How long should a sprint retrospective be?

For a two-week sprint, plan 60–90 minutes. One-week sprints can comfortably fit a 30–45 minute retro. Time-box each phase, gather data, generate insights, decide what to do, so you always finish with action items.

How often should we change retrospective formats?

Rotate every two to four sprints. Repeating the same format invites autopilot answers. A new lens, Sailboat after Mad Sad Glad, for example, uncovers issues a tired team would have skipped.

What is the best retrospective format for a new team?

Start with Mad Sad Glad or Start Stop Continue. Both have low cognitive load, encourage emotional honesty without jargon, and produce clear action items fast.

How do you run a retrospective remotely?

Use a shared real-time board so everyone writes simultaneously, mask cards during writing to avoid groupthink, keep the call audio-on, and time-box every phase. A tool like Retro Harbour does this in the browser with no sign-up.

18 sprint retrospective ideas your team will actually enjoy

Same retro format every sprint? Engagement drops, action items dry up, and the meeting starts to feel like a checkbox. Below are 18 sprint retrospective ideas, battle-tested formats, prompts, and timings, you can run as-is or remix for your team.

By the Retro Harbour team · ~12 min read

What a sprint retrospective actually is

A sprint retrospective is the last ceremony of a Scrum sprint. The whole team, engineers, product, design, sometimes the Scrum Master and a manager, sits together for 30 to 90 minutes to inspect how they worked, not what they shipped. The output is a small number of concrete experiments to try in the next sprint.

That is the formal definition. The harder, more interesting question is: how do you make people actually care about the meeting on a Friday afternoon after a brutal sprint? Answer: change the lens.

How to pick the right format

Use this rough decision tree before each retro:

New team, or team in conflict? Start with Mad Sad Glad or Start Stop Continue. Low cognitive load, high emotional safety.
Stable team, business as usual? Rotate something fresh, Sailboat, 4Ls, or Starfish, to dodge autopilot answers.
End of a release or quarter? Reach for Timeline or Rose Thorn Bud to mine the longer arc.
Last few retros had no follow-through? Try Action-Only or Lean Coffee, both lean hard on decisions over discussion.
Team morale tanking? Appreciations or Hot Air Balloon rebuild trust before tackling the painful stuff.

The 18 retrospective ideas

1. Mad, Sad, Glad

Best for: emotional honesty, newer teams · Time: 45 min

Three columns capturing what frustrated, disappointed, or delighted you in the sprint. The emotional framing pulls out interpersonal and process pain the more clinical formats miss.

"What made you mad this sprint? What made you sad? What made you glad?"

2. Start, Stop, Continue

Best for: action-oriented teams · Time: 30–45 min

The most direct retro format. Every card is already a candidate behaviour change, so converting cards to action items is almost free.

"What should we start doing? Stop doing? Keep doing?"

3. The 4Ls: Liked, Learned, Lacked, Longed For

Best for: after a tough sprint or release · Time: 60 min

Four columns. "Liked" and "Learned" surface positives and growth; "Lacked" and "Longed For" surface gaps and aspirations without finger-pointing.

"What did you like, learn, lack, and long for this sprint?"

4. Sailboat (Speedboat)

Best for: strategic, visual teams · Time: 60 min

Draw a boat with an island goal, wind in the sails, anchors dragging it back, and rocks ahead. Each metaphor invites a different category of feedback, and the visual makes the discussion stick.

"What's the wind pushing us forward? What anchors are slowing us? What rocks are ahead?"

5. KALM: Keep, Add, Less, More

Best for: continuous-improvement teams · Time: 45 min

Less binary than Start/Stop. KALM acknowledges that good practices often need to be tuned in volume rather than switched on or off.

"What should we keep doing, add, do less of, do more of?"

6. Starfish

Best for: nuanced, mature teams · Time: 60 min

Five sections: Keep doing, More of, Less of, Start, Stop. The middle three categories invite gradient feedback, the difference between "stop" and "less of" is often where the real signal lives.

7. Rose, Thorn, Bud

Best for: end of release, quarter · Time: 45–60 min

Roses are wins, thorns are pain, buds are emerging opportunities. The "bud" column is the hidden gem, it forces the team to look forward, not just diagnose the past.

8. DAKI: Drop, Add, Keep, Improve

Best for: process-heavy teams · Time: 45 min

Like KALM but skews more decisive, "Drop" is harsher than "Less of" and forces the team to make real prune-or-keep calls on rituals that have stopped pulling their weight.

9. Lean Coffee

Best for: decision-heavy, time-boxed · Time: 45 min

Everyone proposes topics, the team dot-votes, and you discuss the winners in strict 5-minute chunks with a thumbs-up/down to extend. Almost zero facilitation overhead and brutally focused.

