How We Tried to Introduce E2E Tests and Why It Failed
summary
A frontend team tried to add end-to-end testing for a CMS serving 40+ domains and ~120k JSON files by creating a small 3 MB test repository and running Playwright in parallel (up to 16 workers), achieving about 50 tests in ~3 minutes. They migrated from C# to Playwright, trained content managers to write tests and produced ~200 test cases in a month but only stabilized ~50; major problems were brittle selectors, lack of test abstraction, frequent redesigns breaking tests, and unclear developer ownership and communication.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 6:45Why it failed
- 12:00Testing tool selection
- 17:21Post-redesign issues
- 25:50Branch test runs
- 27:55Project communication role
- 28:59Summary and tips
- 47:02Q&A
related talks
11mHow to make cheap e2e tests
21mAt the top of the pyramid: Playwright testing at scale
35mState of end-to-end testing for modern web applications
41mEnd-2-end testing without the pain
40mEnd to End Automation from a Developer Standpoint
18mAt the top of the pyramid: Playwright testing at scale
49mReal World Lessons from Scaling Test Automation
29mPlaying your tests wright