This webinar presented as part of the free PGCS 2023 Webinar Series looked at two processes that are 'baked into' standard project management estimating and control to show how recommended good practices are still optimistically biased.
- When preparing an estimate good practice recommends using Monte Carlo to determine an appropriate contingency and the level of risk to accept. However, the typical range distributions used are biased – they ignore the 'long tail'.
- When reporting progress, the estimating bias should be identified and rectified to offer a realistic projection of a project outcome. Standard cost and schedule processes typically fail to adequately deal with this challenge meaning the final time and cost overruns are not predicted until late in the project.
This webinar highlighted at least some of the causes for these problems. Solving the cultural and management issues is for another time.
Download the PDF of the slides, or view the webinar at: https://mosaicprojects.com.au/PMKI-PBK-046.php#Process2
Resource constraints are usual in most projects and so they must be taken into account.
Estimating future performance based solely on the past performance is not reliable. Future can contain new risks, some new data like future changes in resource availability can become known, and a lot more.
I would not rely on any method that looks only to the past to forecast project outcome.
We usually explain our customers that 70% probability to meet project target calculated by Monte Carlo or any other method is actually 50% for several reasons. One of them: we missed something almost certainly.
Resource constraints are usual in most projects and so they must be taken into account.
Estimating future performance based solely on the past performance is not reliable. Future can contain new risks, some new data like future changes in resource availability can become known, and a lot more.
I would not rely on any method that looks only to the past to forecast project outcome.
We usually explain our customers that 70% probability to meet project target calculated by Monte Carlo or any other method is actually 50% for several reasons. One of them: we missed something almost certainly.