Theory of Constraints Handbook - James Cox Iii [44]
Copyright © 2010 by Charlene Spoede Budd and Janice Cerveny.
For traditionally-managed projects, two assumptions guarantee project completion delays: (1) project task times can be accurately predicted, and (2) the traditional project management planning and control system is effective (Leach, 2005, 10–11). Resources are asked to provide an estimate of the time required to complete a particular task. Once all project resources have reported their estimated (and safe) times, management frequently requires lower estimates. If those estimates are accepted by all resources (and resources usually have little choice), the estimate becomes a commitment upon which the resource will be evaluated.
Task Duration Uncertainty
We know that task times follow a distribution pattern that is skewed to the right. No task can be completed in zero time, but the maximum possible time can be extremely long. Look at a simple example such as the time required to drive to an important client’s office. Let’s say if you pressed the speed limit (exceeded it by 5 or 6 miles—9 or 10 is more common in Atlanta) and encountered no problems, you might make the trip in 20 minutes (the minimum task time). Normally, however, the trip takes about 30 minutes. If there was an accident on the freeway that you couldn’t avoid, it may take several hours. If you had to promise your client that you would be there at a certain time or lose your account, how much time would you estimate? It would certainly not be 20 or 30 minutes. The same is true for a project resource who must promise to complete a task in a certain amount of time. The estimate typically will be in a range such that the resource has a 90 to 95 percent probability of successful on-time completion.
Since task times follow a skewed distribution, as illustrated in Fig. 3-1, and have unique properties, completion times cannot be estimated with precision. Nevertheless, an estimated time must be provided. Resources operating in traditional project management environments, therefore, are forced to protect their careers by providing times with appropriate safety that will permit them to survive management “adjustments” and to deliver on their promises.
The area under the curve shows the probability of completing the task in a given time estimate. Estimated completion time if resources could dedicate their time to the task, without interruption, most likely would occur somewhere to the left of the longer vertical dotted line in Fig. 3-1 (between the two arrows pointing in opposite directions). A minimum time, the far left point in the distribution curve, can occur, but with very low probability. To provide for interruptions and urgent but unplanned assignments, resources typically elect to provide a time that they are 90 to 95 percent confident they can achieve. In general, if resources deliver on the accepted due date, they receive a good evaluation. If a task is delivered late, their evaluation is diminished, depending on how late a task is delivered. Typically, resources are evaluated based on how well they perform their assignments, independent of other resources working on the same projects.
FIGURE 3-1 Probabilities for a task with a skewed distribution.
In project-based environments, where multiple projects are performed using shared resources, “accurate” task estimates are even more critical for planning achievable schedules. Because making sure that an individual resource is not assigned to work on two tasks at the same time is logistically next to impossible in a multi-project environment (due to task completion uncertainty, where no single point estimate can be correct), only the most sophisticated (project mature) organizations attempt