Practical Software Metrics for Project Management and Process Improvement

What I particularly like about this book includes:

(1) Complete view of metrics that matter, and the chronicle of how these metrics evolved in a large company (Hewlett-Packard).
(2) Recognition that any software metrics initiative extends beyond the project that delivers the software - Grady examines post-production metrics and ties them back to not only the development life cycle, but the product life cycle as well. Ten years after this book was published there are still large organizations that are struggling with doing this, yet Grady's book provides a clear roadmap to achieving this elusive goal.
(3) Continuous improvement is the central theme in this book. Grady does not stop with collecting and analyzing metrics, but how to effectively employ them to spot improvement opportunities and develop a strategy to effect those improvements.

The book is written as both a story of how a successful metrics program evolved, complete with anecdotes that will prove helpful, and as a collection of data that illustrates what is and is not important to a comprehensive metrics program.

Among all of Grady's books I like this one the best; however, I recommend that his other two also be carefully read if software process improvement is your goal. He has much to say and backs it up with data and a chronicle of his experiences from real projects.

The first part starts with a collection of practical rules of thumb for software managers. This collection of heuristics covers every phase of the development life cycle and are backed up with data gathered during 125 software projects at Hewlett-Packard. An example of one of these rules of thumb is that you will find 1 defect after software has been released into production for every 10 defects caught during testing. This, of course, is purely empirical, but is an interesting rule that I mentally filed away. Some highlights of the first part are: a good introduction to the goal-question-metric approach to determining what to measure based on your objectives, and a focus on project goals of maximizing customer satisfaction while minimizing project schedule and costs, and product defects. This is followed by chapters that address each of these goals. One of the best chapters in the first part of this book is work analysis. While I am more focused on the service delivery side of metrics (after the project has produced something that has been released into production), some of the metrics were very valuable to me - especially the ones that revolved around testing and QA.

Part 2 is squarely in my domain - production and application support, and service delivery. The best chapter, Dissecting Software Failures, was one of the most insightful descriptions of the defect life cycle I have ever read. It fully addresses defect data collection and analysis, and how to use this data to effect process and product improvement. Even better is the chapter on investing in process improvement. Here Mr. Grady gives a workable approach to using the defect data to developing a business case for process improvement. He guides you through developing a plan, selecting from among an array of solutions, and case studies.

This book is a quick read. It's main value lies in the many tables and facts provided on nearly every page. I use it as a desk reference, especially the appendices that summarize defect origins, types and modes, and metrics definitions. It spans both project and production metrics, and is as valuable to project managers as it is to application support professionals.