Retrospectives are good. We all agree on that. Scrum emphasizes retro as the main source of Continuous learning and improvement. Prime Directive on Retrospectives states:
“Regardless of what we discover, we understand and truly believe that everyone did the best job they could, given what they knew at the time, their skills and abilities, the resources available, and the situation at hand.”
Is it all this good and beautiful? There are multiple books written on the subject so there must be something in it. On the other hand, I have seen also otherwise. Retrospectives (especially in Scrum context) might become just another ritual (or nice practice in a better case). If organization does not understand the values behind Agile manifesto and is not able to have a long learning strategy then we might face “Inspect and Adapt syndrome”. Small things are discussed, sometimes acted upon and even followed up. The whole and dependencies between cause-effect might disappear and thus organizational learning is not boosted. Team feels that they have done great job but hidden meanings and causal diagrams are not very often used. And this is dangerous. With the help of good coaching majority of these risks could be mitigated, but this luxury is seldom available for all teams.
Teams are also good. We all agree on that. Is team centrism then good? Why should teams be protected from management? Some new coaches take this literally and do what ever they can to avoid good dialogue between managers and teams happen. Management is the entity in the enterprise whose purpose is to enable and strengthen continuous learning and upkeep long and mid-term strategies to support that (there are other purposes as well). Retrospective results are even hidden from the management. How it is then possible for the management to fulfill this purpose? Teams should be protected from interruptions that risk their committed goals, not from the management.
Teams often focus on themselves (this is clear in Scrum) and we should trust them. How is then the enterprise taken into account? Basic I&A might not do the trick here. Scrum considers the teams more or less as a black box and it is up to the team to decide how to improve. In large-scale this is just not enough. Transparency and visibility would help but Scrum with pure I&A retrospectives does not help on this either. At minimum company needs to change Scrum and create some basic rules for visibility and enterprise level learning.
PDCA is a successive cycle which starts off small to test potential effects on processes, but then gradually leads to larger and more targeted change.
Plan, Do, Check, Act is one of the models that could and should be taken into account when considering bigger organizations and their usage of learning methods and models. All models are wrong but some of the are useful when applied to right context. This is again something Scrum is not saying but lean thinking is helping us. Individual teams are required to think a bit deeper, use causal diagrams and add more transparency and visibility on their process improvement. This helps teams to take more responsibility on their own process improvement and eventually whole organization is more empowered. This is something all good organizations should be aiming for, it is the basis for long-term organizational learning.
There is an excellent article on similar subject, written by Alan Shalloway. Please check it out.