Lately I have been in some meetings to discuss if we should automate the tests scenarios or not. After some discuss with my colleagues we came up with this pattern: When reviewing a story instead of asking if we should automated the tests for the story ask: Why not to automate the tests scenarios for this story? If we don't have a really good reason to not automate, let's automate. If the reason is related to the size of the test scenarios, the difficult to write all the scenarios or quantity of scenarios then we should break/rewrite the story. If there's any technical reason, apply the 5 whys analysis to find the root cause, that in most of the cases will be a technical debt . What will left will be just real manual test, like visual validation.
Meus pensamentos...