borghei/Claude-Skills / project-management/sprint-retrospective
sprint-retrospective
VerifiedData-driven sprint retrospectives from git history — velocity, cycle/lead time, contributor insights, and churn hotspots. Use to run a retrospective, analyze team velocity or throughput, or generate a retro report from commit history.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 11.0 KB |
| assets/retro_report_template.md | markdown | 1.9 KB |
| assets/sample_churn.json | config | 2.2 KB |
| assets/sample_contributors.json | config | 2.9 KB |
| assets/sample_previous_retro.md | markdown | 180 B |
| assets/sample_velocity.json | config | 897 B |
| examples/sprint-23-data-driven-retro.md | markdown | 11.2 KB |
| references/red-flags.md | markdown | 10.3 KB |
| references/retrospective_facilitation.md | markdown | 8.2 KB |
| references/velocity_benchmarks.md | markdown | 6.2 KB |
| references/workflows-and-output.md | markdown | 22.2 KB |
| scripts/code_churn_analyzer.py | script | 18.7 KB |
| scripts/contributor_insights.py | script | 18.5 KB |
| scripts/retro_report_generator.py | script | 24.3 KB |
| scripts/velocity_analyzer.py | script | 18.3 KB |
references/velocity_benchmarks.md
View on GitHub# Velocity Benchmarks & Interpretation Guide
Industry benchmarks, healthy patterns, and guidance for interpreting sprint velocity metrics.
## Benchmarks by Team Size
These are practitioner heuristics for git-activity metrics (commits, PR counts and sizes, review and cycle times), not published survey benchmarks. DORA's research does not publish commits/day or PR-size benchmarks; its published measures are deployment frequency, change lead time, change fail rate, failed deployment recovery time and deployment rework rate. Treat the ranges as guidelines, not targets — every team's context is different.
### Small Team (2-4 engineers)
| Metric | Healthy Range | Warning |
|--------|---------------|---------|
| Commits/day (team) | 4-12 | <2 or >20 (noise) |
| PRs merged/week | 5-15 | <3 (bottleneck) |
| Avg PR size (LOC) | 50-200 | >500 (too large) |
| Cycle time (commit→merge) | 2-8 hours | >24 hours |
| Deploy frequency | 1-5/week | <1/week |
| Deep work session ratio | 35-50% | <20% |
### Medium Team (5-9 engineers)
| Metric | Healthy Range | Warning |
|--------|---------------|---------|
| Commits/day (team) | 8-25 | <5 or >40 |
| PRs merged/week | 10-30 | <5 |
| Avg PR size (LOC) | 50-250 | >400 |
| Cycle time (commit→merge) | 4-16 hours | >48 hours |
| Deploy frequency | 2-10/week | <1/week |
| Review turnaround | 1-4 hours | >8 hours |
### Large Team (10+ engineers)
| Metric | Healthy Range | Warning |
|--------|---------------|---------|
| Commits/day (team) | 15-50 | <10 |
| PRs merged/week | 20-60 | <10 |
| Avg PR size (LOC) | 50-200 | >300 (review burden) |
| Cycle time (commit→merge) | 4-24 hours | >72 hours |
| Deploy frequency | 5-25/week | <2/week |
| Knowledge distribution (bus factor) | >2 per area | 1 (critical risk) |
## Healthy vs Unhealthy Velocity Patterns
### Healthy Patterns
1. **Steady throughput with gradual improvement** — Velocity stays within a 20% band sprint-to-sprint, trending slightly upward over quarters.
2. **Balanced commit types** — Mix of feat (30-50%), fix (15-25%), refactor (10-20%), test (5-15%), docs (5-10%). Indicates sustainable development.
3. **Declining cycle time** — PRs merge faster over time. Signals improving review culture and smaller PR sizes.
4. **High deep work ratio** — 35-50% of sessions are deep work (>50 min). Indicates focused, uninterrupted development time.
5. **Consistent PR sizes** — Average PR size stays below 300 LOC with low variance. Small PRs review faster and ship safer.
### Unhealthy Patterns
1. **Velocity spikes** — Sudden 50%+ increases followed by crashes. Usually means crunch followed by recovery or technical debt payment.
2. **Zero refactor commits** — No `refactor:` type commits for 3+ sprints. Technical debt is accumulating silently.
3. **Monotonically increasing velocity** — Teams rarely get faster forever. If velocity only goes up, the definition of "done" may be weakening.
4. **90%+ feature commits** — No time allocated for fixes, refactoring, or documentation. Unsustainable pace.
5. **Growing cycle time** — PRs taking longer to merge each sprint. Signals review bottleneck, unclear ownership, or too-large PRs.
6. **Single-contributor dominance** — One person responsible for 60%+ of commits. Bus factor = 1.
7. **All micro-sessions** — >70% of sessions are under 20 minutes. Indicates excessive context switching, meetings, or interruptions.
## Sprint Predictability
### Measuring Predictability
**Coefficient of Variation (CV):** Standard deviation of velocity divided by mean velocity across sprints.
| CV Range | Predictability | Interpretation |
|----------|---------------|----------------|
| <15% | High | Reliable sprint planning possible |
| 15-25% | Moderate | Plan with buffer, forecasts useful |
| 25-40% | Low | Significant variability, wide confidence intervals |
| >40% | Very low | Planning is guesswork; investigate root causes |
### Improving Predictability
- Break work into smaller stories (reduces per-item variance)
- Consistent sprint length (no variable-length sprints)
- Protect the sprint from scope injection
- Account for PTO, holidays, on-call rotations in capacity
- Track and reduce carryover items
## Burndown / Burnup Interpretation
### Burndown Warning Signs
- **Flat line for 3+ days:** Blocked work or underestimated stories
- **Upward slope mid-sprint:** Scope creep or discovered work
- **Cliff at sprint end:** Batch completion — stories not truly incremental
- **Staircase pattern:** Large stories completing all at once instead of daily progress
### Burnup Advantages
Burnup charts show both scope and completion, making scope changes visible. Preferred for:
- Sprints where requirements evolve
- Stakeholder communication (shows added scope explicitly)
- Long-running releases spanning multiple sprints
## When Velocity Metrics Mislead
### Velocity Is NOT a Performance Metric
Velocity measures team throughput for planning purposes. Using it to evaluate individual or team performance creates perverse incentives:
- Inflated story points (Goodhart's Law)
- Avoiding refactoring and testing (reduces "velocity")
- Splitting work into trivial commits to inflate counts
### Contexts Where Raw Velocity Misleads
1. **New team or new domain** — Velocity will be low while the team learns. This is expected.
2. **Major refactoring sprint** — Velocity may drop while improving long-term health. Track refactor commits separately.
3. **Onboarding sprint** — New team members reduce short-term velocity while increasing long-term capacity.
4. **Infrastructure/tooling sprint** — Investment sprints have lower feature velocity but enable future acceleration.
5. **Post-incident sprint** — Incident response and prevention work may not show up in standard metrics.
### Better Combined Metrics
Instead of velocity alone, consider:
| Combined Metric | What It Shows |
|----------------|---------------|
| Velocity + Defect Rate | Are we shipping fast AND reliably? |
| Throughput + Cycle Time | Are we delivering more items faster? |
| Commit Count + Test Ratio | Are we building features with quality? |
| LOC + Churn Rate | Are we adding stable code? |
| Deploy Frequency + Failed Deployment Recovery Time (MTTR) | How fast do we ship and recover? |
---
**Last Updated:** 2026-03-18
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |