Retrospective Activities

What it is

Retrospectives are the meeting where a team converts experience into adjustment. They're the only routine slot most teams have where the agenda is 'how are we doing this' instead of 'what are we doing.' The format isn't decorative — different formats produce different conversations. Start/Stop/Continue produces commitments. 4Ls produces understanding. Sailboat produces system-level diagnosis. Five Whys and Fishbone aren't retros at all — they're techniques you run inside one. Pick the one that matches the question you need to answer, and rotate to surface what the regular format misses.

When to use

End of every sprint, end of every project, after a release that didn't go to plan, when somebody on the team says 'we should talk about how this is going.' The cadence matters less than the rule that you actually run them — the teams that skip retros when the week is hard are the teams that need them most. Skip the formal retro for one-off meetings or short status calls (use Plus/Delta instead) and skip the big four-quadrant variants for a routine sprint where SSC will do the job in twenty minutes.

Which one to pick

Sprint cadence, healthy team, you want action items

Start, Stop, Continue. The default for a reason — fastest path from cards to commitments.

Stop feels too confrontational for the team

KALM. Less is Stop with the volume turned down. Run it and earn the upgrade to SSC.

End of a sprint where the team hit the goal but the path was ugly

DAKI. Splits Keep from Improve — recognising something can work AND need refining. Force a name and date on each Improve card or drop the column.

End of project or quarter, you want lessons not actions

4Ls. The Lacked column is where the unsaid stuff lives — budget half the meeting there.

Difficult sprint, emotion is the signal you need to hear

Mad/Sad/Glad. Lead with Mad — anonymous. Glad-first manages the room; Mad-first works.

Mixed-discipline workshop or non-software team

Rose/Bud/Thorn. Engineering teams should pick something else — the gardening metaphor lands soft.

Team can feel a structural problem but can't say it

Sailboat. The metaphor lets people name the problem without pointing at a person.

You want anchors only — no positive column to soften it

Speedboat. Otherwise run Sailboat — Sailboat is Speedboat with a wind column.

More topics than time, agenda is contested

Lean Coffee. The five-minute timer and Roman vote are the format.

Post-incident with one likely mechanical cause

Five Whys, inside the post-incident retro. Twenty minutes, not the whole meeting. Ask 'how' not 'why' if a person is at the end of the chain.

Post-incident with multiple contributing causes (most software incidents)

Fishbone, inside the post-incident retro. Localise the categories for software — the 6 Ms are factory-floor relics.

End of project or end of quarter with enough events to plot

Timeline retrospective. The mood line is the artefact — pre-populate the events as facilitator. Don't run it for a sprint.

You want a journal entry, not a retrospective

What Went Well / What Didn't. It's the format with no action column. Pick almost anything else if you ship software.

Common mistakes

• Picking a four-column variant by random — they're not interchangeable. SSC pushes for action, 4Ls for understanding, What Went Well does neither.
• Treating Five Whys or Fishbone as a retrospective format. They're techniques inside a retrospective on a specific incident. Run them for twenty minutes, not the whole hour.
• Letting the metaphor become the meeting. Sailboat, Speedboat, gardening prompts — the visual is a workaround to get the team talking, not the conversation itself.
• Running the same format every sprint for a year. The format shapes the conversation; rotate to surface things the regular format misses.
• Skipping the retro when the week was hard. That's the week the retro most needed to happen.
• Generating cards and stopping there. Without a named owner and a date on each commitment, the retro is theatre.

All retrospective activities