An old Spanish saying says “no one learns a shred from another person’s head.” But let’s try it anyway by reading some experiences other people encountered while managing projects.

In the previous section I’ve told you about my own mistakes in managing projects. In this section I will tell you about the main lessons Jeff Sutherland learned while helping companies like Nokia, Patient Keeper or even Google to adopt Scrum:

In Google, some of the teams adopted Scrum little by little. They did this because they feared resistance from their engineers. They thought that introducing a formal process in teams that were already highly productive would not speed them up but slow them down. They started implementing Scrum using only the most basic techniques. If Google can do it, why can’t we?

In Google, they also had a certain resistance to meeting daily. They thought it would be time spent in vain. They started with meetings that lasted a good deal longer than the initial 15 minutes. At first each engineer took quite a long time to explain what he or she was doing and the problems being faced. After a while, when all the members had explained their job and were aware of the job of the rest of the teammates, the meetings were much more fluid and helpful. To my relief, in their simplified, initial Scrum, they didn’t use a burndown chart for every iteration. They used only one per delivery. That made me feel better. I hadn’t been able to keep the Kanban board updated daily for each sprint (yes, I know, Scrumbutophobia).

The burndown chart helped them realise how long they would spend in developing a feature. For a feature that they initially estimated at 40 points and 3 weeks, they saw that in the first week they had covered only 8 points. In the second week they managed to finish only 7.5 points. But one of the managers thought that they had the feature ready in the third week. Then the graph made it clear that it wouldn’t be ready in that amount of time.

Jeff Sutherland recalls a project for the health sector which would be accessed online simultaneously by hundreds of doctors and other users. In that project, the definition of Done became more and more stringent. At first, they just had to pass the unit tests and acceptance tests. But later on, for something to be really marked as Done or Finished, it had to meet this condition: “deploy to production, and if the phone doesn’t ring for 1 hour, the new delivery is complete.”