Test automation is the first method to reduce costs, but because there is also a high probability of failure, I think that automation should be carefully planned, designed and executed under a well-skilled manager if possible. Otherwise it costs a tremendous cost but it is highly likely to be an entirely useless automated test. Why do not you consult us once before trying to realize automation by all means?
The following figure shows the tool maker with and without automation first.
However, it often becomes more costly to perform the system test manually as shown in the figure below instead of becoming like the above.
At our company we define tests for automation as follows and tests not suitable for automation. So when automating for the first time, we recommend that you test the following for automation. However, at our company, tests of video and audio that are said to be not suitable for automation use AI and have technologies that can be automated, so please contact us.
Let's give some examples of tests and reasons for automation
■ Automated test
- Smoke test Because it
repeats a lot of times.
- Performance test It
is better to simulate many people with tools than to gather lots of people.
- The API test
API does not change so much.
Let's give a test that is not suitable for automation.
■ Test not suitable for automation
- Regression test
Regression test seems to be suitable for automation at first glance, but that is a big mistake. I also automated it in the order in which it was registered in the bug database (it was stupid). In the first few months I felt like 'Good bad, I like it!', But after several months the automation program ceased to be halved due to repeated user interface changes in the program. Still if you automate it, the number of test cases to be automated did not increase at all and only the maintenance work of existing automation code was done.
- How do you automate tests of media-related test sounds and videos such as graphics and sound on the other hand?