Hypothesis-Driven Delivery (HDD)

TL;DR: Applies hypothesis-driven delivery to project decisions, transforming assumptions into testable hypotheses with clear success/failure criteria. Uses build-measure-learn cycles to validate project assumptions before committing full investment, reducing risk through validated learning.

Principio Rector

Toda decisión de proyecto descansa sobre hipótesis. "Los usuarios adoptarán la nueva herramienta" es una hipótesis, no un hecho. HDD hace explícitas estas hipótesis, las ordena por riesgo, y las valida con el mínimo esfuerzo posible. Invertir millones sin validar hipótesis críticas es esperanza, no gestión.

Assumptions & Limits

• Assumes stakeholders have tolerance for experimentation and potential failure [SUPUESTO]
• Assumes sufficient data collection capability exists to measure experiment outcomes [SUPUESTO]
• Breaks if organizational culture punishes failure — HDD requires psychological safety [STAKEHOLDER]
• Scope limited to project-level hypothesis validation; strategic pivots require executive authorization [PLAN]
• Does not replace formal requirements engineering — complements it with validated evidence [PLAN]
• Time-boxed experiments require dedicated budget; unfunded experimentation defaults to opinion-based decisions [PLAN]

Usage

/pm:hypothesis-driven-delivery $PROJECT_NAME --extract-from=assumptions
/pm:hypothesis-driven-delivery $PROJECT_NAME --hypothesis="Users will adopt tool X" --experiment=pilot
/pm:hypothesis-driven-delivery $PROJECT_NAME --rank-by=risk-impact --top=5

Parameters:

Service Type Routing

{TIPO_PROYECTO} variants:

• Agile: Build-measure-learn cycles within sprints; hypothesis cards in backlog; experiment results drive pivot-or-persevere decisions
• SAFe: Innovation and Planning (IP) iteration for hypothesis validation; Lean startup thinking applied at epic level with MVP experiments
• Kanban: Hypotheses as work items with explicit validation criteria; experiment results tracked as flow metrics
• Transformation: Change adoption hypotheses validated through pilot groups; ADKAR-stage assumptions tested before enterprise rollout
• Hybrid: Structured experiments in iterative phases; validated hypotheses feed formal decisions at waterfall gates
• Recovery: Critical assumption re-testing for troubled projects; rapid experiments to validate turnaround strategy viability

Before Testing Hypotheses

1. Read assumption-log — identify assumptions already documented that need validation [PLAN]
2. Glob *risk-register* — high-probability risks often contain untested hypotheses [PLAN]
3. Read stakeholder tolerance for experimentation — confirm organizational readiness for HDD [STAKEHOLDER]
4. Verify experiment budget exists — unfunded experiments produce unreliable results [METRIC]

Entrada (Input Requirements)

• Project assumptions and hypothesis candidates
• Available experimentation budget and timeline
• Success/failure criteria definitions
• Data collection capabilities
• Stakeholder tolerance for experimentation

Proceso (Protocol)

1. Hypothesis extraction — Convert key assumptions into testable hypotheses
2. Hypothesis format — Structure as: "We believe [hypothesis]. We will know we're right when [metric] achieves [target] within [timeframe]"
3. Priority ranking — Rank hypotheses by risk x impact (test riskiest first)
4. Experiment design — Design minimum viable experiments to test each hypothesis
5. Success criteria — Define quantitative pass/fail criteria before testing
6. Execution — Run experiments within time and budget constraints
7. Data collection — Gather measurement data per success criteria
8. Analysis — Evaluate results against pre-defined criteria
9. Decision — Pivot, persevere, or kill based on validated evidence
10. Learning documentation — Document validated/invalidated hypotheses and implications

Edge Cases

1. Experiment results are ambiguous — Redesign experiment with tighter success criteria; if still ambiguous after second run, escalate with documented uncertainty and recommend conservative path.
2. Stakeholders reject experiment results that contradict preferences — Document bias formally; present data without editorial; escalate to governance if decision contradicts validated evidence.
3. No budget for experimentation — Use lightweight validation methods (surveys, desk research, analogous data); tag all conclusions as [INFERENCIA] rather than [METRIC].
4. Critical hypothesis invalidated mid-project — Trigger formal pivot-or-kill decision; document sunk cost vs. future investment analysis for governance review.
5. Too many hypotheses to test within timeline — Strict priority ranking by risk x impact; test only top 5; accept remaining as managed risks.

Example: Good vs Bad

Good Hypothesis-Driven Delivery:

Bad Hypothesis-Driven Delivery: "We tested the tool and people seemed to like it." — No structured hypothesis, no pre-defined success criteria, no quantitative measurement, no clear decision framework. Result: confirmation bias disguised as validated learning.

