Original Post Date: Wednesday, February 23, 2011

In the world of estimating, accuracy is the first question out of people’s mouths.  Above all else they want to know the accuracy of an estimate.  How accurate is that approximate judgment?  Craziness!

True accuracy can only be determined after the project or effort has been completed and a post-audit analysis reconciles what was expected to happen with what did happen.  This is a very expensive, time consuming process that many preach about but few actually attempt. 

In my experience, when people ask about accuracy what they are really interested in is uncertainty.  They want a quantification of any uncertainty in the facts and processes that derived the estimate.  Because naturally, the more uncertainty the less accurate the estimate.   I am using Dale Shermon’s definition of uncertainty as explained in the book, Systems Cost Engineering. Uncertainty means ‘a possible event, the probability and the consequences of which are unknown’.  

Uncertainty about system or project scope, requirements stability, cost or effort drivers, the assumptions, and the range of possible answers all contribute to the individual’s assessment of how much doubt to apply to a response.

In an effort to describe the accuracy of estimates, the industry had added entire disciplines designed to generate all comforting statistics around someone’s approximate judgment.  Risk analysis, confidence levels, technology readiness levels are all valid and useful guides to help the consumer of the estimate understand the estimators assessment of their own accuracy…uncertainty.

What techniques or methods have you found useful to convey your estimate’s accuracy/uncertainty?

Have you experience the same relationship between Accuracy and Uncertainty?