Growth Model

Automated Growth Model Diagnosis

Core Principles

1. Model Matches Product Essence: PLG/SLG/MLG is not a choice but an inevitable result of product characteristics and business model
2. Flywheel Must Be a Closed Loop: The growth flywheel must form a reinforcing loop; an open loop is a chain, not a flywheel
3. Bottleneck Determines Leverage: The current biggest bottleneck determines the highest-leverage investment direction; resources always go to the bottleneck

Interaction Mode

🤖→👤 AI Suggests, Human Approves

Input

| Input Item | Type | Required | Source | Description | |--------|------|------|------|------| | Product characteristics | object | Yes | User provided | Product type, core features, value proposition | | User data | object | Yes | output/pm-metrics-ops/analysis-retention/retention_analysis.json | User behavior, conversion funnel, retention curve | | Business model | object | Yes | User provided | Pricing strategy, target customers, market positioning |

Execution Steps

Step 1: Growth Model Matching Decision Tree [Core]

Analyze the following dimensions to determine the optimal growth model:

PLG (Product-Led Growth) Characteristics

• Product can independently deliver user value
• Users can self-serve registration and usage
• Network effects exist or value increases with usage
• Word-of-mouth is an important acquisition channel

SLG (Sales-Led Growth) Characteristics

• High deal value (complex B2B decisions)
• Requires human demos and customized services
• Sales team is the core acquisition engine
• Customer success is key to retention

MLG (Marketing-Led Growth) Characteristics

• Brand awareness is a prerequisite for purchase
• Content marketing and SEO are important channels
• Requires sustained marketing investment to maintain growth
• Product is relatively standardized

Hybrid Model Determination

• Different user groups adopt different growth models
• Different product lines adopt different growth models
• Different market stages adopt different growth models

Step 2: Flywheel Auto-Modeling [Core]

Based on the identified growth model, build the growth flywheel model:

1. Identify Core Value Loop: Find the core causal chain of product value creation
2. Identify Flywheel Nodes: Key user behaviors and business metrics
3. Identify Reinforcing Loops: Which nodes positively reinforce other nodes
4. Identify Resistance Points: Sources of friction when the flywheel turns

Step 3: Cold Start Threshold Identification [Core]

Analyze the cold start conditions of the growth flywheel:

• How many initial users/revenue are needed to trigger flywheel self-rotation?
• What external resources are needed during the cold start phase?
• How to validate the flywheel hypothesis?

Step 4: Key Leverage Identification [Core]

Based on the flywheel model, identify the highest-leverage growth actions for the current stage:

• Which node, if strengthened first, would bring the greatest flywheel acceleration?
• Which bottleneck, if eliminated, would unlock the most growth potential?
• Where should resources be prioritized?

Output Depth Tiering

| Depth Level | Output Scope | Description | |----------|----------|------| | quick | Growth model diagnosis and bottleneck identification | Core conclusions + minimum viable output | | standard | Full output (current default) | Complete output including all Step outputs | | deep | Full diagnosis + Flywheel modeling inference + Cold start simulation + Growth stage evolution roadmap | Full output + extended analysis + deep inference |

Output

Storage Path: output/pm-growth/growth-model/

Output Files: growth_model.json

Output Schema:

{
  "type": "object",
  "required": ["model", "flywheel", "bottleneck"],
  "properties": {
    "model": {"type": "string", "description": "Growth model: PLG/SLG/MLG/Hybrid"},
    "flywheel": {"type": "object", "description": "Growth flywheel model, including nodes and edges"},
    "key_constraints": {"type": "array", "description": "Key constraint list"},
    "bottleneck": {"type": "string", "description": "Current biggest bottleneck description"},
    "confidence": {"type": "number", "description": "Diagnosis confidence"}
  }
}

growth_diagnosis

{
  "model": "PLG|SLG|MLG|Hybrid",
  "flywheel": {
    "nodes": ["Teachers register and use", "Create and publish courses", "Students join and learn", "Learning data feedback", "Word-of-mouth referral spread"],
    "edges": [{"from": "Students join and learn", "to": "Word-of-mouth referral spread", "description": "The better the student learning outcomes, the more willing teachers are to recommend to peers"}]
  },
  "key_constraints": ["Free version limited to 3 courses, affecting teacher deep usage"],
  "bottleneck": "Teacher activation rate only 35%, course creation barrier too high",
  "confidence": 0.95
}

Diagnosis Output Example

Growth Model: Hybrid (PLG + SLG)

Growth Flywheel:
├── PLG Flywheel: User registration → Use product → Discover value → Word-of-mouth referral → New user registration
├── SLG Flywheel: Marketing campaign → Lead generation → Sales follow-up → Enterprise purchase → Customer success → Upsell

Key Constraints:
1. PLG side: Free-to-paid conversion rate only 2.3%, need to optimize payment funnel
2. SLG side: Average sales cycle 45 days, lead conversion rate 12%

Current Biggest Bottleneck: PLG user activation rate is low (35%), resulting in insufficient word-of-mouth referrals

Recommended Priority Actions:
1. Optimize Onboarding flow, target activation rate increase to 50%
2. Identify common behavioral characteristics of high-activation users
3. Design activation intervention strategy for low-activation users

