Banksalad Ensures Mobile App Stability with Weekly Releases and Automated Testing
Key point
Banksalad has unveiled its collaboration approach for ensuring mobile app stability through weekly releases and automated testing.
Details
Banksalad, considering the irreversible nature of mobile app deployment, ensures stability through a regular weekly release routine and close collaboration between organizations. Similar processes are applied to both iOS and Android, and the company takes a strategy of minimizing bug risk even during large-scale refactoring.
Release Routine and Branch Strategy
Releases follow a week-based version naming convention (e.g., 23.48.*), and after finishing feature development in the first week, a release branch is created on Friday. The develop branch is managed to always remain in a deployable state, and when issues are resolved, they are first merged into develop and then cherry-picked into the release branch, reducing the risk of conflicts between branches. The main branch is used solely as an archive of the currently deployed latest version.
QA and Automated Testing
The internal test build created on Friday afternoon becomes the target of the QA team's automated testing over the weekend. Testing is performed by connecting actual devices in an infrastructure using GitHub Actions and the company's own 3 Mac minis, achieving wide coverage with a small number of people. The QA team completes automated and manual testing over the weekend, resolves major issues by early the following week, and then submits the app for store review.
Phased Rollout and Monitoring
A version that passes store review is first updated to only about 5% of users via phased rollout over two days starting Monday. If there are no metric anomalies or unusual customer inquiries during this period, full deployment proceeds; if problems are found, they are addressed with a hotfix release. This systematic safety net helps reduce the deployment burden on team members and maintain consistent quality.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.