110-ira-framework

IRA Framework: Investigate, Research, Act

• Purpose: Implement a systematic evidence-based approach to problem-solving that reduces ambiguity, integrates conflicting data, and ensures actionable outcomes through iterative validation.

• Requirements:

  • Investigation Phase:
    • Clearly articulate the decision trigger with measurable success criteria
    • Conduct comprehensive evidence audit of existing data and knowledge
    • Identify and document all evidence gaps systematically
    • Reframe vague problems into specific, researchable questions
  • Research Phase:
    • Match research methods to specific evidence gaps (generative vs evaluative)
    • Deploy mixed methods for comprehensive understanding
    • Synthesize findings using structured analysis (affinity mapping, thematic analysis)
    • Prioritize insights by impact and feasibility metrics
  • Action Phase:
    • Co-create solutions with relevant stakeholders
    • Implement controlled pilots with clear boundaries
    • Embed measurement protocols from the start
    • Conduct structured retrospectives for continuous improvement

• Validation:

  • Check: Decision trigger is documented with clear stakeholder mapping
  • Check: Evidence gaps are explicitly identified before research begins
  • Check: Research methods directly address identified gaps
  • Check: Actions include embedded evaluation mechanisms
  • Metric: Time from investigation to actionable insight < 2 weeks

• Examples: <example_correct> Description: Investigating API performance degradation

  # INVESTIGATE: Define the problem space
  decision_trigger = {
      "issue": "API response time increased 40% last week",
      "stakeholders": ["engineering", "customer_success", "product"],
      "success_metrics": {"p95_latency": "<200ms", "error_rate": "<0.1%"}
  }
  
  evidence_audit = {
      "available": ["APM metrics", "error logs", "deployment history"],
      "gaps": ["user impact data", "root cause analysis", "comparative benchmarks"]
  }
  
  # RESEARCH: Fill evidence gaps with targeted methods
  research_plan = {
      "user_impact": "Query customer support tickets + session replays",
      "root_cause": "Distributed tracing + code profiling",
      "benchmarks": "Load testing with previous versions"
  }
  
  # ACT: Implement with measurement
  pilot_rollout = {
      "solution": "Implement caching layer",
      "test_group": "10% of traffic",
      "metrics": ["latency", "cache_hit_rate", "error_rate"],
      "rollback_criteria": "error_rate > 0.5%"
  }
  

  </example_correct>

  <example_incorrect> Description: Jumping to solutions without investigation

  # BAD: No evidence audit or gap analysis
  def fix_performance():
      # Assumption-based solution
      add_more_servers()  # No investigation of actual bottleneck
      
      # No research phase
      deploy_to_production()  # No pilot or measurement
      
      # No reflection mechanism
      close_ticket()  # No learning captured
  

  </example_incorrect>

• Integration Notes:

  • Works best with: OODA loop for rapid iteration cycles
  • May conflict with: Pure brainstorming approaches that skip evidence gathering
  • Performance impact: Adds ~1-2 days upfront but reduces rework by 60%

• Implementation Checklist:

  • Create decision proposal template with stakeholder matrix
  • Set up evidence repository with gap tracking
  • Define research method selection criteria
  • Establish pilot success/failure thresholds
  • Schedule retrospective before starting next cycle

• References:

  • User Interviews Decision-Driven Research
  • Evidence-Based Decision Making in Healthcare
  • Internal: @patterns/research-synthesis-methods.mdc