Start Stop Continue retrospective

Three questions any team can answer: what should we start doing, what should we stop, and what is worth keeping? Here is how to turn the answers into changes that last longer than a sprint.

Start

Things that we think would have a positive impact on our team that aren't already implemented

Stop

Things within our team workflow or process that aren't helping us to achieve goals and should be stopped

Continue

Things that already worked well in the previous iteration and should stay in the workflow to ensure future success

Use this template — it's free

What a Start Stop Continue retrospective is

Start Stop Continue asks the team to look back at the last sprint, month or project and sort its ideas into three piles: things to start doing, things to stop doing and things to continue. Every card is a proposal about behavior, not a report of what happened. The observation "code review takes too long" becomes a card such as "pick up review requests older than a day first thing at stand-up".

That is what sets it apart. Formats like What Went Well or Mad Sad Glad collect observations and leave the team to work out what to change. Start Stop Continue skips straight to the change. You give up some diagnosis and gain speed: by the time the cards are on the board, half of your action items are already drafted.

Why such a plain format keeps working:

• Nobody needs it explained. Three verbs, no metaphor. New joiners, stakeholders and people who dislike retros all know what to write within seconds.
• It talks about habits, not people. "Stop merging on Friday afternoon" describes something the whole team does. It is easier to raise than a complaint about a colleague, and easier to fix.
• It protects what works. The Continue column makes the team name the practices it would miss. Without it, a good habit can quietly disappear under the next deadline because nobody said it mattered.
• It fits in 45 minutes. Three columns mean fewer cards to read and group, so more of the meeting goes to the topics that deserve it.

When to use it, and when not to

Start Stop Continue fits when the team broadly knows what happened and wants to decide what to do about it:

• regular sprint retros for a team that works reasonably well and wants steady, small improvements;
• one-week sprints or busy release weeks, when 45 minutes is all you can get;
• a first retro with a new team or a new facilitator;
• a few weeks after a process change — a new branching model, a new on-call rota, a new planning cadence — when you want a quick verdict on what to keep;
• a review with people from outside the team, such as a client or a partner team, where plain words matter more than a clever format.

Pick something else when:

• Something went wrong and nobody agrees why. Jumping to "start" and "stop" before the team understands a missed release produces fixes for the wrong problem. Lay out the facts first with What Went Well, or use the 4Ls if you also want to capture lessons.
• The mood is the problem. If people are exhausted or frustrated, a tidy list of process tweaks lands badly. Mad Sad Glad gives feelings a place first.
• Every card says "continue, but…". Mature teams often find that most of their practices are fine, just overdone or underdone. The Starfish retrospective adds More of and Less of for exactly that.
• The same cards come back every sprint. The format has stopped producing new thinking. Rotate it out for a few sessions.

The three columns

In ProstoRetro the template opens with three columns in this order: Start, Stop and Continue. Here is what belongs in each, the question that tells it from its neighbor, and cards a real product team might write.

Start: what we don't do yet but should

New habits, experiments, tools or agreements the team believes would help. The test: are we doing this today? If not, it is a Start card. If you do it now and then and want it to become the rule, it still belongs here — write the rule, not the wish.

Good Start cards are small enough to try within one sprint. "Write better tests" is a wish; "add a failing test for every bug before fixing it" is something you can check at the next retro.

• Show unfinished work to design on Wednesdays, not only at the sprint review
• Add a "how to test this" note to every pull request
• Look at the error dashboard together for 15 minutes every Monday
• Tag support tickets by feature, so we can see where complaints come from

Stop: what costs more than it gives

Habits, meetings, rules and shortcuts that are not helping the team reach its goals. The test: would anyone miss it if it disappeared next week? If nobody would, it belongs here. If someone would miss it but there is simply too much of it, write the smaller version as a Start card ("one status meeting a week instead of three").

This is the column where people hold back, because many Stop cards describe something a colleague or a manager asked for. Keep them about the practice ("pinging people for reviews in direct messages") rather than the person.

• Merging straight to main on Friday afternoons
• Pulling work into the sprint after planning without dropping something else
• Status meetings that repeat what is already on the board
• Rerunning the flaky checkout test until it passes instead of fixing it

Continue: what works and should survive

Practices that helped in the last iteration and should stay. The test that separates it from Start: did we actually do this, and did it help? A habit the team only intends to adopt does not go here, however good it sounds.

It is tempting to treat this column as a pat on the back. It is more useful as a list of things to protect. When the next deadline arrives, Continue cards are the habits most likely to be skipped "just this once"; writing them down turns that into a decision rather than an accident.

• Pairing on the first ticket in an unfamiliar part of the codebase
• Product writes acceptance criteria before refinement, not during it
• Rotating who runs stand-up, so it doesn't become a report to the lead
• A feature flag on every release: rolling back on the 12th took two minutes

Use this template — it's free

A 45-minute Start Stop Continue agenda

Timed for five to nine people; add five minutes of discussion for every three extra.

