Start Stop Continue Retrospective: The Simplest Retro Format That Works
Start Stop Continue is the most widely used retrospective format in agile — and the easiest to run well. Three columns, three questions: what should we start doing, what should we stop doing, and what should we continue doing? Here's how to facilitate it so it produces actions, not just complaints.
What is a Start Stop Continue retrospective?
The team reflects on the last sprint (or project, or quarter) and sorts observations into three columns:
- Start — new practices, experiments or behaviors the team wants to try.
- Stop — things that are hurting the team or adding no value and should end.
- Continue — things that are working well and should be protected and kept.
Because every note is already phrased as a change (or a keep), the format moves naturally into action items — which is exactly why it has survived for decades.
Example notes for each column
Start: "Start writing acceptance criteria before sprint planning." / "Start demoing unfinished work to one user every week." / "Start a 15-minute mid-sprint check-in."
Stop: "Stop booking meetings over lunch." / "Stop merging pull requests without a second review." / "Stop letting one person be the only contact for the payment vendor."
Continue: "Continue the Friday demo — it keeps stakeholders close." / "Continue pairing on risky changes." / "Continue posting blockers in the channel early; it worked twice this sprint."
Notice the pattern: specific, observable, and about the team's own behavior — not vague wishes or blame.
How to run it: step by step (~35 minutes)
- Set the stage (5 min) — State the period you're reviewing and the prime directive: everyone acted with the information they had; this is about the system, not people.
- Silent brainstorming (8 min) — Everyone adds notes to all three columns at the same time, one idea per note. Silence prevents the loudest voice from anchoring the room.
- Group and read (7 min) — Cluster duplicates, read each cluster aloud, and clarify anything ambiguous in the team's own words.
- Discuss and vote (10 min) — Dot-vote (3 votes per person) to find the highest-impact items, then discuss the top two or three: root cause, and what "good" looks like.
- Actions (5 min) — Convert the top items into 1–3 actions with an owner and a date. A Start becomes an experiment with a review date; a Stop needs an explicit replacement habit or it quietly returns.
When to use Start Stop Continue
- Your team is new to retrospectives and needs the simplest possible structure.
- You want actions fast — the format is deliberately action-shaped.
- Energy is low and a short, focused retro beats an elaborate one.
Choose a different format (Sailboat, 4Ls) when the team is bored of it, when you need forward-looking risk discussion, or when reviewing a long period where learning matters most.
Common mistakes to avoid
- The Stop column becomes blame. "Stop being slow in reviews" names people; "Stop merging without a second review" names a process. Facilitate toward process.
- Vague starts. "Communicate better" is not an action. Push for observable behavior: "Post a daily async update in the channel by 10 a.m."
- A Continue column full of everything. If everything continues, nothing is being chosen. Ask what the team would protect if time got tight.
- Too many actions. Three committed actions beat ten wishes. Park the rest visibly for next time.
Frequently asked questions
How long does a Start Stop Continue retro take?
About 30–40 minutes for a typical sprint retro with 4–10 people.
How is it different from the Sailboat retrospective?
Start Stop Continue is action-first and looks mainly at the recent past. Sailboat adds the future: its rocks quadrant covers risks ahead, and the island makes the team align on the goal.
Can we run it remotely?
Yes — it translates perfectly to a shared online board where everyone writes simultaneously. Happy Backlog's free retrospective board includes Start Stop Continue; start a room, share one link, no signup.
What if the Stop column is empty?
Probe gently: "What cost us time this sprint that nobody chose?" An empty Stop column usually means low safety, not a perfect sprint.