5 Retrospective Formats That Actually Work (and When to Use Each)

Why the format you choose matters more than you think

The retrospective is supposed to be the engine of continuous improvement. In practice, it's often the meeting that slowly loses its edge. Same format every sprint, same people saying the same things, same action items that never quite get done.

The fix isn't usually more facilitation skill – it's format variety. Different formats surface different kinds of data. Some are better for trust-building, others for structured problem-solving. Picking the right one for the moment keeps things fresh and keeps the insights meaningful.

Here are five formats worth having in your toolkit – with concrete timings and a clear run-of-show for each.

1. Starfish – when you want more nuance than Start/Stop/Continue

Best for

Teams that are past the basics and want a richer conversation. The five categories (instead of three) allow for much more precise feedback – especially useful for mature teams stuck in a rut.

Agenda (90 minutes)

Set the Stage (10 min)
Check-in question: "On a scale of 1–10, how are you feeling coming into today's retro?" One sentence each. The goal is presence, not performance.

Gather Data (20 min)
Draw or display a starfish with five arms. Each arm is a category:

• Keep doing – What's working well and should stay
• More of – What's helping but not happening enough
• Less of – What's costing the team more than it's worth
• Stop doing – What should be cut entirely
• Start doing – New behaviours or practices to introduce

Give the team 10 minutes to add sticky notes silently, then 10 minutes to cluster and explain.

Generate Insights (25 min)
Dot vote on the most important themes. Discuss the top 3–4 items: what's the underlying root cause? What would need to change?

Decide on Actions (25 min)
Turn insights into concrete action items. One owner per item. No more than 2–3 actions per retro – better to do fewer things properly.

Close (10 min)
Read back the actions. Quick appreciation round – one thing you noticed about the team today.

Facilitator tip

The "Less of" category often surfaces more useful data than "Stop doing" because it's less confrontational. If someone raises something in "Stop doing" that's politically sensitive, have them describe the impact first before jumping to the category.

2. 4Ls – when the team needs to dig into learning

Best for

Teams after a high-stakes sprint, a product launch, or any period with significant wins and losses. The focus on learning makes this one especially good after things went wrong – it reframes failure as data rather than blame.

Agenda (90 minutes)

Set the Stage (10 min)
Open question: "What's one word that describes this sprint for you?" Go round the room.

Gather Data (25 min)
The four quadrants:

• Liked – What did we appreciate or enjoy?
• Learned – What did we discover (about the technology, process, domain, team)?
• Lacked – What was missing that would have made us more effective?
• Longed for – What do we wish we had, even if it wasn't realistic this sprint?

10 minutes silent writing, 15 minutes group walk-through.

Generate Insights (25 min)
Focus especially on the overlap between "Lacked" and "Longed for" – that's often where the most actionable insights live. Vote on top themes.

Decide on Actions (20 min)
For each action: what specifically will change, who owns it, when will we review it.

Close (10 min)
Exit ticket: "One word for what you're taking away."

Facilitator tip

"Longed for" often stays empty because it feels abstract. Prompt the team: "If you could have had one thing this sprint that you didn't – what would it be?" Even aspirational answers illuminate systemic constraints.

3. Mad-Sad-Glad – when you want to start with feelings, not facts

Best for

Teams where trust is still developing, or after a difficult sprint where emotions are running high. Starting with how people feel (rather than what happened) creates space for honest conversation before you move into analysis.

Agenda (90 minutes)

Set the Stage (15 min)
Check-in: "What made you mad, sad or glad this sprint?" Ask one person to model vulnerability by going first – ideally the Scrum Master or someone with psychological safety to spare.

Gather Data (20 min)
Three columns:

• Mad – What frustrated you or slowed you down?
• Sad – What disappointed you or didn't go the way you hoped?
• Glad – What energised you or went better than expected?

Silent sticky notes first (8 min), then share and cluster (12 min).

Generate Insights (25 min)
Look for patterns across columns. Mad often points to process issues; Sad can reveal unmet expectations; Glad signals what's worth protecting. Vote on the 3 most important themes.

Decide on Actions (20 min)
Focus on Mad and Sad first – that's where the energy for change usually lives.

