IT Employment

I believe that maintenance cost, normalized to software size, is the best and most easily obtained metric of software quality. Number of FTEs (full time equivalents) needed to maintain 10,000 Function Points will tell you which of your applications are eating up the most maintenance resources. From there you can do the analysis to find out why. An old legacy system that barely runs at all? Bad relations with an end user community that shotguns you with silly change requests? A mission-critical system that simply must be held to a higher quality standard? These are possible legitimate explanations, and some suggest their own solutions, in particular a cost-benefit analysis for reengineering that creaky legacy system and freeing up a dozen despondent maintenance programmers to return to development and get their lives back. But you may also find that your software development process suboptimizes delivery rates at the expense of defect rates -- or that you don't really have much of a process at all. (Can your people spell "QA"?) America's short-sighted fixation with productivity shortchanges quality every time. A hint: don't try to grow your own Function Point counters, that takes two to three years and isn't worth the investment. Check the IFPUG website for a good deal on best-in-class contractors. You should be able to get a good handle on your software portfolio's quality within a few months.