How to Set Up Metabase for Business Analytics

Metabase is an open-source business intelligence tool that any ops manager, analyst, or founder can use to query a database and build dashboards without writing SQL. The setup process has three phases: deploy the application, connect a database, and start building questions and dashboards in the visual query builder. A basic installation takes under 15 minutes using Docker or Metabase Cloud.

What Metabase Is and Who Uses It

Metabase turns database tables into charts, dashboards, and scheduled email reports. Unlike Looker Studio, which is designed around Google data sources, Metabase connects directly to more than 20 databases: PostgreSQL, MySQL, MongoDB, Snowflake, BigQuery, Databricks, SQL Server, MariaDB, and SQLite, among others.

The core unit in Metabase is the "question," which is a saved query paired with a visualization. Questions are combined into dashboards. Dashboards support interactive filters, scheduled subscriptions, and public sharing links that work without a Metabase login.

According to Metabase's own documentation, the platform is used by more than 50,000 organizations. The open-source version is free and covers the full core feature set. Pro and Enterprise plans (starting at $500 per month) add SSO, granular permissions, and white-label embedding.

Choose a Deployment Method

Metabase offers two deployment paths: hosted and self-hosted.

Metabase Cloud is the fastest option. After signup, Metabase provisions infrastructure, handles upgrades, and provides an SSL certificate. A free trial is available; paid production accounts require a Pro license. For small teams that want to be running within 10 minutes and have no preference for self-hosting, Cloud removes all infrastructure overhead.

Self-hosted is the more common choice for teams that want data to stay within their own environment or want to avoid recurring cloud costs. Metabase publishes a Docker image and a JAR file. Docker is recommended because it simplifies upgrades and isolates dependencies from the host system.

Docker Hub pull statistics for the official Metabase image have consistently exceeded 100 million total pulls, reflecting the scale of self-hosted deployments worldwide.

Deploy Metabase with Docker

You need Docker installed on the host machine. Pull the Metabase image and start a container:

docker pull metabase/metabase:latest
docker run -d -p 3000:3000 --name metabase metabase/metabase

Metabase starts on port 3000. Open http://localhost:3000 in a browser. The first startup takes two to three minutes while Metabase initializes its internal metadata database.

By default, Metabase uses an embedded H2 database to store its own metadata: users, dashboards, and saved questions. H2 is suitable for testing but loses all saved data on container restarts or upgrades. For any production or team deployment, configure PostgreSQL as the metadata store before creating dashboards:

docker run -d -p 3000:3000 \
  -e "MB_DB_TYPE=postgres" \
  -e "MB_DB_DBNAME=metabase" \
  -e "MB_DB_PORT=5432" \
  -e "MB_DB_USER=metabase_user" \
  -e "MB_DB_PASS=yourpassword" \
  -e "MB_DB_HOST=your-postgres-host" \
  --name metabase metabase/metabase

This separates Metabase's application metadata from your business data and keeps it intact across container restarts and version upgrades.

Complete the Setup Wizard

On first load, the setup wizard walks through five steps:

1. Create an admin account with an email and password. Record these credentials before continuing since this account manages user access and database connections.
2. Describe your use case (self-service analytics, embedding, or both). This determines which onboarding tutorials appear.
3. Connect a database (you can skip this and use the built-in sample database to explore the interface first).
4. Enter a license key if you purchased a Pro or Enterprise plan.
5. Set usage preferences to allow or block anonymous telemetry.

The built-in sample database is a pre-loaded retail orders dataset containing orders, products, and people tables. It is sufficient to test questions, charts, and dashboards before connecting your real data.

Connect Your Database

Navigate to Admin Settings > Databases > Add a database. Select the database type from the dropdown and complete the connection fields.

For PostgreSQL, the required fields are: display name, host, port (default 5432), database name, username, and password. Metabase also supports SSL certificates and SSH tunnels for databases behind firewalls.

For cloud warehouses, the requirements differ by provider:

• Snowflake requires an account identifier in orgname-accountname format, warehouse name, database, schema, username, and password.
• BigQuery requires a project ID and a service account JSON key with the BigQuery Data Viewer and Job User roles assigned.
• Databricks requires a server hostname and HTTP path from the cluster or SQL warehouse settings, plus a personal access token.

After saving, Metabase scans the database schema. The scan takes 30 seconds to several minutes depending on table count. Once it completes, your tables appear in the data browser under the database display name.

Create a Question

Navigate to New > Question and pick a data source. The query builder presents three panels: Filter, Summarize, and Visualize.

A practical example using an orders table to track monthly revenue:

1. Select the orders table from the data picker.
2. Click Filter and set a date range filter on the order date column (for example, the last 90 days).
3. Click Summarize, choose Sum of the revenue column, then click Group by the order date at monthly granularity.
4. Click Visualize to render a line chart of monthly revenue over time.
5. Click Save, name the question, and assign it to a collection.

For teams comfortable with SQL, the SQL editor accepts raw queries. Use New > SQL query, select a database, and write directly. Results render identically to visual questions and can be added to dashboards.

If your data lives in CSV files rather than a connected database, VSLZ lets you upload files and run the same kind of analysis from plain-English prompts without any database setup or query builder.

Build a Dashboard

Navigate to New > Dashboard, name it, and click create. Add questions using the Add a question button. Each question renders as a resizable card. Cards can be reordered by drag.

To make the dashboard interactive, add a filter card from the toolbar. A date range filter connected to multiple charts lets viewers adjust the time window for all cards at once. To wire a filter to a card, click the filter's gear icon and map it to the relevant date field in each connected question.

Text cards add narrative labels and section breaks, which help non-technical viewers navigate dashboards that contain many charts.

Share and Automate Reports

Metabase supports three methods for distributing dashboards:

Public link: A read-only URL that works without a login. Enable it in the dashboard's share menu. This is useful for sharing with clients or stakeholders who should not have a Metabase account.

Email subscription: Schedule a dashboard to send a PDF or image snapshot daily, weekly, or on a custom cron schedule. Recipients receive the snapshot without needing to log in to Metabase.

Slack integration: Connect Metabase to a Slack workspace and post question results or dashboard snapshots to any channel on a recurring schedule.

Common Setup Issues

H2 in production. If you do not configure an external metadata database before production use and restart the container, all saved questions and dashboards are permanently lost. Configure the PostgreSQL metadata database before inviting team members.

Schema sync delays. Metabase caches the database schema at setup and refreshes it on a background schedule. If you add columns to a table and cannot see them in the query builder, force an immediate sync from Admin Settings > Databases > Sync database schema now.

Query timeouts. Large tables without indexes on filtered columns cause questions to time out. Add database-level indexes on columns used frequently in Metabase filters, particularly date fields and foreign keys.

Open permissions by default. All users can see all data by default. Before inviting team members, create permission groups in Admin Settings > People > Groups and apply table-level and collection-level access rules per group.

Next Steps

After completing setup, the most effective starting point is one dashboard per team function: revenue for sales, pipeline metrics for operations, and spend for marketing. Narrow, purpose-built dashboards produce higher adoption than a single all-in-one dashboard that no single person owns or updates regularly.