Connecting ThoughtSpot to a Semantic Layer: Integration Guide

A semantic layer provides unified metric definitions that work across all analytics tools. ThoughtSpot, with its search-driven approach to business intelligence, gains additional power when connected to a semantic layer that ensures consistent, governed metrics match what other tools report.

This guide covers connecting ThoughtSpot to semantic layer platforms like Codd AI, enabling natural language search against governed enterprise metrics.

ThoughtSpot and Semantic Layers

ThoughtSpot's Approach

ThoughtSpot differentiates through:

• Natural language search for ad-hoc analysis
• SpotIQ AI-powered automated insights
• Search-optimized data modeling (Worksheets)
• LiveQuery for real-time data access
• Embedded analytics capabilities

The Semantic Layer Complement

An external semantic layer enhances ThoughtSpot by providing:

Benefit	Impact on ThoughtSpot
Cross-tool consistency	ThoughtSpot answers match other BI tools
Centralized governance	Metric changes propagate automatically
Pre-computed metrics	Complex calculations ready for search
AI grounding	ThoughtSpot AI uses accurate definitions

When users search in ThoughtSpot, they query the same governed metrics available throughout the organization.

Connection Architecture

Architecture Pattern

User Search → ThoughtSpot → Worksheet → Connection → Semantic Layer → Data Warehouse

ThoughtSpot connects to the semantic layer as a data source. Worksheets define how ThoughtSpot indexes and searches the semantic layer objects.

Key Components

Component	Function
Connection	Database link to semantic layer
Tables	Semantic layer objects imported to ThoughtSpot
Worksheets	Search-optimized models referencing tables
Answers	User-created visualizations from searches
Liveboards	Dashboards combining multiple answers

Setup Guide

Step 1: Prepare Semantic Layer

Configure the semantic layer for ThoughtSpot access:

1. Create service account with read permissions
2. Identify objects to expose to ThoughtSpot
3. Ensure consistent naming and documentation
4. Configure any required network access

Step 2: Create Connection

In ThoughtSpot, create a connection to the semantic layer:

1. Navigate to Data > Connections > Add connection

2. Select connection type matching semantic layer protocol:

   • Snowflake
   • Databricks
   • Google BigQuery
   • PostgreSQL
   • Other supported databases

3. Enter connection details:

   • Host/account information
   • Database and schema
   • Authentication credentials
   • Connection options

4. Test connectivity before saving

Step 3: Import Tables

Import semantic layer objects as ThoughtSpot tables:

1. Navigate to Data > Tables

2. Select the semantic layer connection

3. Choose tables/views to import

4. Configure column settings:

   • Data types
   • Index preferences
   • Attribution settings

5. Run initial data load

Step 4: Create Worksheets

Worksheets optimize data for search:

1. Navigate to Data > Worksheets > Create Worksheet

2. Select tables from semantic layer

3. Configure joins if combining multiple objects

4. Set up columns:

   • Display names (business-friendly)
   • Column descriptions
   • Search optimization settings
   • Default aggregations

5. Define formulas if needed (prefer semantic layer metrics)

6. Save and share the Worksheet

Step 5: Configure Search Optimization

Optimize search experience:

Column settings:

• Set high-value columns as visible
• Configure attribution for dimensions
• Set aggregation types for measures
• Add synonyms for business terminology

Indexing:

• Enable indexing for search-relevant columns
• Configure index refresh schedules
• Monitor index status

Working with ThoughtSpot Features

Natural Language Search

Users search governed metrics naturally:

"revenue by region last quarter"
"top 10 customers by orders"
"monthly revenue trend"

ThoughtSpot translates searches into queries against the semantic layer. The governed metrics ensure accurate, consistent answers.

