A retrospective is a tool in Scrum, to make the team look at how the process can be improved.
This is done by repeatedly looking at how we do things today and see what we can change. By doing this on a scheduled time after the sprint, the team is not distracted and can relax and focus entirely on the improvement.
I have been to a lot of retrospectives that didn’t really provide any use to the team. Sure, it’s nice to get together, especially when most of us work remote. But that’s not why we hold a retrospective. I’ve noticed that so many really doesn’t know why we do them.
Most of us have learned that we should have some sort of structure, usually in the form of three columns on the whiteboard named something like good, bad and investigate. And yes, there are many variations to this. But many of us doesn’t know how to use them. So we fill the good column with “good collaboration”, “fun sprint” and alike. But, how does this help us?
Scrum is a process for constant improvement. This is what gives it the label “agile”. We constantly change how we do things. And in Scrum, the activity to make this happen, because it doesn’t just happen by itself, is the retrospective. This is where we can find out how to improve our ways of working. We want to be faster, better, more accurate, have more fun and so on.
The tool we use is the whiteboard and these three columns, and this is how you are supposed to use them.
This is where we write down things that we think worked well in our process, and that we think we could do more of.
We automated tests. Now we don’t have to test manually any longer, which frees up time.
What more can we automate?
The opposite of good is things that we really don’t want to do. Things that hinder us, that are boring to do or that is prone to errors. We really should try to find other ways to do it.
I had to rewrite three modules just to get this feature through.
How can we isolate the functionality so I only have to rewrite one?
We get new ideas, things that we haven’t tried before, that we don’t know if they will be good or bad. But it could be worth to try.
I have an idea. Is there any benefits in releasing the software as soon as we check in, instead of once a month? Could we maybe get earlier feedback from our customers if we do this?
It doesn’t have to be revolutionary and earth shattering things. It could be just a small detail that improves just a tiny bit of the process. But if we relentlessly do this over and over again, things will rapidly be better. Like interest on interest. If we compare the process now with how we did things a year ago, we will see a profound improvement. And that is the reason for a retrospective!