a5c-ai Product Analytics Skill
Query and analyze product analytics data for metrics definition, funnel analysis, retention curves, and experiment tracking.
Overview
This skill provides comprehensive capabilities for working with product analytics platforms. It enables data-driven product decisions through metric queries, funnel analysis, cohort retention tracking, and dashboard generation.
Capabilities
Analytics Platform Integration
- Query Amplitude, Mixpanel, Heap, GA4 data
- Execute custom event queries
- Pull predefined report data
- Sync metric definitions
Funnel Analysis
- Define and calculate conversion funnels
- Identify drop-off points and friction
- Segment funnels by user attributes
- Compare funnel performance over time
Retention Analysis
- Generate retention curves and matrices
- Calculate cohort retention rates
- Analyze retention by user segment
- Identify retention drivers and predictors
Metric Definition
- Define North Star and supporting metrics
- Create event tracking specifications
- Document metric calculations
- Build metric hierarchies (trees)
Dashboard Configuration
- Generate dashboard layouts
- Configure chart specifications
- Define alert thresholds
- Export dashboard configs
Prerequisites
Analytics Platform Access
Supported Platforms:
- Amplitude (API key required)
- Mixpanel (service account)
- Heap (API access)
- Google Analytics 4 (BigQuery export)
- Posthog (API key)
Configuration
{
"platform": "amplitude",
"credentials": {
"api_key": "${AMPLITUDE_API_KEY}",
"secret_key": "${AMPLITUDE_SECRET_KEY}"
},
"project_id": "123456",
"timezone": "America/Los_Angeles"
}
Usage Patterns
Funnel Analysis Query
## Funnel Definition
### Funnel: Signup to First Value
**Steps**:
1. Page View: /signup
2. Event: signup_started
3. Event: signup_completed
4. Event: first_action_completed
**Filters**:
- Platform: web
- Date range: last 30 days
- New users only
**Segmentation**:
- By traffic source
- By device type
Funnel Query Example (Amplitude-style)
# Funnel analysis query
funnel_config = {
"events": [
{"event_type": "signup_started"},
{"event_type": "signup_completed"},
{"event_type": "onboarding_completed"},
{"event_type": "first_value_action"}
],
"filters": {
"platform": ["web", "ios", "android"],
"date_range": {
"start": "2026-01-01",
"end": "2026-01-24"
}
},
"conversion_window": "7 days",
"group_by": ["platform", "utm_source"]
}
# Expected output format
funnel_results = {
"overall": {
"step_1": {"users": 10000, "rate": 1.0},
"step_2": {"users": 6500, "rate": 0.65},
"step_3": {"users": 4200, "rate": 0.65},
"step_4": {"users": 2100, "rate": 0.50}
},
"overall_conversion": 0.21,
"segments": {
"web": {"conversion": 0.18},
"ios": {"conversion": 0.25},
"android": {"conversion": 0.19}
}
}
Retention Analysis
## Retention Query
### Cohort Definition
- **Cohort by**: signup_date (weekly)
- **Retention event**: any_active_event
- **Time periods**: Day 1, 7, 14, 30, 60, 90
### Output: Retention Matrix
| Cohort Week | Users | D1 | D7 | D14 | D30 | D60 | D90 |
|-------------|-------|-----|-----|-----|-----|-----|-----|
| Jan 1-7 | 1000 | 45% | 30% | 25% | 20% | 15% | 12% |
| Jan 8-14 | 1200 | 48% | 32% | 27% | 22% | - | - |
| Jan 15-21 | 1100 | 46% | 31% | - | - | - | - |
Retention Query Example
# Retention analysis configuration
retention_config = {
"cohort_definition": {
"event": "signup_completed",
"grouping": "week"
},
"retention_event": {
"event_type": "any_active",
"conditions": ["page_view", "feature_used", "content_created"]
},
"periods": [1, 7, 14, 30, 60, 90],
"date_range": {
"start": "2025-10-01",
"end": "2026-01-24"
},
"segments": ["subscription_tier", "signup_source"]
}
# Expected output
retention_results = {
"retention_matrix": [
{
"cohort": "2025-W40",
"cohort_size": 1000,
"retention": {
"D1": 0.45,
"D7": 0.30,
"D14": 0.25,
"D30": 0.20,
"D60": 0.15,
"D90": 0.12
}
}
],
"averages": {
"D1": 0.46,
"D7": 0.31,
"D14": 0.26,
"D30": 0.21,
"D60": 0.16,
"D90": 0.13
},
"trends": {
"D30_trend": "+2%", # vs previous period
"D7_trend": "-1%"
