Logo worldmap
HomeAboutClientsStaffResourcesContact_us
black belt

Black Belt Brief


Project Schedule and Probability

 

 

Probability of achieving a schedule date

The probability of achieving a specific schedule date is represented by a gamma distribution function. Gamma distribution functions have a discrete minimum value and a long roll-off tail. The curves can be short and flat or tall and thin depending on the confidence of the estimate, and is determined by the distribution's standard deviation.

There are four dates of potential interest in the chart below;

 
probability

- The earliest possible date is the most optimistic and least likely date that can be estimated. This date is virtually useless for anything other than an intellectual exercise.

- The most likely single date may be a useful estimate if the standard deviation is small, i.e., the gamma distribution is tall and thin.

- The 50% certainty date (median estimate) is an estimate where 50% of the outcomes will be earlier and 50% of the outcomes will be later.

- The 85% certainty date represents the date where, 85% of the outcomes will occur earlier.

 
 

Chart 1: Probability of completion on a certain date

     
 


Many executives will frame an estimation request in terms like "soonest possible", or "earliest possible" date. A literal response virtually guarantees a failure.

 
cum probability

The cumulative probability chart shows how the confidence of achieving the estimated date increases with more conservative estimates. When the consequences of late delivery are substantial, seasoned development executives seek greater certainty in the estimation.

Best practices suggests that three dates be provided for every prediction; the most likely date, the 50% certainty date, and the 85% certainty date.

The most likely date is used to drive the project team. This should be the internal goal that development build their project plans around.

 
 

Chart 2: Probability of completion by a certain date

     
 


The 50% and 85% date are indicators of confidence in the estimation. If there is considerable time between the most likely date and the 50% date, the confidence in the estimate is low. If the time between the most likely date and the 50% date is short, this is an indication that the estimating team is more confident in the outcome.

Over time and multiple projects, each individual doing estimation will accumulate sufficient data points to demonstrate the accuracy of their 50% and 85% estimations. Chronically inaccurate estimators need to be removed from this process, whether they are optimistic or pessimistic.