The Little Pie Chart that Could
At the XP 2002 conference in Sardinia I presented a little pie chart on the percentage of features and functions that were actually used. It was a last-minute addition to a presentation that was mostly about the CHAOS Research on project delivery. At the time, I thought the data was mildly interesting and was looking forward to gaining some feedback from the group. The research on features and functions started in 1996 with a set of six focus groups with 60 IT executives centered on the topic of Y2K. One of the questions was on the features and functions used by various types of applications, such as internal development, package applications, and PC user applications. In the fall of that year it was also a subject of workgroups with another 60 IT executives at the Standish CHAOS University.
In a parallel project, we were creating a total cost of ownership (TCO) model. There were about 25 organizations that participated in a TCO study from 1999 to 2001 that developed mission-critical applications. We were looking at the cost to maintain these applications as part of the TCO model. The aim was to come up with a general cost based on the size and complexity of the applications. We would later convert that cost into operational metrics by types of applications. During that TCO process we created a spreadsheet for each of the major functions of the application. One of the many columns on the spreadsheet was the frequency of use. Some organizations had only one mission-critical application and some had several. We ended up using 100 of the most complete samples. So, the pie chart created was based on 100 mission-critical applications.
We have many different types of engagements such as TCO studies, project optimization, modernization, single project assessment, project postmortems, and other research cases that cause us to look at this issue. We also have data from workshops and focus groups that from time to time look at this topic. Based on these casual observations, our current estimate of features used for mission-critical applications is 20% often, 30% infrequently, and half hardly ever. I am not sure what the total observation count would be, but it is in the thousands since the little pie chart was presented in Sardinia.
We were surprised that this little pie chart gained so much attention and controversy. It was never a formal study. It is one of the many questions we ask in the course of our research. We did create a tool based on this area, called OptiMix, that helps organizations focus on the value of features and functions so they do not create scope creep or develop more features than they need or with low value. Agile and Scrum are good methods to help focus on high-value features and functions. We are also very big fans of creating the minimal viable product.