Optimum cycle length

What is the optimum cycle length? This puzzled me a long time, until I started doing Evolutionary Delivery (Evo) projects myself.

• Before knowing about the Evo principles, I arrived, from an other point of view, at an optimum cycle length of two weeks. See "Why short cycles".
• When I met Tom Gilb, he was talking about cutting a 50 week project into 50 increments of one week! Based on my earlier view on short cycles, I immediately accepted the Evo way. However, only one week per cycle seemed almost impossible short and too theoretical to me. What can you do in only one week?
• When coaching my first Evo project in 2000, I assumed that nobody would believe that one week cycles would be practical or even possible and that it hence would be impossible to let them start using the Evo way. Even 2 weeks seemed unachievable to me, so I suggested a 3 week cycle.
• The 3 weeks turned out, however, to be too long! Human beings need shorter times to systematize their work.
• In another project, we had only 6 weeks to finish a development. Working in only two 3-week cycles or even three 2-week cycles was hardly useful. So we decided to try 1-week cycles, just as Tom Gilb suggested. And guess what? It worked great!
  The first week, none of the tasks committed were finished. The estimations were way too optimistic. I just showed them a mirror at the end of the first week. They realised that their estimations were impossible: only some 40% was done. I was almost desperate, but kept going because I knew the theory had to work. The second week, none of the tasks were really finished, however, they were almost (80%) done. And from the third week on, all tasks were completely finished at the end of every cycle. People started smiling. They started believing that they could achieve things without bad stress. Motivation returned. Productivity returned. Typical defects vanished. All in a few weeks! Can you imagine?
• In every cycle, we plan 2/3 of the time and we leave the remaining 1/3 for all other things that we have to do in the project, like going to the bathroom, drinking coffee, helping each other, small meetings, planning, email, phone-calls...
  Somebody complained that if you plan only about 3 days (2/3 of the time), and give people 5 days to do it (see "Don't they get lazy?"), it will take 5 days (which actually is true). A lot of managers translate this idea into requiring doing 6-days work in 5 days, to make people "more productive". Reality shows that this does not work. People are not lazy. Stress demotivates, and causes defects that cause even more delays. The satisfaction of getting results motivates, and motivation is the motor of productivity. So, don't be afraid to insist on doing '3 days work' in one week. Don't fool yourself into thinking that you can do 5 days of work within a week, you won't! Most important is that the planned work is done, really done.

Bottom line: Cut your development time in 1-week cycles. At first, it seems impossible. After some excercise, it turns out to be always possible. If you don't believe it, try once more. Most project managers hearing this story tell me that it sounds nice, but that it is impossible in their project. It is always possible. Ask a coach. Ask me. Within a few hours, you will know how you can start the first 1-week cycle! Thanks Tom, you were right!