UI Automation, What is ideal to automate?

Question

So when it comes to UI Automation, there are tons of opinions if it's worth it or not. I've read over: Are Exhaustive "Brute Force" UI Tests Worth It?

And I saw quite a lot of good opinions. I guess im wondering if how I tend to decide "what's worth it" actually makes sense? To me it needs to cover a few rules:

1. Is a regression, IE: Functionality won't be changing anytime soon
2. Doesn't have a ton of dependencies (Which seems hard to avoid)
3. Same as #2 but relating to data cleanup (Obviously should be able to created and destroyed with ease - data wise at least)
4. Can't be covered easier with a Unit/Integration/API test that is more reliable?

What else am I missing? What rules does everyone else have when deciding "Is this actually worthwhile to automate?"

Answer 1

Great questions! And now, here are some practical answers from someone that has been automating for the past 11 years...

1. Absolutely true! Do this all of the time!
2. This is not hard to avoid if you and your team are pragmatic in your approach. With proper planning, you will know what you have in regards to dependencies. You have to ask yourself, "do I/we?"

3. Use your attributes for your tests efficiently. Do you have your tests decorated with:

   [Setup] [TearDown]

If not, DO THAT. Manage your resources accordingly.

4. Do not rely on your dev team to provide unit/integration/API tests. As an automation engineer, that would ideally fall on your shoulders.

I hope this helps!