Analytics Design

Rules: Follow document-writing and text-formatting-ja for Japanese documents.

Workflow

Use references/analytics-design-template.md to document every analysis.

Clarify Purpose: What do you want to know? Why is this analysis needed? Who will use it? One-time or ongoing monitoring?
Discover Data: Explore available datasets and understand schema.
- Ask user for project/dataset context and business background
- Use /bq-query skill for BigQuery schema exploration
- db/schema.rb for Rails projects
- API docs or sample data for external services
- Document schema and table relationships in the report
Build Query: Use /bq-query skill to design and execute queries.
- Requirements and schema from Steps 1-2 provide context
- Interpret results and document findings
Create Dashboard (if ongoing monitoring needed):
- Use references/looker-studio-template.md to design
- Define decisions: What actions will users take based on this dashboard?
- Check existing resources: Similar dashboards or queries already exist?
- Align time granularity with usage frequency (daily/weekly/monthly)
- Design data sources, pages, and charts

Looker Studio Best Practices

Reference Documentation

Data types: Field data types (Number, Text, Date & Time, Currency, Percent, etc.)
Types of charts: Chart types (Time series, Combo chart, Table, etc.)
Parameters: Data source parameters

Settings Documentation

Verify setting names against actual Looker Studio UI before documenting
Use exact terminology from the UI

Data Source Design

One data source per analytical purpose
Pre-aggregate in SQL for performance
Include bucket fields for distribution analysis
Include sort-order fields for proper chart ordering
Descriptive data source names

Report Structure

Separate pages by time granularity (daily/monthly)
Group related metrics per page
Consistent filter scopes within pages

Chart Type Selection

| Purpose | Chart Type | |---------|------------| | KPI current value | Scorecard | | Time series trend | Time series chart | | Category breakdown over time | Stacked area / Stacked bar | | Category comparison | Bar chart | | Composition | Pie chart | | Detailed data | Table | | Distribution (percentile) | Time series (multiple metrics) |