Estimating Testing Effort

The estimated testing effort for a project depends on the factors discussed in the previous section. Most of these factors exert mutual influence on each other and it is virtually impossible to analyze them all exhaustively.

Estimation techniques

There are two basics approaches you can take to better estimate testing effort and costs:

• Expert-based estimation
  The overall testing effort is estimated task by task. This is done either by the task owner(s) or by experts with appropriate experience.
• Metrics-based estimation
  The testing effort is estimated based on the known effort involved in previous or similar current projects, or based on key metrics (for example, the average number of test cases executed per hour).

Even if the test manager doesn’t overlook a single task, an expert-based estimate will usually view the project too “optimistically” and ignore things that could go wrong, thus underestimating the effort required. An estimate based on experience and/or figures from similar projects will usually provide a more realistic assessment.

An estimate for a small project can be made “in one go”, whereas larger projects require a separate estimate for each test level and/or test iteration.

Rule of thumb

If there are no references available, you should always reckon to use 25-50% of the entire development effort for testing (on all levels).

Our Tip Estimating testing effort for agile projects

• The amount of testing that the team can perform for the next iteration can be estimated using metrics based on team productivity (also known as “team velocity”) or using a kind of expert analysis known as “planning poker” (see [Linz 14]).The important thing is to plan testing and all other tasks in the course of the sprint planning—and don’t forget to include test infrastructure maintenance and improvement.