Automation testing is wonderful for all the test cases and issues you anticipate or understand - but it can't deal with the issues people see on real world devices. The right answer, imho, is a portfolio-based approach with unit test, automation and manual testing.
Anyone know an affordable way for a bootstrapped startup to automate this? Testmunk looks like a good platform but the price (at least for us) is out of a small startup's price range.
I know from experience the hassles of point and click testing when a previous company tried to skimp on QA...my job was to write the user stories, but ended up testing half of them myself. Not enough testing at all, especially considering it only consisted of 4 devices, all of the same model. (PM, developer, a secretary and one QA guy later in the process testing on a company given device). The worst part was that even on the same device model, different results were happening. That ain't right. All in all, would have killed for some automated testing on that project. On a side note, wish I could provide some info (a warning)on what the company was that operated in this manner, but I can't.