}
}
Metric Definition Specification
## Metric Specification Template
### Metric: Weekly Active Users (WAU)
**Definition**: Unique users who performed at least one qualifying action in a 7-day period.
**Calculation**:
```sql
SELECT COUNT(DISTINCT user_id)
FROM events
WHERE event_type IN ('page_view', 'feature_used', 'content_created')
AND event_timestamp >= CURRENT_DATE - INTERVAL '7 days'
Qualifying Events:
- page_view (any page)
- feature_used
- content_created
- content_shared
Exclusions:
- Bot traffic (user_agent filter)
- Internal users (email domain filter)
Segments:
- By platform (web, ios, android)
- By subscription tier
- By signup cohort
Alerts:
- Warning: >5% week-over-week decline
- Critical: >10% week-over-week decline
### Event Tracking Specification
```json
{
"event_name": "feature_used",
"description": "User interacted with a product feature",
"category": "engagement",
"properties": {
"feature_name": {
"type": "string",
"required": true,
"description": "Name of the feature used",
"examples": ["search", "export", "share"]
},
"feature_version": {
"type": "string",
"required": false,
"description": "Version of the feature"
},
"action": {
"type": "string",
"required": true,
"enum": ["click", "view", "complete", "cancel"]
},
"duration_ms": {
"type": "integer",
"required": false,
"description": "Time spent on feature"
}
},
"user_properties": {
"subscription_tier": "string",
"signup_date": "date"
}
}
Integration with Babysitter SDK
Task Definition Example
const analyticsQueryTask = defineTask({
name: 'analytics-query',
description: 'Query product analytics data',
inputs: {
queryType: { type: 'string', required: true }, // funnel, retention, metric
config: { type: 'object', required: true },
platform: { type: 'string', default: 'amplitude' },
dateRange: { type: 'object', required: true }
},
outputs: {
results: { type: 'object' },
visualizations: { type: 'array' },
insights: { type: 'array' }
},
async run(inputs, taskCtx) {
return {
kind: 'skill',
title: `Run ${inputs.queryType} analysis`,
skill: {
name: 'product-analytics',
context: {
operation: inputs.queryType,
config: inputs.config,
platform: inputs.platform,
dateRange: inputs.dateRange
}
},
io: {
inputJsonPath: `tasks/${taskCtx.effectId}/input.json`,
outputJsonPath: `tasks/${taskCtx.effectId}/result.json`
}
};
}
});
Dashboard Configuration
Dashboard Specification
{
"dashboard_name": "Product Health Dashboard",
"refresh_interval": "1h",
"layout": {
"columns": 3,
"rows": 4
},
"widgets": [
{
"id": "wau_trend",
"type": "line_chart",
"position": {"row": 1, "col": 1, "width": 2},
"metric": "weekly_active_users",
"time_range": "90d",
"comparison": "previous_period"
},
{
"id": "retention_heatmap",
"type": "heatmap",
"position": {"row": 1, "col": 3, "width": 1},
"metric": "cohort_retention",
"periods": [1, 7, 30]
},
{
"id": "funnel_chart",
"type": "funnel",
"position": {"row": 2, "col": 1, "width": 3},
"funnel_id": "signup_to_activation",
"segments": ["platform"]
}
],
"alerts": [
{
"metric": "weekly_active_users",
"condition": "decrease_percent > 5",
"severity": "warning",
"notification": "slack"
}
]
}
Output Formats
Funnel Analysis Report
# Funnel Analysis Report: Signup to First Value
## Overview
- **Period**: January 1-24, 2026
- **Total Users**: 10,000
- **Overall Conversion**: 21%
## Step-by-Step Analysis
| Step | Event | Users | Conv Rate | Drop-off |
|------|-------|-------|-----------|----------|
| 1 | signup_started | 10,000 | 100% | - |
| 2 | signup_completed | 6,500 | 65% | 35% |
| 3 | onboarding_completed | 4,200 | 65% | 35% |
| 4 | first_value_action | 2,100 | 50% | 50% |
## Key Insights
1. **Biggest Drop-off**: Step 4 (onboarding to first value) - 50% drop
2. **Best Performing Segment**: iOS users (25% overall conversion)
3. **Opportunity**: Mobile onboarding flow optimization
## Recommendations
1. Simplify first value action guidance
2. Add progress indicators in onboarding
3. Implement re-engagement for drop-offs at step 3
Best Practices
- Define Metrics Clearly: Document calculation logic and edge cases
- Use Consistent Time Zones: Align all queries to single timezone
- Segment Everything: Always analyze by key user segments
- Validate Data Quality: Check for tracking gaps and anomalies
- Version Event Schemas: Track changes to event definitions
- Set Appropriate Alerts: Avoid alert fatigue with meaningful thresholds