Tackling Test Automation Part I: What problem are you trying to solve?

For the past seven months, since I came back from maternity leave, I’ve spent the majority of my time focussing on test automation. I have both set up a test automation framework as well as overhauled two existing test automation frameworks for mobile (both Android and iOS).

But one of the struggles I have faced is not getting lost as to fail to see the forest for the trees. During the day to day, it can be easy to be constantly knee deep in code and forget the problem I am actually trying to solve. Why am I actually doing test automation in the first place?

I have found having some focussing questions to be great for both starting a project where I have to write Test Automation but also to come back to these questions throughout a project to make sure I am still on track.

1. What problem am I trying to solve?

This is the main one. For me, at my current project, I want to cover regression so that as the team adds features; makes changes etc. they can do so without worrying about inadvertently breaking something else.

In general, I like to have basic flows covered as well as targeting areas that I have noticed have been a bit problematic in the past (I do this by talking to people in the team and going through the bug tracking tool to see if I can see patterns).

2. Why do I have this test?

Taking a step further and looking at the tests in more detail, I start thinking why do I have this test? What value does this specific test add to my test suite?

3. Does this test still have value?

I also then think, does this test still have value? As changes to our apps are made and I may be looking through our existing suite to see if/how I need to update tests, I may then find that some tests no longer have value.

This is most likely because:

To be honest, I’m pretty brutal, I don\’t like have an excessively high number of tests and I don’t think you need to have heaps and heaps of tests to achieve good coverage. If I can achieve a good amount of coverage, with a reasonable number of tests (where each test still has its specific focus/goal) then I’m all for it.

4. What type of regression bugs can my test automation suite miss?

I then take a step back and ask myself, what type of regression bugs will my test automation suite miss?

Asking myself this question, helps me change my perspective to see what else I need to address. Often when I try to solve problems (not just test automation related), I ask myself questions in different ways to try and figure out what else I could be missing. This question is an example of that.

Follow me on Twitter: @NicolaLindgren