growth-framework

name	growth-framework
description	Build a systematic growth framework covering acquisition, activation, retention, referral, and revenue. Outputs growth model, lever identification, experiment roadmap, and tracking dashboard.
argument-hint	["business model","growth stage","current bottleneck","team size","key channels"]
allowed-tools	Read, Write

Growth Framework

A growth framework makes growth systematic rather than opportunistic. It maps the full customer lifecycle, identifies the biggest bottlenecks, and generates a prioritised experiment roadmap. Growth is not a single tactic — it is a repeatable system for finding, fixing, and scaling the levers that move your North Star Metric.

AARRR Growth Model (Pirate Metrics)

ACQUISITION  → ACTIVATION → RETENTION → REFERRAL → REVENUE
  How do         First        Do they      Do they    Do we
  users find     value        come back?   tell       make
  you?           moment?                   others?    money?

Each stage is a funnel:
- Measure conversion rate at each stage
- Identify the biggest drop-off (the bottleneck)
- Focus experiments on the bottleneck

Growth Model Worksheet

# Growth Model: [Product Name]

Date: 2024-03-15 | Owner: Growth Lead

## Funnel Metrics (current baseline)

| Stage | Metric | Current | Target | Gap |
|-------|--------|---------|--------|-----|
| ACQUISITION | Weekly new trial signups | 450 | 600 | 150 |
| ACTIVATION | Signup → first core action (7 days) | 28% | 45% | 17pp |
| RETENTION | Month 1 → Month 2 | 52% | 65% | 13pp |
| REFERRAL | Referred signups / total signups | 8% | 20% | 12pp |
| REVENUE | Trial → paid conversion (30 days) | 22% | 30% | 8pp |

## Bottleneck Analysis

Weakest conversion: ACTIVATION (28%) — biggest absolute gap from target.

Evidence: Cohort data shows:
  - 62% of trial users never complete the first core action
  - Users who complete first core action in day 1 have 5× higher trial→paid conversion
  - Hotjar shows average user abandons on step 4 of 8-step onboarding

Conclusion: Activation is the primary bottleneck. 
A 10pp improvement in activation → estimated +120 paid accounts/month.

## Primary Growth Lever This Quarter: ACTIVATION

Secondary lever: RETENTION (because Month 1→2 retention affects LTV more than acquisition)

Growth Experiment Framework

# Growth Experiment: Streamlined Onboarding

**Hypothesis:** If we reduce the first-run experience from 8 steps to 3 steps
(focusing on the single action that predicts retention), then activation rate
will increase from 28% to ≥38% because users will reach value before losing momentum.

**Lever:** Activation  
**Primary metric:** % of new users who complete "first core action" within 7 days  
**Guardrail metrics:** Support contacts per user (must not increase), feature adoption (must not decrease)

**Test design:**
- Control: Current 8-step onboarding (50%)
- Variant: 3-step "quick win" onboarding (50%)
- Duration: 3 weeks (for statistical significance at expected volume)
- Sample size required: 850 per arm (at 80% power, α=0.05, min detectable effect: 8pp)

**Expected impact if successful:**
+10pp activation → +45 new activated users/week → +10 paid conversions/week → +$15k MRR

**Implementation:**
- Feature flag in LaunchDarkly: `onboarding_v2`
- Engineering estimate: 5 days (1 sprint)
- Rollout: 50/50 split by user_id hash

**Success criteria:**
- Primary: activation rate ≥ 36% in variant (statistically significant)
- Run for minimum 3 weeks before reading results

**Owner:** @growth-pm | **Engineer:** @growth-eng
**Start:** 2024-03-18 | **Read:** 2024-04-08

ICE Prioritisation Framework

from dataclasses import dataclass

@dataclass
class GrowthExperiment:
    name: str
    impact: int      # 1-10: Expected impact on NSM if successful
    confidence: int  # 1-10: How confident we are it will work (based on evidence)
    ease: int        # 1-10: How easy to implement (10 = trivial)
    
    @property
    def ice_score(self) -> float:
        return (self.impact * self.confidence * self.ease) / 10

experiments = [
    GrowthExperiment("3-step onboarding",          impact=8, confidence=7, ease=6),
    GrowthExperiment("In-app referral program",    impact=7, confidence=5, ease=4),
    GrowthExperiment("Email reactivation campaign",impact=5, confidence=8, ease=9),
    GrowthExperiment("Pricing page redesign",      impact=6, confidence=4, ease=5),
    GrowthExperiment("Integration marketplace",    impact=9, confidence=4, ease=2),
    GrowthExperiment("Contextual upgrade prompts", impact=6, confidence=7, ease=7),
]

 e  (experiments, key= x: x.ice_score, reverse=):
    ()
    

