Quick answer: A session in game analytics is one continuous stretch of a player using the game, from launch (or resume) until they quit or go idle long enough to be considered done. Sessions are a fundamental unit of measurement: session count, session length, and per-session metrics like crash-free sessions all build on this basic concept.
The 'session' is one of the most fundamental units in game analytics, so basic that it is easy to overlook, yet many important metrics are defined in terms of it. A session is essentially one sitting with the game. Understanding exactly what counts as a session, and why it is such a useful unit, clarifies a whole family of metrics, from how long players play to how often the game crashes, that are all measured per session.
What Counts as a Session
A session is a single continuous period of play: it begins when the player starts the game (or brings it back to the foreground) and ends when they stop, by quitting, or by being inactive or backgrounded long enough that the game considers the session over. One session is, roughly, one sitting, one bout of play from start to finish.
The edges have nuance. What ends a session, an explicit quit, the app going to the background, a period of inactivity, depends on definition, and there is usually a timeout: a brief interruption (a quick app switch) might not end the session, while a long absence does. These details matter because they determine how sessions are counted, but the core idea is simple: a session is one continuous engagement with the game.
Why Sessions Are a Key Unit
Sessions are the natural unit for measuring engagement and behavior because they correspond to how players actually use the game, in discrete sittings. Two of the most-watched engagement signals are built on sessions: session length (how long each sitting lasts, a measure of how absorbing the game is) and session frequency (how often players return for a session, a measure of stickiness). Together these describe engagement far better than raw playtime alone.
Sessions are also the denominator for important rate metrics. Crash-free sessions, the share of sessions with no crash, is a per-session stability metric, and many quality and behavior measures are similarly normalized per session. Counting things per session makes them comparable across players and over time, regardless of how much any individual plays, which is why the session is such a foundational analytics unit.
Sessions and Stability Measurement
For stability specifically, sessions are the basis of crash-free sessions and related metrics, which means counting sessions accurately matters for measuring your game's stability. To compute what fraction of sessions crash, you need to know both how many sessions occurred and how many crashed, the session count is half of that equation.
Bugnet's crash reporting captures crashes in the context of play, and combined with session data lets you express stability per session, the crash-free sessions rate, rather than just a raw crash count. This is more meaningful because it normalizes for how much the game is played: a hundred crashes means something very different across a thousand sessions versus a million. Understanding the session as the unit, and tying crashes to sessions, is what turns 'we had some crashes' into 'X% of sessions are crash-free,' a metric you can track over time, compare across versions, and use to gauge whether your stability is genuinely improving.
A session is one sitting with the game. It's the unit beneath session length, frequency, and crash-free sessions, basic, but most metrics build on it.