Start / stop / continue cuts straight to behavioral change. What should we begin doing? What should we quit? What's working and must be protected? Three columns. Clear commitments. Snap it and ship the changes.
Use start / stop / continue when the team needs to make concrete behavioral changes rather than surface problems. It's action-oriented by design — every sticky is implicitly a proposal for what to do differently, not just an observation.
This format is particularly useful when the team knows what the problems are but struggles to commit to specific changes. The three-column structure forces each observation to become a proposed action.
Behaviors, practices, or habits the team should begin. These are things not currently happening that would improve the team's effectiveness. 'Start running five-minute pre-planning grooming sessions before each sprint.' Be specific — vague starts ('start communicating better') don't produce change.
Behaviors, practices, or habits the team should cease. Things currently happening that are creating friction, waste, or harm. 'Stop adding new requirements to stories mid-sprint without updating the sprint commitment.' The Stop column requires the most courage — naming what's not working means acknowledging that you've been doing it.
Behaviors, practices, or habits working well that must be explicitly preserved. This column is as important as the others — without it, improvement efforts often accidentally eliminate what's working. 'Continue the 20-minute end-of-day async update in Slack — it's replaced three status meetings.' Name what you're protecting and why.
Remind the team that every sticky should describe a behavior — something you can actually observe someone doing or not doing. Abstract observations ('morale is low') belong in a different retro format. In Start / Stop / Continue, if you can't describe a specific action, keep thinking.
Everyone writes stickies for all three columns. Encourage people to write at least two stickies per column — teams often underinvest in Continue, which means they risk losing what's working.
Post all stickies. Read each one aloud. Group related stickies within each column. Look for patterns: multiple people proposing the same start suggests a clear team need; multiple people calling out the same stop suggests a clear dysfunction.
Each person gets three votes. Distribute votes across all three columns — at least one vote per column. The highest-voted items in each column become the commitments.
For each top-voted item, define: what specifically will change, who will do it, and by when. Vague commitments don't produce change. Write the commitment in concrete behavioral terms.
Snap the board with BoardSnap. The AI reads all three columns and outputs the commitments as action items — tracked as open until the next retro confirms they've become habit.
Start / stop / continue produces the most immediately actionable retro output of any format. The three columns are already in action-item language — 'start doing X' is a task, not an observation. That directness makes the capture especially important: the stickies shouldn't disappear when the board is erased.
BoardSnap reads the three columns and outputs every commitment as a tri-state action item — open, in-progress, or done — ready to be checked at the next retro. The start commitments from last sprint become the 'did we actually do it?' check at the start of this one.
A standard sprint retro identifies what went well and what didn't. Start / stop / continue focuses entirely on behavioral change — every sticky is a proposal for what to do differently. The format is more action-oriented and less analytical. Use it when you've already diagnosed the problems and need to commit to specific changes.
Disagreement in the Stop column is useful information — it suggests the team has different views on what's causing friction. Use dot voting to surface the most widely-shared stops and debate the contested ones directly. If two people want to stop a practice that three others want to continue, that's a conversation worth having explicitly.
No. If the team can't name anything worth preserving, either the format is being applied too narrowly or the team genuinely has no positive practices — both of which are signals worth examining. Prompt the team: 'What do we do that other teams don't, and that makes us better for it?' There's always something.
Start each retro by reviewing the start and stop commitments from the previous session. If a 'stop' item from three retros ago is still appearing, the team hasn't actually stopped it — and that's the first conversation. BoardSnap's action-item tracking makes this review automatic.
No exporting, no transcription. Snap the board, get the action plan.