In my experience I have seen several methods of deriving task durations related to projects. Some of them include:
1. A cocktail napkin dropped on a project manager’s desk with start and finish date for the project based on last night’s executive dinner and the plans and commitments made.
2. The project manager being told unilaterally to put together a project schedule based on limited information which might include a project charter or other high level information
These methods have always led to poor results and even worse project plans.
My suggestion is that the project manager relies on both a rigorous requirements definition and the expertise of the technical resources that will be doing the actual work on the project. In this method, the project manager will convene at least one meeting in which the specific project tasks will be discussed. During this discussion it is important to break the tasks down to a reasonable level. It has been repeatedly proven that the accuracy of duration estimates improves as the tasks are broken into smaller subtasks. It is also important to include the actual technicians in the planning phase from the “skin in the game” perspective. After all, technicians are far more likely to meet their deadlines if they themselves established them, verses having an executive or project manager dictate them.
My recommendation is to always ensure that the project team works closely together in both the baseline and any updates to the plan as the project continues towards completion.