Salida (Deliverables)

• Hypothesis register with test results
• Experiment design documents
• Validation results and decision log
• Pivot/persevere recommendations
• Validated learning summary

Validation Gate

• Every hypothesis follows the structured format: believe/know/when [PLAN]
• Success criteria defined before experiment execution — no post-hoc criteria
• Hypotheses ranked by risk x impact with quantified scores
• Experiment results evaluated against pre-defined criteria only
• Pivot/persevere/kill decision documented with evidence trace
• All critical project assumptions have corresponding hypotheses
• Experiment methodology appropriate for hypothesis type
• Learning documentation captures both validated and invalidated hypotheses
• Stakeholder decisions traceable to experiment data
• HDD cadence integrated with project delivery rhythm

Escalation Triggers

• Critical hypothesis invalidated
• Experiment results ambiguous
• Stakeholders unwilling to accept experiment results
• Budget insufficient for necessary experiments

Additional Resources

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

Sub-Agents

Experiment Designer

Experiment Designer Agent

Core Responsibility

Designs Build-Measure-Learn experiments. This agent operates autonomously, applying systematic analysis and producing structured outputs.

Process

1. Gather Inputs. Collect all relevant data, documents, and stakeholder inputs needed for analysis.
2. Analyze Context. Assess the project context, methodology, phase, and constraints.
3. Apply Framework. Apply the appropriate analytical framework or model.
4. Generate Findings. Produce detailed findings with evidence tags and quantified impacts.
5. Validate Results. Cross-check findings against related artifacts for consistency.
6. Formulate Recommendations. Transform findings into actionable recommendations with owners and timelines.
7. Deliver Output. Produce the final structured output with executive summary, analysis, and action items.

Output Format

• Analysis Report — Structured findings with evidence tags and severity ratings.
• Recommendation Register — Actionable items with owners, deadlines, and success criteria.
• Executive Summary — 3-5 bullet point summary for stakeholder communication.

Hypothesis Formulator

Hypothesis Formulator Agent

Core Responsibility

Formulates testable hypotheses. This agent operates autonomously, applying systematic analysis and producing structured outputs.

Process

1. Gather Inputs. Collect all relevant data, documents, and stakeholder inputs needed for analysis.
2. Analyze Context. Assess the project context, methodology, phase, and constraints.
3. Apply Framework. Apply the appropriate analytical framework or model.
4. Generate Findings. Produce detailed findings with evidence tags and quantified impacts.
5. Validate Results. Cross-check findings against related artifacts for consistency.
6. Formulate Recommendations. Transform findings into actionable recommendations with owners and timelines.
7. Deliver Output. Produce the final structured output with executive summary, analysis, and action items.

Output Format

• Analysis Report — Structured findings with evidence tags and severity ratings.
• Recommendation Register — Actionable items with owners, deadlines, and success criteria.
• Executive Summary — 3-5 bullet point summary for stakeholder communication.

Measurement Framework Builder

Measurement Framework Builder Agent

Core Responsibility

Builds measurement frameworks for hypothesis validation. This agent operates autonomously, applying systematic analysis and producing structured outputs.

Process

1. Gather Inputs. Collect all relevant data, documents, and stakeholder inputs needed for analysis.
2. Analyze Context. Assess the project context, methodology, phase, and constraints.
3. Apply Framework. Apply the appropriate analytical framework or model.
4. Generate Findings. Produce detailed findings with evidence tags and quantified impacts.
5. Validate Results. Cross-check findings against related artifacts for consistency.
6. Formulate Recommendations. Transform findings into actionable recommendations with owners and timelines.
7. Deliver Output. Produce the final structured output with executive summary, analysis, and action items.

Output Format

• Analysis Report — Structured findings with evidence tags and severity ratings.
• Recommendation Register — Actionable items with owners, deadlines, and success criteria.
• Executive Summary — 3-5 bullet point summary for stakeholder communication.

Pivot Persevere Advisor

Pivot Persevere Advisor Agent

Core Responsibility

Advises on pivot or persevere decisions based on experiment results. This agent operates autonomously, applying systematic analysis and producing structured outputs.

Process

1. Gather Inputs. Collect all relevant data, documents, and stakeholder inputs needed for analysis.
2. Analyze Context. Assess the project context, methodology, phase, and constraints.
3. Apply Framework. Apply the appropriate analytical framework or model.
4. Generate Findings. Produce detailed findings with evidence tags and quantified impacts.
5. Validate Results. Cross-check findings against related artifacts for consistency.
6. Formulate Recommendations. Transform findings into actionable recommendations with owners and timelines.
7. Deliver Output. Produce the final structured output with executive summary, analysis, and action items.

Output Format

• Analysis Report — Structured findings with evidence tags and severity ratings.
• Recommendation Register — Actionable items with owners, deadlines, and success criteria.
• Executive Summary — 3-5 bullet point summary for stakeholder communication.