The second session predicts what the first one cannot

Onboarding effort concentrates almost entirely on the first session. It is where the largest single drop-off happens, it is measurable immediately, and it is where the tutorial lives.

The second session gets far less attention and carries information the first cannot produce, because the two sessions answer different questions.

What each session actually establishes

The first session answers whether the game is comprehensible and whether the moment-to-moment experience is pleasant. A player who leaves during it usually did so because they were confused, bored within a minute, or hit a technical problem.

Those are real failures worth fixing, and they are the ones first-session analysis catches well. But a player completing the first session has only established that nothing repelled them. It says very little about whether they will come back, because coming back requires something the first session cannot test: a reason to return that survives closing the app.

The second session is where that gets answered. The player has been away, done other things, and something brought them back. Whatever that was, it is the thing your game actually offers. A player who returns for a second session has, without being asked, told you the game gave them something worth an intentional return.

That is why the second session is the stronger predictor. The first measures the absence of a problem; the second measures the presence of a reason.

The gap between them is the signal

The interesting measurement is not the second session itself but what happened between the two.

A player who returns within a couple of hours is finishing something. They are still in the first arc of engagement, and the second session is functionally an extension of the first. It predicts less than it appears to.

A player who returns the next day has made a genuine decision. Overnight is long enough for the impulse to fade and for the game to compete with everything else on their device. Returning after that gap means something in the game was worth remembering.

A player who returns after several days has usually been prompted rather than motivated. That is not worthless, but it is a different signal, and treating it as equivalent to a self-motivated return overstates the game's pull.

Splitting second-session returns by this gap tends to separate populations that behave very differently over the following month, and it costs nothing beyond recording the timestamp you already have.

What the second session should do differently

Most games treat the second session as a continuation — the player resumes where they left off, and nothing acknowledges that they made a choice to come back.

That is a missed opportunity for two reasons.

The first is that the player has forgotten things. Whatever they learned in session one has decayed across the gap, and a game that assumes retained knowledge produces confusion at exactly the moment a player is deciding whether this is worth continuing. A brief, unobtrusive re-establishment of context is more valuable in session two than most of what happens in session one.

The second is that this is when a game can start showing what it is beyond the tutorial. Session one is necessarily constrained — it has to teach. Session two can open up, introduce the thing the game is actually about, and give the player the first real taste of the loop they will be doing for months. Games that hold this back until later frequently lose players who never found out what the game was.

The monetisation implication follows the same logic. Session two is generally too early to ask for anything, because the player has not yet decided the game is theirs. Asking then converts a small number and costs a larger number of players who read the request as the game's actual purpose.

See how we time what a player sees to where they are →

Instrumenting it usefully

The second session is easy to measure badly. A session count treats all second sessions identically, and they are not.

The fields worth having are the gap since session one, what the player did in session two as distinct from session one, and whether they reached anything they had not seen before. That last one matters — a second session spent entirely in content the player already saw predicts differently from one that opened something new.

The comparison worth running is between players who returned for a second session and then continued, and those who returned once and stopped. That second group is more informative than players who never returned at all, because they came back — something worked — and then something else did not. Whatever happened in that session is a specific, findable failure rather than a general one.

Most onboarding funnels stop at first-session completion because that is where the biggest number is. The second session is where the smaller number tells you more, because everyone in it has already voted once in your favour.

Talk to us about early-lifecycle design →