How do I determine the schedule for a set of tasks and estimate the time required for each task?

In the process of work, we often face a number of projects or a number of tasks, if not to clear the task, or do not put the time for scientific and reasonable assessment, it is easy to lead to time is not enough, the test is not sufficient, and then the leadership does not approve, his hard work not only not get the reward but the dilemma of pressure.

What should we do?

Talk about my own views, I hope you see more exchanges.

• Each tester's work has a large number of tasks, so it is necessary to develop a checklist of test tasks, which is the first step.
• Some tasks require a general description, and some tasks can be broken down quite finely. Make a separate list of tasks that will take more than one day to complete, depending on your ability.
• Estimate the amount of time each task will take and then add up, plus 25% (or less) of meetings, training, and other non-project work, based on the company's specific circumstances, and estimate the total time required.

The points above are known to all, but in fact it is easier said than done. For example, listing a task list is a simple matter, because it is easy to omit or underestimate the scope of the test (which brings up a question, who should be the time required for the task?) ）。

Talk about my ideas:

• Analogy: If you have done a similar project, you can estimate the time of this task by analogy with previous experience;
• Model estimation: If you understand the length or complexity of a project, and you understand the model based on the data that the current company associates with the length and complexity of the test time, you can use this model for estimation.
• Consider risk: Consider the risk of the project and then list what should be done about the risk (time and Task)
• Other factors: such as the level of understanding of the task, such as the level of skill and rigor of the developer of the task, such as the programmer's proficiency in the application. Also such as the programmer this period of bad state, make more mistakes, also need more testing. If you write a test document, you can also make the test work faster.

Note: Using a similar approach, the Test manager can estimate the number of testers at any point in the progress of the project, and the more it is later in the project (the more information is available), the more accurate the estimate is.

In our company, the test is usually carried out in two rounds, that is, the plan to be two rounds of testing to estimate, this is appropriate?

　　Before I came to the company, I asked the project to do two to one rounds of testing. They thought that the first round would burst out all the problems, and the second and third rounds checked all the wrong modifications. In other words, this is like wishful thinking that the application will not have a change in demand, all defects will be changed one at a time, and other related functions will run very well! In fact, we all know that the product has to be done much more often than in two rounds.

• As product understanding progresses, we will consider new and better tests and find new problems. If only two rounds of arrangement are made, the above-mentioned situation will be suppressed.
• Even if all the problems are found in the first round, the odds of not introducing a new problem after modifying the defect are negligible.
• Moreover, many times the test cases are not executed in the first round and many defects can block the execution of the test.

In fact, I would like to say that the plan changes are not scary and can not be stopped, it is scary that many companies and teams will consider changes as a failure and delay.

　Another thing is that the tester has to decide the time it takes to test the task, and I'll say what I think about that.

As a Test manager, I often use my own time to do a task to ask the team members, but I have to admit that several times I have underestimated the tasks assigned to other people. My approach is that if my assessment is in conflict with the testers ' own assessment, especially when they have a long time to evaluate, listen to their views on the test task and the scope of the test and figure out what causes them to look so long. -This is a great opportunity to help testers grow.

Sometimes I have to revise my estimate and redefine the test task.

It is important to note that do not force testers to accept their own views, everyone is not a fool, this will let oneself lose authority, and the task is so much, the actual need for time is basically fixed. Forcing testers to accept their plans is hard to get a good result.

Of course, I am committed to spending more time on the test plan, rather than let the test task bear the test time, because there are many "distinctive" people in our department, the staff awareness is serious, a 2-hour task can be completed, they can give you an estimate of 2 days.

In my last company, my practice was to let the people who bear the job tell me the time. After bringing people out, they are very relaxed.

All in all, the person who makes the estimate should be the one who takes the most notice, sometimes the manager, sometimes the head of the test, and sometimes no one. It depends on who has more information and who is responsible for estimating the problem. But in either case, do not use "hope" to estimate.

　　

　　
(0 0)
+------OOO---(_)------------+
| |
| "Welcome attention" |
| Miss Zhang's Little black House |
+--------------------OOO-----+
|__|__|
|| ||
Ooo ooo

How do I determine the schedule for a set of tasks and estimate the time required for each task?