Use this when you need to plan and prepare a team retrospective. Helps you choose a format, generates a facilitation script with timing, and prepares discussion prompts based on iteration data. Core team only — no stakeholders.
Related resources:
retro-facilitation-guide.md-- 4-step Quick Start Guide for running retrospectives (standalone, printable).
Process
Step 1: Gather context
Ask the user:
- How many people are attending? (Core team only — PM, Design, Eng)
- How long do you have? (Typical: 30-45 minutes)
- Any format preference? Offer these options:
- Start / Stop / Continue — simplest, good default
- 4 Ls (Liked / Learned / Lacked / Longed for) — richer, good for surfacing what's missing
- Rose / Thorn / Bud — positive framing, surfaces future opportunities
- Sailboat (Wind / Anchor / Rocks / Island) — visual metaphor, good for variety
- No preference — the skill picks based on context
- Iteration data — (optional) stories committed vs. completed, carry-overs, notable events, blockers, wins
- Last retro's action items — (optional) what did you commit to last time? Did it happen?
If the user has no preference on format, recommend one based on context:
- First retro with a team → Start / Stop / Continue (simplest)
- Team has used Start/Stop/Continue 3+ times → rotate to a different format
- Team morale seems low → Rose / Thorn / Bud (starts positive)
- Team needs strategic thinking → Sailboat (surfaces risks and goals)
Step 2: Generate the facilitation plan
## Retrospective Plan — (Date)
### Setup
- **Format:** (Chosen format)
- **Duration:** (Total time)
- **Tool:** (Miro, FigJam, shared doc, or physical board)
### Facilitation Script
**(0:00 - 2 min) Opening**
- "This is a safe space. We're improving the process, not blaming people."
- Check-in: "On a scale of 1-5, how was this iteration for you?" (Quick round-robin)
**(2:00 - 4 min) Psychological Safety Primer — 2 minutes**
- Before brainstorming, explicitly name that half-formed ideas are welcome. Model imperfection: "This is my first stab, it is probably wrong, and I need your help."
- The inner critic gets stronger as people gain more leadership responsibilities. Senior team members may be the ones most likely to self-censor. Name this dynamic so the room gives itself permission to think out loud.
**(4:00 - 9 min) Silent Writing — 5 minutes**
- Everyone adds sticky notes to each column/category
- No discussion yet — just capture
(Format-specific columns here — e.g., for Start/Stop/Continue:)
| Start Doing | Stop Doing | Continue Doing |
|-------------|-----------|----------------|
| (Things to try) | (Things not working) | (Things working — keep them) |
**(7:00 - 10 min) Group and Cluster — 3 minutes**
- Facilitator reads items aloud, groups related ones
- If too many items, dot-vote: each person gets 3 dots
**(10:00 - 30 min) Discussion — 20 minutes**
- Discuss top 3-5 themes (highest votes first)
- For each: "What's the root cause? What can we actually change?"
**(30:00 - 35 min) Last Retro Action Items — 5 minutes**
- Review: did we follow through on last retro's commitments?
- If not: why? Should we recommit or let it go?
**(35:00 - 45 min) New Action Items — 10 minutes**
- Convert discussion into 2-3 specific, owned action items
- Each action item: what, who, by when
- No more than 3 — more than that won't get done
### Discussion Prompts
(Generated from iteration data, if provided)
- (Prompt 1 — based on what went well or poorly this iteration)
- (Prompt 2 — based on patterns like recurring carry-over)
- (Prompt 3 — based on notable events)
### Patterns to Watch
(If iteration data was provided)
- (Pattern 1 — e.g., "This is the 3rd iteration with carry-over on API stories")
- (Pattern 2)
Step 3: Review and adjust
Ask the user:
- Does the timing work for your team? (Adjust if they're faster/slower at discussion)
- Are the discussion prompts relevant? Any to add or remove?
- Do you want to add a specific topic the team needs to address?
- Who is facilitating? (Recommend rotating facilitators every iteration)
Adjust as needed.
Remember: this draft multiplies your output, not your judgment. You own the final narrative.
Output location
Present the facilitation plan as formatted text in the conversation for the facilitator to reference during the retro.
Example Output
Input
- Team size: 6 people (1 PM, 2 designers, 3 engineers)
- Duration: 45 minutes
- Format preference: No preference — skill should choose
- Iteration data: Sprint 14, committed 18 points / completed 11 points; 3 stories carried over (all backend API work); one production incident mid-sprint (payment service outage, ~2 hrs); shipped new onboarding flow on time
- Last retro's action items: (1) Add a 15-min mid-sprint sync to catch blockers early — not done; (2) Write API contract docs before frontend starts — partially done, 1 of 3 written
Output (abbreviated)
Retrospective Plan — Sprint 14
Setup
- Format: Sailboat (Wind / Anchor / Rocks / Island)
- Rationale: Recurring carry-over on API stories and a production incident suggest the team needs to surface systemic risks and reconnect to goals — Sailboat is better suited than Start/Stop/Continue here.
