Discovery Retrospective

TL;DR: Conducts a quantitative post-discovery retrospective analyzing pipeline execution quality, deliverable completeness, stakeholder satisfaction, estimation accuracy, and methodology fit. Produces measurable insights that improve future discovery cycles and calibrate APEX pipeline parameters.

Principio Rector

La retrospectiva de discovery no es un ejercicio de catarsis — es un instrumento de calibración. Cada ciclo de discovery debe producir datos que mejoren el siguiente: cuánto tardamos, cuánto acertamos, qué supuestos se validaron. Sin métricas, la mejora es ilusión. [EXPLICIT]

Assumptions & Limits

• Assumes discovery pipeline has been executed with deliverables produced [PLAN]
• Assumes session changelog and decision log are available for timeline analysis [SUPUESTO]
• Breaks when no session data was captured — retrospective becomes opinion-based, not data-driven
• Estimation accuracy measurement requires post-discovery actuals, which may not yet exist
• Does not replace sprint retrospectives — this is pipeline-level, not ceremony-level
• Calibration updates require organizational authority to modify pipeline parameters [STAKEHOLDER]

Usage

# Run full discovery retrospective
/pm:discovery-retrospective $PROJECT --type=full

# Assess deliverable quality only
/pm:discovery-retrospective $PROJECT --type=quality-audit

# Update pipeline calibration parameters
/pm:discovery-retrospective $PROJECT --type=calibrate --update="true"

Parameters:

Parameter	Required	Description
$PROJECT	Yes	Project identifier
--type	Yes	full, quality-audit, calibrate, timeline-analysis
--update	No	Apply calibration updates (true/false)

Service Type Routing

{TIPO_PROYECTO} variants:

• Agile: Focus on sprint-0/inception quality, backlog readiness, DoR compliance
• Waterfall: Focus on requirements completeness, estimation accuracy, phase-gate quality
• SAFe: Focus on PI planning readiness, ART alignment, architectural runway assessment
• Kanban: Focus on flow establishment quality, WIP calibration, service class definition
• Hybrid: Assess methodology selection decision quality and integration effectiveness
• PMO: Aggregate discovery metrics across portfolio for organizational learning
• Recovery: Assess what the original discovery missed that led to project distress

Before Retrospecting

1. Read all discovery deliverables produced during the pipeline for completeness assessment
2. Read the session changelog to reconstruct timeline and decision history
3. Glob skills/discovery-retrospective/references/*.md for retrospective scoring rubrics
4. Grep for [SUPUESTO] tags across deliverables to assess assumption validation rates

Entrada (Input Requirements)

• All discovery deliverables produced during the pipeline
• Session changelog and decision log
• Time spent per phase and deliverable
• Stakeholder feedback (formal and informal)
• Original discovery scope vs. actual coverage

Proceso (Protocol)

1. Deliverable inventory — Catalog all outputs produced during discovery
2. Completeness scoring — Rate each deliverable against quality checklist
3. Timeline analysis — Compare planned vs. actual time per phase
4. Estimation accuracy — Measure how discovery estimates tracked against actuals (if available)
5. Stakeholder satisfaction — Collect and quantify stakeholder feedback
6. Assumption validation — Review which [SUPUESTO] tags were validated vs. invalidated
7. Pipeline friction — Identify bottlenecks, rework cycles, and blocked phases
8. Methodology fit — Assess whether selected methodology proved appropriate
9. Improvement actions — Define specific, measurable improvements for next discovery
10. Calibration update — Update APEX pipeline parameters based on findings

Edge Cases

1. No session data captured during discovery: Use deliverable timestamps and git history as proxy. Tag all timeline findings as [INFERENCIA]. Recommend session tracking for future cycles. [SUPUESTO]
2. Stakeholder satisfaction below threshold: Deep-dive into dissatisfaction drivers. Separate content quality issues from communication/expectation issues. Design targeted improvements. [STAKEHOLDER]
3. Critical deliverables missing or incomplete: Flag as Critical finding. Root-cause analysis: was it scope creep, capacity issue, or skill gap? Design preventive measure. [PLAN]
4. Estimation variance >50%: Analyze whether variance is from scope change, effort underestimation, or complexity underestimation. Calibrate estimation parameters accordingly. [METRIC]

Example: Good vs Bad

Good Discovery Retrospective:

Attribute	Value
Deliverables scored	100% against quality checklists
Timeline analysis	Planned vs actual per phase with variance
Assumption validation	X validated, Y invalidated, Z still open
Stakeholder satisfaction	Quantified score with driver analysis
Improvement actions	≥5 specific, measurable, assigned
Calibration updates	Pipeline parameters adjusted with rationale

Bad Discovery Retrospective: A meeting where everyone says "it went well" without metrics, scoring, or timeline analysis. No assumption validation. No calibration updates. Fails because feel-good retrospectives produce no data for improvement — the next discovery cycle will repeat the same patterns. [EXPLICIT]

Validation Gate

• Every deliverable scored against its quality checklist with quantified score
• Timeline analysis compares planned vs actual for every phase with variance explanation
• ≥80% of [SUPUESTO] tags reviewed with validated/invalidated/open status
• Stakeholder satisfaction quantified (score or NPS) with satisfaction driver analysis
• ≥5 improvement actions defined with specific, measurable criteria and owners
• Pipeline friction points identified with quantified impact (rework hours, blocked days)
• Methodology fit assessment documented with evidence
• Calibration parameter updates documented with rationale and evidence
• Stakeholders see concrete improvement commitments [STAKEHOLDER]
• Retro format aligned with project methodology [PLAN]

Escalation Triggers

• Discovery quality score below 60% across multiple dimensions
• Stakeholder satisfaction below acceptable threshold
• Critical deliverables missing or fundamentally incomplete
• Estimation variance exceeding 50% on key dimensions

Additional Resources

Resource	When to read	Location
Body of Knowledge	Before retrospecting to understand assessment rubrics	references/body-of-knowledge.md
State of the Art	When evaluating quantitative retro approaches	references/state-of-the-art.md
Knowledge Graph	To link retro to pipeline orchestration	references/knowledge-graph.mmd
Use Case Prompts	When facilitating retrospective sessions	prompts/use-case-prompts.md
Metaprompts	To generate scoring rubric templates	prompts/metaprompts.md
Sample Output	To calibrate expected retrospective report format	examples/sample-output.md

Output Configuration

• Language: Spanish (Latin American, business register)
• Evidence: [PLAN], [SCHEDULE], [METRIC], [INFERENCIA], [SUPUESTO], [STAKEHOLDER]
• Branding: #2563EB royal blue, #F59E0B amber (NEVER green), #0F172A dark