Monday, August 05, 2013

The importance of estimation procedure for traditional projects

One of the major causes for Project delays in Engineering/IT projects is usually poor estimates.  A careful examination of estimation practices is required  to assess whether estimation is the real cause. If the estimate is single point  and only done at the proposal stage of a long duration project,  the estimation will necessarily be poor. 

Steve McConnel's Software estimation book presents 118 tips for improving the estimation after exploring both the art and science of estimation. If I have to pick one best tip, it will be Tip#77, Develop a standardized estimation procedure at the organization level; use it at the project level.

Estimation procedure makes it practical to implement  the cone of uncertainty, which is nothing but how the accuracy of estimate is improved by appropriate selection of  estimation methods and their inputs. Considering NASA SEL estimation procedure example, the project at the end of requirement analysis uses the number of subsystems as input and thumb rules for size and effort estimation along with uncertainity range of  -43% to 75%.  The rule for estimation at end of implementation is to use the current size, effort expended and schedule expended to derive new estimates for size, effort and schedule with an uncertainty range of -9% to +10%. Determining the procedure is difficult, as the  historical data of projects need to be analyzed  and transformed into thumb rules.  In my career, I  did this for few projects I managed to improve the thumb rules  of estimation, spending considerable time to collect the relevant metrics.  By being diligent about updating this year after year, organizations can improve estimation accuracy.

How are you improving the estimation baselines in your organization? Do you have detailed estimation procedure which uses updated baselines? Share your thoughts.

1 comment:

admin said...

This is a fantastic website and I can not recommend you guys enough. fundamental trigonometric identities solver