Monday 11 April 2016

Why Is It That Pretty Much All Projects Overrun?



Why do projects nearly always take longer than scheduled? You know the score - you've seen it many times before: some large firm wins the tender to undertake mass refurbishment on a large government building, or to build a new distributor road stretching several miles, and inevitably the project overruns and the initial budget put aside proves inadequate to the completion of the job.

But why does this happen so frequently? The reason is quite straightforward; it is because the estimated time is a predictable underestimation - a phenomenon that greatly increases with the complexity of a job. And time is, of course, money.

Writing a blog post is quite straightforward, so if I set myself 20 minutes to write it (or less with short blogs like this one) it isn't difficult to stick to the deadline. But projects like big building projects, even when an up-front fee has been agreed and the incentive is to complete it as quickly as possible, are highly likely to overrun - quite simply because our highly consistent underestimation of time at each stage of a project is going to be multiplied with every further stages of the process. It is the individual probabilities of stages multiplied that is the key to the explanation.

This is what is commonly known as the "planning fallacy" - whereby for every step in a project, there is around a 50% chance of that stage being completed over the estimation time. Once you plug in the numbers, that means that for a project of two steps there's a 50% chance multiplied by a 50% chance, which means for two steps the chance of overrunning is now 75%. For 3 steps it's 50% * 50% * 50%, which means there is an 87.5% chance that out of a 3 step program it is going to overrun. Consequently, then, a project in which each constituent step has only a 50% chance of not overrunning, means that for every step the probability increases by x 1/2, which means that projects with multiple steps are nigh-on certain to overrun.

No comments:

Post a Comment

/>