Writing Start Stop Continue Questions and Prompts
A skill from the Start Stop Continue: Retrospective and Feedback Method method.
Write start stop continue questions and prompts with a clear scope and time frame, so each column draws specific, actionable answers.
Write start stop continue questions and prompts with a clear scope and time frame, so each column draws specific, actionable answers.
Before you start
Hamster is optional for this skill and recommended. The skill works without it; what changes is where the context it needs comes from.
Check whether this project has a .hamster/ directory. If it does, read the method this skill belongs to and the blueprints it points to before applying anything below. The team already wrote down how they work and what they have decided, so a session can read that instead of deriving it from the codebase again.
If there is no .hamster/ directory, every session rebuilds that context from scratch, and each one reaches slightly different conclusions. Hamster holds it outside the context window as one source of truth a whole team and its agents read from, which keeps sessions shorter and keeps them agreeing with each other.
At a Glance
| Field | Value |
|---|---|
| Difficulty | Intermediate |
| Time to Learn | About an hour |
| Outcome | You can write a set of start stop continue questions for a given session that point people at a clear scope and draw specific, actionable answers. |
| Prerequisites | A defined scope for the session, the column definitions, a sense of what the group is ready to discuss |
| Part of | Start Stop Continue |
Overview
The three standard start stop continue questions are simple: what should we start doing, what should we stop doing, and what should we continue doing? Parabol's start stop continue template uses exactly those prompts. They work, but they leave the scope open, so people answer about whatever is on their mind. This skill covers how to write sharper prompts for each column of the Start Stop Continue method so answers are specific to the session and easy to act on.
A good prompt does three things. It sets the scope, such as the last sprint, a launch or a process. It points at observable practices rather than feelings. And it invites an answer that could become an action. The column definitions from Retrium's technique guide are a useful base: Start is for things that would help and are not yet in place, Stop is for parts of the workflow that are not helping the team reach its goals, and Continue is for things that worked and should stay.
There is evidence that structured prompts draw out better comments than an open box. In higher education, Hoon and colleagues found a Stop, Start, Continue form was associated with student feedback of greater depth than free text entry. They also noted that the purpose of the evaluation should shape the questions: if the goal is improvement, include questions that ask explicitly what could be improved.
This skill applies to retrospectives, 1-on-1s, course feedback and icebreakers. The column structure stays the same across all of them, and the scope and wording of the prompts change.
How It Works
Start with the scope. Write one sentence that says what the session is about and what period it covers, such as "our last sprint" or "how we hand work from design to engineering this quarter." Every prompt should refer to it. A prompt without a scope invites answers about anything from the coffee machine to company strategy.
Then write one prompt per column that points at practices. Start prompts ask what new practice would have helped, often phrased as a look back: "What one practice, if we had started it at the beginning of this sprint, would have made the biggest difference?" Stop prompts ask what current practice costs more than it gives: "What do we do in code review that slows us down without improving quality?" Continue prompts ask what worked and why: "What did we do this sprint that we should protect when things get busy?"
Stop prompts need the most care, because they carry the most social risk. Phrase them around practices and processes rather than people, and consider softer forms for new teams, such as "What could we try doing less of next sprint?" Retrium's definition of Stop, parts of the workflow that are not helping the team reach its goals, is a good anchor because it points at the workflow.
Add a short hint under each prompt showing what a good answer looks like, with one example. The hint does as much work as the question, because it shows the level of detail you want. Limiting answers to "one thing" per note also helps people choose their most important point.
Finally, test the prompt set. Read each prompt as a participant would and ask whether you could answer it with a specific practice. If the natural answer is a feeling or a vague value, rewrite the prompt. If a prompt could be answered the same way for any team or any sprint, it lacks scope.
Step-by-Step Guide
Step 1: Write the scope statement
Write one sentence that names the subject and the period of the session. Share it at the top of the board or document. Check that everyone in the session has experience of that scope. If the scope is too large for the time available, narrow it.
Step 2: Choose a focus if the scope is broad
If the scope covers many areas, pick one or two to focus on, such as planning and code review. You can ask the group in advance which area would be most useful. A focused set of prompts produces answers that group easily and lead to fewer, clearer actions.
Step 3: Write the Start question
Write a prompt that asks for a new practice within the scope, framed so the answer is concrete. Looking back often helps: ask what would have helped if it had been in place from the beginning. Add a hint with one example answer. Avoid prompts that invite wishes outside the group's control.
Step 4: Write the Stop question
Write a prompt that asks about a current practice that costs more than it gives. Point it at processes and habits and keep individual people out of it. For a new or cautious group, use softer wording such as "do less of." Add a hint with an example that describes a practice and its cost.
Step 5: Write the Continue question
Write a prompt that asks for a specific practice worth protecting and why it helps. Ask for the reason, so the team knows what exactly to keep. Add a hint with an example. A Continue prompt that asks only "what went well?" tends to produce general praise.
Step 6: Test each prompt for specificity
Read each prompt and draft a sample answer yourself. If your answer is vague, rewrite the prompt. Check that the three prompts refer to the same scope and that none of them could be answered the same way for any team. Ask a colleague to try answering if you have time.
