We’ve been trying to use four-week timeboxes for about six months now. But we just can’t finish everything we estimate we can do in those four weeks.
At the end of the first iteration, we had a few days more of work, so we extended the iteration to five weeks. The next iteration, we had more work, too, so we extended the iteration to six weeks. We got up to eight-week iterations, and we still weren’t finishing the work we’d planned for the iterations.
We finally decided to divide by two. That is, make the iterations only two weeks, and see whether our estimates were any good. Turns out, our estimates were good, but because we have support ...