What causes cohort retention mixing different install sources?

Retention is computed over all installs together, so high-intent organic users and low-intent paid users average into a single misleading curve.

How do I catch this when it only happens for some players?

Capture it automatically from the player's device with the full stack trace, the device and OS, the build, and a breadcrumb trail of what they did. That turns a vague complaint into a specific, reproducible issue you can fix without owning the hardware it happens on.

How to Fix Cohort Retention Mixing Different Install Sources

Quick answer: Cohort retention by acquisition source and campaign using attribution captured at first_open, then compare curves per source instead of one blended line.

A blended retention curve hides that organic players retain at 40% while a cheap campaign retains at 5%. Splitting cohorts by install source exposes which channels actually deliver players who stay.

How to fix it

1. Capture source at first_open

Attribute every user to a source and campaign at install and store it immutably on the user. Retention then groups by that stable attribute.

2. Split the curves

Compute D1/D7/D30 per source cohort rather than over all users, and chart them separately. The blended average is what masks bad channels.

3. Control for volume

Show cohort sizes alongside the curves so a tiny high-retention cohort does not mislead spend decisions. Small cohorts are noisy and need a confidence caveat.

Catching the ones you can't reproduce

The hardest version of this to fix is the one you can't reproduce — it only happens on a player's hardware, OS, driver, or save state, under conditions that simply aren't present on your machine. A report that says “it crashed” or “it froze” gives you nothing to act on, so the bug survives release after release while quietly costing you players.

Automatic error capture closes that gap. Each failure arrives with its full stack trace, the device and OS, the build number, and a breadcrumb trail of what the player did right before it broke, so even a failure you have never seen becomes a specific, reproducible issue. Fold identical failures into one signature ranked by how many players each hits, and your worklist sorts itself worst-first instead of arriving as a stream of vague complaints.

This is where a tool like Bugnet earns its place. Its SDK captures every error automatically with the full stack trace plus device, OS, memory, build, and game-state context, folds duplicates into one grouped issue with an occurrence count, and ties each to the build it first appeared on — so you fix the problem that hurts the most players first and confirm it is gone when its signature disappears from the next release.

Reproduce it once with full context and the fix writes itself. The hunt is the expensive part.