The authors report that about one in four data processing projects is completed at the estimated cost. Of projects costing over $50,000, 63 percent are over budget, whereas 14 percent are under (p. 52). These figures provide the motivation for the study undertaken here.
The study highlights how project managers (who, along with estimators and system analysts, are the intended audience) view estimating and system development as an effort to hit a moving target. The development effort often seems to be barely one step behind the changes requested by the business users or clients. Hence, the recommendation is to institute change control to produce more accurate estimates.
One of the best aspects of this report is the way it indirectly addresses the incentives to skew or misrepresent project costs. For example, assigning the initial estimating task to the final developer may prevent skewed lower estimates intended to result in project approval where the estimator need not deliver on the promise (p. 53). Likewise, a surprising result is the claim that the revision of the initial cost estimate does not improve its accuracy. The recollection of the initial estimate, regardless of user changes, continues to be the standard against which project success is evaluated. Hence, the authors recommend undertaking a thorough study before finalizing the initial estimate.
In a study such as this, the reader is likely to occasionally sense that the authors demonstrate an acute grasp of the obvious: for example, more accurate estimates occur when independent auditors participate (p. 57). Even in such instances, the guidelines provide a useful series of reminders for project management purposes.