The more times you do something, the better you get at estimating how long it and similar tasks will take the next time you do them. This suggests that very inexperienced people will typically make bad estimates—which is usually the case.
Edward Russo and Paul Schoemaker, in their book Decision Traps, relate a story about Royal Dutch Shell that provides a nice solution to the problem. Royal Dutch Shell found that its senior geologists were considerably better at analyzing geological surveys to predict where to drill for oil than were its recently graduated geologists. Even the senior geologists have a fairly low "hit rate," but for new graduates the outcomes were much worse.
Royal Dutch Shell started a program in which new graduates were given survey data of areas that had already been drilled. They were then asked to predict the results of drilling in these areas. They were then told what had actually happened. In a very short time, the new graduates were predicting as accurately as the old-timers.
A project should be audited at major milestones, with spreads no greater than three months. Beyond that time, memories are not reliable.
This illustrates a very important point: learning does not take place unless there is feedback on results. If an organization never looks at results and studies the causes for those outcomes, the people involved tend to repeat the same mistakes.
Was this article helpful?