---
name: butterfly-effect
display_name: Butterfly Effect
description: "Activate before making a customer decision to map downstream ripple effects — analyzes how a discount, exception, escalation, or commitment will cascade across other accounts, segments, and precedents. Also activates for 'what happens if I [action]', 'impact of [decision]', 'butterfly effect of [change]', 'ripple effect', 'precedent check'."
icon: "🦋"
trigger: butterfly effect
inputs:
  - name: proposed_action
    description: "The decision or action you're considering (e.g., 'approve 30% discount for Roblox EDP renewal', 'waive overage charges for Acme')"
    type: string
    required: true
  - name: account_name
    description: "The customer/account this action applies to"
    type: string
    required: false
tools: [kg_search, search_all, file_rag_search]
depends-on: [html_design]
---

# 🦋 Butterfly Effect — Impact Mapper

## Overview

Before you send that email, approve that discount, or make that exception — map the downstream ripple effects. The Butterfly Effect skill analyzes a proposed customer decision (discount, exception, escalation, commitment) and traces its cascade across other accounts, segments, and organizational policies. It searches for past precedents, identifies connected accounts that could be affected, quantifies the financial impact in dollars, and visualizes the entire ripple chain in an interactive diagram. Every analysis ends with a clear PROCEED / MODIFY / DON'T DO IT recommendation with specific reasoning.

## Workflow

### Step 1: Parse the Proposed Action

Classify the decision into one of these categories:
- **Pricing/Discount**: Discount percentage, credit, waived fees, custom pricing
- **Exception**: Deviation from standard process, policy override, SLA exception
- **Timeline Commitment**: Delivery date promise, feature timeline, migration deadline
- **Feature Promise**: Committing to roadmap items, custom development, integration work
- **Escalation**: Executive involvement, special handling, priority queue
- **Contractual Change**: Amendment, early termination, renewal terms modification

Extract key parameters:
- **Dollar value** of the action (discount amount, credit value, revenue at risk)
- **Duration/scope** (one-time vs. recurring, single account vs. segment-wide)
- **Reversibility** (can this be undone? What's the exit cost?)

### Step 2: Search for Precedent & Connections

Run parallel searches to build the intelligence picture:

**Search 1 — Past Precedents:**
```
kg_search: "[action type] discount/exception/waiver" + account segment
search_all: "approved discount" OR "exception granted" OR "waived fees" + similar context
file_rag_search: pricing exceptions, discount approvals, policy overrides
```

**Search 2 — Connected Accounts:**
```
kg_search: category="Organization" in same segment/tier/industry as the target account
search_all: other accounts with similar contract terms, size, or relationship
```

**Search 3 — Policy & Contractual Context:**
```
search_all: MFN clause, most-favored-nation, pricing parity, contract terms
file_rag_search: discount policy, approval thresholds, pricing guidelines
kg_search: related opportunities, renewal timelines for similar accounts
```

Look specifically for:
- **MFN (Most-Favored-Nation) clauses** — does any customer have contractual rights to match this deal?
- **Segment precedent** — have we done this before for similar customers? What happened?
- **Internal policy** — does this exceed approval thresholds? Who else needs to sign off?
- **Timing risks** — are other accounts in this segment up for renewal soon?

### Step 3: Map the Ripple Effects

Build a three-layer cascade analysis:

#### First-Order Effects (Direct — This Account)
- Immediate revenue impact (positive or negative)
- Customer satisfaction/retention probability change
- Contractual implications (amendment needed? Term extension?)
- Relationship dynamic shift (does this set expectations for future asks?)

#### Second-Order Effects (Adjacent — Other Accounts)
- **Same-segment customers** who could learn about this and request the same
- **Same sales rep/CSM portfolio** — will they use this as precedent for other deals?
- **Partner/channel impact** — does this affect partner margins or channel pricing?
- For each affected account, estimate:
  - Probability they learn about it (0-100%)
  - Probability they request the same (0-100%)
  - Revenue impact if they get the same treatment

#### Third-Order Effects (Systemic — Organization)
- **Policy changes** needed (do we need to update the discount matrix?)
- **Process changes** (new approval workflows, documentation requirements)
- **Competitive dynamics** (does this leak to competitors or set market expectations?)
- **Team behavior** (does this encourage more exception requests from the field?)

