We have discussed the cost of no automation and reasoned "when to test".
This doesn't automatically solve the test coverage question - the QA teams are swamped enough, and if you ask them to test across the SDLC, that can only lead to testing what is easy or whatever is quick.
Hence the next question - What tests to prioritise?
The testing pyramid was and is gospel, still, and every sprint planning meeting with a QA team included some version of: "we need more unit tests at the base, fewer E2E tests at the top."
The pyramid made complete sense - for its time. I'm not sure it makes sense anymore.
