Scope of This Report
Retrospectives are part of most methodologies, even though there are many different terms. For instance, most waterfall frameworks call them post implementation reviews or postmortems. Each methodology focuses on different nuances. Agile, as a macro set of frameworks, has more aggressively embraced retrospectives than waterfall or iterative frameworks.
Retrospectives in Agile reflect the adoption of the principle of kaizen (Japanese for improvement, often interpreted as continuous improvement). They should be focused on discovering what will make the team or organization deliver more value.
Brian Wernham, in Agile Project Management for Government noted the UK DirectGov project used retrospectives to mold how the worked in order to maximize delivery productivity. While many retrospective techniques posit the questions “what worked well” and “what did not work,” the real reason to do any retrospective is to identify, agree on and plan for what can be done better.
The exact process any team uses is a reflection of the technique the team wants to use, what works in the organization and the specific team situation. For example, the timing of retrospectives varies significantly depending on the framework and the organizational culture. Most waterfall projects do a retrospective at the end of the project (or release), while Agile projects typically do retrospectives at the end of every sprint, at each release, at the end of the project and occasionally on an as needed basis. In Agile, retrospectives occur when change can actually be applied to the project to impact the current delivery.
The power of a retrospective to improve a team’s ability to deliver business value is hard to deny. Given the demonstrable power of the technique, why does it occasionally fail to live up to expectations?
Ritualization can dramatically affect the value of retrospectives. Ritualization can cause the team or organization to enter a downward spiral of disillusionment that will inevitably end with the abandonment of the technique.
The ritualization of retrospectives occurs when the process becomes more important than (or at least as important as) the results. There are two typical reasons that cause ritualization: overcommitted teams who don’t have time to reflect and boredom (wake me up when it's over). Gil Broza in The Human Side of Agile notes that” Works Well and Needs Improvement format gets really old quickly.” Teams that are so overburdened and can’t find a mechanism to get un-overburdened will always stay that way (a “Catch 22”). The ritual of the retrospective will usually be fulfilled so that they team can start planning the next sprint or iteration. In essence, they need to check the process box so they move to the next step.
The Scrum Master or coach needs to help the team address the root cause of the problem, whether that is team-driven over-commitment (taking too much work) to being told to take too much by management. If a team is in an over-committed position more than occasionally, the Scrum Master may be part of the problem. That means that outside coaching is needed. If the root of the problem is technique boredom, the Scrum Master or another team member should learn another technique (such as timelines, the Sailboat, de Bono’s Thinking Hats or the 4 L’s). All Scrum Masters should know at least nine techniques for retrospectives.
A basic process for a typical Agile, end-of-sprint retrospective:
- Set Up: First, create a safe atmosphere (review Norm Kerth’s Prime Directive with the team). Then ground the team by focusing on the current sprint’s results (for example review the Burn-down Chart or have the team develop an annotated sprint timeline). Time box this part of the retrospective to no more than 20 minutes for a two week sprint.
- Idea Generation: Encourage the team to dig in and capture the details. For retrospectives focused on process or flow, use sticky notes to brainstorm, followed by mute mapping to group (affinity diagraming). For team or personnel issues, use storytelling. For example, have subsets of the team describe a fictional scenario based on real life problems and how they would solve the problem. Consider direct discussion as an alternative. This step should be time boxed to 30 minutes for a two week sprint.
- Insight Development: Once the idea generation step is completed, the team reviews the data and comes to a consensus about what it means. One method of analysis is to look for patterns and to determine if there are trends in this stage. The goal is to recognize if there is a problem so you can start to resolve it.
- Identify An Improvement Objective: In many cases a team might have identified a number of ideas for improving their productivity. Focus on the top one or two actionable big wins. The rational for not fixing everything is first that the time need to fix the problem will come from the team’s capacity to deliver business value (there is only so much capacity that the team has at its disposal). Secondly, if the remaining issues are really problems after the one or two most important have been dealt with then the team can decide to address them during the next iteration. Also, trying too many changes at once makes it hard to track cause and effect. This continuous incremental process improvement is one reason team productivity, aka velocity, typically increases from iteration to iteration. After the team selects the issue (or issues) to be tackled, have them add it to the next sprint backlog so that it gets addressed. This step should be time boxed to 30 minutes for a two week sprint.
