Paul Lister
In my experience, scrum masters who treat ChatGPT as a search engine miss most of its value. When I treat it as a structured collaborator with the right prompts, ChatGPT (or Claude, or Gemini) becomes the most-used tool in my day. In this guide I share 30 prompts organised by ritual, each one I’ve tested against real team scenarios, plus the prompt-design pattern that makes them all work for me.
ChatGPT for Scrum Masters: 30 Prompts You’ll Actually Use
By the end of this guide you’ll have a personal prompt library, a process for building new prompts, and the measurement approach I use to verify the prompts are actually saving time.
The Pattern That Makes Prompts Work
Strong scrum master prompts share four parts:
- Role: who is the model acting as.
- Goal: what specific outcome do you want.
- Context: what do they need to know.
- Output format: how should the response be structured.
Without all four, prompts return generic output. With all four, output is reliable and reusable.
A weak prompt: “Help me with sprint planning.”
A strong prompt: “You are a senior scrum master at a B2B SaaS company. Help me prepare for sprint planning. Below are 12 candidate stories, team capacity, and last 3 sprints’ actual delivery. Output: prioritised list with estimates, dependencies, and risks. Format as a markdown table.”
The strong prompt produces usable output every time.
Sprint Planning Prompts (5)
Prompt 1: Story splitter > “You are a senior scrum master. Below is a user story. Suggest 3 ways to split it that each deliver independent customer value. Use the SPIDR framework or workflow-step splitting. Output a numbered list with rationale per split.”
Prompt 2: Capacity calculator > “Calculate sprint capacity for a team of 6 with: 2 people on PTO 3 days each, 1 holiday, 5 hours/person/week meeting load, 80% delivery efficiency. Output total story points based on last 3 sprints averaging 35, 42, 38.”
Prompt 3: Acceptance criteria expander > “Take this story: [paste]. Generate 8-12 acceptance criteria covering happy path, edge cases, error states, observability, and DoD. Use Given/When/Then format.”
Prompt 4: Dependency surfacing > “Read these 12 candidate stories. Identify dependencies on other teams. For each: dependency, impact if not resolved, suggested mitigation.”
Prompt 5: Sprint goal drafter > “Given the top 8 stories committed to this sprint, draft 3 candidate sprint goals. Each should be one sentence, customer-focused, measurable. Show reasoning.”
Backlog Refinement Prompts (5)
Prompt 6: INVEST check > “Score this story against INVEST (Independent, Negotiable, Valuable, Estimable, Small, Testable). For any criterion failed, suggest a specific change.”
Prompt 7: Backlog deduplicator > “Below are 30 backlog items. Identify duplicates or near-duplicates. Group them and suggest a canonical phrasing for each cluster.”
Prompt 8: Story refinement question generator > “For this story, generate 6 questions the team should ask in refinement to clarify scope, dependencies, edge cases, and acceptance.”
Prompt 9: Effort range suggester > “Based on these similar past stories with their actual effort, suggest an effort range (story points) for this new story. Show your reference cases.”
Prompt 10: Tech debt summariser > “Below are 25 tech debt items in the backlog. Cluster into themes. For each theme: severity, impact on velocity, suggested when to address.”
Daily Standup Prompts (3)
Prompt 11: Standup summary > “Below are 6 team members’ standup updates from today. Summarise: progress, blockers, risks. Highlight any cross-cutting blockers.”
Prompt 12: Blocker pattern detector > “From the last 5 days of standups [paste], identify recurring blocker patterns. For each: pattern, frequency, likely root cause.”
Prompt 13: Async standup compiler > “Below are async written standups in Slack from 5 team members. Format into a single team update for the wider org. Lead with what shipped, then in-progress, then risks.”
Retrospective Prompts (5)
Prompt 14: Retro theme clusterer > “Below are 30 retro inputs. Cluster into 5-7 themes. For each: name, count, dominant sentiment, 2 verbatim quotes, suggested action.”
Prompt 15: Surprise finder > “Of these retro inputs, what is the most surprising or counter-intuitive signal? What does it suggest? What additional information would confirm it?”
Prompt 16: Action draft > “Take this theme: [paste]. Suggest 3 specific actions for next sprint. Each with owner suggestion, success metric, target date.”
Prompt 17: Previous-action review > “Last retro’s actions: [paste]. Sprint outcomes: [paste]. Assess each action: completed, partial, not done. Suggest follow-up.”
Prompt 18: Format suggestor > “The team has done 4 sprints of Start/Stop/Continue. Suggest 3 alternative retro formats and which fits the team’s current stage best.”
