High Touch Retrospectives for Distributed Teams Using Trello

Some of us in the Agile community are not fortunate enough as to work within collocated teams all the time (is it most of us already?). Some of us may spend a significant part, or even all, of our effort as part of geographically distributed or dispersed teams. Just one of the myriad difficulties when working in such teams is to replicate the high-touch techniques to increase participation and collaboration in team events, such as planning and retrospectives. Recently plenty of online tools have reached maturity, allowing teams to collaborate in real-time, in simple and effective new ways, that is, closer to the same room experience.

This post describes in detail a real-life example of using Trello to run a retrospective. Please note no technique or tool is universally applicable and these ones are no exception. You will need to check the circumstances and forces influencing your own problem before applying any specific tool to solve it. Expect some story telling ahead.

Before We Start…

This is about an actual project run by a small agile team, comprised of four Team Members plus a Product Owner and myself as Scrum Master helping them get started.  The team had been formed recently, this was our first project together and we have been running Scrum for 7 one-week Sprints. We were a dispersed team, covering four countries and three continents between the 6 of us.

The work the team was doing was more innovative than an average new product development, pushing a bit the limits on existing technology to solve our given problem. We were not tasked with getting a general consumption product, but to “de-risk the area creating something that product development teams could use later on” (sic). The project went smoothly all the way up to sprint 7, we faced some difficulties but no big bumps, which considering the project risk profile was in itself a surprise. The team had impressed our PO and other stakeholders with great progress at a very fast pace.

During sprint 7, things got weird with some functionality developed in previous sprints playing really weird games on us. The team got nervous, and a mixed feeling of anxiety and disappointment was revealing itself all over our communications. It was time for the game changing retrospective. Of course the team had been running short retrospectives every week as part of our sprint change meetings in which actions were taken resulting in welcomed improvements, but this sprint was special. I have noticed along my agile journey that most teams get to a special retrospective when they are forming, one in which they realize the power of a retrospective well done, one which is deemed months later as the defining moment for them. I thought this time has reached for this team and I was eager to facilitate that process. I could not let them down.

Preparation

I prepared an agenda for a ninety minutes retrospective, not our usual thirty minutes time-box and clarified with the team why we were investing longer than usual, it was not that finding the issues was a horrible thing and we needed someone to blame for that, but it was about learning something about our team processes. Once we have done the learning, how to get better will reveal itself. Improvements are not the point of the retrospective, but the learning that enables them. We may need a few experiments before we find something that works for us, and even then we would have found just the starting point, not the end destination. I perceived the team was eager for the exploration and more relaxed once they know there could be a positive outcome after their struggle.

The best way I knew to run the retrospective was using the five steps and the kind of interactive, high-touch, card-based activities described in Esther Derby’s and Diana Larsen’s Agile Retrospectives book. The problem was running this kind of interaction remotely is not an easy thing to do. Fortunately I knew of Trello which I had been using for a while and that would provide us with the kind of interactivity we needed, without having to deal with unnecessary complexity. There are some other online tools providing similar functionality, I just happen to use this one often and I love it.

There was still one thing to clarify with the team, adding cards to a Trello board is not an anonymous activity, history reflects who has done what. Not that I thought this was a showstopper for this team and this retrospective but I double checked just to be sure: first by asking directly and second by adding a safety check to the Agenda.

Here Is the Agenda
Activity Expected Duration
Set the Stage 10 min
Check-in 5 min
Safety Check (anonymous SurveyMonkey survey) 3 min
Prime Directive 2 min
Gather Data 20 min
Timeline, Adding Events (Trello) 10 min
Team Sharing, Reading and Grouping (Trello) 5 min
How Did It Feel? Glad, Sad, Mad (Trello) 5 min
Generate Insights 30 min
Patterns and Shifts (discussion) 15 min
Five Whys (Discussion + MindMap) 15 min
Alternatively for this section we can do a Learning Matrix if we are short on time (Good, Bad, Ideas, Appreciations)
Decide What to Do 25 min
Start, Stop, Keep Gathering (Trello) 10 min
Prioritization (Trello) 10 min
SMART (Trello) 5 min
Close 5 min
Retrospective Plus/Delta (Trello) 5 min

There were a few activities in my Agenda that were thought for single room execution, using post-it notes or index cards and a wall to display them.  My plan was to use Trello for all of them. No complex setup was needed, I created an organization to control the boards visibility and invite all relevant team members. I created three boards and added all team members to each board. I created the right columns within each board and voilá, I was done in less than five minutes. I expected usage to be self-evident for people accustomed to use boards and index cards, so I expected team members to get to add cards and move them around in no time with little guidance. I planned to use some other “fancy” features of Trello like using color tags for cards or member voting, which I planned to give instant instructions for when they were needed. I was setup, ready for some real action.

Running the Retrospective

The team was happy with the board history not being anonymous, they checked-in, they listened to the Prime Directive and agreed to it, they even described their own individual approaches as that of a Explorer. There was a climate of security and discovery.  We were moving full steam ahead.

