5 Mistakes Western Developers Make in China

01

Treating language as geography, and wishlists as demand

A Simplified Chinese review tells you the language of the reviewer, not where they live, where the purchase happened or how much revenue came from the mainland. Steamworks does report sales and wishlists by region, even though the public wishlist page hides it, so use that. Then remember that Chinese players treat a wishlist as a bookmark, rarely read the launch email and often buy without wishlisting: one shooter had 400,000 wishlists and sold over a million copies in week one, with China about 64% of players (游戏陀螺, October 2025). No China wishlist-to-sale norm exists in any language.

The check: pull the regional report, count the Simplified Chinese reviews, look up two or three comparable games with the same genre, price and scope and note how their Chinese review share compares. Keep public signals separate from studio data, with dates. If the regional report is unavailable, say so rather than inventing a market size.

02

Approving words without testing the game

A translation can read well in a spreadsheet and fail in play: labels that do not fit, tutorial instructions that refer to the wrong control, terms that drift between menus and dialogue. And it can be accurate and still wrong. Hollow Knight: Silksong’s Simplified Chinese was criticised in 2025 for a “half-classical” register that players found affected, and the studio ended up adopting the fan community’s own translation. Stardew Valley drew a wave of negative Chinese reviews after a 2024 update quietly changed an existing translation players already knew.

The check: ask for a glossary and a sample inside the build, then have native players test the opening, the common menus and the error states before approving the full job. Check the font (a Mainland Simplified Chinese font licensed for commercial use, such as Source Han Sans; the wrong fallback order substitutes Japanese glyphs). Keep a defect log with screenshots and a studio owner for sign-off. Steam’s localisation documentation distinguishes store and in-game support; your acceptance checks should cover both, and the store page should not claim a language the build does not have.

03

Changing the price after launch

China expects to be the cheapest region on Steam. Valve’s suggested China price has sat at roughly 59 to 63% of the US price, and indies mostly follow it. What gets punished is not the level but the change.

The check: decide the China price with Valve’s March 2026 pricing tool in front of you (the straight exchange-rate method overprices China), write the discount ladder for the first year, and do not touch either after launch. Region-lock keys sold through third parties; one developer found over 90% of early sales from a key reseller were in CNY at half the dollar price.

04

Leaving Chinese reviews unanswered

Because Chinese players cannot easily reach Discord, Reddit or the Steam forums, the review box is how they talk to you. Chinese scores run lower across the board, negatives cluster on days three to six after launch, and many flip once the problem is fixed and acknowledged. Since August 2025 the Chinese score is shown separately once a game has 2,000 reviews overall and 200 in Chinese (Valve via GamingOnLinux), so the slump is visible to every potential buyer, and so is the recovery.

The check: before launch, know who reads Chinese reviews daily in launch week and who can post a Chinese-language reply from the studio’s own Bilibili or Heybox account within a day. Open a QQ group so players have somewhere else to complain. Test connectivity from the mainland; servers unreachable from China produce refunds, not reviews. And never, in any language, say anything that reads as blaming Chinese players.

05

Signing before the hard questions are answered

The last mistake is the one that locks the others in. It takes three forms.

Access. A partner that needs to upload a localised build or run a Steam event needs a task permission, not the app. Steamworks has task-specific permissions for exactly this; an app transfer is a separate, deliberate action a marketing partnership never requires. A request for source code for a Steam Global deal should end the conversation.

Money. A lower share may cost more if it applies to a broader revenue base or permits more deductions. Ask every candidate for a worked statement on the same assumptions, including existing sales, DLC, refunds, Steam’s 30%, tax and the order in which costs are recovered. Our worked example shows a £7,000 swing on £100,000 of sales from the recoupment order alone, with the headline split unchanged.

Route and regulation. Name the storefront and the operating model before agreeing a launch, and do not treat a local partner’s involvement as regulatory clearance. Resolve the licence questions for the proposed activities and record the basis for proceeding. Then agree who owns the Chinese community channels and in whose name, who can change the price, what reverts when the deal ends, and what triggers that reversion. One narrative studio hired a marketer who set up a new “official” Chinese group over a dozen existing fan groups; its recent reviews fell from 92% to 54% positive.

A good first assessment identifies the next decision and the evidence needed to make it, and is willing to recommend a small test or a pause as well as publisher discussions. If you are at that stage, how to choose a China publishing partner is the next read.