I went heavy on Playwright early on. Wrote tests for every button and every form I could. It looked solid until a small UI change broke a dozen of them because I’d used getByText in bad spots. The suite became useless for a while, and we were shipping code without it (so what was the point of all that work?).
Now I focus on automating the things I don’t want to do manually: setup flows, core user paths, data checks. If I can’t explain why a test exists, I don’t write it. I also check UI mocks early so I’m not surprised when labels change or elements move. It saves time and keeps the suite stable.
I keep my suites small on purpose. Five reliable tests are better than fifty flaky ones, and flaky tests destroy trust in QA fast. I also treat tests like real code. I review them, refactor them, and make small commits. If you skip that part, the suite turns into a mess quickly.
I use Playwright every day, but I’m not automating to hit numbers. It’s there to save time and keep focus on users, not on constantly broken tests.