If you “scale agile”, doing sprint retrospectives within development teams might not be enough. We at Adobe Hamburg Shared Cloud are 8 development teams with 42 developers in total. These teams need to work together on a product which is a Platform as a Service for other Adobe products. And since sprint retrospectives work great on team level, we also established them as an “all team” activity on a quarterly cadence. Here is how we like to do it:
Set the Stage (15 minutes)
We do this all together, standing in our largest room: either setting a specific topic, taking a stand, doing a quick Improv Theater exercise, or else getting into the mood. Last activity we did was a temperature reading, creating a histogram of how people think we are doing concerning technical debt.
Gather Data (30 – 60 minutes)
This is done in smaller groups. Usually we ask everyone to build groups of 4 with people not from your own team, then do whatever “gather data” activity they like best (probably 4L, Mad Sad Glad or alike), decide upon 3 main findings.
Liberating Structure 1-2-4-all in action
Sometimes the topic is just “how does our collaboration work for you”, sometimes it is much more specific like “what is your main problem with technical debt”. Then we all get back together, and each sub group presents their 3 main findings.
A much faster way to do this is a liberating structure called 1-2-4-all: first you have 2 minutes to think for yourself about the question. Then you find a partner and discuss with him for another 2 minutes, deciding which topic is most important. Then each pair finds another pair, making it 4 people to discuss their findings and agreeing on one thing, which then is written on a large post-it and presented to the whole group (“all”).
Generate Insights (60 minutes)
Again, this part starts in smaller groups. This time groups form around one of the findings from Gather Data they are interested in working on. They first do a deep dive into the reasons for the finding they work on, probably with asking 5 whys or creating a Cause-Effect-Diagram. In this phase, it is very helpful to have quite a few facilitators helping all the small groups in their work, since as in any retrospective, I consider this the most demanding part to do.
When done, the groups are asked to create an action proposal, leading from problem-space into solution-space. This is expected to be a flip chart with a catchy title, some explanation of the insights, and a concise proposal for action.
Now it is time to vote! All small groups quickly present their proposal poster. Each poster has three areas for votes:
Agreement (“Go for it”)
Engagement (“Count me in”)
Veto (“We need to talk about this”)
Everyone is asked to vote on each poster, whether they agree with the proposal, even want to participate in whatever action is proposed, or vote against doing it. The latter two options are only allowed with your name on it, since the poster designers need to know who to talk to. This way, follow up working groups are composed out of all the people who put their names either into “Engagement” or “Veto”. Proposals without any veto can go into action by all engaged people instantly.
Close the retrospective
After such a long retrospective it’s good to have a short closing. Feedback door works great, and if you do the retrospective on a Friday afternoon, some beers also help to keep people discussing the results a bit longer!
How do you get your team to actually behave like a team instead of a bunch of individuals working separately from each other? While most teams we have worked with have a good degree of teamwork, most of the times there is quite some room for improvement. A fun way to make a team feel what it is like to excel in teamwork is a game, which you can play in an extra session, maybe 60 – 90 minutes of team time: “Escape – The Curse of the temple” by Queen Games is a cooperative and messy game which you can leverage to increase collaboration and coordination in your team! Have fun learning about
why taking time for meaningful communication (often disregarded as “meeting overhead”),
planning cooperation by getting an overview of a complex situation together, and
actual teamwork in terms of “doing things together” instead of just following a similar goal
helps to speed up and reach your goals faster (or actually: at all!).
About the game
The game is cooperative, that means: the players need to “escape” from a temple they have to explore while playing, and they can only win if every player escapes from the temple within 10 minutes. Sounds like a quick game, but you’ll be very exhausted afterwards, because – and that’s the messy part – the game is not played in rounds: each player rolls their dice at the same time, over and over again, as often and quickly as they want to! While they discover new “rooms” all the time on their way to the exit, in some rooms the team can activate gems… and they need to, otherwise they won’t be able to get out.
If you as the facilitator don’t know the game, please take time with friends or family to try and learn beforehand, it is important that you know the rules. The games home page offers PDF versions of the rules in many languages.