Close (10 min)
End with the Glad column – read back the wins before closing the retro.

Facilitator tip

If the Mad column fills up fast and Glad stays empty, that's itself a data point worth acknowledging. Don't force balance – but do make sure every item in Mad has a path to either an action or an acknowledged systemic constraint.

4. Sailboat – when you need a metaphor that builds energy

Best for

Teams that respond well to visual metaphors, or any time you want a retrospective that feels energising rather than draining. Works especially well for cross-functional teams with members who aren't all deep Scrum practitioners.

Agenda (90 minutes)

Set the Stage (10 min)
Draw (or show) a sailboat with four elements: wind (what's pushing us forward), anchor (what's holding us back), rocks (risks we can see ahead), island (our goal or destination). Walk through the metaphor briefly.

Gather Data (20 min)
Silent sticky notes for all four areas (12 min), then share and cluster (8 min).

Generate Insights (30 min)
Prioritise by asking: which anchors hurt the most? Which rocks could derail us if we don't address them? Use the wind items to identify what to protect and amplify.

Decide on Actions (20 min)
One action per major anchor or risk. Keep the island/goal visible throughout – it contextualises why the actions matter.

Close (10 min)
Read back the actions and recap the destination. End with: "What wind do we want to carry into next sprint?"

Facilitator tip

The rocks (risks) section is often neglected in favour of the anchor (team-internal issues). Give it equal attention – surfacing risks proactively is one of the highest-value things a retrospective can do.

5. Timeline – when you want to reconstruct the whole sprint

Best for

Teams at the end of a particularly complex sprint, during a project post-mortem, or when you sense that different team members had very different experiences of what happened. The Timeline format gives everyone a shared, factual picture before you start interpreting.

Agenda (120 minutes)

Set the Stage (10 min)
Remind the team of the sprint dates and any significant external events (releases, incidents, holidays). Explain that you'll build the timeline together.

Build the Timeline (30 min)
On a whiteboard or shared digital board, draw a horizontal timeline for the sprint. Each person adds their sticky notes in chronological order – events, decisions, interactions, anything significant. Encourage a mix of objective facts and subjective experiences.

Identify Themes (25 min)
Look for patterns: where did things slow down? Where did energy spike? Were there points where the team drifted out of sync? Add emotion markers (happy/unhappy face) next to key moments.

Generate Insights (25 min)
Focus on the 2–3 moments the team agrees were most significant. What caused them? What would you do differently, or repeat?

Decide on Actions (20 min)
As always: specific, owned, time-bound. No more than 3.

Close (10 min)
Take a photo of the timeline. It's a useful artefact for onboarding new team members or reviewing patterns across quarters.

Facilitator tip

Leave enough physical/digital space on the timeline for the full sprint – teams consistently underestimate how many events there were. Two metres of wall space for a two-week sprint is not too much.

Choosing the right format: a quick guide

Situation	Recommended format
Team is bored of Start/Stop/Continue	Starfish
High-stakes sprint or launch, lots to learn	4Ls
Trust issues, emotionally heavy sprint	Mad-Sad-Glad
Cross-functional team, need energy	Sailboat
Complex sprint, very different experiences	Timeline

Making any format work: three universal principles

Timebox everything

A retro that runs long is a retro people start dreading. Use a visible timer for each phase. If discussion is still going, explicitly ask: "Do we want to extend this phase by 5 minutes at the cost of action planning?" That question usually moves things along.

Separate data from interpretation

The Gather Data phase should be as neutral and factual as possible. Interpretation happens in Generate Insights. If someone jumps to conclusions in the data phase, gently redirect: "Let's park the "why" for now – what else happened?"

End with fewer, better actions

The most common retrospective dysfunction isn't a bad format – it's too many action items that nobody follows up on. Three specific, owned actions are worth ten vague commitments. Close every retro by reviewing last sprint's actions before generating new ones.

Plan your retrospective on a visual timeline

All five formats work better when you plan them properly – with real time blocks for each phase. Sessionplan's retrospective planner lets you build your agenda visually, share it with co-facilitators and run the session with Live Mode to keep things on track.

If your team also does Sprint Planning, try the ready-made Sprint Planning template to structure that session too.