Start Stop Continue: Retrospective and Feedback Method

Created by Origin undocumented (reference guide: Retrium) - https://www.retrium.com/retrospective-techniques/start-stop-continue

Overview

Start Stop Continue is a feedback and retrospective exercise built on three questions: what should we start doing, what should we stop doing, and what should we continue doing? Participants write answers under the three headings, the group sorts and discusses them, and the session ends with a short list of changes that each have an owner. Teams use it as a sprint retrospective, a project review, a start stop continue meeting between a manager and one report, and a warm-up for new groups. The format needs no training and no special materials beyond three columns.

The three columns have precise meanings. Retrium's technique guide defines the Start list as things the team thinks would help and that are not yet in place, the Stop list as parts of the workflow that are not helping the team reach its goals, and the Continue list as things that worked in the last iteration and should stay. Stop and Continue look back at recent work. Start looks forward to the next stretch of work. That mix is why the guide recommends the format when a team wants to leave with a list of concrete actions.

The origin of the exercise is not documented. The guides from Retrium, Parabol, EasyRetro, Miro and Atlassian describe it without naming an originator. One vendor, TeamRetro, says it was developed as a feedback tool in the 1970s, but it gives no source for that date, so treat it as unconfirmed. TeamRetro also notes that some teams say "Stop Start Continue" and that the order of the three words does not change the exercise.

The firmest evidence about the format comes from higher education. In a paper published online in 2014, Hoon, Oliver, Szpakowska and Newton describe Stop, Start, Continue as a well established way of collecting feedback in many disciplines, and cite a 1999 handbook of formative evaluation by George and Cowan as one example. They compared written student feedback from UK university programmes that used a free text box with feedback from programmes that used a structured Stop, Start, Continue form. The structured form was associated with feedback of greater depth. When one institution switched from free text to the structured form, depth rose again, and students said they preferred it. The study measured course feedback, not team retrospectives, but it supports the idea that three labelled prompts draw out more considered comments than a blank box. The authors also suggest pairing the three boxes with a free text box for comments that fit none of them.

The format fits a wider habit in agile work. One of the principles behind the Agile Manifesto says the team reflects at regular intervals on how to become more effective and then adjusts its behavior. The Scrum Guide gives the Sprint Retrospective the purpose of planning ways to increase quality and effectiveness. Start Stop Continue is one of many activities a team can use to meet that purpose, and its wording pushes toward a decision. Parabol's start stop continue template calls the format action-oriented for that reason, because saying the team is going to start or stop something reads as a commitment more clearly than a general suggestion does.

That strength is also its main risk. Retrium's guide to the five phases of a retrospective calls Start Stop Continue a popular technique that is often misused, because many teams open the retrospective with it and jump to solutions before they have looked at what happened. Retrium places it in the Decide What To Do phase, after the team has found and examined an issue. A team that follows that advice gathers data and discusses causes first, then uses the three columns to choose actions.

The rest of this page covers the principles behind a useful session, the steps to run one, how the format compares with other retrospective formats, and when to pick something else. The skills linked below go deeper on facilitation, one-on-ones, icebreakers, prioritization, templates, writing feedback and writing prompts.

Core Principles

Each column is a different kind of decision

Start adds a practice, Stop removes one, and Continue protects one. Keeping those meanings distinct is what makes the output usable. An item that could sit in two columns is usually two items, or a vague one that needs rewriting. A facilitator who makes sure everyone understands what each column means before the session starts, as Retrium recommends, saves time later.

Decide after you understand

The three questions ask for solutions, so they work best once the team knows what happened and why. Retrium's five-phase guide warns against opening a retrospective with them. When time is short, spend at least a few minutes on facts before anyone writes a Start or Stop item. Otherwise the loudest recent frustration drives the plan.

Everyone writes before anyone talks

Private writing before discussion keeps the first opinion from shaping the rest. Retrium advises keeping notes private during ideation to prevent groupthink, and Parabol's template lets people reflect anonymously. Quiet members and newer members contribute more when they do not have to speak first. The facilitator's job in this phase is to protect the silence.

Continue items deserve the same care as Stop items

The Continue column records practices the team wants to keep when pressure rises. Without it, a retrospective can read as a list of complaints, and useful habits get dropped during the next change. Parabol notes that as teams repeat the exercise the Continue column tends to fill while Start and Stop get lighter. Treat a thin Continue column as a prompt to ask what is working.

A few owned actions beat a long list

A session that produces many items and no owners changes nothing. The Atlassian retrospective play says to assign owners and deadlines to action items and put them in the tool the team already uses. Ben Linders advises asking for volunteers and keeping actions small enough to finish in the next iteration. Choose the few items with the most support and let the rest wait.

