Tuesday, December 30, 2014

Retrospectives - Objective Vs Subjective

Traditionally Agile teams run retrospectives to facilitate learning, development and improvement of the team. My teams have been doing the same over the past ten years. We have tried multiple different formats to get the right outcomes. We have been through it all and we have all been through it - lack of participation, finger pointing, irrelevant topics, lack of prioritization of issues, many a time one of these problems or another hinder the effectiveness of a retrospective. Recently my team moved to a more objective retrospective from a completely subjective form of retrospective.

Subjective Retrospectives

The most common form of retrospective employed is the subjective form. This shows up in numerous variations. It consists of collection and listing of ideas/topics, prioritization, either by moderator or by the participants via voting, discussion and finally extraction of action items. This method, as you can see can range from almost autocratic to a fully team driven setup. There is a middle way, which I like to call guided setup where the facilitator coaxes the team along through the process.

Autocratic Guided Team Driven
Team Lead/Scrum Master/Coach comes up with list of items to be reviewed during the retrospective. Team Lead/Scrum Master/Coach helps seed ideas for discussion and encourages team members to add topics for discussion. Team creates list of items to be discussed with every one having equal access to introducing topics.
Team Lead provides the priority of topics to be covered in the retrospective.Team Lead helps the team prioritize the topics to be discussed.Team votes on the list that they have prepared and generates a prioritized list.
Team Lead presents his/her thoughts on the topics and prompts discussion/feedback on these topics. Team goes through the topics in priority order and Team Lead helps seed discussion if no discussion seems to emerge organically. Team goes through the prioritized topics and discusses/reviews each item to generate next steps.

Regardless of which of these subjective retrospective methods are used the most important piece is to derive action items and to have individuals be responsible for the completion of these. From a casual reading of the above table, it would seem that a team should never use the Autocratic model, should sometimes use the Guided and almost exclusively use the Team Driven approach. That maybe ideal, but unfortunately, is not real. Depending on how mature the team is, any of these approaches would be appropriate. A team with little agile experience might actually benefit a lot more by employing the Autocratic model as they might need more externally imposed discipline and help identifying the issues and the priority of those issues affecting their agility.

Objective Retrospectives

My team has started using metrics for retrospectives. We look at our scatterplot which plots our stories along the axes of dates and days taken to close the story. We then, talk about the stories that are the outliers on the higher end and figure out why they took longer than other stories. This ends up becoming a very direct discussion about why things are taking longer than they should and exposes process inefficiencies very explicitly. I found that this helped the team identify direct impact action items that improve the team's efficiency. For example in the plot below, the stories in the red circles are the ones we would concentrate on in the retrospective.

Objective retrospectives do have the side-effect of the team trying to game the system to make the numbers look good. That, in my opinion is not a bad thing if you choose the metric you are using well. I find days to finish a story to be a good metric as in order to make it look better, your direct options are to either create smaller stories or reduce the number of items you are working on. Both of these are positives in any agile system. Lets say you chose a different metric, velocity for example, gaming the system there could be achieved by overestimation of stories. The idea is to pick a metric that reveals inefficiency and making that metric better improves the team overall.

We did notice an issue with the objective retrospectives. There are certain things, that dont show up as data points on the scatterplot. There are both positive and negative subjective comments that are left over to be covered after the objective retros are done. The team going forward is now doing a split review where we start with the objective and cover what is slowing down progress and then do a shortened subjective Team Driven retrospective to cover any other topics that need to be discussed.


Regardless of the method you employ, it is the willingness of the participants to improve as a team that is critical. That is why sometimes the Autocratic or Guided subjective retrospectives might be the right place to start for teams new to retrospectives as a tool for improvement. Objective retrospectives make sure that you talk about everything that is slowing down the team's work. The priority goes to the issues that create the greatest slowdowns and you can easily see the results the next time you review the data. Subjective retrospectives still have their place, but a mature team should be able to derive most of the value out of a retrospective via the objective method.