How to use Escape in Team Learning
Escape can be played with up to 5 players, but you should plan to have 6 to 10 players for your session. The idea is to let half of the team play, and the other half observes how they play, draw a “burn up chart” for rooms explored (the exit is in the last five rooms, so they need to explore a lot) and “burn down graph” for gems activated (without activating almost all gems, they won’t be able to use the exit). It is key to let observers note actual quotes from the players, since game play is very hectic and quick, they’ll most likely not remember exactly what they did or said. In all of my sessions, the first attempt of the game failed – the team died, and that’s good! Not for spirit, but for learning. Most often, lack of clear communication, lack of explicit collaboration and lack of effective re-planning leads to teams dying in the temple. So, after the first game, let the observers give feedback to the players. Take at least 10 minutes to analyze why the team failed, and speak out precise situations like “when Ken was here, he said ‘I need masks’ but no one responded”, more than general “you did not collaborate”.
Then do another round, switching observers with players. Now that everyone knows what to look after and how to behave, the team will have a better chance of winning. Again, take your time to analyze what happened this time.
Extending the rules to learn about team dependencies
If you have time for a third round, you can take it to another level: let both parts of the team play at the same time (you need 2 games then, obviously, and 2 facilitators help a lot, too). We did that in separate rooms because the game play can get quite noisy. We modified the rules, so that collaboration with other teams (the feared dependencies) is modeled in, too.
There are rooms in this game where the team can decide to activate 1, 2, or 3 gems. With two teams playing simultaneously, the new rule is that
the “1 gem” option can only be solved by the team themselves,
the “2 gems” option can only be solved by going to the other team and asking for the required 7 torches (in this case) from players in that team in the same room, and
the “3 gems” option can only be solved by both teams adding up the 10 required torches from both teams.
If a team chooses to ask for help from the other team, and the other team agrees to help, then the other team can only do that with players who are on the same room tile at that time. That does not need to be a “3 gem room”, any room that they are all in is okay.
Added learning with this modification is that team dependencies, while they should be avoided where possible, can lead to better results if handled collaboratively and cooperatively.
Thanks to Malte Sussdorf, who introduced us to using this game in team building at RFG 2016, and to Florian Noeding, who developed the “dependencies” extension with me!
This article is about an activity you can do with a team to find spots for improvement regarding agility. As an extra, during this activity there is a chance to deepen knowledge about the Agile Manifesto. Read it, try it, and please share whether your outcome was worth it!
What you need
The 12 principles of the Agile Manifesto printed out on cards (feel free to use this template: AgilePrinciples.pdf) *
A flip chart or wall, if you can’t stick the cards to a wall just use the floor
About 30 minutes with your team for this activity
How to play
The goal is to have a distinctly ordered list of the agile principles. The most important part is to ask the best question, which we found to be “how big is our demand of improvement regarding this principle?”, because it makes people think about their situation and yields actionable insights.
Start with a random principle, discuss what it means and how big your demand may be, and place it somewhere in the middle.
Pick the next principle, discuss what it means and sort it relatively to the other principles. Use bubble sort or any other algorithm, I tend to propose a position depending on the discussion and move from there by comparison.
Repeat at 2. until all cards are sorted.
Now that all cards are set, consider the card on top: this is the most needed and most urgent principle you should work on, so prepare to get into “generate insights”:
does everyone still agree?
how do you feel about it?
what are the reasons there is the biggest demand for change here?
should you compare to the second or third most important issue again?
if someone would now rather choose the second position, why?
Put up the result up in the team space. This way, everyone can always re-check and start a conversation about them. You can also get it back into another retrospective to see what changed after measures for the first insight were done. As with any retrospective, it is very important to follow up on measures.
In this activity, you sort the principles along one dimension, the question “how big is our demand of improvement regarding this principle”. You could add another dimension “urgency” or “effort”, so that as a result you get a map of principles with most demanded and urgent or easy to fix principles on the top right. But then, how do you estimate effort without even thinking about measures first, and why should urgency be so much different to demand? This modification didn’t help us, but maybe it helps you.
A better modification might be to sort only the top three principles. Compare each picked principle with the current top three, and if it does not yield more demand of improvement, don’t spend your time deciding whether it is on position 10 or 11. You might lose some insights on what people actually think about these principles, probably also letting down on the educational aspect of the activity. Still, having focus on the top principles should be fine.
No team is perfect, we all know that. And it’s nothing to be ashamed of. We all strive for better agility, because we believe that iterative software delivery in close contact with the customer is a better way of building actually valuable software than waterfallishly keeping progress (or lack thereof) secret until the time frame for learning how to do it better is long gone. In agile teams we do retrospectives in regular intervals to identify measures on how to improve our team performance.
