๐ง๐๐ ๐๐๐๐๐๐ก ๐๐ข๐ฆ๐ง ๐ข๐ ๐๐ฎ๐ ๐ง๐๐ฆ๐ง๐๐ก๐
I thought I was one fix away from a green test suite. I added a selector. I patched a bug. It felt close. I learned a lesson. E2E testing is hard when your UI changes every release.
Playwright codegen looks easy. You record a flow. You use AI to clean it up. The first test works. You trust the tool. Then you see the truth.
Here are four hidden costs:
- Source Prep: Add stable IDs to every button before recording.
- Manual Fixes: AI misses things. You fix selectors by hand.
- Fragile Flows: Long tests break often. One UI change ruins the chain.
- Dirty Data: Tests create data on TikTok or Facebook. Deletion is not reliable.
These costs repeat every release. You do not build it once. You pay a bill every time you change the UI.
I paused E2E investment. We kept the tools. We stopped the build. Knowing when not to build is valuable.