Subscribe to receive all latest blog updates

This white paper concludes our Testing Process Improvement series and considers maybe one of the most difficult questions of test planning: how to estimate time required for the new tasks.


Written by:
Anna Sheretova,
Senior Tester of Network Testing Team


When starting a task, each team member must have a possibility to get the information concerning the time spent on the similar tasks in the project before.

How it was:

To answer the question about how long the new task will take, tester had to:

- find a person, who had made a similar task, and ask him (if he remembered, surely),

- or search for the information about the similar tasks in Time Tracking System.

And only after that, he started to weight up what was the difference between the new task and the previous ones and make estimations.


We were wasting time on the search of the needed information.

The data provided by a specialist, who had made a similar task before, could be possibly mistaken or lacking some important aspects. Also, this specialist could already work in some other company, or be on the vacation, etc. Search in Time Tracking System was complicated by the fact that tester did not know for sure the name of the task and the exact period when team had worked on it.

It was a common situation when statistic data were lost, so we had to estimate task from scratch spending additional time on it.


We introduced the Statistics Table, where we accumulate data about the duration of various testing tasks.

Here is an example of such table:


Task name






Acceptance testing of the whole version



For 5 Browsers



Full testing accordingly to the Test Plan



525 tests






This table includes:

- Product name,

- Type of testing that was conducted,

- Task duration,

- Who was responsible for it,

- Some nuances of the task, which influence the time estimate.

Such table is useful for both test manager and any tester, who needs to estimate testing task.

Each tester after finishes his new task, adds the corresponding statistics to the table. If a similar task appears again, we build our estimates on the basis of that retrospective data. If the task we need to estimate is quite similar to the one in the table and testing is performed under the same conditions, we can simply use the number of working hours from the table. If some conditions have changed, usually we can foresee how it would influence the estimated time using data from the table. For example, for the task similar to the task number 1 in the table above, we will use only 2 browsers instead of 5. Or that task was made by a highly experienced specialist, and now a junior tester is assigned to the similar task. We can easily take these changes into account.

We can also analyze some combinations of tasks or sets of similar tasks selected from the different projects to estimate a new, unfamiliar task.


Each team member, even a newbie, can estimate the time that he needs for the major part of tasks in the project with rather good accuracy, and answer quickly the manager’s question: “When do you finish testing?”


I want to emphasize one more time that, in this article, we’ve described problems and solutions of our projects. Some of the problems were the consequences of the selected tools or project specific character; other ones appear frequently in various companies, which representatives I have talked to. The main thing is to remember that testing process can be improved. Do not give way to a wish to work from cheer habit.

Tatiana Kit,
Team Leader of Network Testing Team