---
name: Kill Check Base Framework
description: Shared evaluation framework for GO/PIVOT/KILL decisions. Extended by phase-specific kill checks (0.4 Validation, 6.4 Launch, 7.6 Growth).
version: 1.0.0
type: engine
---

# Kill Check Base Framework

> Universal evaluation framework for data-backed GO / PIVOT / KILL decisions. This is a base template -- each phase extends it with specific thresholds and data sources.

---

## Overview

Kill checks are structured decision points that prevent wasting time on ideas/products that won't work. They consume real data (not vibes) and output a clear verdict with reasoning.

**Three kill checks in the system:**
- **Phase 0.4** -- Validation Kill Check (should we build this?)
- **Phase 6.4** -- Launch Kill Check (is this getting traction?)
- **Phase 7.6** -- Growth Kill Check (should we keep investing?)

All three share this base framework.

---

## Verdict Definitions

| Verdict | Meaning | Action |
|---------|---------|--------|
| **GO** | Data supports continuing | Proceed to next phase |
| **PIVOT** | Partial signal, wrong angle | Change approach (audience, positioning, channel, pricing), re-run check |
| **KILL** | No viable signal | Stop immediately, document learnings, move to next idea |

---

## Weighted Scoring Framework

Adapted from business-idea-multi-agent (ADR-005).

### Base Dimensions

Each dimension scores 1-10. Weights are phase-specific (override below).

| Dimension | What It Measures | Default Weight |
|-----------|-----------------|----------------|
| **Market Potential** | Size, growth rate, demand signals | 25% |
| **Blue Ocean Score** | Competitive whitespace, differentiation | 25% |
| **Disruption Potential** | How much better than status quo | 20% |
| **Technical Feasibility** | Build complexity, time to MVP | 15% |
| **Capital Efficiency** | Cost to validate / cost to scale | 15% |

### Score Calculation

```
base_score = (
  market_potential * market_weight +
  blue_ocean * blue_ocean_weight +
  disruption * disruption_weight +
  tech_feasibility * tech_weight +
  capital_efficiency * capital_weight
)

risk_adjustment = sum(red_flag_penalties)

final_score = base_score - risk_adjustment
```

### Score Thresholds

| Score Range | Default Verdict | Notes |
|-------------|----------------|-------|
| 7.0 - 10.0 | **GO** | Strong signal across dimensions |
| 4.0 - 6.9 | **PIVOT** | Some signal but not enough; needs repositioning |
| 0.0 - 3.9 | **KILL** | Insufficient signal; move on |

*Phase-specific kill checks may adjust these thresholds.*

---

## Red Flag Taxonomy

Adapted from Critical Briefs (invalidation-first methodology).

### Category A: Fatal Flags (Score Penalty: -2.0 each)

| Flag | Description | Signal |
|------|-------------|--------|
| **No demand signal** | Zero search volume, no forum discussions, no complaints about status quo | Market doesn't exist |
| **Dominant incumbent** | Single player with >70% market share AND high satisfaction | No room to enter |
| **Regulatory barrier** | Requires licenses, certifications, or legal approvals to operate | Impossible for solo/small team |
| **Technical impossibility** | Core feature depends on technology that doesn't exist yet | Can't be built |

### Category B: Serious Flags (Score Penalty: -1.0 each)

| Flag | Description | Signal |
|------|-------------|--------|
| **Crowded market** | 10+ well-funded competitors with similar positioning | Hard to differentiate |
| **Declining market** | Search volume trending down, industry contracting | Shrinking opportunity |
| **High CAC** | Customer acquisition requires paid channels with no organic path | Unsustainable unit economics |
| **Complex sales cycle** | Enterprise sales, multiple stakeholders, long decision time | Solo founder can't sell |
| **Hardware dependency** | Requires physical devices, inventory, or shipping | Capital intensive |

### Category C: Warning Flags (Score Penalty: -0.5 each)

| Flag | Description | Signal |
|------|-------------|--------|
| **Seasonal demand** | Interest spikes only at certain times | Revenue volatility |
| **Platform dependency** | Relies on single platform's API/algorithms | Existential risk |
| **Niche too small** | TAM under $10M | Limited upside |
| **Feature, not product** | Could be a feature in an existing product | Acquisition risk |

---

## Skepticism Protocol

Apply skepticism BEFORE scoring. The default assumption is KILL. Data must PROVE otherwise.

