Lock n Roll, a fun dice game for Android, recently got translated. Here are the stats for how well it went. The app was translated using an online service for accessing human translators. It was translated from English to Spanish, Chinese (Simplified/China and Traditional/Taiwan), Japanese, and Korean. Marketing descriptions were translated as well.
Installs increased noticeably in Spanish and Simplified Chinese languages. This is the installs graph for the ad supported free version on Google Play (it is also in other app stores and there is a paid version and a virtual goods supported version):
Installs didn’t react so much in the other languages. English installs stayed pretty much the same. The picture is slightly different for retained users. In that one Japanese users did tend to be retained better. Below is the Flurry stats for retained users with users who used the app that week counting, so a new user that week counts, but also a user who started before that time frame, but used it again that week. I exported the data and removed languages like English that were too big and made the changes tough to see. I left German in so you can see an untranslated language, which should not change much:
Maybe Japanese users retain better because they enjoy the game more in their native language, or are more likely to read the tutorial/rules instead of skip it if it is in their native language, and end up having more fun.
It’s interesting that marketing description translations help a lot with Spanish speakers, but not much for Japanese speakers. This may be the more important result, because in terms of overall users, less than 1% are Spanish, but over 10% are Japanese, and over 10% are Chinese. So improving retention of user groups that are in the app already can be more valuable than a slightly improved install rate from a rare group of users.
Regarding translation quality, a friend who speaks Korean checked out the app afterwards. He said the translation isn’t perfect, only 90%, but that it is understandable, and that translating it with Google Translate wouldn’t have made any sense. Chinese speaking friends could understand the Chinese translation, but said it was too literal and not short and fun. I did get a few updates from them over the weeks, though, particularly for a mistranslation of suit (of cards) as a business suit.
Cost was $58 USD per language for a 1,180 word strings file. I think I could cut this in half by trimming the market description and reusing strings a little bit better. It might also be possible to only do one Chinese translation and use Google Translate to convert it to traditional characters. There’s a lot of reuse of common terms, like same number and same color, throughout the strings file. I’m also looking forward to seeing the performance in other apps stores that do better in different languages, like Korean.
Combined with an average revenue per user calculation, this should tell you if translating is a good idea for you! Thanks for reading. Give Lock n Roll a try if you need a fun break!