PERT (method for estimating)

Program Evaluation and Review Technique

I first ran into PERT a couple years ago. Any “well known” methodology catches

my attention. I questioned the validity of this method for estimating at first but later realized it is fairly accurate on projects with many activities. I’m not sure you can expect accuracy on a single activity.

Wikipedia “PERT is a method to analyze the tasks involved in completing a given project, especially the time needed to complete each task, and identifying the minimum time needed to complete the total project.”

http://en.wikipedia.org/wiki/Program_Evaluation_and_Review_Technique

The premise is the majority of time lost and time saved for unexpected activity performance can be anticipated. The most optimistic view is weighted against the most pessimistic view then both are tempered by the most likely scenario weighted four times more than the others.

There is no magic or mysticism involved. If there is any magic it is the expert judgment required by the person estimating the activity.

     Optimistic - shortest time in which the activity can be completed.

    Most likely - highest probability guess.

    Pessimistic - longest that might be required.

FORMULA:

                         Optimistic + (4 x Most likely) + Pessimistic

Estimate =     -----------------------------------------------------------

                                            6

TCE (Total Containment Effectiveness)

The effectiveness of a development team to capture defects before product

release significantly reduces the total cost of ownership of an application.  It is far cheaper to correct a defect before product release.

One simple measurement of this effectiveness is TCE (Total Containment Effectiveness).  This is the percentage of defects contained or discovered before a product is released, divided by the total defects both pre and post release.  The higher the percentage the more effective a development team is.

                   preRelease Defects
TCE = ------------------------------------------- * 100%
            preRelease + postRelease Defects

To get the postRelease Defect count the application must be used for 6 to 12 months.  If you track the version a defect was injected into then you can further extend this analysis while other versions of the application are release.  This is helpful in an agile environment where TCE would be more difficult to track because of the number of releases required per cycle.

If you then track the development phase the defect was injected then analysis will lead you to improving the effectiveness of a struggling phase.