Note: This was originally posted on 10/02/2008.
“At regular intervals, the team reflects on how to become more effective, then tunes and adjusts its behavior accordingly.”
Some of you recognize this as the last item principle in the list of principles behind the Agile Manifesto. There are whole books written on the subject.
If you want to really, really know all there is to know about retrospectives, you can go and buy the book. My intent here is to provide a simple overview. I hope you find it useful. (If you do, you can let me know.)
Just to make sure we’re all on the same page here, this post is about Scrum retrospectives. In the classic book, Agile Project Management with Scrum, Ken Schwaber states very simply “The purpose of the sprint retrospective is to inspect how the Scrum process worked during the last sprint and adjust it to improve the next sprint.” Wow, that’s pretty simple. Not a lot of guidance here. It’s not surprising to me that people get so confused about it. I see questions about retrospectives pop up on the Yahoo! ScrumDevelopment board quite often, and there are whole sessions about retrospectives at the large agile conferences, so don’t feel too bad if you’re here trying to figure this out.
Retrospectives aren’t something that we normally do in traditional software development. Sure, we might have a “post mortem” at the very end of a project, where we talk about all the things that went wrong, and vow to never do them again, and a project manager will take notes and maybe even send them out to everyone, but in reality, they get filed
away and forgotten, and we repeat the same mistakes over again on the next project.
Retrospectives don’t have to be difficult, but it is important to keep some things in mind. Scrum is simply a framework for developing software in an agile manner, and it provides some fairly easy rules and guidelines that are based on the principles behind the agile movement. Anytime we do something because Scrum says we should, it would help us to keep the corresponding Agile Manifesto principle in mind. The corresponding principle to the Scrum retrospective is “At regular intervals, the team reflects on how to become more effective, then tunes and adjusts its behavior accordingly.”
So, given that statement, let’s take a closer look at the individual parts of the principle and then go ahead and get the easy one out of the way: The first part is “At regular intervals…”. Since Scrum sprints are normally from one to four weeks, our “regular interval” should be at the end of every sprint, after the sprint review. The retrospective should take place after the sprint review because you will normally have stakeholders, managers, and end users who show up to see what the team has accomplished. These stakeholders will say something about what you are showing them, and you should be paying attention. The feedback you receive should be discussed in your retrospective, whether it is good or bad. If you aren’t responding to feedback early and often, you’re missing one of the key benefits of agile. Retrospectives are a great place to capture real feedback from the people you are working for.
Since we’ve discussed when to have the retrospective, the next question is who should hold the retrospective? Well, according to the principle, the team should be the ones doing the reflecting. Sure, managers, stakeholders, etc can reflect on how the team can be more effective, but the team is required to do this. If someone outside the team has input on how to be more effective, I would encourage the team to listen. Many times, managers see the bigger picture, interact with several other teams, and have great ideas or advice. Since we know it is the team’s responsibility to reflect, a question that sometimes comes up is “Who is the team?”.
Unfortunately, the answer varies from team to team. Ideally, your team consists of the regular team members (developers, testers, designers, etc), but also includes the Scrum Master and Product Owner. Yes, the product owner, but only if they really are a member of the team. By “member of the team”, I mean if they sit with the team, help the team, eat lunch with the team sometimes. Do they have skin in the game and help the team to accomplish the goals, or do they blame the team when things don’t go exactly according to plan? Another way of putting this
is: “Are they a pig or a chicken”? I’ve seen it both ways, but I can promise that I’ve seen it work far better when the Product Owner is a part of the team.
OK, enough about the product owner. Back to “the team”. Who is NOT a part of the team? Managers, stakeholders that happen to be “chickens”, and anyone else that isn’t working together with the team on the project day to day. Retrospectives are a place where transparency is vital. If people are afraid to speak out or speak up, your retrospective won’t be as effective as it could be. This is the place where the team reflects not only on what went right, but what went wrong. Team dynamics, individual weaknesses, team members not doing their best; these are all fair game during a retrospective, and the team should feel free to discuss these without fear of management reprisal. If you don’t think that Ted is pulling his weight, you need to be able to say it (in a thoughtful, kind way) without worrying that a manager will take note. You also want to be able to discuss things that didn’t go well, or could have been done better, and you may not want stakeholders present that wouldn’t understand the day to day details that take place.
Great, we’ve got the when and the who. Now what does the team reflect on? The team should reflect on how to be more effective. Huh? More effective at what? The primary thing to reflect on is “How to satisfy the customer through early and continuous delivery of valuable software”. Come on, that’s your job. Do it well, and look for ways to do it better. Instead of just complaining about outsourcing, find ways to be so
efficient and valuable that you don’t worry about offshoring that much. Sure, you still won’t be completely immune to decisions to outsource you, but the better value you provide, the better chance you have of people up the corporate chain thinking that they don’t need to replace you, especially as more and more companies get bitten by sending their
business know-how out of the country, and find that quality typically suffers.
In addition to looking at how to be more efficient at delivering software, this is a great time for the team to take note of accomplishments, jobs well done, problems solved, and any other good things that happened during the last sprint. If a team member really shows improvement, let them know. If there are things that could make your job better or make the work more fun, bring it up. Sometimes, simple improvements can make a big difference in people’s attitudes or enjoyment of their work, and this is good for everyone. Happy programmers have happier end users, because they are willing to go the extra mile and pay attention to the little things that make big differences.
There are many different ways to reflect on how to be more effective. Different teams use different techniques. What works for one team may not work for another. For the team that I’m on, we’ve settled on this set of questions/statements that we discuss and take notes on during each retrospective:
- What did we like about this sprint or do well? It’s a good idea to get the meeting started in a positive fashion. This is a nice time to pat ourselves on the back and take note accomplishments and work completed.
- Room for improvement/lessons learned:
- What caused us pains this sprint?
- What would we do differently if we started this sprint all over again?
- What are some things that we want to try or do next sprint?
- Concerns (long term, short term, etc)
- Concrete Action Items
The last one is very important. As Bob Schatz once said: “Retrospectives that don’t result in change are sterile and frustrating.” This is where the rubber meets the road, and a good way to implement the part of the principle that says “….then tunes and adjusts its behavior accordingly”. Take a look through the list of things that caused pain, or things that you would like to do differently if you could. Also look closely at the “Things we want to try or do next sprint”. This may be as simple as “show our tester how to do make css changes”, or try pair programming”. If the team really means it, this is their chance. Don’t put eleven different things on your concrete action items list. Pick the top two or three, maybe four if they aren’t real big. Chances are, some of the smaller easier ones will happen over time without you really thinking about them.
Be sure to post these action items, or somehow make sure they really do happen. ( Scrum Masters, pay attention here.) Whatever you do, make sure the team tunes and adjusts their behavior accordingly.
Here’s another retrospective idea that I came across one day. I’d like to give credit for it where it is due, but I can’t remember where I saw it. It’s very simple. Basically, just draw a big chart on a whiteboard that looks like this:
|Liked:||Didn’t Like:||Do Next Sprint|
|-Ted brought donuts||
-Lots of bugs found really late
-Waited too long to test
-Take turns bringing donuts
-Deploy QA drop at end of first week.
The important thing is that the team always tries to be more efficient and better. Don’t become complacent, or settle for “good enough”. As you may have heard before: “Good enough is the enemy of great”.