two producers adapting similar romance titles at once diluted both.
Voicing the backlist synthetically with consent, labels and a way out
By 2024 we had human-narrated audio for 14,000 titles and none for about 58,000 completed books. Human narration cost us roughly RMB1,600 per finished hour; synthetic narration, including editing, checks and narrator fees, about RMB140. Our chief operating officer proposed moving new serials to synthetic voices as well, saving about RMB14 million a year. Tingwei Lou, who leads our narration studio, opposed synthetic voices altogether, concerned for 46 staff narrators and about 380 contracted ones.
I recommended a narrower path. Synthetic narration would cover only completed titles with no human recording and only where the author signed a contract addendum, which added five percentage points to their audio revenue share. Voice models would come solely from narrators who agreed separately, each paid RMB18 per finished hour and free to withdraw on ninety days' notice. Every track would carry a spoken notice and an in-app label. New serials and flagship titles would stay with human narrators. We gave up most of the saving the operating plan had assumed.
Twenty-three narrators agreed. About a third of authors declined the addendum, so roughly 19,000 titles stayed outside the programme. We released the first titles in September 2025, the month China's labelling measures for AI-generated content took effect, and by June 2026 had voiced 6,100 titles. Synthetic tracks account for 12 per cent of listening hours. Audio revenue rose from RMB97 million in 2024 to RMB118 million in 2025, though not all of that came from the new titles, and human commissioning stayed above RMB30 million.
In February 2026 one narrator withdrew. We had to re-voice 340 titles with another licensed voice at a cost of about RMB410,000, and our process for telling listeners proved clumsy.
What I would do differently: I would test the withdrawal procedure, including the notice to listeners and authors, before the first release rather than after the first narrator used it.