Dispersed Retrospectives with CardMeeting and Skype

April 27, 2010 at 8:05 am 4 comments

A few weeks ago I facilitated a retrospective for a dispersed software development team with the help of CardMeeting and Skype. The result was surprisingly usefull – at least when the number of 21 participants is taken into account.

During the preparation of the retrospective I speculated a while about what steps are neccessary and how long I should plan for these. I reproduce the actual timeline here. Perhaps that is useful for others facilitating a dispersed retrospective.

I had created the meeting in CardMeeting.com long before the retrospective took place. I emailed the link and password to all participants and asked them to test if CardMeeting would work on their computers. CardMeeting uses Java Applets and doesn’t run on every system straightaway. But in the end everybody managed to get cardmeeting running on his PC – there was no pattern I could understand: some participants had to use IE on Windows while for others only Firefox worked. I for myself had to switch to Chrome on Ubuntu since Firefox didn’t want to start the applet correctly.

At the PC of one participant CardMeeting broke down ten minutes after the retrospective had started. He then joined a collegue in the next office and they paired on front of one PC during the retrospective.

10 minutes: Introduction
After I welcomed everybody to the retrospective we did the privacy contract: everything discussed during the retrospective would belong to the team privately unless everybody agrees to another handling of the informations at the end of the retrospective.

Then I gave a short introduction to CardMeeting. I showed how to create, edit, move and trash cards. (We didn’t use the billboards to keep things simple.)

For the rest of the retrospective I took orientation from retrospectives.com and used the proposed four key questions.

10 minutes for question 1: What did we do well, that if we don’t discuss we might forget?
I created a white card with the question. The participants added green cards with their thoughts and placed them around my white card.

Then a funny problem occurred. In CardMeeting a newly created card is visible to the other participants immediately even if the creator hasn’t written something on it. When 20 people create cards in parallel it is hard to tell which card belongs to whom. So we observed some card robberies. But the participants adapted fast: Immediately after one had created a new card he moved the card far away from the center of the action, edited the card and then moved it back to the center of the action. That worked out sufficiently well. I even think of the card robberies as a welcomed little entertainment.

10 minutes for question 2: What did we learn?
I created a white card with the question. The participants added their thoughts on blue cards.

10 minutes: Read and understand the cards
Everybody read all cards and asked questions via Skype audio for understanding.

10 minutes: Cluster the cards
The participants clustered the green and blue cards from the first two questions. They used red cards to give the clusters headlines. The clustering went surprisingly well without any intervention from my side.

15 minutes for question 3: What should we do differently next time?
Again I wrote the question on a white card. The participants added their thoughs on yellow cards.

We needed 15 minutes for this question (in contrast to the 10 minutes for the first two questions) to create focused cards. Some of the originally created cards didn’t describe what the team should do differently but what outsiders (managers, customers etc.) should do differently. Some other cards were somewhat fuzzy in a “we must try harder” manner. I commented on these cards that they didn’t match the question exactly.

5 minutes: Read and rework cards
All participants read all cards and the authors integrated the comments. In this step identical and similar cards were merged so that we didn’t need an additional clustering step.

5 minutes: Prioritize the cards
The previous step resultet in 17 yellow cards. For the prioritization every participant was allowed to attach two green cards to his favourite yellow card(s). Everybody had the possibility to attach red veto cards to yellow cards.

After the voting we ordered the green cards by the number of votes.

15 minutes: Select cards and define action
With respect to the votes there was a clear gap between the most voted six cards and the remaining eleven cards. Every of the six cards had a lot of votes. The eleven cards had no votes or just one or two. Therefore the participants decided to discard the eleven cards and work with the high voted six cards.

For the six yellow cards we created action items on blue cards and attached them to the yellow cards. That was pretty easy since the yellow cards were already very close to action items.

BTW: The previous step created one veto. Since the veto card was attached to one of the discarded cards we didn’t have to discuss the veto.

10 minutes for question 4: What still puzzles us?
I created the question on a white card and the participants added their thoughts on yellow cards.

5 minutes: Cluster the cards
The participants clustered the cards and created a red headline card for every cluster. This step was shorter than the clustering done before since there were very few yellow cards to consider.

5 minutes: Closure
In the end I initiated a discussion about the privacy policy we defined in the beginning. The participants were very open and wanted to put all the cards in their wiki. Unfortunately there is no export function in CardMeeting. One has to transfer the cards one by one with copy&paste. Therefore the participants decided to only transfer the committed action items to the wiki.

All in all the retrospective lastet nearly two hours.

After the retrospective I sent a Google-Docs feedback form to all participants and asked for their impressions. The participants gave the retrospectice a positive ROTI (return of time invested) – the retrospective was worth it. But the energy level and the quality of the results were a bit lower than in a conventional face-to-face retrospective.

I think there is a common pattern of distributed work that applied to the retrospective: it is possible but it is less effective than co-located work.

The partipants found CardMeeting useful and Skype acceptable.

Personally I observed a few differences compared to face-to-face retrospectives:

  1. CardMeeting doesn’t attach the author of a card to the card. Therefore it is impossible to see who wrote which card. That leads to a higher degree of anonymity.
  2. There is less discussion via Skype audio compared to a face-to-face retrospective. And more cards are written. Maybe this allowes the shyer people to participate stronger.
  3. As the moderator I had to adapt some of my moderation techniques (e.g. voting with dots vs. voting with cards).
  4. I felt the need to prepare the retrospective more detailed and plan for shorter steps. In hindsight I think the detailed planning and the shorter steps were useful. Since there were less discussions the energy level decreased fast during the “write cards” steps. With the short steps there was an intervention by me every few minutes that raised the energy level again at least to a medium level.

What would I do differently next time?

  1. I would like to use an alternative tool to CardMeeting – one reason is that I have the impression that CardMeeting isn’t developed any further. Unfortunately I don’t know of other tools that may work as a replacement.
  2. For a single session two hours is too long. It is hard to concentrate and my headset caved an interesting pattern in my head. I will plan for a five minute break for the next time.

Entry filed under: it-agile-blog-planet. Tags: .

Lernen im 1. Quartal 2010 JAX-Vortrag über Scrum-But(t) online

4 Comments Add your own

  • 1. Phillip  |  May 2, 2010 at 11:05 pm

    We do distributed retrospectives as well, and use Mindmeister. It’s a fairly fancy JavaScript/canvas based tool which allows concurrent editing by many users. Here’s a demo I set up for your questions: http://www.mindmeister.com/49435948/demo-retrospective
    This works fairly well for our small team, YMMV with 20 people.

  • 2. Alex Bepple  |  May 3, 2010 at 8:37 am

    Why did call this a dispersed retrospective? Why not ‘distributed’? Did you pick up the term ‘dispersed’ somewhere?

  • 3. stefanroock  |  May 3, 2010 at 2:12 pm

    I stole the term from Jutta Eckstein who used it years ago in an email discussion.

  • 4. stefanroock  |  May 3, 2010 at 2:13 pm

    Interesting idea. I will think about that.

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 )

Google photo

You are commenting using your Google 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 )

Connecting to %s

Trackback this post  |  Subscribe to the comments via RSS Feed

%d bloggers like this: