Creator Analytics Flywheel

We'll show you how learners interact with your content through our proven 12-cell mastery system, guide you on what to build next based on real coverage gaps, and gradually connect your content to downstream trading improvements—all while keeping the system fair and explainable.

The Analytics Loop

Core Infrastructure

12-Cell Framework

4 domains (ANALYSIS, STRATEGY, RISK, EXECUTION) × 3 Bloom levels (Apply, Analyze, Evaluate)

Item Health System

Success rate, discrimination index, abandon rate, time Z-scores with automatic status progression

Mastery Tracking

Beta model per cell: E[θ] = α/(α+β) with spaced review triggers

Creator Analytics

Exposure-weighted impact, difference-in-differences uplift, coverage gap identification

Progressive Dashboard Views

View 1: Getting Started (Days 1-30)

For new creators with < 100 learners or in their first 30 days

┌─────────────────────────────────────┐
│ Your Content Health Score: 72/100   │
│ Active Learners (7d): 47            │
│ Items Published: 23                  │
│ Best Performing Cell: RISK-Apply    │
└─────────────────────────────────────┘

Coverage Grid (Your 12 Cells):
┌─────────┬────────┬─────────┬──────────┐
│         │ Apply  │ Analyze │ Evaluate │
├─────────┼────────┼─────────┼──────────┤
│ANALYSIS │   ✓3   │   ✓2    │    ⚠️1    │
│STRATEGY │   ✓4   │   📍0    │    📍0    │
│RISK     │   ✓5   │   ✓3    │    ✓2    │
│EXECUTION│   📍0   │   📍0    │    📍0    │
└─────────┴────────┴─────────┴──────────┘
✓ = Healthy  ⚠️ = Needs revision  📍 = Gap

Top Priority: Add EXECUTION content (67% need this)

Getting Started Dashboard

Health Score Calculation

Item Quality (40 pts)

Average discrimination × (1 - abandon_rate)

Coverage (30 pts)

Cells with ≥2 healthy items / 12

Engagement (30 pts)

Completion rate × median session length factor

View 2: Growing (100-500 learners)

Unlocked after 100+ learners OR 30+ days of activity

Mastery Flow Visualization

ANALYSIS-Apply (87%) → STRATEGY-Apply (73%) → RISK-Apply (68%)
                                ↓
                    [Struggling Here - Suggest Content]
                                ↓
                        RISK-Analyze (42%)

Your Impact: +8% mastery gain vs baseline
(confidence: medium, n=127)

Learner Journey Analysis

Item Performance Table

Item Title	Health Status	Success Rate	Discrimination	Action
Stop Loss Basics	stable	72%	0.31	—
Position Sizing	needs_edit	45%	0.08	[Revise]
Risk/Reward	stable	68%	0.28	—

View 3: Established (500+ learners)

Full analytics suite with behavioral bridge metrics

┌──────────────────────────────────────────┐
│ Your RISK-Apply Modules Improved:        │
├──────────────────────────────────────────┤
│ • Rules Compliance: +31% (89 competitors)│
│ • Loss Exit Discipline: +23% (147 comp.) │
│ • Position Size Consistency: +18% (112)  │
│                                          │
│ Your RISK-Analyze Modules Reduced:       │
│ • Revenge Trading: -42% (203 competitors)│
│ • Pressure Performance Delta: -15% (156) │
│                                          │
│ Sample: 12 competitions, 2,847 trades    │
│ Confidence: HIGH (structured constraints) │
└──────────────────────────────────────────┘

Competition Attribution Dashboard

Pressure Performance Delta measures behavioral consistency between practice and competition trading. A -15% improvement means 15% less deviation under competitive pressure.

Content Creation Flow

Spec-First Generation

{
  "domain": "RISK",
  "bloom": "Analyze",
  "constraints": ["policy-level only"],
  "grounding": ["doc://your_material#section3"],
  "expected_impact": {
    "addresses_gap": true,
    "learner_need": "68% struggle here after RISK-Apply",
    "competition_relevance": "reduces oversizing violations"
  }
}

AI Pipeline Integration

Publishing Timeline

Immediate (Same Day)

View attempt rate, completion rate, cell coverage

After 50 Attempts (3-7 days)

Initial health metrics, discrimination index, time normalization

After 200 Attempts (2-3 weeks)

Full health status, revision suggestions, impact preview

The Quality Flywheel

Stage 1: Measure

Automatic tracking of attempts, mastery updates, gap identification

Stage 2: Analyze

Weekly confusion reports, health monitoring, coverage analytics

Stage 3: Recommend

Real-time suggestions based on gaps, audience progression, behaviors

Stage 4: Create/Revise

One-click templates, revision workflow, A/B testing

Stage 5: Validate

Performance comparison, mastery improvements, behavioral tracking

Implementation Phases

Phase A: Foundation (Months 1-2)

Focus on basic metrics and coverage

Impact MVP with sample gates
Coverage bounties for underserved cells
Basic creator following
Rules-based struggle detection

Phase B: Behavioral Bridge (Months 3-6)

Connect content to trading behaviors

Revenue share pilot with quality multiplier
Impact dashboards
Asynchronous cohorts
Mastery flow visualization

Phase C: Full Attribution (Months 6-12)

Complete impact measurement and prediction

Collaborative filtering
Uplift modeling
Performance forecasting
Competition impact attribution

What Remains Unchanged

These core systems stay separate to maintain fairness

XP System

XP rewards engagement only
Never tied to difficulty or creator
No "bonus XP" items

Privacy & Fairness

All views are aggregated
Minimum sample requirements
Market condition normalization
No individual learner data

Core Classification

12-cell matrix unchanged
Hybrid classifier consistency
Precedence rules prevent drift

Success Metrics

Platform Health (Monthly)

Classification reliability: κ & AC1 ≥ 0.75
Item health: < 5% in quarantine
Coverage: All cells have ≥10 healthy items

Creator Success (Quarterly)

80% of creators check dashboard weekly
60% of flagged items revised within 14 days
40% improvement in underserved cells

Learner Outcomes

D7 retention ≥ 40%, D30 ≥ 20%
+8-12% mastery gain in targeted cells
< 5% recommendations to unhealthy items

Communication Framework

Always Tell Creators

About Their Content

Exact health status and why
Sample sizes behind metrics
Confidence levels for impact
Platform fit assessment

About The System

Classification changes
Flagging reasons
How recommendations work
Revenue sharing drivers

Other creators' exact metrics
Unsubstantiated causation claims
Predictions without confidence intervals

Technical Integration

Data Sources

-- From questions table
health_discrimination, abandon_rate, 
median_time_ms, health_status

-- From user_stats
mastery_by_cell (Beta parameters)

-- From creator_analytics tables
creator_exposures_daily, creator_impact_daily

Dashboard Architecture

Data Layer

Pull from existing analytics rollups

Confidence Layer

Apply established calculations and sample thresholds

Presentation Layer

Cache for performance, progressive disclosure

Recommendation Engine

Hook into coverage analytics and mastery flows

This approach uses everything already built in the quiz architecture, presents it progressively so creators aren't overwhelmed, and maintains system integrity while scaling naturally with the Phase A→B→C plan.