App description translation: conversion vs keywords
You wrote a good English description. Now you want it in German, Japanese, and Portuguese, and the obvious move is to run it through a translator and paste the result into both App Store Connect and Play Console. The text is the same on both stores in English, so why would the translation be any different?
Because the description isn't doing the same job on both stores. It reads identically to a human, but each store's search algorithm treats it in a completely different way. Get the translation right for one and you can still hand the other a weaker listing than the one you started with.
On the App Store, the description is a sales page
Apple does not index your description for search. The words in it, however good, do not affect which searches your app shows up for. Ranking on the App Store comes from the app name, the subtitle, and a dedicated keyword field — a hidden, comma-separated field with a 100-character limit, and Apple gives you a separate one for each locale you localize into.
So on iOS the description has exactly one purpose: conversion. It runs after someone has already found your app and is deciding whether to install. It needs to be clear, persuasive, and easy to skim — especially the first few lines that show before the "more" fold. When you translate an iOS description, you translate for a reader, not an algorithm. Natural phrasing beats a keyword you crammed in, because that keyword earns you nothing here anyway.
On Google Play, the description is also the keyword field
Google Play has no separate keyword field at all. It reads your title, your short description, and your full long description, and it pulls ranking terms straight out of that text. The whole long description — up to 4,000 characters — is fair game for indexing.
That flips the job of the words. On Android the description still has to convince a human, but it also has to carry the keywords you want to rank for, in the language of that market. A common ASO rule of thumb is to keep your primary term at roughly 2–3% density — repeated naturally a handful of times across the description rather than stuffed. Treat that as guidance from practitioners, not a number Google publishes; the underlying point is solid, though: on Play, the terms have to actually be present in the translated text, or you don't rank for them.
The same sentence, two jobs
Here's the trap. Say your English description leans on the phrase "habit tracker". On the App Store, that phrase belongs in your keyword field, and the description can describe the app in whatever natural language reads best. On Google Play, "habit tracker" has to survive translation as a searchable term in the target language — the German equivalent people actually type, not a literal word-for-word rendering that no one searches for.
A generic translation optimizes for reading smoothly. It has no reason to preserve the exact market-specific term a Play searcher would use, and often it won't — it'll pick a synonym, or a more formal register, or drop the phrase into a construction that reads fine but no longer contains the keyword. On iOS that's harmless. On Android you just translated your rankings away.
Why "translate once, ship to both" is the mistake
The instinct to run the description through a translator a single time and paste the identical output into both stores is efficient, and for the iOS description it's mostly fine. The problem is the Android side. The same output that reads well on the App Store is the output that has stopped carrying keywords on Google Play, and nothing warns you — the listing looks translated and professional. You only notice months later when the Android app never ranks in that language.
So the two translations should not be identical. The iOS description gets translated for conversion and clarity. The Android description gets translated with the target-market keywords kept intact and placed naturally in the running text. Same source, two outputs, on purpose.
What keyword-aware translation looks like
A translation that respects Google Play does three things a plain one doesn't. It identifies the terms you're actually trying to rank for. It finds the phrase real users in that market search with — which is a localization decision, not a dictionary lookup — and it keeps that phrase present in the translated description at a sane density instead of paraphrasing it out. The result still reads like a person wrote it; it just doesn't accidentally discard your keywords on the way.
For iOS you do the mirror of that: translate the description for the cleanest, most convincing read, and move your keyword effort into the 100-character keyword field for that locale, where Apple will actually use it. Two stores, two strategies, one source text.
How Mokbi handles it
Mokbi translates your whole listing — title, description, and captions — into 50 languages, and it uses an ASO-aware model (Claude) rather than a raw translate-and-forget pass. That means the Android description keeps its keyword coverage in each language, while the iOS description is translated to read for conversion. You're not manually diffing two versions of the same paragraph in twenty markets; the split between "rank" and "convince" is built into how the text comes out per store.
One honest caveat: AI translation is a strong starting point, not a final sign-off. For your highest-value markets — the two or three languages that drive most of your installs — a native speaker reviewing the output still catches the phrasing an algorithm can't judge, especially whether a translated keyword is the term locals genuinely search. Use the automatic pass to cover all 50 languages cheaply, then spend a human review pass where the revenue is.