It is important to not do the same retrospective over and over again. It is very valuable to try different formats every now and then, to let the team take different view points and identify different impediments. Diversity is your friend! An excellent resource to find new things to do at retrospectives is the Retromat by Corinna Baldauf. It not only helps to be inspired for new activities to try in retrospectives, it even let’s you design whole plans for retrospectives and share these with your colleagues.
This article describes the “Find your focus principle” activity. The goal is to identify a teams weakest spot regarding the principles of the agile manifesto. A sub goal is to make the team aware of these principles. While the 4 values of the agile manifesto are quite present and understood by most of the people I worked with, the 12 principles seem to be a list of things one might consider once in a while. Some people call these principles “the better manifesto” (although being part of the original one), so take your time to consider them.
This activity covers the “gather data” part of a retrospective and shouldn’t take longer than 30 minutes. You can combine it with any activity for setting the stage, generating insights, deciding what to do, and closing the retrospective. If you need ideas on how to do that, either use the Retromat or read the book on Agile Retrospectives by Esther Derby and Diana Larsen.
* We found it useful to put a catch phrases on each card which is easier to read from a distance than the whole text of the principle. It is important to not confuse the tag line with the principle itself, it just makes the card easier to handle (read / talk about). Thanks a lot to the participants of the Retrospective Facilitators Gathering 2016 for feedback on the catch phrases!
No, not people, people aren’t resources, they’re people. I’m talking about our favorite books, websites, and blogs related to agile. Read on and feel free to recommend additional sources in the comments.
These align nicely with Dan Pink’s autonomy (decentralized control), mastery (technical competence), and purpose (clarity of the mission). Thanks for the excellent analysis! I’ll be sharing this.
The funny thing is, I’ve probably watched the RSAnimate video of Dan Pink(@DanielPink – check it out below – well worth the time) more than 100 times since I often show it in training classes, and I’ve read the book (Drive) twice. I’m a big fan of Dan’s podcast (Office Hours) and his other books, and I still didn’t make that link! Thanks to Rob for connecting the dots for me between Turn the Ship Around (experience) and Drive (behavioral economics/neuroscience).
Here’s the Dan Pink Ted talk animated, just in case you haven’t seen it yet!
I recently finished reading former U.S. Navy Submarine Commander David Marquet’s book “Turn the Ship Around”. It is a powerful story of learning what leadership means and the struggles Marquet had putting it into place in his role as commander of the Los Angeles-class fast attack submarine USS Santa Fe (SSN 763). Marquet proposes that Leadership should be defined as:
“Embedding the capacity for greatness in the people and practices of an organization, and decoupling it from the personality of the leader”.
The paradox is that more traditional leadership creates more unthinking followership; less top-down leadership creates more engaged leadership – at every level of an organization.
Commander Marquet with Stephen Covey aboard the Santa Fe
Leadership and productivity guru Stephen Covey took a tour of Marquet’s submarine in 2000, a couple of years into Marquet’s command, and reported that it was the most empowered organization he’d ever experienced, of any type, and wrote more about it in his book “The 8th Habit”.
The hyper-quick summary of Marquet’s approach involves three pillars: Control, Competence, and Clarity. These form the basis for what he calls “Leader-Leader” behavior, as opposed to the much more common “Leader-Follower” culture found in most organizations.
Marquet talks about shifting the psychological ownership of problems and solutions using a simple change in language. I’ll attempt to illustrate the evolution of leadership behavior through a series of conversations:
Traditional leader-follower pattern:
Captain: “Submerge the ship”
Subordinate: “Submerge the ship, aye”
To push Control down in the organization, Marquet began using the following speech pattern:
Captain: “What do you think we should do?”
Subordinate: “I think we should submerge the ship, sir”
Captain: “Then tell me you intend to do that”
Subordinate: “Captain, I intend to submerge the ship”
Captain: “Very well”
Giving control without an assurance of competence could lead to disaster on a nuclear submarine, and so over time, the pattern evolved to include an assurance of technical Competence, becoming:
Subordinate: “Captain, I intend to submerge the ship.”
Captain: “What do you think I’m concerned about?”
Subordinate: “You’re probably concerned about whether it’s safe to do so”
Captain: “Then convince me it’s safe”
Subordinate: “Captain, I intend to submerge the ship. All crew are below decks, the hatches are shut, the ship is rigged for dive, and we’ve checked the bottom depth.”
