It has been twenty years in the making. The Standish Group found that common measurements such as thousand lines of code (KLOCS) and function points (FPS) were neither consistent across organizations nor fully descriptive to form true normalization for effective cross organization or project comparatives. In addition, sometime there is no code. For example, packages are purchased off-the-shelf. Other times there is a mix of components and development. Yet, these projects are included in the CHAOS database and need to be measured in a consistent fashion to be able to truly gauge their effectiveness. Therefore The Standish Group has created a new normalization metric called Stanmets. Stanmets is shorthand for Standish Metrics.
Stanmets uses, when available, FPS, KLOCS, requirement sets, functions, complexity, money, number of users, skill levels, and work efforts. In assigning the Stanmets rating a Standish adjudicator also considers the number and skills of executive sponsors and stakeholders. Each project is assigned Stanmets points based on available data and reference projects. Stanmets points allow us to measure all and/or any projects against each other. The abbreviation for a Stanmets is SNS. Our current cost estimate for one SNS is $100. Therefore a project that cost one million dollars in labor should have10,000 SNS in it. We can also consider the average costs, estimated cost, and real cost across various methods, industries, and project styles.
For example, we estimated that healthcare.gov to be a 300,000 SNS project. This means that this project has average cost of $30 million. The fact that the initial bid was close to $100 million and actual cost was over $600 million is of grave concern. The logical question is what and where was the extra $570 million cost. This of course does not count any of the hidden costs such congressional oversight, President Obama and his staff’s time and expense. The loss of faith in the program, the political and real capital that surrounded this challenged project could easily be in the billions of dollars.
Now healthcare.gov is an extreme example. More common examples show a wide discrepancy in projects with similar SNS ratings based on various methods, industries, and project styles. Some of these variations include the hidden and non-budgeted costs. We are currently working on the July research paper with the working title: MoneyCall: The True Cost of a Project. While we may not have the full 50,000 projects coded in the CHAOS database by July, we already have a significant representative sample to be able to make some reasonable conclusions.