I have always tracked the remaining effort to complete an engineering task in ideal hours rather than ideal days. However, the current project I’m working on is using real days as the unit. The switch back to real time from ideal time is consistent with Kent Beck’s latest approach to estimation. Regardless of whether ideal or real time is used, I prefer tracking the remaining effort in hours and not days.
My development style is test-driven but I have a ‘track’ action at the end of my test-code-refactor cycle. Within a disaggregated user story, I approach each engineering task by formulating a number of code-and-test goals that build on one another to evolve the task to completion. When I attain each goal I record my effort and re-assess the remaining effort to get the task completed in my tracking data (I use V1:Scrum). My goals are small enough that, on average, they take me between 30 minutes and 2 hours to achieve. Therefore, tracking this frequently and in hours generates more accurate data from which to draw my future empirical estimates.
The engineering tasks and associated tracking data are private to the team and are not part of management reporting. I report progress to management in story points.