Captain: “Very Well”
The final evolution of the language added the third pillar – Clarity of mission, becoming:
Subordinate: “Captain, I intend to submerge the ship. All crew are below decks, the hatches are shut, the ship is rigged for dive, and we’ve checked the bottom depth.”
Captain: “Is it the right thing to do?”
Subordinate: “Yes sir, our mission requires that we submerge now in order to (classified reason (-: ) ”
Captain: “Very Well”
The book is highly engaging and I found it to be a fascinating model of leadership extremely well-tuned to the needs of leading complex organizations in the knowledge work era.
So, what does this have to do with Agile?
Empowerment is a core concept of agility, and specifically the scrum framework, but it is something that can be a major challenge to get working well in organizations without decentralized control, insurance of competency, and clarity of mission. Marquet’s approach provides a simple pattern to follow in empowering teams.
Interestingly, empowerment is a term that Marquet dislikes, since it implies that individuals can only be “powerful” once it has been granted by a leader. His claim, and one that I agree with, is that all human beings are naturally powerful, they don’t need to be “empowered”. Rather, leaders simply need to remove cultural norms and processes that are meant to exert control, resulting in people tuning out and becoming disengaged. When the right leadership behaviors are in place, people will naturally bring their whole selves to their jobs. From a lean standpoint, such controls can be viewed as creating waste – people that show up and go through the motions, rather than devoting their creativity and energy to their jobs, and the lean leader’s job is to remove waste from the system.
Empowered Product Owners & Teams
Scrum is fundamentally based on the idea that a Product Owner is the single accountable person for setting the priorities of the team(s). Leaders can ensure that Product Owners have this accountability by using the “Intend To” language.
Product Owner: VP, I intend to move this new feature to the top of the Product Backlog and deprioritize this other feature that was in our original plan. Customer validation tests indicate that the new feature would increase retention of existing users by around 4%, more than any other feature we’ve tested, aligning with our highest priority goal for this quarter of increasing existing subscriber retention rates. The team has done some high level scoping and forecast that this feature would be completed within two sprints, a similar size to the feature that we’ll be cutting.
VP: Very Well
The leader gets what they really want: an assurance that the Product Owner is aware of the business concerns and have done their due diligence to address those concerns. The Product Owner gets what they want – mentoring to understand what business leaders are most concerned about (a great career development aspect of this approach), with the autonomy to meet the business need however they see fit.
Agile Leadership is the missing link for many organizations
Agile has had a major impact on some organization’s capability to balance delighting customers, keeping people engaged at work, and delivering great business results. It has, however, struggled to make an impact in many organizations where Cargo Cult Scrum, “scrum-but” and other half-hearted implementations of agile are the norm. The difference is in the leadership of these organizations. Where agile is seen as the latest trend, something the developers do, or a bandaid to fix some specific annoyance, agile will have a marginal (if sometimes still improved) result. Where agile is viewed as a mindset for both teams and leaders, it can have a profound impact. Marquet’s book provides some simple rules that leaders can apply to start seeing that bigger impact of agile at the organizational level.
Check out the animated talk on this topic by Marquet:
Scrum teams conduct a Sprint Retrospective at the end of every Sprint to find ways to improve the way they work together. It turns out, having really effective retrospectives requires some specific conditions to exist to allow improvement to happen. All too often, the team’s environment doesn’t support them expending effort to make real, substantial improvement. Here’s a simple test for a Scrum Master: “Is my team improving every sprint?” Sounds obvious but it needs to be answered, and answered honestly. If the answer is “no” or the answer is unclear, there are some simple things that can be done to help (well, some are not so simple).
Common Retrospective Patterns
First, I want to share some of the anti-patterns I have experienced with some teams I’ve worked with:
There is already too much work in the sprint to make improvements.
The team rushes into problem solving right away without much discussion leading to weak outcomes.
The same impediments are discussed at every retrospective.
The same activities/questions are used for each and every retrospective; they have become boring and repetitive.
The impediments that are discussed are too big for the team to solve.
The team has stopped having retrospectives.
There are many others, but this is a good, common list of retrospective anti-patterns that I often hear about when I work with teams.
To help alleviate some of these problems, here are 6 actions you can take:
3 actions to allow continuous improvement to happen and
3 actions that help retrospectives work better.
3 Actions to Allow Continuous Improvement to Happen
These 3 are the “not so simple” actions since they require a real change to how work is planned and who plans it. This requires a commitment from leaders to create an environment of continuous improvement.
