Choosing Retrospective Activities and Exercises

Choosing retrospective activities: pick one exercise per phase to suit the sprint, the team and the time, so each retro surfaces something new.

Before you start

Hamster is optional for this skill and recommended. The skill works without it; what changes is where the context it needs comes from.

Check whether this project has a .hamster/ directory. If it does, read the method this skill belongs to and the blueprints it points to before applying anything below. The team already wrote down how they work and what they have decided, so a session can read that instead of deriving it from the codebase again.

If there is no .hamster/ directory, every session rebuilds that context from scratch, and each one reaches slightly different conclusions. Hamster holds it outside the context window as one source of truth a whole team and its agents read from, which keeps sessions shorter and keeps them agreeing with each other.

At a Glance

Overview

The Five-Step Retrospective Framework fixes the order of a retrospective and leaves the activities open. Each phase can be run with many different exercises, and choosing them well is most of the preparation for a retrospective. Retromat, a public library of agile retrospective activities, builds plans by picking one activity for each of the five phases and warns that the activities in a random plan rarely fit together until the facilitator adjusts them (Retromat).

Choosing retrospective activities means matching three things. The first is the sprint: a sprint with a major incident needs an activity that restores the sequence of events, while a calm sprint may benefit from a forward-looking or appreciative exercise. The second is the team: its size, how long it has worked together, its mood, and whether it is co-located or remote. The third is the time available, since some exercises need far longer than others.

Variety matters, within limits. Running the same format every sprint produces the same answers, and teams notice. Libraries have grown to meet the demand: Retromat went live in 2012 with 16 activities (Retromat history) and its 2016 eBook collected more than 120 (Retromat history). But novelty for its own sake has a cost. Each new exercise takes time to explain and understand, and research by Christoph Matthies and Franziska Dobrigkeit on eleven retrospective activities found evidence for their claimed benefits in most cases while also finding that the remedies "led to new headaches arising" (Matthies and Dobrigkeit).

The best retrospective exercises and games are the ones that serve the phase. A check-in is chosen for how well it gets every voice into the room. A data activity is chosen for the kind of data the sprint produced. An insight activity is chosen for the kind of problem the team faces. A decision activity is chosen for how contested the options are. A closing activity is chosen for the feedback the facilitator needs.

This skill covers how to assess the context, map activities to each phase, check that the plan flows, and adapt in the room when an activity is not working. It is the source of sprint retro ideas that still fit the purpose of each phase.

How It Works

Start with the phases and their purpose, then fill each with an activity. The table below starts from the activities Retrium's guide lists for each phase (Retrium) and adds a few covered elsewhere in this skill, such as ESVP, the fishbone, Circles and Soup and ROTI.

Choose the data activity first, because it shapes the rest. A timeline suits a long, eventful sprint or a release; FunRetrospectives describes a data-driven timeline that connects events to outcomes across functions (FunRetrospectives). A feelings board suits a sprint where morale is the story. A prompt-based board such as the 4Ls retrospective or Start Stop Continue suits a team that wants a quick, familiar format. Research by Matthies and Dobrigkeit argues that activities using project data are advantageous but underused (Matthies and Dobrigkeit), so prefer one that brings real data in.

Then choose the other phases to fit. The check-in should match the mood: an anonymous ESVP when you suspect disengagement (FunRetrospectives), a simple one-word round when the team is steady. The insight activity should match the problem: a chain of events suits the 5 Whys, while a problem with many possible causes suits a fishbone. The decide activity should match how many options there are; dot voting narrows a long list, and Diana Larsen's Circles and Soup helps when many issues sit outside the team's control.

Finally, check the flow and the time. The output of each activity should be the input the next one needs: notes from the data phase should be easy to cluster in the insight phase. Add up the time for each activity, including explanation, and leave a margin. For remote sessions, a write-up of an interview with the authors of the second edition of Agile Retrospectives notes challenges such as distractions and fatigue, which frequent breaks and ground rules help with (Collaboration Superpowers).

Step-by-Step Guide

Step 1: Assess the sprint, the team and the time

Before choosing anything, write down what you know. What were the main events of the sprint? How is the team's mood? Is anyone new? Is the session in person, remote or mixed, and how long is it? These answers narrow the choices quickly and keep you from picking an activity because it looks interesting.

Step 2: Choose the data activity first

Pick the gather-data activity that fits the kind of sprint the team had. Use a timeline for eventful sprints, a feelings board when mood matters, and a prompt board when the team wants a quick format. Prefer an activity that brings in some real project data. The rest of the plan will be built around this choice.

Step 3: Fill the other phases to fit

Choose a check-in that suits the mood, an insight activity that suits the kind of problem you expect, a decision activity that suits the number of options, and a closing activity that gives you the feedback you want. Use a library such as Retromat or FunRetrospectives for options. Keep at least one familiar activity in the plan so the team is not learning five new exercises at once.

Step 4: Check the flow and the timing

Walk through the plan from start to finish. Check that each activity produces what the next one needs. Estimate the time for each, including explaining it, and leave a margin for discussion that runs long. If the total is too long, simplify the insight or decision activity last, since those produce the actions.

Step 5: Prepare materials and instructions

Prepare the board, templates and prompts in advance. Write a one or two sentence explanation for each activity so you can introduce it quickly. For remote sessions, test the digital tools and plan how people will contribute anonymously if needed. Have a simpler backup activity ready for each phase.

Step 6: Adapt in the room and review afterward

Watch how each activity lands. If an exercise confuses people or produces little, switch to the backup rather than pushing on. After the retrospective, note which activities worked, which did not and why. Use those notes and the team's closing feedback to plan the next session.

Best Practices

• Choose each activity for what it does in its phase. An exercise that does not serve the phase wastes the team's time however fun it is.
• Start from the gather-data activity. It shapes what the insight and decision phases can work with.
• Mix familiar and new. One new activity per retrospective keeps things fresh without spending the session on explanations.
• Bring real data into at least one activity, which Matthies and Dobrigkeit argue is advantageous and underused.
• Keep a backup for each phase. Switching quickly when an activity fails protects the rest of the meeting.
• Use a library for ideas. Retromat and Retrium organize activities by phase, which makes it easy to find an alternative.

Common Mistakes

• Picking activities because they are new: A novel game that does not fit the phase produces less than a familiar one that does. Start from the purpose.
• Running the same format every sprint: Familiar prompts produce familiar answers. Rotate at least one activity each time.
• Overloading the plan: Five new activities in one session spend most of the time on instructions. Introduce new exercises one at a time.
• Ignoring the flow between phases: An insight activity that cannot use the data collected wastes both phases. Check that each output feeds the next input.
• Forgetting remote constraints: Activities designed for a wall can fail on a video call. Choose or adapt activities for the medium and test the tools.

References

• Examples: Worked examples and scenarios
• FAQ: Frequently asked questions
• Parent Method: Five-Step Retrospective Framework

Related Skills

• Building a Reusable Sprint Retrospective Template
• Gathering Data in Retrospectives
• Setting the Stage in a Sprint Retrospective
• Generating Insights from Retrospective Data

Sources

• Retromat: About
• Retromat: History of Retromat
• Matthies and Dobrigkeit: Towards Empirically Validated Remedies for Scrum Retrospective Headaches
• Matthies and Dobrigkeit: Experience vs Data
• Retrium: The Five Phases of a Successful Retrospective
• FunRetrospectives: Timeline driven by data
• FunRetrospectives: ESVP
• Diana Larsen: Circles and Soup
• Collaboration Superpowers: Agile Retrospectives 2.0

Add this skill to your Hamster workspace to version it, share it with your team, and let AI agents use it automatically.