Did you know that I could test your software with just one test case? The test case does not even need to be complicated.
So here we go. I will now demonstrate a test case I planned for Twitter.com a mere moment ago.
Step 1: Test Twitter.com
Expected result 1: Twitter.com works
Nothing mysterious there. With this remarkable test case, you get a requirement coverage of 100% and equally impressive test coverage too. With this one case, we can also measure the system’s functionality reliably. Right?
Any project leader who has any knowledge of testing would of course find my logic faulty. You just cannot measure any system with one test case. You don’t even get any metrics from that!
Ok. So if my one case isn’t enough, how many would you like to have? 10, 100, 1.000 or 10.000? How many would you need, to get good metrics? What would be a high enough resolution for your expectation?
I don’t know the answer for Twitter. I don’t even have an idea about the scale. 1.000 cases might be just as correct as 10.000. And I’m pretty sure that the project leader doesn’t know the answer either.
You can try this out for yourself. Write test case like the one above. You can choose the target freely. Then start doubling the resolution. Split each test case into two new ones and see how far you can get before you start to get frustrated. I bet you didn’t even get to FullHD here.
The number of test cases is an arbitrary figure. It’s subjective. And this excercise is a tool for you to demonstrate it to others.
Testcase is like a suitcase. You will never know what’s inside.