Allow the team to determine their Sprint capacity.
I know many teams, including their managers, who say that the team determines their own capacity, but when you dig into it, we see that there are other things at work undermining the team’s autonomy in determining their capacity. Many teams are hesitant to push back on the Product Owner (PO) who is asking for a particular story in a particular sprint, even though they don’t believe they can finish it. In fact, any time the team is taking on work that they are not convinced they can get to “done” in a sprint, that is a sign that they are not truly determining their own capacity. The Scrum Master plays a key role in helping the team only agree to the work they can complete and no more.
I often tell teams “you should be happy when Sprint planning is over, convinced you will get everything done by the end of the Sprint.” When I say this, I get puzzled looks or disbelief. If your team isn’t leaving Sprint Planning very excited, convinced they will succeed with their Sprint plan, this is a clear sign that they don’t have control over the amount of work they commit to.
I recently surveyed a large group at Adobe that has many teams, distributed around the world. One question I asked was whether they put all work that the team does in the Product Backlog and the answer was mostly “yes”. I was very encouraged that they were making the team’s work visible. However, the results of the next question were very telling about what was really going on. “Is work added during the Sprint?” The answer was also a resounding “yes”! What was going on? I asked around and discovered that planned work went in the backlog, but many unplanned items were creeping into the Sprint and coming from many directions; sometimes from the PO, sometimes from a manager, and sometimes from the business-side. This is in direct contradiction to how Scrum is supposed to work. We make all the work visible by putting it in the Product Backlog and we let teams focus on their Sprint goal by not interrupting them. This also undermines item #1 above since the team now has no way to truly determine their capacity if they aren’t the ones pulling the work into the Sprint during Sprint Planning.
In a talk I attended some years back by Jeff Sutherland, the co-creator of Scrum, he stated that one of the primary reasons for having uninterrupted periods of work (Sprints) was to stop the thrashing that often goes on within teams and derails their focus and productivity. Teams who can really focus become much more productive, not to mention, happier and more fulfilled in their work. If you want to understand this concept more, check out this TED talk by Mihaly Csikszentmihalyi on Flow: http://www.ted.com/talks/mihaly_csikszentmihalyi_on_flow and also, The Progress Principle by Teresa Amabile and Steven Kramer.
The team has permission to add work to the Sprint to make improvements.
I regularly facilitate retrospectives for teams and when we get to the part where they are deciding what they are going to improve in the next Sprint, they often express concern that there is no time since they have too much work. How do they know how much work they will have in the next Sprint if they haven’t done Sprint Planning yet? This concerns me because in Scrum, self-organizing teams determine what gets into the Sprint Backlog so they should have direct control over that.
Also, if the leadership has not given the team permission to spend time improving, then when will it happen? Later? In the competitive world we live in, there is no “later”. We will always have very urgent and important features to implement. Improvement is a investment you make to go faster later. If you don’t pay it and your competitors do…well, you know what happens.
Here is a simple agreement that teams and leaders should discuss: every Sprint, the team puts one improvement at the top of the Sprint Backlog before any product work is planned. This reinforces its importance and ensures there is time to make it happen.
3 Actions that Help Retrospectives Work Better
I too frequently encounter a level of frustration with retrospectives from teams who regularly do them, but have found them unsatisfying. What I often discover, after discussing how they conduct them, is a lack of the experience and skills required to do them well. For my first 3 years of implementing Scrum, I followed the usual retrospective pattern and asked the following 2 questions:
What went well?
What do we want to improve?
After a while, these retrospectives become mind-numbingly boring. The stubborn keep at it and the disenchanted abandon them altogether. I was one of the stubborn ones 😉 Others learn a few basic activities like sailboat or +/delta and then repeat them, over and over again, every retrospective.
I had the wonderful privilege to learn retrospective facilitation from one of the co-authors of Agile Retrospectives book, Diana Larsen. This class completely changed how I saw retrospectives and when I put into practice the techniques I learned, I saw much better outcomes.
Here are 3 things you can try to improve your retrospectives:
Follow the retrospective pattern.