1. Warm-up, 4 minutes. One quick question to get every voice into the room — choose one from the icebreaker questions list. ProstoRetro can open the board with an icebreaker round that picks the question and who answers it.
2. Review last time's actions, 4 minutes. What did you agree to start or stop at the last retro, and did it happen? ProstoRetro starts the retro with the open action items, so this does not depend on anyone's memory.
3. Silent writing, 8 minutes. Everyone writes cards in all three columns, one idea per card. Ask for at least one Continue card each; it is the column people skip.
4. Read and group, 6 minutes. Read the cards out column by column and merge duplicates.
5. Vote, 3 minutes. Everyone puts their votes on the topics they most want to act on.
6. Discuss, 14 minutes. Take the top three topics. The question depends on the column: for Start, "what is the smallest version we could try next sprint?"; for Stop, "what replaces it, if anything?"; for Continue, "what could make us drop this, and how do we prevent that?"
7. Agree on actions, 4 minutes. One owner and one due date per action. Two or three actions are plenty.
8. Close, 2 minutes. Read the actions out loud and check that every owner agrees to theirs.

Short on time? Skip the warm-up, give writing six minutes and discuss only the top two topics; the whole retro then fits in half an hour. Or give the board a 45-minute meeting length when you create it: ProstoRetro divides the time between the phases and runs each timer itself.

Running it with a remote team

• Keep cards hidden while people write. On a shared board, the first visible Stop card sets the agenda and everyone else writes a variation of it. In ProstoRetro cards stay hidden from others until grouping, so you get independent opinions instead of one opinion and several echoes.
• Remind people that cards are anonymous. Anonymity is on by default, and each person can decide per card. It matters most in the Stop column: "stop booking meetings over lunch" is easier to write when your name is not on the card.
• Group by practice, not by column. "Write release notes as we go" in Start and "writing release notes the night before" in Stop are one topic; merge them so their votes add up. ProstoRetro can suggest groups of similar cards, and you keep or undo each suggestion.
• Vote for change, not for agreement. Say before voting that a vote means "I want us to act on this", not "I agree this is annoying". Votes are anonymous, so a junior developer can back a card that challenges a senior's habit.
• Write actions during the discussion. As soon as the team agrees on something, add it as an action item with an owner and a due date. If the call goes quiet, ProstoRetro can suggest actions based on the topic. Owners get an email reminder a week after the retro and again before the due date.
• Make guests welcome. If the room's access is set to Anyone, a stakeholder or a colleague from another team joins with just a name, without creating an account.

Variations

Aim it at one topic

Instead of the whole sprint, point the three questions at a single area: code review, the on-call rota, how the team works with design. A narrow scope gives sharper cards, and the session can end with a new working agreement for that one area.

Feedback for a team lead or a role

The same three questions work for feedback outside a retro: what should the team lead, the product owner or the QA role start, stop and continue doing? It works because it asks for behavior, not personality. Do it only with the person's agreement, keep the cards anonymous, and ask them to write their own cards first — the difference between how they see their role and how the team sees it is the useful part.

Add more columns

Some teams add a fourth column such as Change for things that should stay but work differently. Others split Continue into Keep, More of and Less of — that version is the built-in Starfish. For any other mix, create a custom template in ProstoRetro: your own name and up to five columns with your own descriptions and colors.

Common mistakes

• A Start column that reads like a wish list. Fifteen new practices will not all start. Vote, pick one or two, and make them small enough to check at the next retro.
• Solutions without problems. "Switch to a new test framework" might be the right answer to the wrong question. In discussion, ask which pain each Start card is meant to remove before you agree to it.
• An empty Continue column. A board with nothing worth keeping tells the team that nothing they do works. That is rarely true, and it makes the next retro harder to start.
• Names on Stop cards. Rewrite them as practices before the discussion. The person will recognize themselves anyway; the team can now talk about the habit instead of the colleague.
• Never checking what you stopped. Stopped habits creep back. Open each retro with the previous actions and ask whether the thing really stopped.
• Running it every sprint for a year. Cards get shorter and vaguer until the board says "fewer meetings, more focus" every time. Change the format every few sessions, or keep the columns and change the topic.

FAQ

What are the three questions in a Start Stop Continue retrospective?

What should we start doing that we don't do yet? What should we stop doing because it doesn't help? What should we continue because it works? Each card answers one of them.

How long does a Start Stop Continue retrospective take?

About 45 minutes for a team of five to nine. Trimmed to the top two topics, it fits in half an hour.

What is the difference between Start Stop Continue and What Went Well?

What Went Well records what happened and leaves the team to decide what to change; Start Stop Continue records the changes directly. Use What Went Well when you still need to understand a sprint, and Start Stop Continue when you already do and want to act.

Is Start Stop Continue the same as the Starfish?

No. The Starfish keeps Start, Stop and Keep and adds More of and Less of, which suits teams whose practices mostly work but need adjusting. Start Stop Continue is the quicker, coarser version.

Can I run a Start Stop Continue retrospective online for free?

Yes. Start Stop Continue is one of the built-in templates in ProstoRetro, which is free to use; people open the room link in a browser on a laptop or a phone, with nothing to install. For a walkthrough of a whole session, read how to run a great retro, or compare formats in the retrospective templates library.