Review Project Audit Results
Key Question: Are the task estimates based upon a known history or are they just guesses?
"Probably the most difficult function in building a project plan is estimating the work effort. Most people tend to estimates on a 'personal experience basis.' Another term for this is 'my guess'. Personal experience, dead-reckoning, or guesstimating work well if the estimator has the appropriate baseline definition documents and is doing the work personally. Quite often in a runaway project, a senior architect will have the responsibility for the technical estimate. This person tends to have a great deal of experience in performing the technical tasks to be estimated and produces an estimate based upon that experience. Unfortunately, the technical team tends to be less experienced than the architect - sometimes fare less experienced. This leads to a negative variance in time and effort. Slowly but steadily the project gets farther and farther behind. Left unchecked, more and more time and money are poured into the project and you have the classic runaway project."
"An even worse scenario exists related to estimating has led to the demise of the well-intended project. In this case, the project team did all the right things and produced a defensible estimate of the work effort and subsequent schedule and budget. However, either the budget or the schedule or both was unacceptable to the project's executive sponsor, for real or perceived reasons. To make the project more acceptable, the estimate was reduced by an arbitrary amount without reducing the scope of the project. Unfortunately, these team members were only fooling themselves. The work is the work. Not surprisingly the project falls behind and is more in line with the original estimated effort, budget, and schedule."
"Once you get the plan's activities, tasks and dependencies in place, the proper way to estimate your project is to use a proven development methodology or a set of proven project plan templates with experience-based work effort estimates as the initial guide. These estimates are only a start to your estimating process. Working with the project team, determine the best case, worst case and most likely case."
"Start by putting the template estimates in the Most Probable Case column and adjust it according to the experience of the planned resource. Next, working with the IT team, add the best case and worst case scenarios. Risk factors such as level of specification, physical resource availability, geographic location of the extended team, and so on will help derive the best case and worst case scenarios. Next calculate the weighted average using the following formula:"
Weighted Average = (BC + (MP * 4) + WC)/6
"This gives the greatest weight to the most probable case, giving a 67 percent probability of the weighted average falling within +/- 1 sigma. The Weighted Average represents probability that the task can be completed at or less than the Weighted Average. Of course, the real value here is the variance - the greater the variance, the greater the uncertainty of the estimator. The variance is what you want to reduce. The lower the variance the greater the estimator's confidence in the estimate and the more stable the overall project estimate will be. Use these weighted averages as the task estimate. Finally, refine the estimate by computing the standard deviation of each task using the following formula:"
Standard Deviation = (((WA - BC)2 + (WC - WA)2)/6).5
"Review any components having a standard deviation greater than 2. This indicates a low level of requisite knowledge for that task and the worst case estimate should be used for that task instead of the weighted average. The new estimate for the project would be the sum of the refined task estimates."
The above is an excerpt from a book written by Sanjiv Purba and
Joseph Zucchero, published by McGraw-Hill/Osborne, 2100 Powell Street,
10th Floor, Emeryville, California 94608 U.S.A. Sanjiv has over 20
years of experience managing large projects and many years engaged in
rescuing ailing projects.