I won’t go into too much detail about what pattern Diana Larsen and Esther Derby have defined for good retrospectives. You can read it yourself in their book. I will focus on the most important step so that you may understand why the basic “two-question” approach is lacking in effectiveness. Here is the basic pattern they define:
Set the Stage
Decide What To Do
The “two-question” approach mentioned above are two questions that should be asked in the “Generate Insights” portion of the retrospective. Diving right into those two questions skips right past “Set the Stage” and “Gather Data”. What exactly is “Gather Data” then? This is where the team answers the “What?” question. In other words, “What happened in the previous Sprint?” Not “What do we think about what happened?” or “What do we want to do about it?”, just the facts, the events, the good, the bad, and the indifferent. Some examples might be: the build broke in the first week, bug counts went up, Sue got sick, Jayanth and Alexei paired on a really difficult story, or whatever. Just record the things that happened. You should spend the bulk of the time (~40%) on this step. Why so much? In order for groups of people to come to a common understanding of what happened and to make good decisions about what to do next, they need time. Lots of it. Spend the time on this part and the next two stages move more quickly. Which also means that you need to make sure your retrospectives are long enough for this to happen. How long? 75-90 minutes is a minimum for good improvements to emerge. Don’t believe me? Try it and find out for yourself.
Mix things up.
At Diana Larsen’s urging, I went on to become a Human Systems Dynamics Professional with the Human Systems Dynamics Institute (HSD) led by Glenda Eoyang. The organizational change work they do is rooted in the idea of human systems as complex systems. They have learned from complexity science about how attractors work in complex systems. There are a number of attractors in complex systems but one type is called a periodic attractor. Scrum is a periodic attractor because of the iterative nature of Sprints. Every two weeks (or so), we review, we plan, and we work. There is a danger with periodic attractors though. In Coping with Chaos, Glenda Eoyang writes that periodic attractors build resistance. When I read that, my head nearly exploded. Does this mean that holding regular retrospectives can cause resistance to change? Then it dawned on me why the Derby and Larsen book recommends changing the activities for retrospectives on a regular basis; to mix things up a bit. The analogy that came to mind is of someone rubbing the same spot on their arm. After a while, they don’t feel anything. Doing the same thing for each retrospective, over and over again builds resistance in people. They become immune to taking real action and getting to the heart of their problems. They go through the motions and get out of the meeting as quickly as possible. Groups like these often rate retrospectives as their least favorite Scrum meeting.
(Aside: HSD has an activity called “Adaptive Action” which is virtually identical to retrospectives, following the same basic pattern, albeit with different terms.)
A good Scrum Master will seek out training and other resources for retrospective activities to continuously improve their retrospective facilitation. As with all things agile, we continuously seek ways to improve our implementation of agile methods. A good starting place is the Agile Retrospectives book but there are other resources you can draw on such as Retr-O-Mat, Tasty Cupcakes, Retrospective Wiki, and many others. I strongly suggest learning the pattern and the purpose of each step first. The other resources then can fill in when you are designing your retrospectives.
Focus on one improvement at a time.
In fast-paced, complex work, there will always be many problems uncovered, issues to resolve, and improvements to be made. It is the very nature of the work itself. During retrospectives, teams will regularly identify many issues. In fact, the list can be quite long and daunting, making it seem as though nothing will ever improve. A very common pattern with team retrospectives is that they choose a list of actions to take, sometimes 5, or even 10 actions. Sadly, because of time pressure, many of these actions are not taken, often none at all. During the next retrospective, the team finds themselves talking about the same issues, Sprint after Sprint. This can really demoralize a team that is struggling to improve.
The concept of “small batch size” from the Theory of Constraints by E. M. Goldratt teaches us that to increase speed (and here, I am referring to the speed of improvement), small batches move faster through a system and don’t cause the work in progress to become a bottleneck. Small incremental improvements are easier to focus on, as well.
As an experiment, for your next few retrospectives, choose only one improvement. If you want, you can place all the other actions in an “improvement backlog” but only commit to one action for the next Sprint. Most importantly, make it actionable by being very explicit about what exactly the team is committing to try. Don’t commit to things like “the team will write more unit tests” or something as equally fuzzy.
Answer these 5 questions to make a weak retrospective outcome stronger:
What? – identify exactly what experiment you are going to run (e.g. Dave and Jo will pair program on the riskiest story in the next Sprint)
Why? – state the hypothesis of why this experiment can be an improvement (e.g. Pair programming will reduce the risk of error)
Who? – what single individual will ensure that this happens (e.g. Jo will own this action)
When? – what date will this be done by (e.g. Jo will invite the team to a meeting at the end of the first week)
How? – how will the team know that this happened (e.g. at the meeting, Jo will demo all the passing tests and discuss other technical details at the meeting)
Some will recognize that this is very similar to SMART goals. I prefer this set though because they are easier to remember (for me at least) and they make it clear that this is an experiment, i.e. it can fail, too! We learn the most from those experiments.
