Lightdash
Lightdash is an open-source BI platform that builds its semantic layer directly from a dbt project — dimensions and metrics are declared in the dbt YAML an analytics team already maintains, and every chart and dashboard is generated from those definitions.
LicenseOpen source (permissive)· MIT
DeploymentSelf-hostedManaged
PricingFreeSubscription
Operational complexityMedium
WorkloadInteractive
LanguageTypeScript
Use it when
- The warehouse is modelled in dbt: dimensions and metrics come from the YAML the team already maintains, so there is no second modelling layer to keep in sync.
- Metric definitions must live in git under review rather than in a BI tool's own database.
- Embedded analytics needs row-level security without accounts: JWT-carried attributes substitute into the dbt sql_filter on the warehouse.
- Flat pricing appeals: Cloud Pro is $3,000 a month with unlimited users, and self-hosting is free.
Think twice when
- There is no dbt project; without one there is nothing for Lightdash to read.
- The warehouse is outside its ten supported ones (BigQuery, Snowflake, Databricks, Redshift, Postgres, Trino, ClickHouse, DuckDB, Athena, Supabase).
- Threshold-triggered alerts matter: only scheduled deliveries exist.
How it runs
Self-hosted (MIT, except one enterprise-licensed backend package), or Lightdash Cloud. Connect a warehouse and point it at a dbt repository or dbt Cloud project; queries compile to warehouse SQL and push down; charts and dashboards round-trip as YAML for version control.
Details
CompareHow Lightdash answers the questions Business Intelligence turns on.
Business Intelligence
| How it works | |
| Authoring model | Point-and-click over dbt fields, or the SQL runner |
| Semantic layer | The dbt project: dimensions and metrics come from dbt YAML |
| Query execution | Pushdown to the warehouse: dbt models compile to warehouse SQL |
| Chart library | Preset chart types plus custom charts written in Vega-Lite, which give up drill-down, viewing underlying data and cross-filtering in exchange; no plugin marketplace |
| Definitions in | dbt repository for the model; charts and dashboards download and upload as YAML |
| Alerts and delivery | Scheduled deliveries only: image, CSV, XLSX or PDF to email, Slack, Teams or Google Chat on a cron, with AI-written summaries; no threshold-triggered alert exists |
| Connections | |
| Connectors | Ten warehouses of its own choosing (BigQuery, Postgres, Supabase, Redshift, Snowflake, Databricks, Trino, ClickHouse, DuckDB and Athena) not simply whatever dbt has an adapter for |
| Embedding | JWT-signed URLs, an iframe or a React SDK, for Cloud and Enterprise on-prem; the token carries user attributes that substitute into the dbt sql_filter, so embedded viewers get row-level security without accounts |
| Access | |
| Access control | Row-level security is a sql_filter in the dbt YAML that runs on the warehouse and cannot be bypassed; column access comes from required_attributes or any_attributes, both driven by user attributes |
| Cost | |
| Billing unit | Free self-hosted; Cloud Pro is a flat $3,000 a month for unlimited users with no per-seat pricing at all, and Enterprise is quoted |
Alternatives to Lightdash
