automating a test case usually means that you do not have to re write the same test cases over and over again.some of the criterias for running automation would be

1. Resuable scripts

2. test scripts which need to run on multiple operating system or different browsers

3. test cases based on requirements which do not change in short time

4. test cases which needs to be run with multiple inputs

 for example in the case of regression testing you do not need to re write the test case again as it would cause a lot of hassle. in case of mercury tool, if you have written a test case in quality center and given that you have QTP installed in your system, you can always link the two and run the automation via quality center and record and run the scripts by calling QTP from Quality center itself. you need to configure QTP to run the application that you are testing for and that QTP records the step that you have written. You can always enhance the recorded scripts by setting checkpoints (there are multiple checkpoints for eg standard checkpoints, page checkpoints,  table checkpoint, xml checkpoint etc etc which you can choose depending upon what you are testing for) in the keyboard view and run the test cases to see if it passes these criterias.  You can also choose to parametarize the scripts, meaning you can run the script with different data input. All of these can be done in QTP's keyboard view or in other words you do not need to have any programming skills to do any of it.

You can also use the expert view, given that you know Visual Basic script, to program the test scripts. this is also know as discriptive programming.

For most part of testing you will always have to rely on yer manual skills, for about 30% you might have to rely on automation. Most places i have worked for only worked with record and play part of automation. I had seldom used the expert view to program a script.

Let me know if this helps. please feel free to comment and correct me if i am wrong. thanks and hope this helps.