My thanks to Colm Baldwin of the Innovation Value Institute for his blog post bringing my attention to the September 2011 Harvard Business Review article by Professors Flyvbjerg and Budzier on the above topic. The key point they make is that while the average overrun of IT project costs in their study was 27 percent, one in six of the projects were "black swans," with cost overruns averaging 200 percent! As Colm highlights, some of the recommendations from the article to the reduce the chances of being a "one in six" are:
- Using new forecasting techniques, such as reference class forecasting, to test your cost and schedule estimates and really find out what you are possibly getting into!
- Breaking the project down into limited-sized chunks (using size, complexity and duration as a guide).
- Making proper, tested and workable contingency plans to deal with risks.
- Resisting change to the original scope and sticking to a single target.
The 2nd and 4th points here help to make the case for an Agile SCRUM-based approach.
What are your thoughts?