Modeling the Rate

The rate at which software is developed also needs more precision than “Projects per Year.” Whatever your measurement of “size,” you need some idea of progress over time. If you’re using a simple one-dimensional sizing model, such as the count of user screens or User Stories, then you can use the average time per screen or story as your rate, assuming the effort applied remains constant.

Advantage of Functional Slices

One of the advantages of decomposing the project into functional slices (see ​Which Way to Slice?​) is that you can test whether that slice works or not. This gives you a measurable indicator of progress that you can check at intervals to calculate your rate of progress. Whether you measure your progress by Story Points per Sprint, as many Scrum teams do, or Count of Stories per Iteration or per unit of calendar time, as I prefer, you’ve got a reliable if imprecise indicator of your development rate. This works well as long as you maintain the same team working in the same fashion.

Some organizations try to optimize employee utilization by swapping developers in and out of the team based on perceived need for their particular skills, or relative priorities of different projects in progress at the same time. This throws a monkey wrench into the presumption of a constant rate of accomplishing work. If you do this, you now need a parameter in your model for how many people are working on the project.

There are many confounding factors in modeling a changing makeup of the development team. Programmers are not fungible and interchangeable. Changing the number of programmers, or even replacing one with another, is not an easily calculated effect. Consider also that a well-functioning team is more than the sum of its members. Every time you change the membership of the team, it has to revisit the process of team formation to some degree. Of course, if you reshuffle the team frequently enough, you’ll inhibit the team formation and the rate will become more predictable again, although slower than you might otherwise achieve.