Search Suggestions

ThoughtSpot suggests terms based on indexed data:

1. Users type partial queries
2. ThoughtSpot suggests completions
3. Suggestions reference semantic layer columns and values
4. Business terminology from semantic layer appears in suggestions

SpotIQ Insights

Enable AI-powered insights on semantic layer data:

1. Configure SpotIQ on Worksheets
2. SpotIQ analyzes governed metrics
3. Automated insights surface anomalies and trends
4. Users receive proactive analysis on standardized data

Liveboards

Build dashboards combining semantic layer queries:

1. Create Answers from searches
2. Pin Answers to Liveboards
3. Configure interactions between visualizations
4. Set refresh schedules based on data freshness needs

Optimization Strategies

Data Modeling for Search

Optimize Worksheets for search performance:

Denormalized structures:

• Pre-join related semantic layer objects
• Reduce runtime joins for faster search

Search-ready columns:

• Business-friendly names
• Clear descriptions
• Appropriate indexing

Pre-aggregated metrics:

• Import pre-computed aggregations from semantic layer
• Faster response for common queries

Connection Performance

Configure connections for optimal performance:

LiveQuery vs. Import:

• LiveQuery: Real-time access, best for governed metrics that change
• Import: Pre-loaded data, faster search for large datasets

Query optimization:

• Use semantic layer query caching
• Configure connection pooling
• Set appropriate timeout values

Index Management

Manage ThoughtSpot indexes effectively:

• Schedule index updates during low-usage periods
• Monitor index size and performance
• Balance freshness needs with system load

Access Control

Permission Alignment

Coordinate access between semantic layer and ThoughtSpot:

Semantic Layer Access	ThoughtSpot Configuration
Full data access	Read access to all Worksheets
Filtered by region	Row-level security by region
Specific metrics only	Limited column access in Worksheet
No access	No Worksheet sharing

Row-Level Security

Implement RLS in ThoughtSpot or semantic layer:

ThoughtSpot RLS:

• Configure rules on Worksheets
• Filter data based on user attributes
• Apply consistently across answers

Semantic Layer RLS:

• Filter at query time
• Consistent across all tools
• Centralized management

For comprehensive security, implement RLS at the semantic layer level.

Sharing and Collaboration

Control content sharing:

1. Share Worksheets with appropriate groups
2. Set Liveboard permissions
3. Configure embedding permissions
4. Monitor usage and access

Integration with ThoughtSpot Everywhere

Embedded Analytics

Embed ThoughtSpot search in applications:

// Embed search component
const searchEmbed = new SearchEmbed("#container", {
  frameParams: {
    width: "100%",
    height: "600px"
  },
  dataSources: ["worksheet-guid"], // Semantic layer Worksheet
  searchOptions: {
    enableSearchAssist: true
  }
});
searchEmbed.render();

Embedded experiences use the same governed metrics as standalone ThoughtSpot.

Liveboard Embedding

Embed Liveboards built on semantic layer data:

1. Create Liveboards with semantic layer Answers
2. Configure embedding security
3. Embed in applications
4. Consistent metrics in customer-facing and internal tools

Codd AI Integration

Codd AI provides optimized ThoughtSpot integration:

• SQL-compatible endpoints for standard connectivity
• Pre-calculated metrics ready for search indexing
• Governance enforcement on all queries
• Consistent definitions with other connected tools

Connect ThoughtSpot to Codd AI using supported database connectors for immediate access to governed metrics.

Troubleshooting

Issue: Search Returns Unexpected Results

Symptoms: Search answers differ from expectations

Solutions:

• Verify Worksheet configuration matches semantic layer
• Check column aggregation settings
• Review join relationships
• Validate data freshness

Issue: Slow Search Response

Symptoms: Searches take long to return

Solutions:

• Review indexing configuration
• Consider pre-aggregated semantic layer objects
• Check connection performance
• Optimize Worksheet structure

Issue: Data Discrepancies

Symptoms: ThoughtSpot values differ from other tools

Solutions:

• Ensure same semantic layer version
• Check ThoughtSpot formulas vs. semantic definitions
• Validate data refresh timing
• Compare raw queries between tools

Best Practices Summary

1. Connect to semantic layer as primary source for governed metrics
2. Create Worksheets that reference semantic layer objects directly
3. Keep complex calculations in semantic layer - use ThoughtSpot for search optimization
4. Configure search-friendly metadata with business terminology
5. Align permissions between semantic layer and ThoughtSpot
6. Monitor data freshness and configure appropriate refresh schedules
7. Leverage SpotIQ for AI insights on governed metrics

Connecting ThoughtSpot to a semantic layer combines natural language search capabilities with enterprise metric governance, enabling anyone to find accurate answers to business questions instantly.