To gather information about what happened, the sequence of events and how the team felt the team built a timeline. They did it in approximately the time I had estimated, but surprisingly to me the glad, mad and sad feelings-recording exercise took a bit longer than in my previous experiences. To do this the team is asked to mark the cards in the timeline with three different colors depending on how that particular event made them felt. According to the team the main reason for my estimate to be too low was I didn’t account for the time it took them to think how they actually felt at the time of the event, not how they did feel about the event at the time of the retrospective. Something missing when you use Trello labels feature as we did is some sense of quantity: if three participants mark one card red it will look exactly the same as if only one of them marked it. That did not make our exercise less useful as its main purpose was to spark the conversation which happened.

After gathering all the data, the team was in a calmed but energetic mood which allowed me to move to the analysis part: generate insights. The timeline helped us getting started and the whole team debated the patterns and shifts observable in the timeline. The tool was proving its value again. The team got engaged in a lively discussion of a specific sequence of events, discerning the important and the accessory without any intervention on my side. It was taking longer than my estimate but what was happening there was clearly good for the team so I let them go on for a while. Once the silence marked me the team was ready, I asked why, looking for a sequence of causes that lead us to the learning we were after. It was time well spent but it has just passed the time I planned to finish, so I asked the team what they wanted to do. The answer was immediate, unanimous and pretty sound, they wanted to get to the bottom of their investigation.  We agreed on a thirty five minutes extension, and got into action identification mode.

Within the next half an hour the team discussed and agreed to SMART goals. The ambient was good, we identified a few issues and we had some experiments to run to validate our theories. It was time to learn about running retrospectives. We spent five minutes discussing what was to be repeated (plus) and to be changed (deltas) in the subsequent retrospectives. We agreed it was effective but we felt we needed to add some more time to the estimate or plan and keep ourselves accountable for not breaking the time-box.

Records
Activity Actual Duration Records
Set the Stage – Safety Check (anonymous survey) 5 4 = E (Explorer). here to explore both my own and others thoughts and feelings. Happy to talk about anything.
0 = S (Shopper). Happy to observe and interested in what people have to say about all topics. If something that I care about is discussed then they will get involved and talk.
0 = V (Visitor). Not keen on talking, just happy to sit back and observe.
0 = P (Prisoner). Not comfortable being here at all. Only here because I have to be, if I had a choice I would be somewhere else.
1 = No answer (Javier sent it too late)
Gather Data – Timeline 29 Timeline Board
Generate Insights – Root Cause(s) 60  Identified Root Cause: Testing quality

  • Not enough detail in test case scenarios – requirements. Lack of input from customers for test case detailed definition. What is in scope? What assumptions are reasonable?
  • Lack of full test automation coverage (leading to not enough tests run every sprint, both for regression and new functionality). Affecting forces:
    • Difficulty of the environment for automation.
    • Uncertainty about the final goal (sample, toolkit, full integration) made it difficult to judge whether investing in thorough testing and automation was necessary.
Decide What to Do – Actions 30 Start, Stop, Keep Doing BoardConclusions:

  1. Add a check box to our DoD: for each test case, identify a possible automation strategy. It may be different for each case and it may be just support not full automation.
  2. During each Sprint Planning verify the quality of test scenarios for each story. Walk the entire functional flow. Describe the different paths as test cases. Exceptional scenarios (out of the happy case) should be covered.
  3. Add an entry to DoD covering updates to the User Manual and Deployment Guide with every story (only if needed).
Close – Retrospective Plus/Delta 5 Plus/Delta BoardConclusions:

  1. Very effective retro.
  2. Javier to increase adherence to time-box next time.

After the retrospective

Just in case you are interested in knowing the end of the story, the weird issues found by the team were caused by a weird combination of previously untested scenarios and an unexpected hardware failure. They could have been avoided if the actions agreed during the retrospective were adopted a couple of weeks before. A few things changed in the team that day: they started considering more carefully the scenarios to be covered for each functional story, testing quality got pumped, automation tools were created when possible… but above all the most important discovery the team did was how learning (or inspect and adapt, or PDCA, or the scientific method, or… you name it) will transform them as a team and the work they do.

I have learned a couple of things as well that day. I can actually use Trello as a tool in this kind of environment, not only for retrospectives but for card-based collaborative games to be played online, but I just need to allow a little extra time to account for being remote to each other.

References

My boards:

Even though I haven’t used them when preparing my retrospective here you have a couple of useful links I could find on the web when deciding if I was to write this post:

Advertisements

2 thoughts on “High Touch Retrospectives for Distributed Teams Using Trello

  1. Wow, I love Trello and using it for retrospectives like this is genius!!
    Even though I don’t work with a distributed team, I’m wondering if I could use it to collect retrospective items through the sprint to combat the issue of getting to the retrospective and team members not being able to remember what they wanted to bring up. Obviously talking is better, but if it was done carefully, it might allow us to have an underlying current of reflection and improvement without being too disruptive to the work.

    • If I’m understanding you correctly your idea is to use trello to build the timeline or a list of items to discuss “as the sprint goes”. It sounds interesting please do let me know how it goes if you try it.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s