10 Retrospective Formats, and When to Actually Use Each One

Most teams pick one retrospective format and stick with it. Usually Start / Stop / Continue. It works well enough that they never reconsider, until the sessions start to feel stale and someone suggests "mixing it up" without knowing what they are mixing it up with.

Format choice is not trivial. The structure you use shapes what gets said. A format that centers emotions surfaces different content than one that centers actions. A format that works for a new team can flatten a mature one. Running the same structure indefinitely trains the team to give the same kind of answers indefinitely.

This is a guide to all ten, organized by what each one is actually good at — and where it falls short.

The Workhorses

These two formats run most retrospectives in the industry. There are good reasons for that.

Start / Stop / Continue

The default for most teams. Three columns: what we should start doing, what we should stop doing, what is working and should continue.

The structure is productive because it generates concrete candidates for action items in every column. Start and Stop cards are already halfway to a commitment. Continue cards validate what not to change, which matters — teams that only surface problems and never reinforce strengths create a distorted picture of reality.

Use it when: the team needs directional clarity. New teams, new projects, any sprint where the signal-to-noise ratio was low.

Skip it when: the team has run it enough times that cards are becoming rote. When everyone already knows what the Start and Stop columns will say, the format is no longer generating insight — it is generating familiarity.

Good / Bad / Improve

A compressed version of Start / Stop / Continue. Good, Bad, Improve instead of Continue, Stop, Start — the framing is evaluative rather than directive.

The difference is subtle but real. "What was bad?" is a different question than "What should we stop?" The evaluative frame opens up cards that are complaints without a clear action attached. That is useful when the team needs to surface frustration before they can discuss fixes. It is less useful when the goal is direct output.

Use it when: the team needs to vent before they can problem-solve. Good for emotionally charged sprints or teams coming out of a difficult delivery.

Skip it when: the session needs to produce action items. The Improve column gets vague faster than the directional columns in Start / Stop / Continue.

Formats That Center Emotion

Run these when the team's morale or psychological state is the thing that needs attention.

Mad / Sad / Glad

Three columns mapped to emotional states rather than categories of work. What frustrated you, what disappointed you, what made you happy.

This format does something the others do not: it legitimizes emotion as data. Teams that run structured formats exclusively sometimes develop a culture of over-professionalism in retrospectives — cards are written clinically, feelings are abstracted into process observations. Mad / Sad / Glad breaks that pattern. A card that says "I was frustrated that we found out about the scope change the day before the deadline" is more honest than "better communication around scope changes."

The emotional framing also tends to surface things that would not appear in a category-based format. The Sad column in particular — disappointment is a specific emotion that points to unmet expectations, which is often where the real problem is.

Use it when: the sprint was emotionally difficult, the team has been under pressure for multiple cycles, or you suspect that the real issues are not making it onto the board in more structured formats.

Skip it when: the team is not psychologically safe enough for it yet. This format only works if people trust that expressing frustration will be received constructively. In a new team or a team with existing tension, it can go sideways.

Hopes & Fears

Three columns: what are you hoping for next sprint, what are you worried about, what will we commit to as a team. Unlike the other formats, this one is forward-looking — it belongs at the boundary between sprints rather than purely in the past.

The Fears column is unusually valuable. Getting a team to name what they are worried about before the sprint starts, in a structured setting, converts implicit anxiety into explicit risk. Items in the Fears column are often the most actionable things in the entire session.

Use it when: transitioning into a high-stakes sprint, starting a new project phase, onboarding new team members, or whenever the team's energy toward the next sprint needs a reset.

Skip it when: the team is mid-cycle with no inflection point. Running Hopes & Fears in an ordinary sprint without a transition context can feel forced.

Formats That Improve Precision

These formats are more granular. They generate better action items than the workhorses, at the cost of more cognitive load.

Starfish

Five columns: Stop, Less of, Keep, More of, Start. An expansion of Start / Stop / Continue that adds a gradient.

The Less of and More of columns are the differentiator. They capture degrees of change rather than binary switches. "We should do less of synchronous standups" is a more nuanced — and more actionable — card than "We should stop standups." Continuous improvement rarely involves eliminating things entirely; it usually involves calibrating them.

Starfish tends to generate the most specific action items of any format. The tradeoff is complexity. Five columns slow the session down and can overwhelm teams that are not used to it.

Use it when: the team is mature enough to be precise, or when a previous retro identified a problem but produced too-vague action items to resolve it. The action item clarity problem is partly a format problem — Starfish helps.

Skip it when: the team is new, the sprint was chaotic, or time is short. The session needs room to breathe with this format.

DAKI (Drop / Add / Keep / Improve)

Drop, Add, Keep, Improve. Similar in spirit to Starfish but without the gradient columns — this one is binary, which makes it faster.

