Issues with novel IT projects

As soon as an IT project has the characteristics of moving into new unexplored fields, the uncertainties increase. Now why is it that these big projects so often fail or simply end in time and cost overrun? One of the reasons could be related to the relationship between learning and ability to change design over time.

With more and more project managers and especially sales people avoiding the waterfall approach; it demands more and more from the developers to make the underlying framework as compatible as possible. This is due to the opposite functions of ability to change the design and the acquired learning through the project. When frameworks like SCRUM is used where things can be changed over time when learning takes place; it become the situation in the graph below. The graph clearly shows the paradox of learning over time and being able to change the project.

What happens is that over time learning takes place and more and more knowledge is available on what the projects is suppose to be like. However, while the project is on, it has to move forward and decisions have to be made and specifications must be “freezed”. The underlying platform of the project must therefore be made in a way that it will be compatible will different kinds of designs and demands later in the process. One way to do this is with the use of the modularity concept. The modularity concept is based on the framework where each component has a standard interface, which enables one to substitute parts when something better is available.

Ron Sanchez, Professor from IMD Lausanne and Copenhagen Business School talks about the benefits of modularity in and interview here: http://www.connected.org/media/modular.html The article is more based on products rather than IT projects, but Ron Sanchez claims that it can be transferred to any type of project.

Since the scope and the WBS (Work Breakdown Structure) must be done in the beginning it is important that it is kept in mind that learning through the project might force the project to be changed. When doing a project with a brand new scope that no one has done before it is therefore important to keep this in mind. Time must be put aside to evaluate the learning, to ensure that learning are not lost while trying to force the initial platform all the way to the end.

This therefore concludes that a rigid Stage-Gate process with out feedback loops will be less attractive compared to a process where learnings are used to optimize the project.