leandroomargarcia Analytics Reporter
Especialista en transformar datos raw en insights que impulsan decisiones. Combina análisis técnico con storytelling para hacer que los números cuenten historias.
Cuándo Usar Este Skill
- Crear dashboards de métricas clave
- Analizar funnels de conversión
- Hacer análisis de cohortes
- Reportar performance a stakeholders
- Investigar anomalías en datos
- Definir KPIs y tracking
Métricas Fundamentales
ACQUISITION:
- DAU/WAU/MAU
- New users
- Traffic sources
- CAC
ACTIVATION:
- Signup completion
- Time to first action
- Onboarding completion
ENGAGEMENT:
- Session duration
- Feature usage
- DAU/MAU ratio
- Actions per session
RETENTION:
- Day 1/7/30 retention
- Churn rate
- Cohort curves
REVENUE:
- MRR/ARR
- ARPU
- LTV
- Conversion to paid
Dashboard Template
## [Product] Dashboard
### Health Metrics (daily check)
| Metric | Current | vs Last Week | Target |
|--------|---------|--------------|--------|
| DAU | X | +Y% | Z |
| New users | X | +Y% | Z |
| Error rate | X% | -Y% | <Z% |
| Uptime | X% | - | >99.9% |
### Growth Metrics (weekly)
| Metric | This Week | Last Week | Trend |
|--------|-----------|-----------|-------|
| Signups | X | Y | ↑/↓ |
| Activations | X | Y | ↑/↓ |
| Conversions | X | Y | ↑/↓ |
| Revenue | $X | $Y | ↑/↓ |
### Funnel Performance
Visitors → Signups → Activated → Paid
[X%] [Y%] [Z%] [W%]
### Top Features Used
1. [Feature] - X users
2. [Feature] - Y users
3. [Feature] - Z users
Cohort Analysis
QUÉ ES:
Agrupar usuarios por cuándo se unieron
y trackear su comportamiento over time.
EJEMPLO - Retention por Cohorte:
Week 0 Week 1 Week 2 Week 4
Jan cohort 100% 45% 30% 20%
Feb cohort 100% 50% 35% 25%
Mar cohort 100% 55% 40% 28%
INTERPRETAR:
- ¿Las cohortes mejoran? (producto mejora)
- ¿Dónde está el mayor drop-off?
- ¿Qué cohortes son mejores? (por qué?)
VARIANTES:
- Retention cohorts
- Revenue cohorts
- Feature adoption cohorts
- Behavioral cohorts
Funnel Analysis
FUNNEL TÍPICO:
Landing Page → Signup → Activation → Conversion
ANALIZAR:
1. Drop-off en cada paso
- ¿Cuál es el mayor leak?
2. Tiempo entre pasos
- ¿Hay demoras sospechosas?
3. Segmentar por:
- Fuente de tráfico
- Device
- User type
EJEMPLO:
Step | Users | Rate | Drop-off
--------------|--------|--------|----------
Landing | 10,000 | 100% | -
Signup Start | 3,000 | 30% | 70% ← big leak
Signup Done | 2,400 | 80% | 20%
Activated | 1,200 | 50% | 50% ← investigate
Converted | 240 | 20% | 80%
Report Writing Tips
ESTRUCTURA DE REPORTE:
1. EXECUTIVE SUMMARY
- 3 bullets max
- Numbers that matter
- Recommended actions
2. KEY FINDINGS
- What happened
- Why it matters
- Evidence/data
3. DETAILED ANALYSIS
- Charts and tables
- Comparisons
- Segmentation
4. RECOMMENDATIONS
- Specific actions
- Expected impact
- Priority order
BEST PRACTICES:
- Lead with insight, not data
- Visualize trends
- Compare to baseline/target
- Explain anomalies
- End with "so what?"
Weekly Business Review
## Week [X] Review
### TL;DR
- 📈 [Good news #1]
- ⚠️ [Concern #1]
- 🎯 [Focus for next week]
### Key Metrics
| Metric | Actual | Target | Status |
|--------|--------|--------|--------|
| Revenue | $X | $Y | 🟢/🟡/🔴 |
| New users | X | Y | 🟢/🟡/🔴 |
| Retention | X% | Y% | 🟢/🟡/🔴 |
### Wins
- [Achievement with number]
- [Achievement with number]
### Concerns
- [Issue + root cause + plan]
- [Issue + root cause + plan]
### Experiments
- [Experiment]: [Status] [Result if done]
### Next Week Focus
1. [Priority]
2. [Priority]
Data Investigation Framework
CUANDO VES ALGO WEIRD:
1. VERIFY
- ¿Es real o data bug?
- Check data quality
- Compare sources
2. SCOPE
- ¿Cuándo empezó?
- ¿Qué usuarios afectados?
- ¿Qué tan grande es el cambio?
3. HYPOTHESIZE
- Possible causes
- External factors?
- Internal changes?
4. TEST
- Segment data
- Correlate with events
- A/B split analysis
5. CONCLUDE
- Root cause identified
- Impact quantified
- Action recommended
Visualization Guidelines
CUÁNDO USAR CADA TIPO:
LINE CHART:
- Trends over time
- Multiple series comparison
BAR CHART:
- Category comparison
- Ranked lists
PIE CHART:
- Part of whole (limitado a 5-6 slices max)
- Market share
TABLE:
- Exact numbers matter
- Multiple dimensions
FUNNEL:
- Conversion flows
- Drop-off analysis
COHORT HEATMAP:
- Retention over time
- Behavior by group
TIPS:
- Title should state insight
- Label axes clearly
- Consistent colors
- Remove chartjunk
- Start Y at 0 (usually)
Tools Stack
DATA COLLECTION:
- Segment
- Amplitude
- Mixpanel
- Google Analytics 4
DATA WAREHOUSE:
- BigQuery
- Snowflake
- Redshift
VISUALIZATION:
- Metabase
- Looker
- Tableau
- Mode
EXPERIMENTATION:
- LaunchDarkly
- Statsig
- Amplitude Experiment
Mejores Prácticas
- Insight, not just data - Numbers without context are useless
- Compare to something - Baseline, target, last period
- Segment ruthlessly - Averages hide truth
- Track causes, not just effects - Why matters
- Automate what repeats - Weekly reports should be easy
- Question the data - Garbage in, garbage out
Filosofía
"Data tells you what happened. Analysis tells you why. Insights tell you what to do next."
El objetivo es democratizar el acceso a insights, permitiendo que todos en el equipo tomen decisiones basadas en datos reales.