Choosing only one action increases the chance that it will actually get done and over time, an accumulation of single actions start to add up to a lot of improvement. Choosing only one improvement aligns very well with one of the 5 Scrum values: Focus.
Leaders who make it clear that investing in continuous improvement is extremely important and not optional enable their teams to creatively accelerate their work and increase speed to value. Scrum Masters who invest in improving their retrospective facilitation skills will increase the speed at which the team learns.
Lastly, since you read all the way down to the bottom, here is a funny video from the Netherlands on improving your retrospectives:
So I started thinking about recent conversation I had with someone regarding the discipline with adoption of Scrum within a large group, with 5-7 teams. As we touched on her observations and current condition as experienced by the teams, she pointed out the fact that these teams valued autonomy as such one expression of this was they had freely established their own sprint cadence and were unlikely to give this up. This was in the context of the suggestion that it would be valuable to establish a single sprint cadence across the group, be it mapped out across two or more weeks, as it will remove all the wasteful activities needed to synchronize across these teams.
Steven Johnson, author of How We Got to Now was being interviewed by Jon Stewart. He explained that his book was about the history of inventions and a look back as to how we got to now given an outcome we see today and often take it for granted. He points out, as in the case of clean water, that for most of us is a simple act of filling a glass at the faucet, is build on top of hundreds of inventions that proceeded in history. He then pointed out that we wouldn’t be able to TiVo shows like this had it not been the simple invention of standard time.
It works out that until 1918 every town in America had defined there own time standard, there was no concept of standard time across U.S.A. Would you believe if it wasn’t for the railway system we probably wouldn’t have seen the need for standard time. Even then, it wasn’t until 1883 every railroad had their own time standard, not to mention every town on the line defining its own to complicate the simple matter.
People regularly lament about meetings. I am guilty of it myself. I used to cynically and sarcastically say, after spending time in a long and fruitless meeting, “Meeting rhymes with beating!” Many would often shake their head in agreement. “There are too many meetings” is a common refrain at many companies. Interestingly, I often hear this after groups have adopted Scrum, “Why are there so many meetings in Scrum?” “Isn’t Scrum supposed to be lightweight? Agile?” Let’s quickly review what meetings Scrum defines and then we will see what happens in practice for many teams, especially when starting out with Scrum.
Purpose: Development Team commits to work they plan to get to the Definition of Done by the end of the Sprint (the Team’s Sprint Goal).
Outcome: A solid Team commitment to a Sprint Goal represented by their Sprint Backlog (containing all the Product Backlog Items (PBIs) and, possibly, task breakdown).
Duration: 2 hours per week of sprint, often less, as teams mature and improve.
When: At the start of a Sprint (after the Sprint Review and the Sprint Retrospective).
Who: Development Team, Product Owner should be available to clarify PBIs, Scrum Master facilitates.
Purpose: Daily meeting for the Development Team to inspect and adapt how to best achieve their Sprint Goal.
Outcome: Impediments get surfaced and improvements for the day’s work are agreed upon.
Duration: 15 minutes per day.
When: Every day of the Sprint.
Who: Development Team, others can attend but only listen, Scrum Master facilitates.
Purpose: Product Owner helps the Development Team to understand the work coming in the next Sprint or two. Sometimes this involves writing acceptance criteria, slicing items smaller, sizing, estimating, or anything that helps the team prepare for the next Sprint Planning.
Outcome: Upcoming PBIs are ready for the upcoming Sprint Planning Meeting.
Duration: Varies by team but is often 1-2 hours per week of the Sprint.
When: Varies, but preferably not immediately followed by Sprint Planning. Try to allow a day or two in between these meetings.
Who: Development Team and Product Owner, anyone else who can increase the Team’s understanding, Scrum Master facilitates.
Purpose: The Development Team demonstrates the PBIs they believe have achieved the Definition of Done to the Product Owner and other stakeholders.
Outcome: Feedback for the Development Team on what they built, often resulting in the generation of new PBIs. Also, a broadly understood view of current progress.
Duration: One hour per week of Sprint, but can vary depending on the number of teams demoing and how big the audience is.
When: At the end of the Sprint, before the Sprint Retrospective and the Sprint Planning Meeting.