Candor needs safety

Stop items can point at leaders, norms or colleagues, so people write them only when speaking up feels safe. Google's research on team effectiveness found psychological safety was the most important of five dynamics it identified. Some facilitators read Norm Kerth's Prime Directive at the start to frame the discussion around systems rather than blame. If the Stop column is empty, suspect safety before you conclude that everything is fine.

The next session starts with the last one

The exercise pays off through repetition. Ben Linders starts each retrospective by checking the previous actions and keeps them visible on the team's board or backlog between sessions. Reviewing last time's Start and Stop items tells the team whether its decisions stuck. Items that come back cycle after cycle point to a problem the team cannot solve alone.

Steps

1. Set the scope and read the column definitions Name the period or project under review and the decision the session should produce. Read the Start, Stop and Continue definitions aloud so everyone uses the same meaning. State ground rules: describe behaviors and practices and leave individuals out of it, and assume good intent, which is the idea behind the Prime Directive. Tell the group how long the session will run. Retrium suggests 30-60 minutes depending on team size.

2. Review what happened Before anyone writes proposals, spend a few minutes on facts: the sprint goal and whether it was met, notable events, and the status of last session's actions. This follows Retrium's advice to use the format only after the team has examined what happened. Ask questions about causes where the facts are surprising. Keep this phase short, but do not skip it.

3. Write items in silence Give everyone a few minutes to write items under each heading on their own. Use sticky notes, a shared board or a start stop continue template with the three columns. Ask for one idea per note, written as a behavior someone could see. Keep notes hidden until time is up so early ideas do not steer the rest.

4. Reveal, read and group Reveal all notes at once and have each author read theirs briefly. The facilitator groups duplicates and related items into themes, following Retrium's grouping step. Ask clarifying questions, but hold debate until later. Name each group with a short label that says what the items share.

5. Vote on what to discuss When there are more themes than time, use dot voting. Nielsen Norman Group suggests giving each person votes equal to roughly a quarter of the options and voting quietly with no lobbying. Count the votes and order the themes. The top few get the discussion time.

6. Discuss and turn themes into actions Take the top themes in order and timebox each one. Retrium suggests 5 or 10 minutes per topic or a single block for the whole discussion. For each theme, agree what the team will actually do, rewriting vague items into specific actions. Record each action with an owner who volunteered and a date to review it.

7. Close and record Read the actions back so everyone hears the commitments. Put them where the team tracks work, as the Atlassian retrospective play recommends, rather than leaving them on a board nobody opens. Thank people for their candor. Ask for a quick comment on how the session went so the next one can improve.

8. Open the next session with the last actions At the next session, review each action first: done, in progress, or dropped, and why. Move completed Start items that the team wants to keep into Continue. Raise any action that failed twice as a theme of its own. Over several cycles, this review shows which decisions the team kept.

Start Stop Continue Compared with Other Formats

Start Stop Continue is one of several column-based formats. Each asks a different question, so the choice depends on what the team needs from the session.

Parabol contrasts the two most common choices directly: compared with the decisive language of Start, Stop, Continue, the 4Ls allow more neutral and far-reaching feedback and do not ask for immediate solutions. The formats also combine. A team can gather data with the 4Ls retrospective or a timeline, then use Start Stop Continue in the decide phase of the five-step retrospective framework.

When to Use

• At the end of a sprint or milestone when the team has already reviewed what happened and needs to agree on a small set of changes for the next iteration.
• In a one-on-one or review conversation where a manager and a report want specific behaviors to add, drop and keep, and a record they can revisit next time.
• When recurring problems keep surfacing, such as missed deadlines or work carried over sprint after sprint, which Parabol lists as a signal to run the exercise.
• With a new or forming team as a low-stakes warm-up, because the three questions are easy to explain and show newcomers that candid feedback is expected.
• For course or program feedback, where the Hoon et al. study found the structured form drew deeper written comments than a free text box.

When Not to Use

• When people do not feel safe to speak. The Stop column will be empty or polite, and the session will suggest that all is well when it is not.
• As the opening activity of a retrospective about a hard or confusing period. The format asks for solutions, and the team first needs time to establish what happened and why.
• For a deep causal problem, such as a production incident. Use a root cause technique first, then use Start Stop Continue to choose actions.
• When nobody can act on the output. If the decisions sit outside the group's control, the exercise produces a list that breeds cynicism.
• When the team wants to reflect rather than decide, for example after a project ends with no next iteration. A neutral format such as the 4Ls fits better.