Every time we move to faster sprint cycles, we hit the same wall: the regression suite becomes too slow to run on every PR, but skipping it feels like playing Russian roulette with the production environment.

I’ve been looking into ways to make our regression testing agile without just "throwing more hardware" at it. One approach that caught my eye is focusing on modular test suites and impact analysis instead of the usual monolith.

This breakdown actually covers some solid strategies for keeping it lean: https://testomat.io/blog/agile-regression-testing/

Curious to hear from others — how do you decide what not to test in a sprint? Do you rely on automated impact analysis, or is it still a manual call by the QA lead?