Evidence: BI as Code With SQL & Markdown
What it is: Evidence (evidence.dev) is an open-source "BI as code" framework: you write reports and dashboards as Markdown files with embedded SQL queries and chart components, and Evidence compiles them into a fast, interactive website. Instead of clicking through a drag-and-drop dashboard builder, everything lives in version-controlled text files - closer to how a static site generator works than a traditional BI tool.
What It Does Best
Git-native dashboards. Because reports are plain Markdown and SQL files, they get real code review, diffs, and version history instead of living only inside a proprietary BI tool's UI.
SQL-first authoring. Anyone comfortable writing SQL can build a full report - query results are referenced directly in Markdown to drive text, tables, and charts.
Fast, static-first output. Reports compile down to a static site, which tends to load and navigate quickly compared to dashboards that hit a live database on every interaction.
Key Features
Markdown + SQL reports: author dashboards as text files with embedded queries
Built-in chart components: drop-in components for common chart types driven directly by query results
Templated pages: generate many similar report pages from one template plus a list of parameters
Multiple data source connectors: connects to common warehouses and databases, including DuckDB
Git-based versioning: reports live in your repository, reviewed and deployed like normal code
Pricing: Free Self-Hosted, Seat-Based Cloud
Evidence's core framework is open source and MIT-licensed, so self-hosting is free. The company also offers Evidence Cloud, a managed hosting option with seat-based pricing plus usage, aimed at teams that don't want to run their own deployment pipeline. Because Evidence Cloud pricing can change, treat any specific number as approximate and check evidence.dev directly for current tiers.
Self-hosted / open source: free, MIT-licensed, deploy anywhere static sites can run
Evidence Cloud: seat-based pricing plus usage for managed hosting and collaboration features
When to Use It
✅ Your team is comfortable writing SQL and wants reports as version-controlled code
✅ You want dashboards reviewed via pull requests like any other code change
✅ You need fast, static-feeling reports rather than dashboards constantly re-querying a live database
✅ You want a free, open-source, self-hostable BI option
✅ You're building similar report pages repeatedly and want to template them
When NOT to Use It
❌ Your users need to build their own ad hoc dashboards without writing SQL or Markdown
❌ You need real-time, live-querying dashboards rather than periodically rebuilt static reports
❌ Your team wants a drag-and-drop interface, not a code-first authoring workflow
❌ You need enterprise governance features that are more mature in established BI platforms
❌ Non-technical stakeholders need to self-serve new reports without engineering help
Common Use Cases
Internal metrics reports: recurring reports built and reviewed like code, not clicked together in a UI
Client-facing reporting: polished, static-feeling reports generated from templates for multiple clients
Data documentation: narrative Markdown combined with live query results to explain a dataset
Analytics engineering deliverables: a natural pairing with dbt-modeled data, published as readable reports
Evidence vs Alternatives
vs Metabase: Metabase is a drag-and-drop, no-code BI tool aimed at letting non-technical users self-serve; Evidence is code-first, aimed at teams that want dashboards as version-controlled files rather than UI-built objects.
vs Superset: Superset is also open source but UI-driven with a large chart library; Evidence trades that visual builder for a Markdown/SQL authoring workflow that fits naturally into a git-based engineering process.
Rule of thumb: pick Metabase or Superset when non-technical users need to build their own dashboards; pick Evidence when your reports are best treated as code, written and reviewed by people comfortable with SQL.
Unique Strengths
Reports as code: full git history, diffs, and review for dashboards, not just for the data pipeline feeding them.
Free and open source: MIT-licensed core removes any barrier to self-hosting at zero cost.
Fast static output: compiling to a static site keeps report browsing snappy compared to live-query-heavy dashboards.
Bottom line: Evidence is a great fit for engineering-minded data teams who want dashboards treated like real code - free, open source, and reviewable in a pull request. It's the wrong tool if non-technical users need to build their own reports without writing SQL.