One hundred percent test automation is a nice thought in hypothesis, but it might not yet be possible. Some code is too unmanageable to pen mechanization for, some codification frankly isn ’ t important plenty to write mechanization for. We need to leverage a risk-based approach to test automation that focus on the about crucial pieces of the software, this is how we generate the best return on investing.

Let go of the pipe ambition of 100 % automation. We need to acknowledge that we will never have decent time to make this happen to a product that is constantly in a province of modification. The best goal is to concentre on that ROI, things that are going to make your boss and CEO happy. What level of coverage do you really need? Well, that depends on your appetite for danger and the type of package we are developing. We want to deliver a premium user experience and cater digital confidence for our customers, but retrieve all bugs are not created equal. A small typo is not travel to change the exploiter experience in the same way a full-blown crash would.

For autonomous testing across multiple user personas, check out SUSATest — it explores your app like 10 different real users.

For more on tryout mechanisation and how we should approach it within our overall testing goals check out this picture.

video

Come back next week for our penult episode when I dispel the rumor that testing Chicago once we get into production. Spoiler alert: we ’ re never do examination.