- Wrap-up: Spend 5 – 10 minutes reviewing the session so that the next retrospective will be even more effective.
Retrospectives are a tool that the team uses to identify what they can do better. The basic process - making people feel safe then generating ideas and solutions so the team can decide on what they think will make the most significant improvement - puts the team in charge of how they work. When teams are responsible for their own work, they will be more commitment to delivering what they promise. The retrospective process is focused on increasing the team’s capacity, rather than trying to generate lessons learned for the next project.
Even though there are many different techniques for executing retrospectives, many teams find one or two techniques they like, and then they ride that horse until it collapses. Every Scrum Master and team should have a broad array of retrospective techniques, such as the sailboat technique, the Four L’s or the timeline. This provides at least two benefits. First, knowing many techniques means that you can match the technique to the particular teams. For example, many techniques use physical sticky notes, which are difficult for distributed teams. So, the Scrum Master needs to know that you can substitute an on-line mind-mapping tool for sticky notes. Second, having a wide range of techniques (and using them) is a formula for beating technique fatigue and the resulting obstacles.
List Generation Techniques
List generation techniques are the most popular class of retrospective techniques. The list generation techniques are popular because they take very little set-up, are easy to explain, easy to facilitate and get results. Listing techniques build on well understood brainstorming techniques to ensure the whole team has a voice. We discussed the basic Affinity Diagraming technique above.
Another technique in the listing category would be the “Sailboat” technique. This method uses a nautical metaphor. The boat moves through the water toward a goal (the team delivering functionality), the wind pushes the boat forward. As the boat moves through the water, it List generation techniques are the most popular class of retrospective techniques. The list generation techniques are popular because they take very little set-up, are easy to explain, easy to facilitate and get results. Listing techniques build on well understood brainstorming techniques to ensure the whole team has a voice. We discussed the basic Affinity Diagraming technique above.
Another technique in the listing category would be the “Sailboat” technique. This method uses a nautical metaphor. The boat moves through the water toward a goal (the team delivering functionality), the wind pushes the boat forward. As the boat moves through the water, it encounters resistance which slows its progress. Examples of resistance might include conflicts for needed resources or conflicting organization goals. Here is the process:
Set-up: Start by drawing a picture of a sailboat in the water on your white board or flip chart. Explain to the team that some things push forward, like the wind, and some things slow your progress down, like an anchor.
Idea Generation: Ask the team to identify what those items were. List one item per sticky note, which are then placed on the boat. As a facilitator, continue to tweak the seed questions you are using to keep the team thinking about the sprint from different angles. You are done when the team is done.
Insight Development: Have the team review the data and group ideas based on how they see the relationships between individual ideas. Techniques like Mute Mapping (grouping without talking) help to maximize team participation while minimizing the chance of a single person dominating. Once the grouping is done, ask the team to name each group. This helps to cement the group’s understanding of the groupings of ideas that they have generated.
Identify An Improvement Objective: Select a group or specific idea to fix. There are a number of techniques to select the improvement objective. Discussion followed by group consensus is one method (use this when it is apparent that the group is close to consensus). Another method is to vote using dots or post-it flags. In this method give each member a fixed number of flags and then ask them to vote (they can use all votes on one item or spread them). The item or group with highest number of votes gets fixed first.
Examples of other listing techniques include:
The Four Ls – Use four categories: liked, learned, lacked, and longed for to generate ideas. Write these titles on four flip charts and place around the room. Have each person silently generate ideas based on those categories. When the team is done (i.e. everyone stops writing) have the team place their ideas (written on sticky notes) on the appropriate flip chart. Once the team has come up with their lists, identify the improvement objective, usually from the lacked or longed for category.
What Went . . . – Use four flip charts, put one of the following titles on each flip chart: what went well, what did not go well and what should we do more of and what should we do less of. Brainstorm ideas to put on each flip chart. Put one idea or statement on each sticky. Depending on the group, this method can be done non-verbally (everyone puts their ideas on a set of “stickies” like the Four L’s) or have the team write ideas down and then shout them out (more akin to classic brainstorming). Insight development and identifying the improvement objective would follow a similar path to what was described above.