- Duration: 45 minutes
- Tool: FigJam (recommended) or physical board with four quadrants
Facilitation Script
(0:00 – 2 min) Opening
- "This is a safe space. We're improving the process, not blaming people — including around the payment incident."
- Check-in round-robin: "On a scale of 1–5, how was Sprint 14 for you?" Go around the room, no commentary yet.
(2:00 – 4 min) Psychological Safety Primer
- Name it explicitly: "Half-formed ideas are what we want. If you have a hunch but can't articulate it yet, say it anyway."
- Model it yourself: "I'll start — I think our API coordination is broken, but I don't know exactly where. Help me figure it out."
- Note for senior members: the inner critic gets louder with seniority. If your tech lead or principal eng is quiet, check in directly.
(4:00 – 9 min) Silent Writing — 5 minutes
- Everyone adds sticky notes. No discussion yet.
| 💨 Wind (What's pushing us forward?) | ⚓ Anchor (What's slowing us down?) | 🪨 Rocks (What risks could sink us?) | 🏝️ Island (What's our goal?) |
|---|---|---|---|
| Onboarding flow shipped on time | API work blocking frontend repeatedly | Undocumented service dependencies | Stable, fast API layer by Q3 |
| Team stepped up during incident | Mid-sprint sync never happened | Incident response has no runbook | Onboarding conversion > 30% |
(9:00 – 12 min) Group and Cluster — 3 minutes
- Facilitator reads items aloud and groups related notes.
- If more than 10 items total, dot-vote: 3 dots per person, vote on what matters most.
(12:00 – 32 min) Discussion — 20 minutes
- Work through top 3–4 clusters, highest votes first.
- For each: "What's the root cause? What's actually in our control to change?"
Suggested discussion order based on iteration data:
- API carry-over (3rd sprint in a row) — Anchor + Rock overlap
- Payment incident response — Rocks quadrant
- Onboarding win — Wind quadrant (don't skip this — it matters for morale)
- Mid-sprint sync commitment we dropped — Anchor
(32:00 – 37 min) Last Retro Action Items — 5 minutes
- Review both commitments honestly:
- Mid-sprint sync: not done. Ask — "What got in the way? Do we still believe in this?"
- API contract docs: 1 of 3 written. Ask — "Is the process working for the one that got done? What blocked the other two?"
- Recommit with specifics or explicitly let it go. Carrying over vague commitments erodes trust.
(37:00 – 45 min) New Action Items — 8 minutes
- Max 3 action items. Each needs: what, who, by when.
| Action | Owner | Due |
|---|---|---|
| Write incident runbook for payment service | Backend lead | End of Sprint 15 |
| Pilot mid-sprint sync (Weds, 15 min) — one person owns the invite | PM | Sprint 15, Day 1 |
| Complete remaining 2 API contract docs before frontend picks up next API story | Backend + Design to agree on trigger | Sprint 15 kickoff |
Discussion Prompts
Based on Sprint 14 data:
- On carry-over: "This is the third sprint where API stories didn't complete. What's the earliest point in the sprint we knew they were at risk — and what stopped us from raising it?"
- On the incident: "The payment outage cost us ~2 hours and likely distracted the team mid-sprint. If it happens again next sprint, what's the first thing we do in the first 10 minutes?"
- On the win: "Onboarding shipped on time despite a rough sprint. What made that work when other things didn't? Can we replicate it?"
- On dropped commitments: "We said we'd add a mid-sprint sync and didn't. Was it the wrong idea, or did we just not assign a clear owner?"
Patterns to Watch
- ⚠️ API carry-over is now a trend, not a one-off. Three consecutive sprints. This is worth a dedicated conversation outside the retro if 20 minutes doesn't resolve it today.
- ⚠️ Action item follow-through is slipping. 0 of 2 commitments fully completed from Sprint 13. If this continues, the retro loses credibility with the team. Consider reducing to 1 action item next sprint to rebuild the habit.
- ✅ Design/frontend delivery is holding. Onboarding shipped on time. Worth naming explicitly so it doesn't get buried under the noise.
Next step: Review the discussion prompts before the session and remove any that feel redundant given what you already know. Consider whether you or a teammate should facilitate — rotating the role every sprint keeps the PM from always steering the conversation.