The belief I never examined
For years I carried a piece of local knowledge I had never once checked: Chinese users would rather spend an afternoon cracking a piece of software than pay for it. Everyone in Chinese tech circles just knew this. Beliefs like that are load-bearing — they quietly decide which markets you build for and which you never even consider.
This week I finally asked for the statistics instead of the folklore. The belief has expired — but only in part, and the part that survives is the useful part.
The consumer side moved
The old data was real. BSA's unlicensed-software rate for China was 77% in 2011. But the series ends at 66% in 2017, and everything since points the same way. China passed the US in iOS App Store revenue in the third quarter of 2016 and has been the world's number two ever since — on iOS alone. Tencent Music has 127.4 million paying users; the share of its monthly actives who pay rose from 3.6% in 2018 to 24.1% at the end of 2025. Tencent Video counts 113 million subscribers, iQIYI passed 100 million back in 2019, and WPS — a mere productivity tool — has 41.79 million annual paying users at home. Domestic games took in ¥325.7 billion in 2024.
The denominator is the whole argument
Absolute counts prove nothing in a country of 1.4 billion; that population is not the market for anything. The honest number is payers divided by the people the product is actually for:
Divide Tencent Music's payers by its own actives and you get 24.1% — against Spotify's 39%, a gap that has narrowed from roughly twelvefold in 2018 to about 1.6-fold, closing rather than a wall. What remains genuinely different is price: revenue per paying user at Tencent Music is about a third of Spotify's. More payers, at local prices.
Nor is a single-digit ratio some Chinese defect: YouTube's paid tiers reach only about 4.9% of its users worldwide, and Duolingo — the best-monetizing language app alive — converts 9.2% of its yearly actives. Low conversion is simply what free looks like, everywhere.
The bases differ — actives for the music, video and learning services, broadband households for Netflix, ad-reach users for YouTube — and NetEase stopped disclosing after 2022. Read the chart as orders of magnitude, not league positions.
That is the part my old belief got exactly backwards. A market priced out of reach does not pirate your product; it ignores it. Piracy was the symptom of an era when there was no one-tap way to pay and nothing priced for the people being asked.
What did not change
Enterprise seats. The last clean comparison put China at roughly 16% of world GDP and about 4.2% of world enterprise IT spending, and its software market still runs on customization, projects, and services instead of standardized per-seat subscriptions. The stereotype survives there — but that is a procurement machine, not a consumer instinct, and it says nothing about whether individuals pay.
My own track
I sell into language learning, so I asked for that slice too. The best conversion cell in RevenueCat's entire dataset is education in South Korea: 10.7% of downloads become payers, against a 1.7% global median for subscription apps. Duolingo has 12.2 million paying subscribers, yet the US share of its revenue fell from 45% to 38% in two years — paying is diffusing outward, not concentrating. In China, the observed price of reading is ¥19.9 a title, and the country's adult-English champion, Liulishuo, peaked at about 1.1 million paying users in a single quarter before distribution and retention killed it. Not unwillingness to pay. Distribution and retention.
The updated rule
So the question was never "do people there pay?" It is two questions: of the people this is actually for, what fraction pay — and at what local price? I had the answer to the first question memorized from a decade ago and was still making plans with it. The data had moved. The belief hadn't. That lag, not any market, was the thing costing me.