At a past job I learned my lesson the hard way to not use the “G” word when referring to project estimates. What’s the “G” word?: Guess.
But the reality is that making an estimate for the task is still just a guess, and sometimes it’ s a lucky one and sometimes not. If it’s done with proper methods it will probably be luckier than if it’s done without.
In a recent blog post, Kailash Awati details six ways in which project estimates can go wrong, but he is clear in saying that this is only six of many. Among the six are false analogies; making the estimate on historical data that does not exactly match what the current project conditions are. How can we ensure that we don’t make this mistake? We can start by keeping accurate records of how conditions in the project changed and what led them to change, by keeping an estimation history for each task and comments to show what made the estimate change as the project progressed.
Another reason that Kailash lists is false precision, stating that “an estimate quoted as a single number is almost guaranteed to be incorrect.” How can we avoid this problem? We can by using a project management tool that is designed to accept a ranged estimate and will calculate an appropriate project schedule based on this information. All it takes is one time sitting with a programmer and asking her to think through the scenario of “what’s the longest this task will take, if everything that can go wrong, does.” to make her realize that most of the time when she gives a single point estimate she is really only thinking about the best possible scenario and rarely thinking about all of the things that could potentially delay or extend the work.
Another common mistake that Kailish mentions is getting the estimate from someone other than the person doing the work. If you are off somewhere in your own Microsoft Project file land and are not on a shared system that your team members can view and collaborate with you on, its easy to just guess some of those estimates by yourself. Better to use a tool that everyone can easily access and update, where your task owners can not only make their own estimates, but update them as they go along.
Again, these are only three of many reasons why project estimates fail, but if we can arm ourselves with the proper tools we’ll have a chance at getting a little bit closer to an excellent guess.