andurilcode Scenario Planning
Core principle: The future is uncertain. Don't predict which future happens — map plausible futures and test strategies against each. Robust strategies perform reasonably across multiple scenarios. Fragile strategies only work if one specific future occurs.
The Core Process
Step 1: Define the Decision or Strategy
- Time horizon (6 months / 2 years / 5 years)?
- Decision needed now?
- What does "good" look like across futures?
Step 2: Identify Key Uncertainties
The 2–3 factors that are both:
- Highly uncertain (we genuinely don't know how they'll unfold)
- Highly impactful (they'd significantly change the right strategy)
These are the scenario axes. Skip certainties and minor factors.
Step 3: Build the Scenarios
3–4 distinct, internally consistent futures. Each:
- Plausible: not science fiction
- Distinct: meaningfully different from the others
- Challenging: at least one uncomfortable for the current plan
Two-axis matrix (two key uncertainties):
High market adoption
|
Slow tech ──┼── Fast tech
change | change
|
Low market adoption
→ 4 quadrant scenarios.
Three narrative scenarios:
- Expected: the future most are implicitly planning for
- Optimistic: uncertainties resolve favorably
- Pessimistic: uncertainties resolve adversarially
- Wild card: low-probability, high-impact disruption
Step 4: Stress-Test the Strategy
For each scenario:
- Current strategy: well / adequately / poorly?
- What would need to change to work in this scenario?
- Cost of being wrong about which scenario occurs?
Step 5: Identify Robust Actions
Actions that perform well across multiple scenarios — highest-confidence regardless of future.
Plus hedging options — small bets that preserve flexibility, cost little in expected case, pay off in unlikely ones.
Output Format
Key Uncertainties
| Uncertainty | Why it matters | Range |
|---|---|---|
| [Factor 1] | [Strategy impact] | [X to Y] |
| [Factor 2] | [Strategy impact] | [A to B] |
The Scenarios
For each (3–4 total):
Scenario Name (memorable)
- Description: 2–3 sentences
- Key conditions: what's true here?
- Probability estimate: rough (sum to ~100%)
- Key signals: early indicators we're in this scenario
Strategy Stress Test
| Scenario | Strategy performs | Why | What must change |
|---|---|---|---|
| A | Well | [Reason] | Nothing |
| B | Adequately | [Reason] | [Adjustment] |
| C | Poorly | [Reason] | [Major pivot] |
| D | Catastrophically | [Reason] | [Fundamental rethink] |
Robust Actions
Cross-scenario actions:
- [Action 1] — works because [cross-scenario reason]
- [Action 2] — works because [cross-scenario reason]
Scenario-Specific Actions
| Scenario | Trigger signal | Response |
|---|---|---|
| B | [Leading indicator] | [Action] |
| C | [Leading indicator] | [Action] |
Fragility Assessment
- Worst-handled scenario for current plan?
- Single assumption the plan most depends on?
- Cheapest hedge against the worst scenario?
Pitfalls
- Too many scenarios: 3–4 is ideal.
- Scenarios too similar: each must require different strategies.
- Anchoring on the expected: give real attention to uncomfortable ones.
- No early warning signals: every scenario needs leading indicators.
- Planning for the scenario, not the strategy: goal is resilience, not prediction.
Thinking Triggers
- "What are we implicitly assuming about the future?"
- "What's the world where this strategy fails completely — how likely?"
- "What would we do differently if we knew we were in Scenario C?"
- "What signals tell us which future we're heading into?"
- "What's the cheapest move that protects us in bad scenarios without sacrificing good ones?"
Example Applications
- Product roadmap: AI commoditizes the core feature? Regulation shifts? Major platform behavior change?
- Architecture: Traffic 10x? Third-party API sunsets? Team doubles?
- Business strategy: Funding tightens? Competitor undercuts? Adoption 5x faster?
- Agent pipeline: LLM costs drop 90%? Unlimited context? Single agent replaces pipeline?