There are many other listing techniques each using a different set of seed questions (e.g. What worked well? or What would you like to do more of?) or different metaphors (e.g. sailboat, motorboats or trees). The questions or metaphors exist to help the team focus their discussion. The metaphor or seed questions that are used need to make sense to the team’s culture, and also solicit areas that they are concerned about or can be improved. The Scrum Master or facilitator needs to use their observations about the team to select the retrospective technique that will provide the greatest benefit. All of these techniques can work, selecting which you will use is matter of team culture.
There are other, more specialized techniques like the Timeline Retrospective, which is useful for long-running releases or in projects were a retrospective has not occurred in recent memory. These techniques deal with more complex issues than can be tackled using simple lists. These more complex methods can also be used to spice up a more basic fare of listing techniques to keep teams involved and interested in the retrospective process.
These techniques are more complex to execute. Let’s explore a few examples of this class of retrospective.
The Timeline Retrospective uses the following process:
Goal: The Timeline Retrospective technique develops a visual overview of the events that occurred during the period under investigation. This technique identifies and isolates the events that impacted the team’s capacity to deliver over a set period of time. It uses distinct colors to identify events (avoid typical red – green colors as color blind participants may have difficulty).
When To Use: The Timeline Retrospective is useful for refreshing and re-grounding the memories of team. Other circumstances in which this may be a useful technique:
- If there have not been any intermediate retrospectives;
- To provide context to program-level (i.e. multiple projects) retrospectives;
- If the team has not been working on the project over the whole life cycle, and
- An end of project retrospective.
Set Up: Draw a timeline that represents the period since the last retrospective on a white board (or several pieces of flipchart paper). Make sure there is room above and below the line. Secure dry erase markers in a few colors and sticky notes in three colors. The three sticky note colors will represent:
- Blue represents good events;
- Yellow represents significant events that are neither good nor bad, and
- Red represents problem events.
Use the colors that you feel the most comfortable with and that you have in sufficient supply.
The process is as follows:
- Have each team member silently write down on sticky notes the major events, from their perspective, using the color code from above.
- Have each team member put their events on timeline chronologically, placing positive events above the timeline, neutral on or near the timeline and negative events below the timeline
- Throw out duplicates.
- Have the team select someone to walk through the final timeline.
- Using the dot voting technique (provide each team member with three dots) rank the event that slowed the project down the most to date.
- Identify tasks and actions that could be taken to solve the problems. Pick the top two or three.
- Have the team tell the story of the project for the next sprint or release, if they took the identified actions. This will help validate the choice of the proposed changes.
Another example of non-list retrospectives is the 6 Thinking Hats Retrospective (based on De Bono’s Six Thinking Hats). Use this type of approach when the team has experienced significant challenges, has not established norms on how to interact or tends to be dominated by one or two personalities. In this technique, the team uses a structured approach to discuss the period since the last retrospective. The team “wears” one of De Bono’s “hats” at a time, which means all participants talk about a specific topic area at a time. Each hat represents a particular way of thinking. Using the hats forces the team to have a focused discussion (this is called collective thinking in the literature). Until you are comfortable with this type of technique, use a facilitator. The facilitator should ensure that the comments are appropriate to the “hat” that is currently being worn. Here is the order of the “hats”:
1. Blue Hat (5 minutes) – focus on discussing session objectives.
2. White Hat (10 minutes) – discuss or identify FACTS or information since the last sprint (e.g. we had a hurricane during this sprint).
3. Yellow Hat (10 minutes) – talk only about the good things that happened since the last retrospective.
4. Black Hat (10 minutes) – talk only about the bad things that happened since the last retrospective.
5. Green Hat (10 minutes) – talk only about ideas to solve the identified problems or ideas that would add more significant value in the Product Owner’s perception.
6. Red Hat (5 minutes) – Have each team member come to the white board or flip chart and write two emotive statements about the project during this period. Do this fast and with very little preparation. You want gut reactions.
Finally, have the team review the emotive statements to identify clusters of comments or trends that can be combined with the issues in green group. From the identified issues pick one or two actions that will improve the ability of the team to deliver to add to the backlog for the next sprint.
Other techniques in this class include:
Emotional Trend Line – This is many times combined with the Timeline technique. It provides an estimate of the team’s emotional state since the last retrospective.
Complexity Retrospective – Draw a complexity radar plot with at least five axes. Engage the team to determine what each axis should be labeled (e.g. data, workflow, code, business problem) and then engage the team to rate each axis. If an axis is rated as complex ask the team to identify actions to reduce complexity.
