There is much written about estimating.  There is much to study.  In some disciplines there are even computer tools to aid in the process.  As far I can tell every methodology has a prescription for estimating cost.  There are generally two parts; the accounting part and the lessons learned part.  The accounting part is simply adding up how much of each resource is required and how much that resource will cost.  The lessons learned process is to check actual costs against estimated and examine the mistakes.  This sounds reasonable.  But in practice, what really happens is, "We were 20% low in our estimate last time, so next time we will add 20%." Is that really sensible? It is reasonable only if the next project is just like the last one including the part where the estimator makes a mistake.  But, the purpose of lessons learned is to prevent the mistake.  Therefore, the next time we do not need to add 20%. 

It does seem that underestimating is extremely common and over estimating is rare.  Is this really so, or is it just a perception re-enforced because we hear about underestimating mistakes and we do not hear about over estimating mistakes? My simulations lead me to support the common perception.  We are more likely to underestimate than over estimate with one caveat.  If we are truly just guessing, then we are just as likely to be over as under.  But we do not generally just guess.  We know something about what will be done and we can estimate it.  Our estimate of what we know will tend to be as likely over as under, once we become very skilled.  But, there will always be the potential of something we do not know.  Since we cannot know what we do not know, we cannot estimate it.

The simulations also give a very simple and, in retrospect, obvious reason.  I call it the discovery dynamic.  Every effort to reach some objective will be a path we must travel with some things very well known, and some things to be discovered.  Some tasks travel well known paths.  Some take a less traveled road. 

I have a friend who is a project manager with the California Department of Transportation.  He once commented to me, "You think that with more than 2000 years of experience building roads going back to the Romans, we would know enough about estimating the cost of a new road that we would get it right.  Yet it seems there is always something that we did not expect that either causes a delay or increased costs." There is always the potential of something that we did not know; something we will discover.  It is, however, only a possibility, therefore sometimes nothing new occurs.  One fundamental thing about what we don't know is that we don't know it.  We simply do not know what we do not know.  It may seem that we could assign a probability to discovering something new.  Later when I discuss the theoretical basis for modeling, simulating, and forecasting, I will explore this problem in more detail.  It is like predicting the size of the next earthquake.  We don't know how big the next one will be.  But, it would be foolish not to plan for it.

When estimating the cost of your design calculate the cost for that which you know and then add something as a contingency.  Keep the two separate and clear in your mind.  Afterwards take feedback on the accuracy of that part of your estimate for what you believed you knew.   Make corrections as appropriate.  Then after many design efforts take feedback on your contingency planning.  Contingency planning is a statistical factor and you cannot take it into account on a one at a time basis.  Remember that you cannot simply take the average amount spent on contingency during several projects and use it in the future.  That would be like saying the average size of earthquakes in an area for 5 years has been 3.45; therefore the average size for the next 5 years will be 3.45.  This is not true.  You must study this deeply.  It is not a simple matter.  I will discuss it in Forecasting.  There are some things you can forecast and some things you cannot.  You must know the difference.

When including a contingency component in the cost of your project you are not providing an estimate, you are negotiating an agreement that the cost versus benefit trade-off of wants will not be re-evaluated as long as the contingency budget is not exceeded.  You must appreciate why this is true. 

Now suppose you are coming to the end of your task and you realize you will not spend the entire contingency fund.  What do you do? I am not going to answer this question, but you should.  If you are to be an extraordinary contributor to an optimally achieving organization you need to be able to answer such questions.  And it needs to be an exceedingly good answer.  Consider this practice or a challenge.  If you find it easy to answer, think about it again. 

ç  Prior Page of Text     Next Page of Text è
(C) 2005-2014 Wayne M. Angel.  All rights reserved.