Plan B

Plan B Framework

When to Apply

• After finalizing any Strategy Kernel (mandatory)
• During periodic environmental scans
• When early warning KPIs trigger
• When buffer enters YELLOW or RED
• When the founder is emotionally attached to a single plan

Core Framework

6 Environmental Forces

Monitor continuously. Any shift can invalidate the current strategy:

| Force | What to Monitor | Trigger | |---|---|---| | Technology | Competitor launches, new tools, paradigm shifts | New direct competitor or substitute | | Cost structure | API costs, tool pricing, entry barriers | Unit economics change >20% | | Customer needs | Support themes, feature requests, win/loss | NPS drop >10 or new theme >30% of tickets | | Target customer | Cohort behavior, demographic changes | Best-retaining cohort shifts segment | | Regulation | Laws, compliance, industry rules | New regulation affecting operations | | Social mood | Sentiment, brand perception, trends | Sentiment spike or channel backlash |

Multi-force escalation: 2+ forces shifting simultaneously = "Strategy revalidation recommended."

Pre-Mortem (5-Step Process)

1. Assume total failure: "It is 6 months from now. This strategy has completely failed."
2. Enumerate 3-5 specific failure causes (concrete, not vague)
3. For each cause define: prevention, Plan B, early warning KPI, trigger threshold
4. Rank by likelihood x impact
5. Embed triggers into monitoring for automatic Plan B activation

Scenario Planning with Auto-Triggers

Four scenarios with QUANTITATIVE trigger points:

| Scenario | Trigger | Auto-Response | |---|---|---| | OPTIMISTIC | Exceeds target by 20%+ | Continue scaling, increase targets | | BASE | 80-120% of target | Maintain strategy, increase constraint effort | | PESSIMISTIC | Buffer RED + growth 90% + negative growth for 2 weeks | Full strategy reset |

Auto-activation removes bias. Pre-commit to switching. Don't decide in the moment when sunk cost fallacy, loss aversion, and status quo bias are strongest.

Devil's Advocate (On Every Proposal)

Standard output, not optional add-on:

Proposed: "Launch LinkedIn content campaign targeting CTOs"

Devil's Advocate:
  1. "LinkedIn organic reach declining ~15% YoY."
  2. "Segment data shows email outperforms for >$50K deals."
  3. "Requires ~10h/week founder time — 25% of capacity."

  Risk: MODERATE
  Plan B if underperforms: Warm intros via mutual connections.
  Review gate: 30 days.

100-Day Gate Reviews

After any Plan B activation:

Day 30:  [milestone] -> continue / modify / exit?
Day 60:  [milestone] -> continue / modify / exit?
Day 100: [milestone] -> continue / modify / exit?

Each gate has explicit, measurable decision criteria. Prevents drifting in the new plan.

S-Curve Awareness

Monitor where each Driver/channel sits on its lifecycle:

| Stage | Agent Behavior | |---|---| | Introduction | High effort, low returns — expected. Keep investing. | | Growth | Returns accelerating — prepare Plan B NOW | | Maturity | Returns plateauing — activate diversification | | Decline | Returns dropping — Plan B should already be active |

Critical insight: Plan B must be PREPARED during the growth phase. During decline, it's too late — you're reacting, not preparing.

5 Hurdles to Plan B Execution

Obstacles that prevent organizations from switching to Plan B:

| Hurdle | Description | Countermeasure | |---|---|---| | Sunk cost attachment | "We've invested too much to stop" | Pre-committed stop conditions (defined at launch) | | Identity attachment | "This is my idea / my baby" | Reframe: "The goal is the outcome, not the method" | | Status quo bias | "Maybe it'll turn around" | Quantitative triggers remove judgment from decision | | Loss aversion | "Stopping feels like losing" | Frame as "redirecting resources to higher-impact approach" | | Social commitment | "We told investors/team this plan" | Distinguish pitch narrative from operational reality |

Boomerang Prevention

When switching to Plan B, ensure you don't oscillate back to Plan A:

Plan B Activation:
  1. Log WHY Plan A was abandoned (specific data, not feelings)
  2. Define conditions under which Plan A would be reconsidered
     (must be DIFFERENT conditions, not just "things got better")
  3. Set minimum commitment period for Plan B (typically 60-90 days)
  4. Track Plan B independently — don't compare to Plan A's
     peak performance (cherry-picking)

The agent monitors for oscillation patterns (A->B->A) and halts: "You're oscillating. Each reversal wastes accumulated progress. Commit to the current plan for [minimum period] before reconsidering."

Decision Rules

1. Plan B is mandatory — no Strategy Kernel without contingency
2. Auto-triggers are pre-committed — quantitative, not judgment-based
3. Devil's Advocate on every proposal — standard output, not optional
4. Prepare during growth — S-curve timing is critical
5. 100-day gates after activation — prevent drifting
6. No oscillation — minimum commitment period before switching back
7. 5 hurdles are psychological — countermeasures are structural (pre-commitment, triggers)

Anti-Patterns to Detect

| Anti-Pattern | Signal | Response | |---|---|---| | Single-plan thinking | No Plan B defined | "What happens if this fails? Define triggers and alternatives now." | | Late preparation | No contingency until crisis | "S-curve: prepare Plan B during growth, not decline." | | Bias-driven switching | Emotional reaction to single bad week | "Check against pre-committed triggers. Single-week data is noise." | | Oscillation | A->B->A pattern | "Each reversal wastes progress. Commit for [minimum period]." | | Plan B avoidance | Triggered conditions ignored | "Pre-committed conditions met. Present the switching decision." | | Missing Devil's Advocate | Proposal without counter-arguments | "Every proposal includes structured counter-arguments. Add them." |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: masterleopold
• Source: masterleopold/anty-framework
• License: MIT
• Homepage: https://4n7y.com

Install and usage instructions live in the source repository linked above.