Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Structure a product data analysis, metric deep-dive, funnel analysis, or cohort study. Use when asked to analyse product metrics, investigate a drop in conversion, explain a data change to stakeholders, or find the root cause of a metric movement. Produces a structured analysis with question, root cause, confidence level, and recommended action.
.claude/skills/mohitagw15856-data-analysis-standard/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 39% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 83% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 81% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 99% | 0% |
Convierte números crudos en decisiones de producto. Estructura cada análisis con una pregunta clara, metodología, hallazgo y acción recomendada.
Todo análisis comienza aquí:
Nunca entregues datos sin responder las cuatro preguntas. Un gráfico sin narrativa no es un análisis.
Usa cuando una métrica se haya movido inesperadamente:
MÉTRICA: [Nombre]
MOVIMIENTO: [X% de cambio durante Y período]
LÍNEA BASE: [Cuál era lo normal]
VALIDACIÓN POR SEGMENTACIÓN:
- ¿Por plataforma (iOS / Android / Web)?
- ¿Por cohorte de usuario (nuevos / recurrentes / power users)?
- ¿Por canal de adquisición?
- ¿Por geografía?
- ¿Por plan/tier?
HIPÓTESIS DE CAUSA RAÍZ:
1. [Explicación más probable] — Evidencia: [punto de dato]
2. [Explicación alternativa] — Evidencia: [punto de dato]
3. [Descartando] — Eliminada porque: [razón]
CONCLUSIÓN: [Respuesta en una oración a "¿por qué cambió esto?"]
CONFIANZA: [Alta / Media / Baja] — basada en [datos disponibles]| Etapa | Métrica | Actual | Benchmark/Meta | Caída % | Notas | |---|---|---|---|---|---| | Inicio del funnel] | Usuarios] | N] | N] | — | | | Paso 2] | Usuarios] | N] | N] | X%] | | | Paso 3] | Usuarios] | N] | N] | X%] | | | Conversión] | Usuarios] | N] | N] | X%] | |
Mayor caída: Paso X → Paso Y] — Hipótesis: razón] Investigación recomendada: consulta específica o test]
Siempre define:
Entrega una tabla de retención de cohortes y anota:
Pregunta que se responde: Pregunta específica en lenguaje claro] Período de tiempo: Rango de fechas] Fuente de datos: De dónde vienen los datos]
Hallazgo: > Resumen de 1–2 oraciones en lenguaje claro de qué muestran los datos]
Gráfico/tabla clave: Incluir o describir]
Causa raíz: Mejor explicación con evidencia]
Nivel de confianza: Alto / Medio / Bajo] — razón]
Acción recomendada:
Qué este análisis NO nos dice: Salvedad importante — qué datos faltan o qué no se puede concluir]
Pregunta al usuario por esto si no está proporcionado:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 18,657 | 22,902 | +23% | 1 | 1 | 0% | 3,335 | 5,247 | +57% | 0 | 0 | — |
case-14 | pass→pass | 9,807 | 10,318 | +5% | 1 | 1 | 0% | 1,659 | 3,270 | +97% | 0 | 0 | — |
case-02 | fail→pass | 19,772 | 16,712 | -15% | 1 | 1 | 0% | 3,507 | 4,033 | +15% | 0 | 0 | — |
case-03 | fail→pass | 20,252 | 23,008 | +14% | 1 | 1 | 0% | 3,625 | 5,042 | +39% | 0 | 0 | — |
case-04 | pass→pass | 13,900 | 30,560 | +120% | 1 | 1 | 0% | 2,483 | 3,521 | +42% | 0 | 0 | — |
case-05 | pass→pass | 19,305 | 19,969 | +3% | 1 | 1 | 0% | 3,596 | 5,471 | +52% | 0 | 0 | — |
case-06 | pass→pass | 11,399 | 11,471 | +1% | 1 | 1 | 0% | 2,273 | 3,730 | +64% | 0 | 0 | — |
case-07 | pass→fail | 13,535 | 12,056 | -11% | 1 | 1 | 0% | 2,201 | 3,306 | +50% | 0 | 0 | — |
case-08 | fail→fail | 17,910 | 16,544 | -8% | 1 | 1 | 0% | 2,981 | 3,888 | +30% | 0 | 0 | — |
case-09 | fail→pass | 13,733 | 14,833 | +8% | 1 | 1 | 0% | 2,319 | 4,245 | +83% | 0 | 0 | — |
case-10 | fail→fail | 17,929 | 14,031 | -22% | 1 | 1 | 0% | 3,122 | 3,992 | +28% | 0 | 0 | — |
case-11 | fail→pass | 11,269 | 14,227 | +26% | 1 | 1 | 0% | 1,912 | 3,469 | +81% | 0 | 0 | — |
case-12 | pass→pass | 11,923 | 13,644 | +14% | 1 | 1 | 0% | 2,171 | 3,761 | +73% | 0 | 0 | — |
case-13 | fail→pass | 9,292 | 11,817 | +27% | 1 | 1 | 0% | 1,685 | 3,355 | +99% | 0 | 0 | — |
case-15 | fail→fail | 14,342 | 11,628 | -19% | 1 | 1 | 0% | 2,675 | 3,217 | +20% | 0 | 0 | — |
case-16 | pass→pass | 12,161 | 15,550 | +28% | 1 | 1 | 0% | 2,049 | 3,846 | +88% | 0 | 0 | — |
case-17 | fail→fail | 15,607 | 14,080 | -10% | 1 | 1 | 0% | 2,505 | 3,899 | +56% | 0 | 0 | — |
case-18 | fail→pass | 10,261 | 10,102 | -2% | 1 | 1 | 0% | 1,833 | 3,135 | +71% | 0 | 0 | — |
case-19 | fail→pass | 12,253 | 12,014 | -2% | 1 | 1 | 0% | 2,068 | 3,102 | +50% | 0 | 0 | — |
case-20 | pass→pass | 12,231 | 13,647 | +12% | 1 | 1 | 0% | 2,057 | 3,772 | +83% | 0 | 0 | — |
case-21 | pass→pass | 10,966 | 14,484 | +32% | 1 | 1 | 0% | 1,922 | 3,783 | +97% | 0 | 0 | — |
case-22 | pass→pass | 11,315 | 9,472 | -16% | 1 | 1 | 0% | 1,960 | 2,933 | +50% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +27 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.