10. Appreciations

Best for: morale rebuild, after launch · Time: 20–30 min

Each person calls out a teammate and a specific moment they appreciated this sprint. Skip the "what should we improve" column entirely. Run this maybe once a quarter, it pays back in trust.

11. Timeline retrospective

Best for: long sprints, incidents · Time: 75–90 min

Draw a horizontal line for the sprint. Everyone plots events above the line (positive) or below (negative) with a short label. The visual reveals patterns, clusters of pain on Wednesdays, or every deploy day, that a column-based format would miss.

12. Hot Air Balloon

Best for: strategy resets · Time: 60 min

The balloon is the team. Sandbags drag you down (blockers), the burner lifts you up (energy sources), storm clouds threaten (risks), the destination is the goal. Same family as Sailboat with a slightly more aspirational framing.

13. The Three Little Pigs

Best for: quality and tech-debt focus · Time: 45 min

Three columns: house of straw (will collapse soon), house of sticks (shaky but standing), house of bricks (solid). A great way to talk about systems and code quality without the meeting devolving into a tech-debt rant.

14. Stop, Keep, Try

Best for: experiment-minded teams · Time: 30 min

Variant of Start Stop Continue with "Try" swapped in for "Start". The framing nudges the team toward small, time-boxed experiments instead of permanent commitments.

15. Action-Only retro

Best for: recovering from action-item drift · Time: 30 min

Skip discussion columns entirely. Each person writes one action item they think the team should commit to. Group, vote, pick the top two. Use when previous retros generated lots of conversation but zero follow-through.

16. Five Whys

Best for: after a clear incident · Time: 45 min

Pick the one biggest pain from the sprint and ask "why?" five times in a row. Each answer becomes the next question. Cuts through symptoms to the actual root cause, useful as a deep-dive after a regular retro flags something gnarly.

17. Superheroes and supervillains

Best for: creative, high-trust teams · Time: 45 min

What superpower would have rescued the sprint? Who was the supervillain? Lightweight, slightly ridiculous, and surprisingly good at surfacing systemic blockers a tired team would rationalise away.

18. One-word retrospective

Best for: 30-minute slot, energy check · Time: 15–30 min

Each person writes one word for the sprint. Reveal at the same time. The cluster of words becomes the agenda, pick the two or three loaded ones and dig in. Perfect for the sprint after a heavy retro, when the team doesn't need another full process pass.

Remote vs in-person retrospectives

Every format above works remote, but three things change:

Use a real-time shared board. Sticky notes on a Zoom whiteboard don't scale. A board where everyone writes simultaneously, like Retro Harbour, keeps the energy of a co-located room.
Mask cards during writing. When people see others' ideas as they type, you get groupthink and silent agreement. Hide cards until the reveal phase.
Time-box ruthlessly. Remote retros stretch. Use a visible countdown on writing, grouping, and voting phases.

Facilitation tips that quietly do the heavy lifting

Rotate the facilitator. The Scrum Master shouldn't own every retro. Rotating spreads ownership and changes the dynamic naturally.
Start with a one-word check-in. 60 seconds, sets the emotional baseline.
Cap action items at two. A retro that produces eight action items produces zero done by next sprint. Pick two, assign owners, agree on what "done" looks like.
Read last sprint's actions first. Five minutes. If you don't, the team learns the retro doesn't matter.
End with a one-sentence summary. The facilitator says back what was decided. If it sounds vague, refine before closing.

Run any of these formats in 10 seconds

Retro Harbour is a free, real-time retrospective board. No sign-up. Share a link. Templates for every format above are built in.

Start a free retro

Frequently asked questions

How long should a sprint retrospective be?: For a two-week sprint, plan 60–90 minutes. One-week sprints can comfortably fit a 30–45 minute retro. Time-box each phase, gather data, generate insights, decide what to do, so you always finish with action items.
How often should we change retrospective formats?: Rotate every two to four sprints. Repeating the same format invites autopilot answers. A new lens, Sailboat after Mad Sad Glad, for example, uncovers issues a tired team would have skipped.
What is the best retrospective format for a new team?: Start with Mad Sad Glad or Start Stop Continue. Both have low cognitive load, encourage emotional honesty without jargon, and produce clear action items fast.
How do you run a retrospective remotely?: Use a shared real-time board so everyone writes simultaneously, mask cards during writing to avoid groupthink, keep the call audio-on, and time-box every phase. A tool like Retro Harbour does this in the browser with no sign-up.
What if our retros never produce action items?: Try the Action-Only or Lean Coffee format for one sprint. Both formats force decisions over discussion. Cap action items at two, assign owners, and read them back at the start of the next retro.