Most of the players you lose, you lose before they have played properly. Worth treating that first minute as its own piece of design.
Get to play immediately. Every screen between launch and doing something costs you people: splash screens, logos, login walls, a settings tour, a permissions prompt with no context. If it can happen later, make it happen later.
Teach by doing, not by telling. A modal explaining three mechanics is read by almost nobody. Put the player in a situation where the only available action is the one you want to teach, and let them discover it. Then get out of the way.
One idea at a time. If your first level introduces movement, an enemy, a currency and a shop, most of that will not land. Space them out across sessions.
Do not ask for anything early. Notification permission, an account, a rating prompt, a purchase, all before there is any reason to care. Ask after they have felt something worth caring about.
Make the first success arrive fast and be obvious. People continue because something felt good, not because they understood the systems.
How to find your actual problem: instrument the funnel. Log first launch, tutorial start, tutorial complete, first level complete, and session two. The biggest single drop tells you where to work. Almost every team is surprised by which step it is.
A useful test costing nothing: hand your phone to someone who has never seen the game, say nothing at all, and watch. Where they hesitate is your onboarding bug. The hard part is genuinely staying silent.
Where do your numbers drop off?
Get to play immediately. Every screen between launch and doing something costs you people: splash screens, logos, login walls, a settings tour, a permissions prompt with no context. If it can happen later, make it happen later.
Teach by doing, not by telling. A modal explaining three mechanics is read by almost nobody. Put the player in a situation where the only available action is the one you want to teach, and let them discover it. Then get out of the way.
One idea at a time. If your first level introduces movement, an enemy, a currency and a shop, most of that will not land. Space them out across sessions.
Do not ask for anything early. Notification permission, an account, a rating prompt, a purchase, all before there is any reason to care. Ask after they have felt something worth caring about.
Make the first success arrive fast and be obvious. People continue because something felt good, not because they understood the systems.
How to find your actual problem: instrument the funnel. Log first launch, tutorial start, tutorial complete, first level complete, and session two. The biggest single drop tells you where to work. Almost every team is surprised by which step it is.
A useful test costing nothing: hand your phone to someone who has never seen the game, say nothing at all, and watch. Where they hesitate is your onboarding bug. The hard part is genuinely staying silent.
Where do your numbers drop off?