Non-list based retrospectives are generally more complicated to apply due to the formal structure they use to guide teams toward discovery. For example the De Bono’s Six Thinking Hats will require active facilitation (at least until everyone is comfortable with the process). These techniques are generally used to address special or specific circumstances. The structure of the techniques has been designed (or in some cases these techniques were adopted from other disciplines) because they help to focus the retrospective participants on a type of problem. The goal of any retrospective, list or non-list, is to help the team to discover how they can learn to be better during the next iteration.
Coaching a Retrospective
Facilitation skills, choice of technique and the tools that are used, in that order, will impact the effectiveness of retrospectives. The degree of team distribution will cause the degree of importance of the attributes to vary. For instance, distributed teams will have to lean on communication tools to a greater extent.
At their heart, all retrospectives are social exercises. Even in well-honed teams it is an ongoing challenge to keep team member talking and sharing in a manner that will convey information without damaging relationships. As Naomi Karten pointed out in her interview on the Software Process and Measurement Cast, there are a relatively high proportion of introverts in IT who need help in order to be drawn out. The facilitator’s skill at getting people to interact is more important as the degree of team is distribution increases. In distributed teams, the facilitator needs to find ways to make the team’s interactions more personal. For example, making sure everyone talks and that location bias does not set in. An interesting technique to defeat this issue is to pair individuals from different locations as homework for the retrospective. In one example DCG facilitated, a Scrum Master asked each pair to collaborate and identify five ideas to improve collaboration, while another time (with different sets of pairs) the Scrum Master asked the groups to do a five minute overview on an upcoming local holiday. The goal was to make sure the locations were talking and that the whole team was exposed to the different cultures on the team, which fosters deeper communication.
Different retrospective techniques will evoke different responses from participants. For example, a timeline retrospective will focus on events. A classic list generation-based retrospective will focus on process. Picking the right type of retrospective gives a facilitator the chance at opening up the team. The technique selected for a retrospective is generally a balance between the focus and technique satisfaction (i.e. how many times you have used the technique as over use causes boredom). When coaching long-term projects, teach the team a variety of techniques and then let them select the technique that they want to use. Remember to add techniques or remove techniques from the team playbook as the situation warrants.
Every retrospective requires some sort of tool. Tools can be as simple as a white board and markers or as complex as mind-mapping and screen-sharing software. When a team is distributed, screen sharing and teleconferencing/videoconferencing tools are necessities. The combination of technique and level of team distribution will influence tool selection. Likewise, tool availability will influence technique selection. For example, use a mind mapping tool and screen sharing when executing a listing retrospective for a distributed team so that each location can see the ideas and participate. If the distributed team could not use those tools, you will have to find a different approach. Generally the technique defines the toolset, but that is not always the case. When everyone is in the same room sticky notes are great but when team members are teleconferencing into the retrospective electronics are required.
The retrospective can’t become ritualized to the point that it lacks meaning. Each retrospective needs to provide a platform for the Agile team to reflect on their performance and to determine how they can achieve more. This is a team activity that requires a free flow of conversation and ideas in order to maximize effectiveness. That means someone needs to facilitate the process and police the boundary. No team is perfect and all teams can learn and improve on a continuous basis. Most obstacles to effective retrospectives are solvable with a bit of coaching and education, if you recognize the obstacles before you abandon the technique. Facilitation skills, retrospective techniques and tools are all important for an effective retrospective. The technique is driven by needs of the team. The coach/facilitator needs to be aware of the needs of the team and the proper tools to facilitate the technique. If they are not available, pick another technique. However once the retrospective begins, facilitation skills are always the most important factor. Even with the best technique and tools, retrospectives are all about the people.
- Agile Project Management for Government, Brian Wernham, Maitland and Strong, 2012, P 195
- Norm Kerth’s Prime Directive, http://www.retrospectives.com/pages/retroPrimeDirective.html
- The Human Side of Agile, Gill Broza, 3P Vantage Media, 2012, P. 172
- The Agile Mind-Set, Gil Broza, 3P Vantage Media, 2015, 149
- De Bono’s Six Thinking Hats, https://en.wikipedia.org/wiki/Six_Thinking_Hats
Download this Trusted Advisor report here.