Task D2.2: Extract Experiments from Investigations

What is Task D2.2?

Task D2.2 is part of the "Act" phase in the Viability Canvas methodology, specifically within the "Energy Map" step (Step D2). This task instructs you to "Extract experiments and, ideally, fixed activities from the Investigations, with which we can achieve more clarity about how to proceed."

Purpose of Extracting Experiments from Investigations

The purpose of this task is to convert larger, more ambiguous "investigations" (identified in the Decide phase) into concrete, manageable experiments that can be implemented and evaluated. This approach serves several important functions:

1. Breaking down complexity: Large-scale changes are converted into smaller, testable hypotheses
2. Reducing risk: Instead of committing to a full implementation, you first test ideas on a smaller scale
3. Generating learning: Experiments provide actionable data about what works and what doesn't
4. Building momentum: Small experiments can show early results and build support for larger changes
5. Creating clarity: Experiments help resolve uncertainties and answer open questions

This task is fundamental to an iterative approach to change, where you "probe, sense, and respond" rather than trying to plan everything in advance.

Understanding Investigations vs. Experiments

In the Viability Canvas methodology:

• Investigations are larger areas of inquiry identified during the "Decide" phase that require further exploration before full implementation. They represent potential improvements that have merit but contain significant uncertainties or unknowns.
• Experiments are small-scale, time-bounded tests designed to generate specific learning about how a potential change might work in practice. They have clear hypotheses, defined scopes, and evaluation criteria.

The key difference is that investigations are questions, while experiments are structured ways to answer those questions.

How to Complete Task D2.2

To extract experiments from investigations:

1. Review your investigations from the Decide phase. These are the improvement opportunities that were deemed too large, uncertain, or complex to implement directly.
2. For each investigation, identify key uncertainties that prevent you from moving forward with full implementation:
   • What don't you know?
   • What assumptions need validation?
   • What variables could impact success?
3. Design specific experiments to address these uncertainties:
   • Define a clear hypothesis (we believe that...)
   • Specify what you'll measure to evaluate the experiment
   • Determine the scope (who, what, where, when)
   • Set a timeframe (typically 2-6 weeks)
   • Identify resources needed
   • Clarify what success looks like
4. Prioritize experiments that:
   • Address the most critical uncertainties
   • Can be implemented with minimal resources
   • Will generate learning quickly
   • Have high potential impact if successful
5. Document each experiment with:
   • The originating investigation
   • The specific hypothesis being tested
   • Success criteria
   • Required resources
   • Timeline
   • Expected learning outcomes

Example from Canned Tornado

In the Canned Tornado case study, they identified experiments such as:

• Pilot project for self-controlling teams in a production line
  • From investigation: "Reorganization in value stream teams"
  • Hypothesis: Self-managing teams will improve coordination and reduce delays
  • Scope: Single production line
  • Timeline: 4 weeks
  • Metrics: Throughput time, quality metrics, team satisfaction
• Testing Kanban systems between selected process steps
  • From investigation: "Improved material management"
  • Hypothesis: Visual pull systems will reduce material shortages
  • Scope: Between cell production and module assembly
  • Timeline: 3 weeks
  • Metrics: Material availability, buffer sizes, stockouts
• Prototype of a digital dashboard for real-time data
  • From investigation: "Comprehensive ERP customization"
  • Hypothesis: Real-time visibility will improve decision-making
  • Scope: Key metrics for one production area
  • Timeline: 2 weeks
  • Metrics: Manager decision time, quality of decisions, user feedback

By extracting these specific experiments, Canned Tornado could begin testing key elements of their larger change vision without committing to full-scale implementation before validating their assumptions.

Best Practices for Designing Experiments

When extracting experiments from investigations:

1. Keep experiments small and focused – they should be completable within a few weeks
2. Ensure they test a specific hypothesis – not just implementing a solution
3. Design for learning – even "failed" experiments should generate valuable insights
4. Consider both technical and social dimensions – test both the solution and how people respond to it
5. Define clear success criteria upfront – know how you'll evaluate results
6. Select diverse experiments – cover different aspects of your larger investigations
7. Build in reflection time – schedule time to process learnings and implications

Well-designed experiments allow organizations to learn rapidly while minimizing risks, providing the clarity needed to determine whether to scale, modify, or abandon potential change initiatives.