Soft Launch a Mobile Game in Cheaper Countries

What is a soft launch for a mobile game?

A soft launch is a limited release of your game in a few chosen countries before the worldwide launch. You use it to find crashes, measure whether players come back, and check that the game earns money, while installs are cheaper and mistakes are seen by few people. When the numbers meet the targets you set in advance, you launch everywhere; when they do not, you fix the game or stop before the expensive part.

This guide covers the plan: what to test, where, how to set it up in the stores, and how many players you need before a number means anything. For promotion tactics on a small budget, see How to promote a mobile game on a small budget.

Three goals, tested in order

A soft launch that tries to answer everything at once usually answers nothing. Most studios split it into phases, and each phase has one question and one number that decides the next step.

The lengths are a planning range, not a rule. A phase ends when its number is clear, not when the calendar says so. If day 1 retention is far below your target, there is no point testing monetization yet: fix the first session and run phase 2 again.

Write your go and stop numbers before the first install. For example: crash-free sessions above an agreed level, day 1 retention above the figure your genre needs, and a cost per retained player you can afford. Our app retention benchmarks show published numbers by genre to help you set realistic targets.

How to choose soft launch countries

A good test country behaves like the markets you will launch in, but costs less to buy players in and has a smaller audience that will see your rough first version. Three things matter:

• Similar players. Language, devices, payment habits, and genre taste should resemble your target markets. Retention measured in one region does not always carry over to another.
• Lower cost per player. You will buy many players to get clean cohorts, so each one should cost less than in your main market. Cost per install by country shows published prices.
• Enough volume. A country so small that you cannot buy a few hundred players a week will stall your test.

Industry coverage has named the same few markets for years. A 2016 PocketGamer.biz study of soft-launched games found Canada used as a test country in 72% of the games it studied, followed by Australia, the Philippines, Sweden, and New Zealand. The reasoning still holds: Canada, Australia, and New Zealand resemble the US and UK at a lower cost, and the Philippines combines strong English use with low prices, which suits technical tests. Treat this as a starting point, not a recommendation for your game.

Pick by the language of the build too. If your game is not localized yet, test where players use the language it ships in, or the results will measure the translation gap instead of the game.

Limit availability in the stores

Both stores let you release to chosen countries only, so the game is simply not available elsewhere during the test.

• Google Play. Play Console Help explains that you can target a release in production, open testing, or closed testing to users in specific countries, from the Countries/regions tab of each track. Testing tracks follow the production country list by default unless you unsync them. Google notes that targeting is based on the user’s Play country, where their account is registered, not their current location, so a traveler will not see your game and a VPN does not change it.
• App Store. App Store Connect Help describes the App Availability section under Pricing and Availability, where you choose All Countries or Regions or Specific Countries or Regions. Select only your test countries, and add the rest at global launch.

Two more store details to plan for:

• New Google Play developer accounts must pass a closed test before production access. Read Google Play’s 12-tester rule and budget the time before your soft launch date.
• Early ratings stay. Ratings left during the test remain on your listing in the test countries after global launch, so a buggy phase 1 can follow you there. This is one more reason to start with a small technical test.

The metrics that decide go or stop

Always read these by cohort: players who installed in the same week, from the same source, in the same country. A blended number mixes a good week with a broken build, or a strong country with a weak one.

How many players do you need?

Retention is a share of a group, so its precision depends on how many players are in the group. A common way to estimate it is the margin of error for a proportion: 1.96 × √(p × (1 − p) ÷ n) at 95% confidence, where p is the rate and n the number of installs.

The 35% and 12% are example rates, not benchmarks. The pattern is what matters: with 200 installs, a day 1 result of 35% could really be anything from about 28% to 42%, which is too wide to compare two builds. A few hundred to a thousand installs per cohort is a practical range for retention decisions.

Monetization needs far more. If 2% of players buy something, 1,000 installs give you about 20 payers, and the true rate could be anywhere from roughly 1% to 3%. Revenue per payer also varies widely, so plan larger or longer cohorts for phase 3, and judge it on several weeks combined.

Buying players per result, country by country

Soft launch spend goes to two things: enough players for clean cohorts, and a fair read of what each country will cost at scale. Paying per result fits the first, because you set a price per country and pay only when a player does what you asked.

• Pay for a goal, not an install. A result such as finishing the tutorial or reaching level 5 tells you the player saw the game. It also gives you a clean cost per engaged player to compare across countries.
• Bid per country. Set a higher price where players are worth more to you, and stop the countries that bring players who do not stay.
• Keep sources apart. Rewarded players behave differently from players who found the game in the store. Track them as their own cohort, and do not use them to judge your organic retention.
• Never pay for ratings or reviews. Both stores forbid it, and fake early reviews hide the problems a soft launch is meant to find.

Soft launch mistakes to avoid

• No stop number. Without one, every result looks “promising” and the soft launch runs for a year.
• Changing the build mid-cohort. Release changes at the start of a new cohort, or you cannot tell what caused the difference.
• Testing monetization too early. If players leave in the first session, purchase data says nothing.
• Judging by cost per install. The cheapest install is often the most expensive retained player.
• Forgetting server and support load. A soft launch is also a rehearsal for your backend, analytics, and player support.

Frequently asked questions