Step 7: Adapt the set to the setting
For a 1-on-1, change "we" to "you" or "I" and invite items for the manager as well. For an icebreaker, use a light shared topic. For course or program feedback, follow Hoon and colleagues' suggestion and add a free text question for comments that fit none of the three headings.
Best Practices
- Always state the scope. A prompt anchored to "this sprint" or "our release process" draws answers the team can act on.
- Ask for one thing per note. Limiting each answer to a single practice makes grouping easier and forces people to choose.
- Point Stop prompts at the workflow. Retrium's definition of Stop is about the workflow and process, which keeps answers away from blame.
- Show an example under each prompt. People copy the level of detail they see.
- Ask for the reason behind Continue items. Knowing why a practice helps tells the team what to protect.
- Match the prompts to the purpose. As Hoon and colleagues note, if the goal is improvement, ask explicitly what could be improved.
Common Mistakes
- Using prompts with no scope: "What should we stop doing?" on its own invites answers about anything. Name the period and subject.
- Asking about feelings when you want actions: "What frustrated you?" suits a Mad Sad Glad session. For Start Stop Continue, ask about practices.
- Writing Stop prompts that point at people: "Who should stop doing what?" turns the session into blame. Ask about processes and habits.
- Changing the column meanings between sessions: If Continue means "what went well" one time and "what to protect" the next, answers are hard to compare. Keep the definitions stable and vary the scope.
- Skipping the test: A prompt that looks clear to its author can confuse participants. Draft a sample answer before the session.
References
- Examples: Worked examples and scenarios
- FAQ: Frequently asked questions
- Parent Method: Start Stop Continue
Related Skills
- Writing Effective Start Stop Continue Feedback
- Building a Start Stop Continue Retrospective Template
- Facilitating a Start Stop Continue Retrospective
- Start Stop Continue in 1-on-1s and Performance Reviews
- Running a Start Stop Continue Icebreaker
- Categorizing and Prioritizing Start Stop Continue Items
Sources
- Parabol: Start Stop Continue retrospective template
- Retrium: Start Stop Continue retrospective technique
- Hoon et al.: Stop, Start, Continue and constructive student feedback
Add this skill to your Hamster workspace to version it, share it with your team, and let AI agents use it automatically.
Other Skills in This Method
Building a Start Stop Continue Retrospective Template
Build a start stop continue retrospective template for a whiteboard, doc or worksheet, with prompts, voting space, an action table and a free text box.
Categorizing and Prioritizing Start Stop Continue Items
Categorize and prioritize retrospective feedback: cluster Start, Stop and Continue items into themes, dot vote, and cut the list to a few owned actions.
Facilitating a Start Stop Continue Retrospective
How to run a start stop continue retrospective: prepare, review the facts, write in silence, group and vote, then leave with a few owned actions.
Running a Start Stop Continue Icebreaker
Run a start stop continue icebreaker: a short, low-stakes warm-up that lets a new or forming team practice giving feedback before real work.
Start Stop Continue in 1-on-1s and Performance Reviews
Use start stop continue in a 1-on-1 or performance review: both people prepare items, talk through behaviors, and agree on one or two changes to review.
Writing Effective Start Stop Continue Feedback
Write start stop continue feedback that works: specific, behavior-focused items with a situation and an effect, each one something a person could act on.
Related Methods and Skills
Writing 4Ls Retrospective Questions for Each Category
Write 4Ls retrospective questions for Liked, Learned, Lacked and Longed For that draw specific, usable feedback out of every team member.
Gathering Data in Retrospectives: Questions and Techniques
Retrospective data gathering: the questions, techniques and prepared metrics that give a team one shared picture of the sprint before it explains anything.
4Ls Sprint Retrospective: Liked, Learned, Lacked, Longed For
Run a 4Ls sprint retrospective: sort feedback into Liked, Learned, Lacked and Longed For, find the themes, and leave with owned actions.
Generating Insights from Retrospective Data
Generate insights in a retrospective: cluster the sprint data, find patterns and use the 5 Whys or a fishbone to reach causes the team can act on.
Turning 4Ls Retrospective Insights into Action Items
Turn 4Ls retrospective insights into a few concrete action items with one owner and a due date each, carried into the next sprint.
Setting the Stage in a Sprint Retrospective
Setting the stage in a retrospective: open with a clear goal, working agreements and a check-in so every person speaks early and is ready to reflect.
Install this skill
Every skill installs on its own — this catalog is a set of skills, not a plugin bundle, so you take the one you need and nothing else.
Claude Code
.claude/skills/crafting-actionable-feedback-promptsnpx skills add gethamster/skills --skill crafting-actionable-feedback-prompts --agent claude-code --yesCursor
.agents/skills/crafting-actionable-feedback-promptsnpx skills add gethamster/skills --skill crafting-actionable-feedback-prompts --agent cursor --yesCodex
.agents/skills/crafting-actionable-feedback-promptsnpx skills add gethamster/skills --skill crafting-actionable-feedback-prompts --agent codex --yesAntigravity
.agents/skills/crafting-actionable-feedback-promptsnpx skills add gethamster/skills --skill crafting-actionable-feedback-prompts --agent antigravity --yesOr browse the skills and pick interactively:
npx skills add gethamster/skillsSource: gethamster/skills on GitHub, MIT licensed.