The application has to be automated. What test cases I will consider to automate .
A generic answer here is:
- Essential stuff like login. It's a start.
- Easy to automate functions to acquire some coverage, fast.
- Bugs as you encounter then, this is good practice and already receiving attention anyway.
- Fragile functionality as determined by your developers.
This list goes on, of course. In the end 100% coverage is the goal. Which never will be reached. But that is another story.
When trying to figure out what to automate I like the questions (and advice) from this Elemental Selenium Tip. It basically says
A great way to increase your chances of automated web testing success is to map out a testing strategy. And the best way to do it is to answer these four questions:
- How does your business make money?
- How do your users use your application?
- What browsers are your users using?
- What things have broken in the application before?
Then you map out a strategy based on these ideas. I like to throw my strategy into a Mindmap so I can share it with the rest of the team.
Sure you can automate easy things but ideally you'd be able to start with the easy as you work towards the important things.