The Drop column is notably different from Stop. "Stop" implies something that should end. "Drop" is more decisive — it implies something that does not belong at all. That framing tends to produce more honest answers about practices the team has been tolerating rather than finding useful.

Use it when: the team has accumulated technical debt in its processes — meetings that no one values, ceremonies that serve no current purpose, habits that outlived their context. DAKI is good at clearing that kind of clutter.

Skip it when: the team's main need is emotional processing or strategic thinking. DAKI is an operational format; it does not stretch well into those spaces.

Formats That Encourage Reflection

These formats slow the session down in productive ways. They are better for depth than breadth.

4Ls (Liked / Learned / Lacked / Longed For)

Four columns: what did you like, what did you learn, what was missing, what did you wish for.

The Learned column is what makes this format different. Asking teams what they learned — not just what went wrong — changes the framing from retrospective to reflection. It is particularly effective for teams that are in a growth phase, onboarding new technology, or navigating a significant change. It also tends to surface positive signal that other formats miss: insights that came out of failure, processes that worked unexpectedly well, skills the team did not know it had.

Longed For is the other unusual column. It opens up wishes rather than complaints, which can surface aspirational thinking that would never appear in a "what went wrong" column.

Use it when: the sprint involved significant learning — new tech, new domain, new team members. Also useful when the team's retros have become complaint-heavy and need a reset toward a more balanced frame.

Skip it when: the sprint was a grind and the team needs to surface operational problems. The reflective framing can feel detached when the issues are concrete and frustrating.

Rose / Bud / Thorn

Rose (what went well), Thorn (what was a challenge), Bud (what opportunity or idea emerged).

The Bud column is the point of differentiation. It asks the team to look forward from the session — not just what was good or bad, but what possibility exists that we have not pursued. This is the most generative of the three columns and often the most surprising. Buds can come from any direction: a conversation that happened in passing, a tool someone tried, a pattern noticed in the data.

Use it when: the team has been in execution mode for several sprints and needs to surface new ideas. Also good for teams that are high on delivery but low on innovation — the Bud column creates structured permission to think differently.

Skip it when: the sprint was a crisis. When the team is in damage-control mode, the forward-looking column can feel tone-deaf.

Strategic Formats

Use these at natural inflection points — end of quarter, project launch, major delivery milestone. Running them mid-cycle without a strategic context usually produces shallow results.

Sailboat

Wind (what is driving us forward), Anchor (what is slowing us down), Rocks (what risks lie ahead), Island (what is our goal).

The metaphor is the feature. Teams that respond poorly to clinical formats often engage more with this one — the visual framing gives them a different way in. The Island column forces a goal articulation that rarely happens in standard retrospectives: not just what is wrong, but what are we actually trying to reach.

The Rocks column is underused. Most retro formats are backward-looking; Rocks is explicitly a risk register for the next sprint. Getting the team to name obstacles before they encounter them tends to produce better preparation than discovering them mid-delivery.

Use it when: the team needs a strategic reset, or when there is genuine uncertainty about direction. Also useful for teams that find standard formats dry — the metaphor increases engagement for some team compositions.

Skip it when: the team is mature and precise. The metaphor that helps some teams can feel like an obstacle to teams that prefer direct language.

SWOT

Strengths, Weaknesses, Opportunities, Threats. The business strategy framework applied to team retrospectives.

SWOT works when the scope extends beyond a single sprint. It is designed for assessing position — which makes it excellent at quarter boundaries, project kickoffs, or any moment when the team needs to think about its trajectory rather than just the last two weeks.

The Opportunities and Threats columns are the reason to use it. They pull in external context — the competitive environment, upcoming changes to the roadmap, dependencies on other teams — that purely internal formats miss.

Use it when: the team is planning beyond the next sprint, running a program-level retro, or needs to explicitly incorporate external factors.

Skip it when: it is a routine sprint retro. SWOT is a strategic tool; using it for operational review produces either shallow results or analysis that is too abstract to act on.

On Rotating Formats

The case for variety is not just about keeping sessions fresh. Different formats surface different types of signal. A team that only runs Start / Stop / Continue will consistently generate one type of card — directional, operational — and consistently miss emotional undercurrents, learning moments, and strategic risks.

A practical rotation: run your workhorse most of the time, switch to an emotional format after difficult sprints, switch to a reflective or strategic format at quarter boundaries. Let the team's history guide it. Teams that track which formats correlated with better follow-through over time have a data-driven basis for those decisions — rather than guessing.

The format is a lever. Knowing which lever to pull and when is one of the most underrated facilitation skills in the room.

Retromate includes all ten formats and recommends which one to use based on your team's history — completion rates, sentiment trend, and which formats drove the best follow-through for your specific team. Free for teams of 5 or fewer.