When to Localise Your App, and What to Change First

You look at your dashboard and a country you have never thought about is in the top five. Indonesia, Brazil, Mexico. Real installs, almost no revenue.

So you open a translation tool, run the App Store description through it, and ship. That is what I would have done too, and it is the version of localisation that changes the least.

Here is the part that costs you the money. The thing people in that market are reacting to is usually not the language. It is a price set for a different country, in a currency written the wrong way round, paid for with a card they may not have.

So here are the four layers in the order they pay back, with six real screens. Including the one app that does all four at once, in Indonesian rupiah. Hold it up against whatever somebody in Jakarta is seeing on your paywall right now.

The order to do it in

Localisation is four separate jobs that get talked about as one. They cost wildly different amounts and they do not pay back equally.

What all four layers look like at once

Shopee is the clearest example I have found of an app that did not stop at translation.

None of those is a translation decision. Every one of them is a decision about how money is written and handled where the user lives.

And look at the price. Rp14.900 is roughly ninety US cents a month. If Shopee had set one global number, that plan would not exist.

The format layer, which costs nothing

This is the cheapest thing on the list and the one I see skipped most often, because it is invisible to whoever is shipping it.

A euro storefront writes 6,99 €. A comma for the decimal, and the symbol after the number. An Indonesian one writes Rp14.900, where the dot is a thousands separator, so the same number written the American way would read as fourteen rupiah and change.

Get this wrong and you have not just looked foreign. You have published a price that means something else.

When the storefront does it and the app does nothing

Most apps you look at in a non-US store are not localised at all. They are the same app, with the price converted by Apple.

Give the user the control, or make them find it

Two more screens, and the gap between them is the whole argument about where this belongs.

When not to do any of this

Localisation is a lot of work and it is not always the work in front of you.

If a market is not already showing up in your installs, translating will not create it. Localisation makes an existing audience convert better. It is weak at making a new one appear.

If your retention is bad everywhere, a new market gives you the same bad retention in another language. Ekaterina Gamsriegler, a B2C growth advisor previously at Mimo, lists “never localised” as one of only four situations where she thinks cutting a price is right. She will not run that test on the localisation signal alone, though. She wants a bad trial or purchase conversion rate sitting next to it.

If you cannot support the language, a translated listing produces support tickets you cannot answer and reviews you cannot read. That is a worse outcome than not being there.

Where to start, concretely

Three to five markets, not twenty. India, Brazil, Indonesia and Mexico are reasonable first picks, because the install volume tends to arrive before anybody targets it.

Do layer one and layer two, wait for a full billing cycle, and look at conversion rather than installs. Installs will move on metadata alone, which is exactly why they are the wrong number to judge it by.

If you want a starting price rather than a guess, roughly half the North American number is where the apps in RevenueCat’s dataset already sit for India and Southeast Asia.