App-review volume is noisy. A handful of angry reviews can look like a regression, while a real recurring problem can be buried across countries and app versions.

For BugChorus, I only compare two observed versions when both samples clear minimum review thresholds. Even then, I label the result as an investigation signal not proof that the release caused the issue.

If you monitor mobile releases, what would make a review-based alert worth investigating: a minimum review count, repeated-issue count, percentage change, or something else?

The launch is live now, and I m using the answers to tighten the default evidence rules.