Brace yourself. What follows is sound logic behind using low-side estimates for project plan task scheduling — assuming you also adopt a breakthrough method to deal with project plan uncertainty.
Conventional wisdom tells us to protect a project from uncertainty by “adding” EXTRA TIME into each task estimate. That’s often called “safety.” While seeming perfectly logical, this guidance is fundamentally flawed because we have no way of knowing how much added time each task needs to be properly protected. To use this “added safety” method is to estimate uncertainty for each task, where it’s difficult enough to estimate durations of actual work. What this means is that the estimate is pretty much someone’s guess.
Fortunately, there is a superior way of protecting projects from uncertainty. Rather than protecting each task, let’s accumulate that “safety” and place it at the end of chains of dependent tasks. These are called “buffers.” We’ve now ‘banking’ the necessary project protection in a more strategic location in the project schedule. We’ve placed it in a place where it won’t be easily squandered.
We now have a project schedule that is brilliantly designed to be completed as quickly as possible (taking advantage of good fortune), yet is also protected from “bad surprises” by the buffers. The project end point is accurately shown as occurring over a range of time — which is absolute REALITY. Best and worst case dates are known. And in a future blog I can show how to find the statistical expected value (most likely) date at any time during the project life.
Can the safety still be “wasted”? Not without seeing it happen in the metric. Since tasks are scheduled without safety, people drive towards an optimistic target — which is great! However if they need more time, they can use a portion of the buffer. When that happens, everyone sees the impact in a beautiful project status metric called the buffer chart. For Six Sigma fans, this is effectively a Project Control Chart. (See image above.)
I’ve found this metric to be a thing of beauty. Easily understood and interpreted. As sensitive as the weekly update process. But there’s one more thing. Projects will complete faster. Much faster. Simulations calculate the difference to be about 30% faster. Here’s why:
The amount of safety that’s needed to protect each and every task is MORE than the amount of safety needed to protect chains of dependent tasks (or the project overall). This is a statistical law when summing distributions. Remember that task durations fall within a distribution, therefore have a “variance.” Tasks are also dependent upon each other. The statistical law states that the amount of variance needed to protect a chain of dependent tasks is LESS than the variance needed to protect each task. That means projects scheduled in this manner will complete quicker.
Amazing. Incredible. Mind-boggling.
And worth trying out, no? #GPSChallenge
