Most retros produce the same action items every time and nothing changes. The format is rarely the problem. The problem is what happens after the session ends.
This guide covers the four most useful retrospective formats with copy-paste templates, plus a practical fix for the action item follow-through problem.
The Four Most-Used Formats
Format 1: Start / Stop / Continue
The simplest format. Best for new teams or when you want a fast, low-friction session.
Template:
START: What should we begin doing?
- What's missing from how we work?
- What have we seen other teams do that we should try?
- What process gap keeps causing problems?
STOP: What should we stop doing?
- What's wasting time without producing value?
- What meeting, process, or habit isn't pulling its weight?
- What are we doing out of habit rather than intent?
CONTINUE: What's working and should be protected?
- What went well this sprint that we should repeat?
- What processes are working even if they're not perfect?
- What should we be careful not to drop when we make changes?
How to run it: Give everyone 5 minutes of silent writing (each item on a separate sticky or card). Then group similar items, dot-vote on the highest-priority ones, and discuss the top three to five. End by converting selected items into action items.
Format 2: 4Ls (Liked / Learned / Lacked / Longed For)
Better for teams that want to capture growth and learning alongside process issues.
Template:
LIKED: What went well this sprint?
- What are you proud of?
- What felt good to deliver?
LEARNED: What did you learn?
- About the product, the codebase, the users, or each other?
- What would you do differently because of what you now know?
LACKED: What was missing?
- Information you needed but didn't have?
- Support, resources, or clarity that would have helped?
LONGED FOR: What do you wish had been different?
- What would make the next sprint better?
- What's the gap between how we work and how you'd like to work?
How to run it: Same as Start/Stop/Continue. Silent writing first, then group, vote, discuss, action.
Format 3: Mad / Sad / Glad
Emotionally honest format. Useful when a sprint was rough or when the team needs to process a difficult outcome.
Template:
MAD: What frustrated you?
- What made work harder than it needed to be?
- What would you change immediately if you could?
SAD: What disappointed you?
- What didn't go the way you hoped?
- What's a near-miss or missed opportunity worth noting?
GLAD: What are you happy about?
- What went better than expected?
- What are you grateful for this sprint?
How to run it: This format can surface stronger emotions than Start/Stop/Continue. Give extra time for the Sad and Mad columns. Facilitate carefully: acknowledge feelings before jumping to solutions.
Format 4: Sailboat (or Speedboat)
Visual format that works especially well for remote teams using Miro or FigJam.
Template:
Draw or display a sailboat with four elements:
WIND (what's pushing us forward): What's accelerating the team? What's giving us momentum?
ANCHOR (what's slowing us down): What's dragging on our speed or quality? What's holding us back?
ROCKS (risks ahead): What could go wrong in the next sprint or quarter? What risks do we see coming?
ISLAND (our destination): What are we aiming for? What does success look like in the next sprint?
How to run it: Works best with a visual tool. Post the image, give 5-7 minutes for silent addition of stickies to each element, then discuss the anchors and rocks first (they usually drive the action items).
The Action Item Problem
Most retro actions die because they lack an owner, a deadline, and a review mechanism. "We should improve our communication" is not an action item. It's a feeling.
A good action item has three parts:
- What: A specific, concrete change (not a vague aspiration)
- Who: One named person responsible (not "the team")
- When: A date by which it will be done or reviewed
Example: "Add a 15-minute weekly async update to the team Slack channel every Friday. Owner: [Name]. Starting this Friday."
At the start of every retro, spend 5 minutes reviewing the action items from the previous sprint. Did they happen? If not, why not? This single habit is more valuable than any format change.
For teams working on stakeholder communication as a retro action item, see the stakeholder management guide.
Time-Boxing the Session
For a 75-minute retro:
- 5 min: Review last sprint's action items
- 5 min: Set the stage / check-in question
- 15 min: Silent input (everyone writes their items)
- 10 min: Group similar items
- 20 min: Dot-vote and discuss top items
- 15 min: Agree on action items (owner + deadline)
- 5 min: Close / appreciation round
If you're short on time, cut the discussion time before cutting the action item time. The actions are the output.
When to Switch Formats
If your team gives the same feedback every sprint and action items don't move, the format isn't the problem. But switching formats can reset the energy and surface different perspectives.
Try the Sailboat when the team needs to look forward as much as backward. Try Mad/Sad/Glad after a hard sprint. Return to Start/Stop/Continue when you want speed.
Pay attention to participation. If the same two people dominate every retro discussion, the format may be too open-ended. Silent-first formats like dot-voting tend to level participation.
Remote Retrospective Tips
Async option: For distributed teams across many time zones, run the input phase async. Post the prompts in Slack or a shared doc 24 hours before the live session. People add their items on their own time. The live session focuses entirely on discussion and action items.
Tools: Miro and FigJam work well for visual formats (Sailboat especially). For text-heavy formats (4Ls, Start/Stop/Continue), a simple shared Google Doc or Notion page works fine. Don't over-engineer the tool choice.
Fatigue: Remote retros drain energy faster than in-person ones. Keep them tighter: 60 minutes is the ceiling for most remote teams.
For connecting retrospective improvements to your product planning cycle, see the complete guide to product roadmaps and prioritization guide for frameworks to triage the work that comes out of these sessions.