The following are some general considerations to be taken into account when initiating a project automation software.

The first step is to analyze the feasibility of the automation.

Consider the following:

* The limitation of resources,
* The budget,
* The schedule and;
* The knowledge available.

Therefore it is expected that not all combinations and permutations of use cases / test in an application can be tested, and therefore not all tests can be automated.

Will require a great effort from the team, set priorities for automation.

One way to do this task is to create an "index of automation."

If the project is to automate processes that were hitherto tested manually, the calculation should be done as follows: Number of test cases to automate viable dividing the total number of test cases.

The term "number of test cases to automate viable" is used because, while almost all the test cases can be automated, given the limitations set forth above, not all can be automated effectively. Therefore, it is very important to distinguish the cases that can be automated for those who are really practical to do so. (Here we should also add as an ingredient of the domestic policy issue in most situations influences and impacts throughout project).

The analysis of all potential test cases to be automated through the use of this list will help reduce the likely number and focus on those cases that are most viable to be automated.

After reaching this first step and the first result, we must consider whether the experience and resources of the organization will implement the process of automation.

After this stage, we must consider very carefully the criteria referred to the risks and complexity in the test cases, and from personal experience that in some situations are more difficult to establish automation, so the project leaders must carefully assess the balance between the development of the command to run the test automation against time and cost required to develop the functionality itself.

After reaching this stage and have received a number of cases candidates, the next step is to analyze the feasibility of automation, as if some functionality is in development stage there will not be stable enough in the flow or workflow to consider and therefore some of the commands built for the run of the test will generate errors or warnings.