### Pre-Scoring Questions

1. **Am I seeing what I want to see?** -- Confirmation bias check
2. **Would I invest $10K of my own money based on this data alone?** -- Skin in the game test
3. **If a friend showed me this data, what would I tell them?** -- Outside view
4. **What's the strongest argument AGAINST this?** -- Steelman the opposition
5. **What data would change my mind?** -- Falsifiability check

### Invalidation-First Process

```
1. Start with assumption: KILL
2. Look for reasons to promote to PIVOT
3. Look for reasons to promote to GO
4. Each promotion requires SPECIFIC DATA (not feelings)
5. Any fatal flag locks verdict at KILL regardless of score
```

---

## Output Format

All kill checks produce this structured output:

```json
{
  "verdict": "GO | PIVOT | KILL",
  "score": {
    "final": 0.0,
    "breakdown": {
      "market_potential": { "score": 0, "weight": 0.25, "evidence": "" },
      "blue_ocean": { "score": 0, "weight": 0.25, "evidence": "" },
      "disruption": { "score": 0, "weight": 0.20, "evidence": "" },
      "tech_feasibility": { "score": 0, "weight": 0.15, "evidence": "" },
      "capital_efficiency": { "score": 0, "weight": 0.15, "evidence": "" }
    },
    "risk_adjustment": 0.0,
    "red_flags": []
  },
  "reasoning": {
    "strongest_signal": "Best data point supporting the idea",
    "biggest_risk": "Most concerning finding",
    "skepticism_check": "What the pre-scoring questions revealed"
  },
  "next_actions": {
    "if_go": ["Specific next steps"],
    "if_pivot": ["Specific pivot suggestions with rationale"],
    "if_kill": ["Document learnings", "What to explore next"]
  },
  "data_sources": ["List of MCP servers and data points used"],
  "confidence": "HIGH | MEDIUM | LOW (based on data quality/quantity)"
}
```

---

## Phase-Specific Extensions

### Override Points

Each phase-specific kill check overrides:

1. **`[WEIGHTS]`** -- Dimension weights (must sum to 100%)
2. **`[THRESHOLDS]`** -- Score ranges for GO/PIVOT/KILL
3. **`[DATA_SOURCES]`** -- Which MCP servers provide the input data
4. **`[CRITERIA]`** -- Phase-specific pass/fail criteria
5. **`[CADENCE]`** -- How often to re-run (one-time, 48hr, monthly)
6. **`[RED_FLAGS]`** -- Additional phase-specific red flags

### Phase Map

| Kill Check | Phase | Key Data Sources | Cadence |
|-----------|-------|-----------------|---------|
| Validation (0.4) | Pre-build | Reddit Research, Google Trends, Exa.ai, Firecrawl | One-time |
| Launch (6.4) | Post-launch | GA4, PostHog, Stripe | 48-hour intervals |
| Growth (7.6) | Ongoing | GA4, Stripe, Metricool, social platform analytics | Monthly |

---

## Decision Tree

```
START
  |
  v
[Collect Data from MCP Servers]
  |
  v
[Run Skepticism Protocol]
  |
  v
[Check for Fatal Flags (Category A)]
  |
  +-- Any fatal flag found? --> KILL (immediate, no scoring needed)
  |
  v
[Score Each Dimension 1-10 with Evidence]
  |
  v
[Apply Weights]
  |
  v
[Calculate Base Score]
  |
  v
[Apply Risk Adjustment (Category B + C penalties)]
  |
  v
[Calculate Final Score]
  |
  v
[Apply Phase-Specific Thresholds]
  |
  +-- Score >= GO threshold --> GO
  +-- Score >= PIVOT threshold --> PIVOT
  +-- Score < PIVOT threshold --> KILL
  |
  v
[Generate Structured Output with Evidence]
  |
  v
[Include Next Actions for the Verdict Given]
  |
  v
END
```

---

*This base framework is extended by: `phase-0-validation/0.4-kill-check/SKILL.md`, `phase-6-launch/6.4-launch-kill-check/SKILL.md`, `phase-7-growth/7.6-growth-kill-check/SKILL.md`*