Who: Development Team, Product Owner, generally, anyone else can attend, especially desired are stakeholders, Scrum Master facilitates.
Purpose: The Development Team gets together to examine how they have worked together during the previous Sprint and what they will try in the next Sprint to improve.
Outcome: One specific experiment the Development Team will put at the top of their next Sprint Backlog.
Duration: One hour per week of Sprint.
When: At the end of the Sprint, after the Sprint Review and before the Sprint Planning Meeting.
Who: Development Team, Scrum Master facilitates.
There are a large number of variations on the above descriptions but I think of these as a good starting point. If I customize the process, I don’t lose sight of the purpose or the “why” of each meeting.
The Meetings of a Typical 2-Week Sprint
Special thanks to ICAgile and Ahmed Sidky for the inspiration for these two graphs.
If you add up all the meeting time vs. time for building the product or service (or doing agile vs being agile), you get: That’s larger than a 6:1 ratio. Looks very lightweight to me. Where does the complaint about too many meetings come from then?
Common Scrum Meeting Anti-Patterns
There are many symptoms of dysfunctional Scrum meetings. Here are just a few.
Symptom: Planning takes too long
When Teams start out, this meeting seems to take forever, wearing the Team down so they agree to any amount of work. My first Sprint Planning meeting dragged on for 3 days!!!
Cause: Poorly refined Product Backlog
A Product Backlog that is not well-refined will not be well-understood by the Team. They will need time to gain enough understanding to make a solid commitment to the Sprint Goal. This refinement ends up taking up most, if not all, of the planning time. Time usually runs out and a rushed Sprint Goal is created. So much work gets packed into the Sprint, during the Sprint there is no time to refine the Backlog for the next Sprint, perpetuating the problem. Sprint Goals in this circumstance are rarely reached.
Cure: Weekly Product Backlog Refinement Meetings
Make a serious commitment to a timeboxed, Product Backlog Refinement meeting every week. Make sure the right people are in the room. Ensure that the goal of this meeting is to be ready for the next Sprint Planning.
Symptom: Lasts longer than 15 minutes
Cause: Poor meeting facilitation and/or weak team agreements
This meeting is truly for the Team to synchronize and coordinate their efforts for the day so that the Team can be as effective as possible. But, other items leak in like reporting status to a leader, deep-diving on problem solving, staff meeting agendas, etc.
Instead of trying to describe what needs to happen, let Jeff Sutherland, the co-founder of Scrum, describe the Daily Scrum in a two-minute video:
Notice he mentions the type of agreements Teams can have, e.g. “We can talk about one issue for no more than 60 seconds…” When Teams hold each other accountable to their own agreements can lead to much more effective and dynamic Daily Scrums and eliminate much waste.
Symptom: Team demos unfinished, non-shippable, product increments
The Team shows everything they have completed whether it meets the Definition of Done or not.
Cause: Not focusing on the Agile principle of “working software as the primary measure of progress”
Many traditional project management views about reporting progress are rooted in the idea that you can estimate the percentage of completion of work. Unfortunately, reporting that something is 50% finished lacks information needed to use for decision making. It makes it difficult to answer the “Are we ready to ship?” question. Also, people want to be recognized for the effort spent on the work to date so they hesitate leaving anything out.
Cure: Only demo “potentially release product increments”
By only demoing the PBIs that meet the Definition of Done, it is very clear to everyone in attendance what real progress has been made. A decision to ship is now much easier to make. This gives the organization real business agility. Teams improve from rarely completing anything in their Sprint Goal to completing nearly everything after making this change.
Symptom: Too many improvements to make
When presented with the real challenges facing the Team, they will regularly want to solve many problems right away. They often will commit to too many improvements.
Cause: Not strictly prioritizing improvements
One of the most difficult activities is deciding what is not going to get done. Being crystal clear on what is most important is difficult when faced with so many choices.
Cure: Choose one actionable improvement per Sprint
By only choosing one experiment to try in the next Sprint and making it specific and actionable, we increase the odds of actually making improvements. It gives Teams something to reflect on during the next Retrospective. Over time, repeatedly making small improvements will add up to many significant improvements for the Team willing to learn and experiment.
Good meeting facilitation skills are critical for keeping meetings focused on their purpose. Training and supporting Scrum Masters to develop these skills has invaluable benefit to the Team and the organization as a whole.
While there are many other dysfunctions, I’ve highlighted the ones I hear about most often. If there is enough interest, I’ll add more.