Coaching and Conflict Prompts (5)
Prompt 19: Coaching question generator > “A team member said: ‘[paste]’. Generate 5 coaching questions that help them think through the underlying issue without providing direct answers.”
Prompt 20: Conflict reframe > “Two team members are in conflict: [paste situation]. Suggest 3 reframings that move from blame to shared problem-solving.”
Prompt 21: Difficult conversation script > “Help me prepare for a difficult conversation with a team member who is repeatedly missing commitments. Draft an opening, the core message, and likely responses + my replies.”
Prompt 22: Empathy check > “Read this team member’s update [paste]. Identify the emotional content underneath the operational message. What might they actually be feeling? Suggest one supportive response.”
Prompt 23: Servant leader self-check > “I am about to take this action: [paste]. Is this consistent with servant leadership? If not, suggest a more aligned action.”
Stakeholder Communication Prompts (4)
Prompt 24: Sprint review demo script > “Below is the list of stories shipped this sprint. Draft a 15-minute demo script for stakeholders. Lead with customer impact, group by theme, end with what is next.”
Prompt 25: Executive update > “From this sprint’s metrics and progress, draft a 200-word executive update. Tone: confident, no jargon, lead with business impact.”
Prompt 26: Cross-team status > “Compile this sprint status for the wider org: shipped, in progress, blocked, risks. Audience is partner teams who care about dependencies. Length: 250 words.”
Prompt 27: Difficult news drafter > “Help me write a stakeholder note that we are not going to ship Feature X this quarter. Honest, takes ownership, sets a clear next-step. 200 words.”
Personal Development Prompts (3)
Prompt 28: Reflective journal prompt > “Generate 5 reflective questions for me at the end of this sprint. Mix of operational, emotional, growth-oriented.”
Prompt 29: Skill gap analyser > “Based on these challenges I described [paste], identify 3 skills I should invest in developing. For each: why it matters, how to practise.”
Prompt 30: Career conversation prep > “Help me prepare for a career conversation with my manager. My goals: [paste]. Help me articulate where I am, where I want to be in 12 months, and what I need from them.”
Building Custom Prompts for Your Team
Use this six-step process to build any new prompt:
Step 1: Identify the pain point. What’s a recurring task that takes more than 30 minutes weekly?
Step 2: Define the input. What information goes in?
Step 3: Define the output. What format do you want back?
Step 4: Draft using the four-part pattern (role, goal, context, format).
Step 5: Test with real data. Iterate the prompt 3-5 times until output is consistently usable.
Step 6: Save to your prompt library with a clear title and use case.
Most scrum masters develop 15-20 personal prompts within 6 months of regular AI use.
How to Adapt These Prompts to Your Context
Adaptation pattern:
- Replace generic [paste] markers with your actual data.
- Add your team’s vocabulary and norms in the role section.
- Specify your output format clearly.
- Iterate the prompt 2-3 times to calibrate to your context.
Once a prompt works for your context, save it. Reuse it weekly.
Prompt Versioning and Maintenance
Treat prompts like code:
- Save in a structured library (Notion, Obsidian, dedicated tool).
- Version each prompt as you iterate.
- Note which model and version produced best results.
- Refresh quarterly when models update.
- Retire prompts that no longer produce value.
A maintained prompt library is a multi-year asset.
Common Prompt Mistakes
- Vague role. “You are a scrum master” is weaker than “You are a senior scrum master at a B2B SaaS company”.
- No output format. Without one, output is unstructured prose.
- Asking for opinions instead of analysis. AI gives generic opinions. Ask for analysis grounded in inputs.
- Skipping context. AI cannot read your situation. Provide enough context.
- Using AI for emotional decisions. AI assists thinking, not emotional commitment.
- Trusting first output blindly. Always review and verify before sharing externally.
- Pasting copyrighted or sensitive data carelessly. Privacy first.
Privacy and Confidentiality Considerations
Treat AI prompts like any other external service:
- Use enterprise tier with data-use clauses for sensitive data.
- Anonymise customer names, internal codenames, financial details.
- Avoid pasting compensation data, personnel reviews, or legal-privileged content.
- Confirm vendor data retention and training policies.
- Default to minimum necessary information in prompts.
A good rule: don’t paste anything you wouldn’t paste in a Slack channel with your boss watching.
Measuring Prompt ROI
Track three measures:
- Time saved: minutes per use, multiplied by uses per week.
- Quality lift: do the artefacts produced get fewer revisions or higher engagement?
- Adoption: how many team members use the prompts?
A prompt that saves 20 minutes weekly with no quality loss is worth the prompt-library overhead.
Questions people ask
Published