In 2015, the Systems Engineering Research Center (SERC) published the Software Cost Estimation Metrics Manual for Defense Systems. This manual presents the results of research conducted by SERC Researchers Ray Madachy and Brad Clark, who with the help of Barry Boehm, have documented the core metric definitions along with relevant and useful cost estimating relationships for software intensive defense systems. The e-book is available for download at the following link…

What’s great about this manual is that it covers best practices specific to software systems in the areas of measurement, data collection, normalization, analysis, benchmarking, and CER development.  The manual is based on analysis of data collected through the Software Resources Data Report (SRDR) which is required to be submitted for all contractors development software with projected software costs greater than $20M on ACAT I and ACAT 1A programs.  These reports need to be filed with both the estimate at contract award and the actual costs when the program is completed.

Based on analysis of this data the authors are able to provide guidance on ways to collect, assess, stratify and analyze data in ways that purpose the data to support forward looking estimation of projects in the planning stages.  They have taken the data collected to date through the SRDR process and have determined that the best CERs are those that are customized to represent the varying Operating Environments and Application Types for typical defense systems. 

In addition to providing a set of CERs based on the analysis of data, the authors go further by delving into modern estimation challenges and providing insight and methodology to addressing these challenges within a software estimation.  Some of the topics covered include evolutionary Development, Emergen Requirements, Agile, Kanban, Software System Assurance, Net-Centric Systems of Systems, Model Driven Development and Commercial off the Shelf (COTS) or Not Development Internally (NDI) intensive software systems, etc.

If you and your team are involved in the development of software for defense (or related) systems, and are expected to participate in or support the estimation process, this manual should be an important part of your estimation toolkit.  Not only does it present analysis of data collected for actual software programs, but it also presents thoughtful guidance on how to use this analysis as well as how to repeat the analysis with data collected from actual program you and your team have delivered.