| Estimating the probabilities of success for your | | | | you will have in your overall project-value ranking in |
| projects is necessary for calculating the expected | | | | your project portfolios. |
| value of a project and is an essential part of project | | | | Developing more accurate project risk estimates |
| portfolio management (PPM). Unfortunately, most | | | | requires 4 basic activities: |
| project managers and project management offices | | | | 1) Identifying the key drivers of cost, time, and |
| (PMO) don't do this very well. They could learn to do | | | | resource risks in completing project tasks. |
| it better by looking at how meteorologists predict | | | | 2) Preparing a database of these tasks that includes |
| the probability of rain. | | | | the corresponding cost, time, and resource estimates |
| So, just what does it mean when a meteorologist | | | | assigned to each project and the basis for those |
| says "the chance of rain today is 60%?" | | | | estimates at the beginning of the project. |
| Each day in the United States, a massive amount of | | | | 3) Tracking the actual costs, times, and resources |
| data is collected from weather stations, satellites, and | | | | used performing the task as each task is completed. |
| weather balloons from around the world and sent to | | | | 4) Comparing the actual costs, times, and resources |
| the National Meteorological Center near Washington, | | | | with the starting estimates. |
| D.C. The data is processed to give a multi-dimensional | | | | After you have maintained this database for your |
| picture of global atmospheric conditions, and then it is | | | | project portfolio for a period of time, you will be able |
| analyzed using various algorithms to develop local | | | | to plot the actual versus the predicted results. This |
| weather forecasts and predictions. | | | | plot will show you the accuracy of your cost, time, |
| But this isn't how they make the "percent chance of | | | | and resource estimates as well as revealing the |
| precipitation" predictions. Even with the massive | | | | distribution of the actual results. (You will probably |
| amount of data and super computer speed, their | | | | learn that your cost estimates were too low, your |
| predictive algorithms alone just aren't good enough. | | | | time estimates were too short, and your resource |
| So they use comparisons to historical data. | | | | estimates were for too few. And that is a good |
| Basically, they take the current atmospheric | | | | thing to learn.) Eventually, you will be able to use the |
| conditions and compare them with days in the past | | | | actual results data as a basis for future probability |
| that had very similar conditions. So when they say | | | | predictions because patterns will emerge. The data |
| that "the chance of rain today is 60%," it means that | | | | will also give you an understanding the uncertainty in |
| it rained on 60% of the days in the comparison set. | | | | those estimates. |
| And guess what? Assuming the data was entered | | | | I saw the data of one major pharmaceutical |
| properly, these predictions are 100% reliable all the | | | | company who did this for their project "percent |
| time. Why? Because they are only predictions of | | | | probability of success" estimates over a number of |
| probability - they aren't "wrong" on a particular day, | | | | years. The data between 20 and 85% was |
| whether it rains or not. But whether they are | | | | surprisingly linear; for example, about 50% of the |
| accurate or not in the long term is an entirely | | | | projects that had "percent probability of success |
| different question. | | | | estimates" of 50% were ultimately successful. It also |
| The only way to determine if the predictions are | | | | showed that all projects that had an estimated |
| accurate is to collect the data and plot the actual | | | | "percent probability of success" of 85% or greater |
| versus the predicted conditions over time to learn | | | | succeeded and all that had an estimate of 20% or |
| the margin of error. If it only rained on 30% of the | | | | less failed. |
| days that the prediction was 60%, then there is a | | | | If you're involved in project portfolio management |
| problem with the data or the data processing. | | | | and you're looking for ways to improve your project |
| You can do the same type of probability prediction | | | | planning, compiling and analyzing your historical data is |
| and testing with your business projects, too. The | | | | a great way to test and improve your future |
| more accurate your estimates, the more confidence | | | | estimates. |