Experiment Roadmap Template

# Growth Experiment Roadmap — Q2 2024

## In Progress
| Experiment | Stage | Hypothesis | Start | Read | Owner |
|-----------|-------|-----------|-------|------|-------|
| 3-step onboarding | Running | +10pp activation | Mar 18 | Apr 8 | @growth-pm |
| Email trigger: day 3 nudge | Design | +5pp activation | Apr 1 | Apr 22 | @lifecycle |

## Backlog (prioritised by ICE)
| # | Experiment | ICE | Expected impact | Confidence |
|---|-----------|-----|----------------|-----------|
| 1 | Contextual upgrade prompt | 29 | +2pp trial→paid | High |
| 2 | Reactivation email sequence | 28 | +50 churned accounts | High |
| 3 | Referral reward program | 14 | +10% viral coefficient | Medium |
| 4 | Pricing page A/B | 12 | +3pp conversion | Medium |

## Completed (Last 90 days)
| Experiment | Result | Impact | Decision |
|-----------|--------|--------|---------|
| Social proof on signup page | Win (+4pp activation) | +18 paid/mo | Shipped |
| Progress bar in onboarding | No effect (p=0.31) | — | Reverted |
| Annual plan discount offer | Win (+8% annual %) | +$8k MRR | Shipped |

Viral / Referral Loop Design

## Referral Program Design

### Why users would refer:
Evidence: Top referral motivation in NPS verbatims: "My team needs to use this too"
Insight: Referrals are product-driven (need colleagues) not incentive-driven

### Referral mechanics:
1. SHARING PROMPT: After first successful workflow, show:
   "You've automated [X] hours this month! Share with your team →"
   
2. REFERRAL INCENTIVE:
   Referrer: 1 free month for each referred user who activates
   Referee: 30-day extended trial (vs standard 14)
   
3. IN-PRODUCT VIRAL:
   Shared workflow links require signup to view
   PDF exports include "Made with [Product]" watermark (with opt-out)
   
4. TRACKING:
   referral_link_created: when user generates referral link
   referral_link_clicked: when referee clicks
   referral_signup: when referee signs up via link
   referral_activated: when referee completes first core action

### K-factor calculation:
K = (Invites sent per user) × (Conversion rate of invites)
K > 1 = viral growth (referrals > churn)
Current K = 0.22 × 0.08 = 0.018 (well below viral threshold)
Target K = 0.40 × 0.20 = 0.08 (still not viral but meaningful contribution)

Growth Dashboard

## Weekly Growth Metrics — Week of 2024-03-11

### Funnel
| Stage | This Week | Last Week | WoW | Target | Status |
|-------|-----------|-----------|-----|--------|--------|
| New Trials | 487 | 451 | +8% | 600 | 🟡 |
| Activation Rate (7d) | 31% | 28% | +3pp | 45% | 🟡 |
| M1→M2 Retention | 54% | 52% | +2pp | 65% | 🟡 |
| Trial→Paid (30d) | 23% | 22% | +1pp | 30% | 🟡 |

### Experiments Running
- Onboarding v2: Day 12 | Variant ahead (34% vs 28%) | Not significant yet (p=0.14)

### This Week's Key Action
Activation rate improving — attributed to Day 3 email nudge experiment (shipped last week)
Continue monitoring; schedule qualitative interviews with non-activating users

Anti-Patterns to Avoid

Anti-Pattern	Problem	Fix
Optimising acquisition when activation is broken	More users entering a leaky funnel	Fix the biggest drop-off first
Running too many experiments	Split attention; insufficient traffic per experiment	2-3 running at a time maximum
Reading results too early	Novelty effect and sampling errors	Define minimum run duration before starting
No guardrails	Activation improves but NPS drops	Define guardrail metrics upfront
No experiment velocity tracking	Teams do experiments but not systematically	Track experiments per quarter; target 4-6/quarter
Ignoring qualitative data	Quantitative shows what, not why	Pair every experiment with user interviews

10 Rules

1. Fix the biggest bottleneck first — optimising a later funnel stage while losing at an earlier one is waste.
2. Impact × Confidence × Ease (ICE) prioritises experiments — gut feel doesn't scale.
3. Define success criteria before starting an experiment — never read early.
4. Run 2-3 experiments maximum at once — more is noise, not velocity.
5. Every experiment has a hypothesis, primary metric, and guardrail metric.
6. Sample size is calculated upfront — running until you see what you want is p-hacking.
7. Failed experiments are valuable — they rule out hypotheses and guide the next test.
8. Ship winning experiments immediately — learning without shipping is waste.
9. Retention compounds more than acquisition — a 10% improvement in retention beats a 30% increase in acquisition over 12 months.
10. Growth is a system — acquisition, activation, retention, referral, and revenue are interconnected; optimise the system, not just one stage.