Cone of Uncertainty

Uncertainty of project estimates (Specially the software projects) is the biggest problem we face when planning a project ahead. I’ve never made a perfect plan for a project (The preliminary plan to be valid till the project is completed and to complete the project successfully with the same exact number of hours I planned initially?? No I have never been lucky).. So how do we deal with this uncertainty of projects when planning? Yeah I hear you .. Agile .. Scrum.. NO in this post I’m not talking about that.. Im just talking about our normal pre planned projects.
“Cone of uncertainty” (research done by NASA) reflects the uncertainty at different points of sequential development of a project. It shows that
In feasibility study stage schedule estimate can be different in a range of 60% - 160%
After the requirement specifications - + or – 15%
Likewise when the project progress each stage of the project reduces the uncertainly of the estimate. However uncertainty of estimate becomes 0 when the software is accepted.
Ok this is a long discussed theory.. then what does PMI say about the project uncertainty ? I wanted to refresh my memory J
PMI also believe in progressive accuracy of estimation. PMI defines some asymmetric way of viewing the uncertainty of estimates.
Initial order of magnitude estimate can have a difference from +75% to -25%
Budgetary estimate can have a difference from +25% to -10%
Final definitive estimate reduce the difference up to +10% to -5% Most the time at the level of the budgetary estimate , we sign up the contract with the customer. So we have somewhat risk in the estimations to take in to consideration when preparation of the estimates.
Twice I worked with some pretty large projects where we had to agree on penalties for each day of the schedule over run if occurred in the project.. Both times I could manage it within the agreed schedule.. (So didn’t pay any penalty to the customer) But I didn’t complete the project with expected profit to the last cent. Because I had to do certain resource utilization variance to deliver a project which was confirmed based on the budgetary estimate.
But now that I need to think beyond what I see I just like to have some feedback from other readers about this situation.. One solution which we can do is to apply risk factor to the budgetary estimate and have the luxury of that provision. The other solution is revising the estimate in selected milestones. I have done that many times.. but this needs establishing trust , loads of negotiations and convincing with the customer and other stakeholders as most of the stakeholders don’t care about cone of uncertainly..:-) The Million dollar question is that how do you handle this situation in traditional project management methods without hurting your project bottom line ?