Output Validation Rules

| Field Path | Type | Required | Description | |----------|------|------|------| | model | string | Yes | Growth model, only allows PLG/SLG/MLG/Hybrid values | | flywheel | object | Yes | Flywheel model, must include nodes and edges | | flywheel.nodes | array | Yes | Flywheel node list, at least 4 nodes | | flywheel.nodes[].nodename | string | Yes | Node name, cannot be empty | | flywheel.edges | array | Yes | Flywheel edge list, at least 2 edges, must include from/to/description | | flywheel.edges[].from | string | Yes | Source node, cannot be empty | | flywheel.edges[].to | string | Yes | Target node, cannot be empty | | flywheel.edges[].description | string | Yes | Causal relationship description, cannot be empty | | keyconstraints | array | Yes | Key constraint list, maximum 5 | | keyconstraints[].constraint | string | Yes | Constraint description, cannot be empty | | keyconstraints[].impact | string | No | Impact assessment | | keyconstraints[].suggestedaction | string | No | Suggested action | | bottleneck | string | Yes | Bottleneck description, cannot be empty | | confidence | number | Yes | Diagnosis confidence, range 0-1 |

Decision Rules

| Condition | Decision | |------|------| | Product self-service completion rate ≥60% + Viral coefficient K>1 | Recommend PLG model | | Deal value ≥50K CNY + Sales cycle ≥30 days | Recommend SLG model | | Content-driven acquisition proportion ≥40% | Recommend MLG model | | None of the above conditions clearly met | Recommend hybrid model, note needs validation | | Growth flywheel self-drive score ≥7/10 | Mark as "can auto-execute" | | Growth flywheel self-drive score <7/10 | Mark as "requires human intervention", human final confirmation of growth model | | Bottleneck constraints ≥3 | Prioritize resolving the 1 with highest constraint degree, rest on watch list | | North Star metric misaligned with current growth model | Recommend re-evaluating growth model |

Quality Checks

P0 Checks (must pass for quick/standard/deep)

• [ ] North Star metric is directly linked to ≥1 OKR Objective
• [ ] Growth model includes ≥3 quantifiable variables with clear causal relationships between variables

P1 Checks (must pass for standard/deep)

• [ ] Input variables are 100% trackable (have data source or collection plan)
• [ ] Each diagnostic recommendation cites at least 1 data point
• [ ] Growth flywheel includes ≥4 nodes and forms a closed loop
• [ ] Bottleneck constraints identified ≤5, each with quantified impact assessment

P2 Checks (only deep must pass)

• [ ] Extended analysis is complete (deep inference and roadmap generated)
• [ ] Decision records are complete (key decisions have rationale and alternatives)

Degradation Strategy

Upstream File Missing Degradation Plan

| Missing Upstream Input | Degradation Plan | Output Impact | Data Acquisition Instructions | |----------|----------|----------|------------| | Product characteristics missing | User describes product → Diagnose growth model based on description | Product characteristics based on user description, diagnosis precision limited | Request user to provide product description (what the product is, what problem it solves, core value proposition) | | User data missing | Skip data-driven growth stage determination, infer based on user description | Growth stage determination based on qualitative description | Request user to provide current growth stage and core growth metrics (e.g., DAU, GMV, MRR, etc.) | | Business model missing | Use generic business model template, mark as "to be confirmed" | Business model fit may not be high | Request user to provide business model type (subscription/transaction/advertising/platform, etc.) and revenue sources | | Product characteristics + User data + Business model all missing | User describes product → Diagnose growth model based on description | Output is growth diagnosis based on description, key parameters marked as "to be confirmed" | Request user to provide product description, growth stage, and business model information |

Data Acquisition Instructions

When upstream files are missing, the user needs to provide the following information to support degraded generation:

• Product Description: What the product is, what problem it solves, core value proposition
• Current Growth Stage (optional): Product is in exploration/growth/maturity/decline phase
• Core Growth Metrics (optional): Currently most important growth metrics (e.g., DAU, GMV, MRR, etc.)

Upstream Change Response

Upstream Change Impact Table

| Upstream Source | Change Type | Impact Scope | Response Action | |----------|----------|----------|----------| | analysis-retention | Retention curve shape change | Growth model determination and flywheel modeling | Re-evaluate growth model, adjust flywheel nodes | | User provided - Product characteristics | Major product feature change | PLG/SLG/MLG model matching | Re-run decision tree, update model determination | | User provided - Business model | Pricing or target customer change | Growth model matching and bottleneck identification | Re-evaluate business model fit |

Downstream Notification Mechanism Table

| Downstream Consumer | Notification Condition | Notification Method | Notification Content | |------------|----------|----------|----------| | growth-strategy-report | Growth model or bottleneck change | Write to output file | New growth model, flywheel model, and bottleneck identification | | acquisition-orchestrator | Growth model change | Output file updated | Model diagnosis completion status and key conclusions | | activation-orchestrator | Growth model change | Output file updated | Model diagnosis completion status and key conclusions | | retention-orchestrator | Growth model change | Output file updated | Model diagnosis completion status and key conclusions | | revenue-orchestrator | Growth model change | Output file updated | Model diagnosis completion status and key conclusions |

Source & license

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

• Author: LuckyOneTwoThree
• Source: LuckyOneTwoThree/vibe-skill
• License: MIT
• Homepage: https://luckyonetwothree.github.io/all-skill-html/

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