A Classification of Post-Release Stability Monitoring Practices for Mobile Applications with a Large User Base
A mobile release is not finished when a version reaches the store: users decide when to update, several versions stay in use, and failure signals arise on devices the release team does not control. Post-release observation is therefore a distinct engineering practice, yet research treats its objects separately, as problems of crash analysis, lifecycle testing, review mining or flaky tests. This review classifies post-release stability monitoring practices by the visibility mode of a failure, that is, by how the failure becomes observable once a version is in users' hands. The classification draws on a bounded literature corpus, platform documentation and the author's own release practice on a retail application with a large audience. Five classes are distinguished: self-announcing failures (crashes and hangs), silent failures (unexpected states; lifecycle and restoration failures), baseline-relative failures (regressions), instrument failures (analytics integrity) and gate-signal failures (flaky tests before release). A synthesis matrix crosses the classes with four release horizons. Audience scale makes only self-announcing failures visible early in a phased rollout; silent and instrument failures need instrumentation designed before release. A flaky test is treated by removing its cause, not by reruns.
Authors
- Mikita Burkouski (ORCID: https://orcid.org/0009-0001-5723-5068)
Publication Details
- Journal
- International Journal of Computer (IJC)
- Published
- 2026-11-20
- Primary Topic
- Software Engineering Techniques and Practices
- Type
- article
- Field-Weighted Citation Impact
- 0.00