Introduction
Retrospectives are where teams convert experience into improvement. Done well, they are the highest-leverage practice in Agile — a regular, structured reflection that compounds learning over months and years into measurable change in how the team works. Done poorly, they devolve into vent sessions, perfunctory check-ins, or repeat performances of the same format until nobody bothers to surface anything new.
The retro is not really about the meeting. It is about what changes between now and the next one. Every format below exists to make that more likely — to draw out the real signal, to surface what isn't yet visible, and to leave the team with a concrete experiment to run rather than a list of complaints.
Topics
Classic Structures
The reliable workhorses — Start/Stop/Continue, Mad/Sad/Glad, 4Ls, Plus/Delta. Useful as defaults, easy to overuse.
Read →Emotional Techniques
Formats that make room for what teams feel about their work — energy maps, emotional timelines, sailboat retros.
Read →Creative / Visual Techniques
Drawing, metaphor, and prop-based formats that bypass over-rehearsed discussions and produce fresh observations.
Read →Analytical Techniques
Root-cause-oriented formats — 5 Whys, fishbone diagrams, anchor incidents — when the team needs to dig deeper than feelings.
Read →Safety Builders
Opening formats for new, shaken, or distributed teams that need to feel safe enough to surface real issues at all.
Read →Structured Conversation Formats
Fishbowl, Lean Coffee, Liberating Structures-style retros for topics that require careful conversational scaffolding.
Read →Making Retros Worth the Hour
The single best predictor of a useful retro is variation. A team running the same format every sprint stops noticing anything new — the same five complaints surface, the same vague action items get noted, and the meeting becomes the cost of doing Agile rather than the engine of getting better at it. Rotate formats, rotate facilitators, vary the focus.
The second best predictor is follow-through. A clear, owned, time-boxed experiment is worth more than three pages of insights nobody acts on. Pick one thing to try. Bring it to the next retro and ask whether it worked. That is the practice that compounds.
Back to Hubs