Quick answer: Your crash-free rate is dropping because crashes are occurring more often relative to sessions, and the usual causes are a new crash introduced by a recent update (a regression), an existing crash spiking (more players hitting it), or an influx of new players on hardware/conditions that surface crashes. Version-tagged crash data tells you which, a crash appearing on the new version is a regression you just shipped.

Your crash-free rate, the percentage of sessions without a crash, is a key stability metric, and a drop means stability is getting worse. The good news is that a dropping rate usually traces to a specific, identifiable cause, often one crash, which version-aware crash data can pinpoint.

Why the Crash-Free Rate Drops

Crash-free rate drops when crashes happen more often per session. The common causes: a regression from a recent update, an update introduced a new crash (or reintroduced an old one), so a crash that wasn't there before starts occurring, dragging the rate down, this is the most common cause of a sudden drop. An existing crash spiking, more players hitting a particular crash (perhaps a condition became more common). And new players on new conditions, an influx of players (a sale, a launch) on untested hardware or doing new things surfaces crashes, lowering the rate.

A sudden drop right after an update strongly indicates a regression; a drop coinciding with a player influx may be new-condition crashes. Either way, the drop is usually driven by one or a few specific crashes you can identify.

How to Diagnose and Fix It

Use version-tagged, grouped crash data to find what's driving the drop. Look at which crashes are increasing and whether they're new on the latest version (a regression) or an existing crash spiking. Bugnet tags crashes by version and groups them by signature with occurrence counts, so a crash that appears on the new version (and wasn't on previous ones) stands out immediately as the regression dragging your rate down, and a spiking existing crash shows in its rising count. This connects the aggregate metric (the dropping rate) to the specific crashes causing it.

Fix the crashes driving the drop, starting with the highest-occurrence ones (and any regression from your last update, which you might roll back if severe). As you fix the top crashes, the crash-free rate climbs back. See our guides on measuring whether your fixes are working and tracking which update introduced a bug.

A dropping crash-free rate means more crashes per session, usually a regression from your last update or a spiking crash. Version-tagged crash data pinpoints the specific crash driving it, fix that and the rate recovers.