How to Get Started with ThoughtSpot AI Analytics

ThoughtSpot is a search-driven analytics platform that lets business users query live cloud data warehouses using plain English, without writing SQL or waiting for a developer to build a custom report. You type a question, ThoughtSpot generates the query, runs it against your warehouse, and returns a chart or table. This guide walks through setup from a new account to a working shared dashboard.

What ThoughtSpot Does and Who It Is Built For

ThoughtSpot is built around two concepts: search and Liveboards. Search means typing a business question into a bar, not clicking through filters or waiting on an analyst. Liveboards are ThoughtSpot's term for dynamic dashboards that refresh against live warehouse data each time someone opens them.

The platform connects directly to cloud data warehouses -- Snowflake, Google BigQuery, Amazon Redshift, Databricks, Azure Synapse -- and queries them in real time. It does not copy or store your data, so your existing data governance and access controls remain unchanged.

This makes ThoughtSpot most practical for operations managers, analysts, and founders who need fast answers from structured business data and already have that data centralized in a cloud warehouse. End users face a low learning curve. Most of the setup work lives in the data model configuration, covered below in Step 3.

In February 2026, ThoughtSpot announced an expansion of its Spotter AI feature into agentic data preparation -- the platform moving beyond query and visualization into identifying data quality issues and suggesting transformations upstream of the query layer. According to the announcement, this capability was in active rollout across ThoughtSpot Analytics Cloud accounts as of early 2026, marking a meaningful expansion of where the tool fits in a data team's workflow.

Step 1: Start a Free Trial

Go to thoughtspot.com and click "Free Trial." You will select your cloud data platform from a dropdown -- Snowflake is the most commonly used -- and complete a short sign-up form. A confirmation email arrives within a few minutes with a link to activate your account.

The trial lasts 30 days and includes access to all core features, including Spotter. No credit card is required. ThoughtSpot also provides sample datasets inside the trial environment, so you can explore the interface before connecting real data.

Step 2: Connect Your Data Warehouse

After logging in, open the Admin Console from the top navigation. Go to Data, then Connections, and click "New Connection."

Select your warehouse type. For Snowflake, the required fields are:

• Account identifier, found in your Snowflake URL before ".snowflakecomputing.com"
• Username and password, or an OAuth token if your organization uses single sign-on
• Database name and schema name

ThoughtSpot tests the connection immediately. If it succeeds, you will see a list of tables available in the selected schema. Select the tables you want to make queryable and proceed.

For BigQuery, you authenticate via a service account JSON key and specify the project ID and dataset. For Redshift, you provide the cluster endpoint, port, database name, and credentials. The connection wizard prompts for each required field in sequence.

If you do not yet have a cloud warehouse, ThoughtSpot includes built-in sample datasets covering retail sales, human resources, and support data. These are sufficient for testing the full feature set before committing to a warehouse connection.

Step 3: Build a Data Model

Once tables are connected, go to Data, then Models, and click "Create Model." This step is where you define relationships between tables, set column display names, and configure which columns appear as filterable dimensions.

Defining table joins is the most technically involved part of setup, but it does not require SQL. You select the join type (inner, left, right) and specify which fields link the tables. For example, linking an "orders" table to a "customers" table on a shared customer ID field takes about two minutes in the visual interface.

The most impactful thing you can do at this stage is rename columns with business-friendly labels. ThoughtSpot's natural language engine works better when column names reflect how people actually talk about the data. Rename "rev_usd_gross" to "Revenue" and "acct_mgr_id" to "Account Manager" before saving the model. Users who search for "revenue by account manager" will get correct results immediately; searching for raw field names will not.

You also set column types at this stage: measures (numeric values to aggregate), attributes (dimensions to group or filter by), and dates. ThoughtSpot infers these automatically but any setting can be overridden. Plan to spend 15 to 20 minutes on this step for a model covering three to five tables.

Step 4: Search Your Data

With a model published, return to the home screen and use the search bar. ThoughtSpot interprets natural language queries that combine column names, values, time periods, and aggregation keywords. Queries that work well:

• "revenue by region last quarter"
• "top 10 customers by order value year to date"
• "monthly new signups trend last 12 months"
• "average order value by product category"

As you type, ThoughtSpot surfaces autocomplete suggestions based on your model's column names and known dimension values. This feedback helps users correct intent before submitting a query.

ThoughtSpot selects a default chart type based on query structure. Trend queries return line charts. Ranked queries return bar charts. Raw lookups return tables. You can switch the chart type with one click without re-running the query. Save any result to a Liveboard using the "Pin" button in the top right of the result view.

Step 5: Create a Liveboard

Go to Liveboards and click "New Liveboard." Give it a name that describes its purpose, such as "Weekly Revenue Summary" or "Customer Retention Overview."

Add tiles by clicking "Add Visualization" and either running a new search directly from the Liveboard editor or selecting a previously saved search result. Each tile operates independently as a live query. When a viewer opens the Liveboard, every tile queries the warehouse and returns current data.

Add Liveboard-level filters by clicking the Filter button in the top bar. Filters apply across all tiles simultaneously. Setting a date range filter lets viewers switch between last week, last month, and last quarter without opening each tile individually.

Share the Liveboard with specific users or groups via the Share button. ThoughtSpot supports role-based access: users only see data they have permission to view in the underlying warehouse.

Step 6: Use Spotter to Investigate Anomalies

Spotter is ThoughtSpot's AI reasoning engine. It goes beyond retrieving a metric and attempts to explain why a pattern exists.

From any Liveboard tile or search result, click "Ask Spotter." Type a question in plain English: "Why did support ticket volume spike in week 8?" or "What is driving the decline in average order value since January?" Spotter analyzes the data across multiple dimensions, tests hypotheses, and returns a structured explanation with supporting visualizations. It identifies which variables correlate most strongly with the pattern and ranks them by explanatory weight.

Spotter is most useful when you have already spotted an anomaly and do not know where to start. Manual investigation would mean slicing the data across product lines, regions, sales reps, and time periods one by one. Spotter runs this analysis in parallel and returns a summary in seconds.

What ThoughtSpot Does Not Handle

ThoughtSpot is a query and visualization tool. It does not transform, clean, or load data. If your source tables contain inconsistencies, duplicate records, or poorly formatted values, those issues should be resolved in the warehouse or in a transformation layer such as dbt before ThoughtSpot queries them.

It also requires a cloud warehouse as a prerequisite. Teams working primarily from spreadsheets will need to load their data into a warehouse first. If your team is working directly from uploaded files and wants to ask plain English questions without any infrastructure setup, VSLZ AI handles that from a file upload with no warehouse or configuration needed.

Practical Summary

To get from zero to a working ThoughtSpot setup:

1. Sign up for the 30-day free trial at thoughtspot.com
2. Connect your cloud data warehouse in the Admin Console under Data, then Connections
3. Import tables and build a data model with clear business-friendly column names
4. Run natural language searches to create charts and tables
5. Pin results to a Liveboard and share with your team
6. Use Spotter to explain anomalies and identify root causes automatically

The full setup, from account creation to a shared Liveboard, typically takes under two hours for a team with an existing cloud warehouse. The ongoing value is in how quickly individual contributors can get answers without waiting for a data analyst to build a custom report each time a question comes up.