#### The "Do Nothing" Baseline
Always model what happens if you DON'T take the action:
- Churn probability if the request is denied
- Revenue at risk from the current account
- Relationship damage estimate
- Alternative paths (counter-offer, partial approval, delayed approval)

### Step 4: Quantify the Impact

Build a financial summary:

| Scenario | This Account | Segment Ripple | Systemic Impact | Net Impact |
|----------|-------------|----------------|-----------------|------------|
| Approve as-is | -$X direct | -$Y if N accounts request same | -$Z policy cost | -$Total |
| Modified version | -$X' reduced | -$Y' smaller ripple | Minimal | -$Total' |
| Decline | -$R churn risk | $0 | +$P policy integrity | Net +/- |
| Do nothing (delay) | -$D uncertainty | $0 short-term | $0 | -$D |

Where data is unavailable, use ranges and state assumptions clearly.

### Step 5: Visualize the Butterfly Effect

Create an HTML artifact using the html_design skill guidelines. The visualization must show:

**Layout:**
- **Center**: The proposed action in a prominent card with the account name and action summary
- **Ring 1 (First-Order)**: Direct effects on this account, shown as nodes connected to center
- **Ring 2 (Second-Order)**: Effects on other accounts/segments, connected to Ring 1 nodes
- **Ring 3 (Third-Order)**: Systemic/organizational effects, connected to Ring 2 nodes

**Each node shows:**
- Brief description of the effect
- Dollar impact (color-coded: red for negative, green for positive)
- Probability percentage
- Expected value (impact × probability)

**Design requirements:**
- Use concentric ripple rings emanating from center (like a butterfly wing or water ripple)
- Color intensity indicates severity (deeper red = higher negative impact)
- Interactive: hover on any node to see detail
- Summary cards at top: Total Expected Impact, Accounts Affected, Risk Level
- Use `var(--color-*)` theme tokens from html_design
- D3.js for the ripple/network visualization OR pure CSS/SVG concentric rings
- Responsive and clean — follow the html_design "Content First" principle

**Top summary cards:**
```
[Net Revenue Impact: -$47K] [Accounts at Risk: 3] [Precedent Risk: HIGH] [Recommendation: MODIFY]
```

**Bottom section:**
- Recommendation with reasoning
- Suggested modified approach (if applicable)
- Key risks to monitor if proceeding

### Step 6: Present the Recommendation

End with a clear, actionable recommendation:

**PROCEED** 🟢 — The ripple effects are contained and acceptable. State why.

**MODIFY** 🟡 — The action has merit but should be adjusted to limit downstream effects. Provide specific modifications:
- "Reduce discount from 30% to 20% to stay below the MFN trigger threshold"
- "Add a confidentiality clause to prevent precedent-setting"
- "Make it a one-time credit instead of a recurring discount"
- "Require a term extension in exchange"

**DON'T DO IT** 🔴 — The ripple effects significantly outweigh the benefits. Explain the cascade and offer alternatives:
- "Counter-offer with X instead"
- "Escalate to Y for a structural solution"
- "Defer until Z contract terms are renegotiated"

## Key Lessons

1. **Always quantify in dollars** — "this sets a bad precedent" is weak; "-$47K net revenue impact across 3 accounts" is actionable.
2. **Check for MFN clauses first** — most-favored-nation clauses in contracts can turn a single discount into a segment-wide price reduction automatically.
3. **Model both upside AND downside ripples** — sometimes an exception creates positive ripple effects (case study, reference customer, upsell opportunity).
4. **Account for the "do nothing" baseline** — the cost of inaction is often higher than the cost of the action. Always compare against it.
5. **Time-bound the analysis** — a discount during renewal season has different ripple effects than one in Q1 when no one else is negotiating.
6. **Consider information spread probability** — enterprise customers talk to each other at conferences, in Slack communities, and through mutual contacts. Estimate how likely the decision leaks.
7. **Separate one-time from recurring impact** — a one-time credit has much smaller ripple effects than a permanent price reduction.
8. **Flag irreversible decisions** — some actions can't be undone (public pricing change, contractual amendment). These deserve extra scrutiny.

## Output Format

Every Butterfly Effect analysis must include:
1. **Action Summary** — What's being decided, for whom, and why
2. **Precedent Check** — What we found from past similar decisions
3. **Ripple Map** — The three-ring cascade visualization (HTML artifact)
4. **Financial Impact Table** — Scenarios with dollar estimates
5. **Recommendation** — PROCEED / MODIFY / DON'T DO IT with reasoning
6. **If MODIFY**: Specific adjustments and their expected effect on the ripple map